1 2017-11-03 Wilco Dijkstra <wdijkstr@arm.com>
3 * config/aarch64/aarch64.c (aarch64_override_options_internal):
4 Set PARAM_SCHED_PRESSURE_ALGORITHM to SCHED_PRESSURE_MODEL.
6 2017-11-03 Kito Cheng <kito.cheng@gmail.com>
8 * config/riscv/riscv.c (riscv_legitimize_move): Handle
9 non-legitimate address.
11 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
13 * config/rs6000/rs6000.md (*lt0_disi): Delete.
14 (*lt0_<mode>di, *lt0_<mode>si): New.
16 2017-11-03 Segher Boessenkool <segher@kernel.crashing.org>
18 * config/rs6000/rs6000.md (move_from_CR_ov_bit): Change condition to
21 2017-11-03 Siddhesh Poyarekar <siddhesh.poyarekar@linaro.org>
22 Jim Wilson <jim.wilson@linaro.org>
24 * config/aarch64/aarch64-cores.def (saphira): New CPU.
25 * config/aarch64/aarch64-tune.md: Regenerated.
26 * doc/invoke.texi (AArch64 Options/-mtune): Add "saphira".
27 * gcc/config/aarch64/aarch64.c (saphira_tunings): New tuning table.
29 2017-11-03 Cupertino Miranda <cmiranda@synopsys.com>
31 * config/arc/arc.c (arc_save_restore): Corrected CFA note.
32 (arc_expand_prologue): Restore blink for millicode.
33 * config/arc/linux.h (LINK_EH_SPEC): Defined.
35 2017-11-03 Richard Sandiford <richard.sandiford@linaro.org>
38 * config/i386/i386.c (ix86_vector_duplicate_value): Use
39 gen_vec_duplicate after forcing the scalar into a register.
41 2017-11-02 Segher Boessenkool <segher@kernel.crashing.org>
43 * combine (try_combine): Print the insns input to try_combine to the
46 2017-11-02 Steve Ellcey <sellcey@cavium.com>
49 * config/aarch64/aarch64-c.c (aarch64_pragma_target_parse):
50 Remove second argument from aarch64_process_target_attr call.
51 * config/aarch64/aarch64-protos.h (aarch64_process_target_attr):
53 * config/aarch64/aarch64.c (aarch64_attribute_info): Change
55 (aarch64_handle_attr_arch): Remove second argument.
56 (aarch64_handle_attr_cpu): Ditto.
57 (aarch64_handle_attr_tune): Ditto.
58 (aarch64_handle_attr_isa_flags): Ditto.
59 (aarch64_process_one_target_attr): Ditto.
60 (aarch64_process_target_attr): Ditto.
61 (aarch64_option_valid_attribute_p): Remove second argument.
62 on aarch64_process_target_attr call.
64 2017-11-02 David Malcolm <dmalcolm@redhat.com>
66 * diagnostic.c: Include "selftest-diagnostic.h".
67 (selftest::assert_location_text): New function.
68 (selftest::test_diagnostic_get_location_text): New function.
69 (selftest::diagnostic_c_tests): Call it.
71 2017-11-02 David Malcolm <dmalcolm@redhat.com>
73 * Makefile.in (OBJS-libcommon): Add selftest-diagnostic.o.
74 * diagnostic-show-locus.c: Include "selftest-diagnostic.h".
75 (class selftest::test_diagnostic_context): Move to...
76 * selftest-diagnostic.c: New file.
77 * selftest-diagnostic.h: New file.
79 2017-11-02 James Bowman <james.bowman@ftdichip.com>
81 * config/ft32/ft32.c (ft32_addr_space_legitimate_address_p): increase
82 offset range for FT32B.
83 * config/ft32/ft32.h: option "mcompress" enables relaxation.
84 * config/ft32/ft32.md: Add TARGET_NOPM.
85 * config/ft32/ft32.opt: Add mft32b, mcompress, mnopm.
86 * gcc/doc/invoke.texi: Add mft32b, mcompress, mnopm.
88 2017-11-02 Wilco Dijkstra <wdijkstr@arm.com>
90 * config/aarch64/aarch64.h (MALLOC_ABI_ALIGNMENT): New define.
92 2017-11-02 Jeff Law <law@redhat.com>
94 * gimple-ssa-sprintf.c (sprintf_dom_walker): Remove
95 virtual keyword on FINAL OVERRIDE members.
97 * tree-ssa-propagate.h (ssa_propagation_engine): Group
98 virtuals together. Add virtual destructor.
99 (substitute_and_fold_engine): Similarly.
101 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
103 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
105 2017-11-02 Richard Biener <rguenther@suse.de>
107 PR tree-optimization/82795
108 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
110 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
112 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
114 (gcc_GAS_CHECK_FEATURE): Remove.
115 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
116 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
117 * configure: Regenerate.
119 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
121 * config/arc/arc.c (hwloop_optimize): Account for empty
124 2017-11-02 Richard Biener <rguenther@suse.de>
127 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
128 Truncate ARRAY_REF index and element size.
130 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
132 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
134 2017-11-01 Jeff Law <law@redhat.com>
136 * tree-ssa-ccp.c (ccp_folder): New class derived from
137 substitute_and_fold_engine.
138 (ccp_folder::get_value): New member function.
139 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
140 (ccp_fold_stmt): Remove prototype.
141 (ccp_finalize): Call substitute_and_fold from the ccp_class.
142 * tree-ssa-copy.c (copy_folder): New class derived from
143 substitute_and_fold_engine.
144 (copy_folder::get_value): Renamed from get_value.
145 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
146 * tree-vrp.c (vrp_folder): New class derived from
147 substitute_and_fold_engine.
148 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
149 (vrp_folder::get_value): New member function.
150 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
151 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
152 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
153 provide a class interface to folder/substitute routines.
154 (ssa_prop_fold_stmt_fn): Remove typedef.
155 (ssa_prop_get_value_fn): Likewise.
156 (subsitute_and_fold): Remove prototype.
157 (replace_uses_in): Likewise.
158 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
159 Renamed from replace_uses_in. Call the virtual member function
160 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
161 (substitute_and_fold_dom_walker): Remove initialization of
162 data member entries for calbacks. Add substitute_and_fold_engine
163 member and initialize it.
164 (substitute_and_fold_dom_walker::before_dom_children0: Use the
165 member functions for get_value, replace_phi_args_in c
166 replace_uses_in, and fold_stmt calls.
167 (substitute_and_fold_engine::substitute_and_fold): Renamed from
168 substitute_and_fold. Remove assert. Update ctor call.
170 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
171 (ssa_prop_visit_phi_fn): Likewise.
172 (class ssa_propagation_engine): New class to provide an interface
174 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
176 (ssa_prop_visit_phi): Likewise.
177 (ssa_propagation_engine::simulate_stmt): Moved into class.
178 Call visit_phi/visit_stmt from the class rather than via
179 file scoped static variables.
180 (ssa_propagation_engine::simulate_block): Moved into class.
181 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
182 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
183 set file scoped statics for the visit_stmt/visit_phi callbacks.
184 * tree-complex.c (complex_propagate): New class derived from
185 ssa_propagation_engine.
186 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
187 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
188 (tree_lower_complex): Call ssa_propagate via the complex_propagate
190 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
191 ssa_propagation_engine.
192 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
193 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
194 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
195 * tree-ssa-copy.c (copy_prop): New class derived from
196 ssa_propagation_engine.
197 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
198 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
199 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
200 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
201 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
202 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
203 (execute_vrp): Call ssa_propagate from the vrp_prop class.
205 2017-11-01 Jakub Jelinek <jakub@redhat.com>
207 PR rtl-optimization/82778
208 PR rtl-optimization/82597
209 * compare-elim.c (struct comparison): Add in_a_setter field.
210 (find_comparison_dom_walker::before_dom_children): Remove killed
211 bitmap and df_simulate_find_defs call, instead walk the defs.
212 Compute last_setter and initialize in_a_setter. Merge definitions
213 with first initialization for a few variables.
214 (try_validate_parallel): Use insn_invalid_p instead of
215 recog_memoized. Return insn rather than just the pattern.
216 (try_merge_compare): Fix up comment. Don't uselessly test if
217 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
219 (execute_compare_elim_after_reload): Remove df_chain_add_problem
222 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
223 Alan Hayward <alan.hayward@arm.com>
224 David Sherwood <david.sherwood@arm.com>
226 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
227 aarch64_hard_regno_nregs to get the number of registers
230 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
231 Alan Hayward <alan.hayward@arm.com>
232 David Sherwood <david.sherwood@arm.com>
234 * config/aarch64/constraints.md (Upl): Rename to...
236 * config/aarch64/aarch64.md
237 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
240 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
241 Alan Hayward <alan.hayward@arm.com>
242 David Sherwood <david.sherwood@arm.com>
244 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
245 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
248 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
249 Alan Hayward <alan.hayward@arm.com>
250 David Sherwood <david.sherwood@arm.com>
252 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
253 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
254 (aarch64_evpc_dup): Generate rtl direcly, rather than using
256 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
258 * config/aarch64/iterators.md: Add a comment above the permute
259 unspecs to say that they are generated directly by
260 aarch64_expand_vec_perm_const.
261 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
263 2017-11-01 Nathan Sidwell <nathan@acm.org>
265 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
267 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
269 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
270 and medany code models, and describe what they do.
272 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
274 Revert accidental duplicate:
276 * combine.c (can_change_dest_mode): Reject changes in
277 REGMODE_NATURAL_SIZE.
279 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
281 PR rtl-optimization/64682
282 PR rtl-optimization/69567
283 PR rtl-optimization/69737
284 PR rtl-optimization/82683
285 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
286 register mentioned in the note, drop the note, unless it came from I3,
287 in which case it should go to I3 again.
289 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
291 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
292 and return false if not.
293 (clear_bytes_written_by, live_bytes_read): Update accordingly.
295 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
297 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
298 range is known to be empty.
300 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
301 Alan Hayward <alan.hayward@arm.com>
302 David Sherwood <david.sherwood@arm.com>
304 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
305 and CONST_VECTOR_NUNITS instead of computing the number of units from
306 the byte sizes of the vector and element.
307 (simplify_binary_operation_1): Likewise.
308 (simplify_const_binary_operation): Likewise.
309 (simplify_ternary_operation): Likewise.
311 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
312 Alan Hayward <alan.hayward@arm.com>
313 David Sherwood <david.sherwood@arm.com>
315 * var-tracking.c (INT_MEM_OFFSET): Replace with...
316 (int_mem_offset): ...this new function.
317 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
318 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
319 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
322 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
323 Alan Hayward <alan.hayward@arm.com>
324 David Sherwood <david.sherwood@arm.com>
326 * lower-subreg.c (interesting_mode_p): New function.
327 (compute_costs, find_decomposable_subregs, decompose_register)
328 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
329 (resolve_clobber, dump_choices): Use it.
331 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
332 Alan Hayward <alan.hayward@arm.com>
333 David Sherwood <david.sherwood@arm.com>
335 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
337 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
338 Alan Hayward <alan.hayward@arm.com>
339 David Sherwood <david.sherwood@arm.com>
341 * alias.c (find_base_value, find_base_term): Only process integer
342 truncations. Check the precision rather than the size.
344 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
345 Alan Hayward <alan.hayward@arm.com>
346 David Sherwood <david.sherwood@arm.com>
348 * machmode.h (is_narrower_int_mode): New function
349 * optabs.c (expand_float, expand_fix): Use it.
350 * dwarf2out.c (rotate_loc_descriptor): Likewise.
352 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
353 Alan Hayward <alan.hayward@arm.com>
354 David Sherwood <david.sherwood@arm.com>
356 * rtl.h (narrower_subreg_mode): New function.
357 * ira-color.c (update_costs_from_allocno): Use it.
359 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
360 Alan Hayward <alan.hayward@arm.com>
361 David Sherwood <david.sherwood@arm.com>
363 * optabs-query.h (convert_optab_p): New function, split out from...
364 (convert_optab_handler): ...here.
365 (widening_optab_handler): Delete.
366 (find_widening_optab_handler): Remove permit_non_widening parameter.
367 (find_widening_optab_handler_and_mode): Likewise. Provide an
368 override that operates on mode class wrappers.
369 * optabs-query.c (widening_optab_handler): Delete.
370 (find_widening_optab_handler_and_mode): Remove permit_non_widening
371 parameter. Assert that the two modes are the same class and that
372 the "from" mode is narrower than the "to" mode. Use
373 convert_optab_handler instead of widening_optab_handler.
374 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
375 instead of widening_optab_handler.
376 * expr.c (expand_expr_real_2): Update calls to
377 find_widening_optab_handler.
378 * optabs.c (expand_widen_pattern_expr): Likewise.
379 (expand_binop_directly): Take the insn_code as a parameter.
380 (expand_binop): Only call find_widening_optab_handler for
381 conversion optabs; use optab_handler otherwise. Update calls
382 to find_widening_optab_handler and expand_binop_directly.
383 Use convert_optab_handler instead of widening_optab_handler.
384 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
385 find_widening_optab_handler and use scalar_mode rather than
387 (convert_plusminus_to_widen): Likewise.
389 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
390 Alan Hayward <alan.hayward@arm.com>
391 David Sherwood <david.sherwood@arm.com>
393 * machmode.h (fixed_size_mode): New class.
394 * rtl.h (get_pool_mode): Return fixed_size_mode.
395 * gengtype.c (main): Add fixed_size_mode.
396 * target.def (get_raw_result_mode): Return a fixed_size_mode.
397 (get_raw_arg_mode): Likewise.
398 * doc/tm.texi: Regenerate.
399 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
400 * targhooks.c (default_get_reg_raw_mode): Likewise.
401 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
402 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
403 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
404 (msp430_get_raw_result_mode): Likewise.
405 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
406 * dbxout.c (dbxout_parms): Require fixed-size modes.
407 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
408 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
409 * omp-low.c (lower_oacc_reductions): Likewise.
410 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
411 (simplify_subreg): Update accordingly.
412 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
413 (force_const_mem): Update accordingly. Return NULL_RTX for modes
414 that aren't fixed-size.
415 (get_pool_mode): Return a fixed_size_mode.
416 (output_constant_pool_2): Take a fixed_size_mode.
418 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
419 Alan Hayward <alan.hayward@arm.com>
420 David Sherwood <david.sherwood@arm.com>
422 * doc/rtl.texi (vec_series): Document.
423 (const): Say that the operand can be a vec_series.
424 * rtl.def (VEC_SERIES): New rtx code.
425 * rtl.h (const_vec_series_p_1): Declare.
426 (const_vec_series_p): New function.
427 * emit-rtl.h (gen_const_vec_series): Declare.
428 (gen_vec_series): Likewise.
429 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
430 (gen_vec_series): Likewise.
431 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
432 * simplify-rtx.c (simplify_unary_operation): Handle negations
434 (simplify_binary_operation_series): New function.
435 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
436 (test_vector_ops_series): New function.
437 (test_vector_ops): Call it.
438 * config/powerpcspe/altivec.md (altivec_lvsl): Use
439 gen_const_vec_series.
440 (altivec_lvsr): Likewise.
441 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
443 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
444 Alan Hayward <alan.hayward@arm.com>
445 David Sherwood <david.sherwood@arm.com>
447 * doc/rtl.texi (const): Update description of address constants.
448 Say that vector constants are allowed too.
449 * common.md (E, F): Use CONSTANT_P instead of checking for
451 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
452 checking for CONST_VECTOR.
453 * expmed.c (make_tree): Use build_vector_from_val for a CONST
455 * expr.c (expand_expr_real_2): Check for vector modes instead
456 of checking for CONST_VECTOR.
457 * rtl.h (const_vec_p): New function.
458 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
459 (unwrap_const_vec_duplicate): Handle them here too.
461 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
462 David Malcolm <dmalcolm@redhat.com>
463 Alan Hayward <alan.hayward@arm.com>
464 David Sherwood <david.sherwood@arm.com>
466 * rtl.h (vec_duplicate_p): New function.
467 * selftest-rtl.c (assert_rtx_eq_at): New function.
468 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
469 (assert_rtx_eq_at): Declare.
470 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
471 * selftest-run-tests.c (selftest::run_tests): Call it.
472 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
473 (simplify_unary_operation_1): Recursively handle vector duplicates.
474 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
476 (simplify_subreg): Handle subregs of vector duplicates.
477 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
478 (selftest::simplify_rtx_c_tests): New functions.
480 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
481 Alan Hayward <alan.hayward@arm.com>
482 David Sherwood <david.sherwood@arm.com>
484 * emit-rtl.h (gen_const_vec_duplicate): Declare.
485 (gen_vec_duplicate): Likewise.
486 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
488 (gen_const_vector): ...here.
489 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
490 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
491 whose elements are all equal.
492 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
493 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
494 (simplify_relational_operation): Likewise.
495 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
497 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
498 (aarch64_expand_vector_init): Likewise.
499 * config/arm/arm.c (neon_vdup_constant): Likewise.
500 (neon_expand_vector_init): Likewise.
501 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
502 (arm_block_set_unaligned_vect): Likewise.
503 (arm_block_set_aligned_vect): Likewise.
504 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
505 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
506 (expand_vec_perm_even_odd_pack): Likewise.
507 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
508 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
509 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
510 gen_const_vec_duplicate.
511 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
512 * config/mips/mips.c (mips_gen_const_int_vector): Use
513 gen_const_vec_duplicate.
514 (mips_expand_vector_init): Use CONST0_RTX.
515 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
516 (define_split): Use gen_const_vec_duplicate.
517 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
518 (define_split): Use gen_const_vec_duplicate.
519 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
520 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
521 * config/spu/spu.c (spu_const): Likewise.
523 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
524 Alan Hayward <alan.hayward@arm.com>
525 David Sherwood <david.sherwood@arm.com>
527 * combine.c (can_change_dest_mode): Reject changes in
528 REGMODE_NATURAL_SIZE.
530 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
532 * configure.ac (--enable-libssp): New.
533 (gcc_cv_libc_provides_ssp): Check for explicit setting before
534 trying to determine target-specific default. Adjust indentation.
535 * configure: Regenerated.
536 * doc/install.texi (Configuration): Expand --disable-libssp
539 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
541 config/i386/i386.c (ix86_expand_epilogue): Correct stack
544 2017-10-31 Martin Jambor <mjambor@suse.cz>
547 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
549 2017-10-31 David Malcolm <dmalcolm@redhat.com>
551 * auto-profile.c (autofdo_source_profile::read): Use
552 UNKNOWN_LOCATION rather than 0.
553 * diagnostic-core.h (warning_at_rich_loc): Rename to...
554 (warning_at): ...this overload.
555 (warning_at_rich_loc_n): Rename to...
556 (warning_n): ...this overload.
557 (error_at_rich_loc): Rename to...
558 (error_at): ...this overload.
559 (pedwarn_at_rich_loc): Rename to...
560 (pedwarn): ...this overload.
561 (permerror_at_rich_loc): Rename to...
562 (permerror): ...this overload.
563 (inform_at_rich_loc): Rename to...
564 (inform): ...this overload.
565 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
566 (diagnostic_n_impl_richloc): Rename to...
567 (diagnostic_n_impl): ...this rich_location *-based decl.
568 (inform_at_rich_loc): Rename to...
569 (inform): ...this, and add an assertion.
570 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
571 (warning_at_rich_loc): Rename to...
572 (warning_at): ...this, and add an assertion.
573 (warning_at_rich_loc_n): Rename to...
574 (warning_n): ...this, and add an assertion.
575 (warning_n): Update location_t-based implementation for removal of
576 location_t-based diagnostic_n_impl.
577 (pedwarn_at_rich_loc): Rename to...
578 (pedwarn): ...this, and add an assertion.
579 (permerror_at_rich_loc): Rename to...
580 (permerror): ...this, and add an assertion.
581 (error_n): Update for removal of location_t-based diagnostic_n_impl.
582 (error_at_rich_loc): Rename to...
583 (error_at): ...this, and add an assertion.
584 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
585 (driver::do_spec_on_infiles): Likewise.
586 * substring-locations.c (format_warning_va): Update for renaming
587 of inform_at_rich_loc.
589 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
591 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
592 _Float<N>X built-in functions so that the variant without the
593 "__builtin_" prefix is only enabled for the GNU C and Objective C
594 languages when they are in non-strict ANSI/ISO mode.
595 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
596 * target.def (floatn_builtin_p): Add a target hook to control
597 whether _Float<N> and _Float<N>X built-in functions without the
598 "__builtin_" prefix are enabled, and return true for C and
599 Objective C in the default hook. Include langhooks.h in
601 * targhooks.h (default_floatn_builtin_p): Likewise.
602 * targhooks.c (default_floatn_builtin_p): Likewise.
603 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
604 floatn_builtin_p target hook.
605 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
607 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
608 Eric Botcazou <ebotcazou@adacore.com>
610 PR rtl-optimization/81803
611 * lra-constraints.c (curr_insn_transform): Also reload the whole
612 register for a strict subreg no wider than a word if this is for
613 a WORD_REGISTER_OPERATIONS target.
615 2017-10-31 Jason Merrill <jason@redhat.com>
617 * gdbinit.in: Skip over inlines from timevar.h.
619 2017-10-31 Martin Liska <mliska@suse.cz>
621 * doc/gcov.texi: Document new option.
622 * gcov.c (print_usage): Likewise print it.
623 (process_args): Support the argument.
624 (format_count): New function.
625 (format_gcov): Use the function.
627 2017-10-31 Martin Liska <mliska@suse.cz>
629 * gcov.c (struct name_map): do not use typedef.
630 Define operator== and operator<.
631 (name_search): Remove.
633 (main): Do not allocate names.
634 (process_file): Add vertical space.
635 (generate_results): Use std::find.
636 (release_structures): Do not release memory.
637 (find_source): Use std::find.
639 2017-10-31 Martin Liska <mliska@suse.cz>
641 * gcov.c (struct line_info): Remove it's typedef.
642 (line_info::line_info): Add proper ctor.
643 (line_info::has_block): Do not use a typedef.
644 (struct source_info): Do not use typedef.
646 (get_cycles_count): Likewise.
647 (output_intermediate_file): Iterate via vector iterator.
648 (add_line_counts): Use std::vector methods.
649 (accumulate_line_counts): Likewise.
650 (output_lines): Likewise.
652 2017-10-31 Martin Liska <mliska@suse.cz>
654 * gcov.c (struct source_info): Remove typedef.
655 (source_info::source_info): Add proper ctor.
656 (accumulate_line_counts): Use struct, not it's typedef.
657 (output_gcov_file): Likewise.
658 (output_lines): Likewise.
659 (main): Do not allocate an array.
660 (output_intermediate_file): Use size of vector container.
661 (process_file): Resize the vector.
662 (generate_results): Do not preallocate, use newly added vector
664 (release_structures): Do not release sources.
665 (find_source): Use vector methods.
666 (add_line_counts): Do not use typedef.
668 2017-10-31 Martin Liska <mliska@suse.cz>
670 * doc/gcov.texi: Document that.
671 * gcov.c (add_line_counts): Mark lines with a non-executed
673 (output_line_beginning): Handle such lines.
674 (output_lines): Pass new argument.
675 (output_intermediate_file): Print it in intermediate format.
677 2017-10-31 Martin Liska <mliska@suse.cz>
679 * color-macros.h: New file.
680 * diagnostic-color.c: Factor out color related to macros to
682 * doc/gcov.texi: Document -k option.
683 * gcov.c (INCLUDE_STRING): Include string.h.
684 (print_usage): Add -k option.
685 (process_args): Parse it.
686 (pad_count_string): New function.
687 (output_line_beginning): Likewise.
688 (DEFAULT_LINE_START): New macro.
689 (output_lines): Support color output.
691 2017-10-31 Martin Liska <mliska@suse.cz>
693 PR gcov-profile/82633
694 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
695 their interaction with GCOV infrastructure.
696 * configure.ac: Add -fkeep-{inline,static}-functions to
698 * configure: Regenerate.
700 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
703 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
705 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
708 * config/rs6000/rs6000.md (allocate_stack): Force update interval
709 into a register if it does not fit into an immediate offset field.
711 2017-10-31 Olivier Hainque <hainque@adacore.com>
713 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
715 2017-10-31 Julia Koval <julia.koval@intel.com>
717 * config.gcc: Add gfniintrin.h.
718 * config/i386/gfniintrin.h: New.
719 * config/i386/i386-builtin-types.def
720 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
721 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
722 __builtin_ia32_vgf2p8affineinvqb_v32qi,
723 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
724 __builtin_ia32_vgf2p8affineinvqb_v16qi,
725 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
726 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
727 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
728 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
729 V64QI_FTYPE_V64QI_V64QI_INT): New types.
730 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
731 * config/i386/immintrin.h: Include gfniintrin.h.
732 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
734 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
736 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
738 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
740 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
741 (arm_ashldi3_1bit): Remove pattern.
742 (ashrdi3): Remove shift by 1 expansion.
743 (arm_ashrdi3_1bit): Remove pattern.
744 (lshrdi3): Remove shift by 1 expansion.
745 (arm_lshrdi3_1bit): Remove pattern.
746 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
747 cost of ashldi3 by 1.
748 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
749 (<shift>di3_neon): Likewise.
751 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
753 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
754 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
755 and (*aarch64_simd_mov<VQ:mode>).
756 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
758 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
759 attributes to match pattern alternative.
761 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
763 * config.gcc (powerpc*-*-*): Add emmintrin.h.
764 * config/rs6000/emmintrin.h: New file.
765 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
767 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
769 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
770 * (movdi_vfp_cortexa8): Remove pattern.
772 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
774 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
776 (Specific, alpha*-dec-osf5.1): Remove.
777 (Specific, mips-sgi-irix5): Remove.
778 (Specific, mips-sgi-irix6): Remove.
780 2017-10-30 Jakub Jelinek <jakub@redhat.com>
783 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
784 arguments to clear_bit_region_be.
786 2017-10-30 Jim Wilson <wilson@tuliptree.org>
788 * gimplify.c: Include memmodel.h.
790 2017-10-30 Martin Jambor <mjambor@suse.cz>
792 * omp-grid.c (grid_attempt_target_gridification): Also insert a
793 condition whether loop should be executed at all.
795 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
797 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
798 gimple folding of vec_madd() intrinsics.
799 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
800 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
801 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
803 2017-10-30 Richard Biener <rguenther@suse.de>
805 PR tree-optimization/82762
807 2017-10-23 Richard Biener <rguenther@suse.de>
809 PR tree-optimization/82129
811 2017-08-01 Richard Biener <rguenther@suse.de>
813 PR tree-optimization/81181
814 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
815 (compute_antic): ... end of iteration here.
817 2017-10-30 Joseph Myers <joseph@codesourcery.com>
819 * doc/invoke.texi (C Dialect Options): Document -std=c17,
820 -std=iso9899:2017 and -std=gnu17.
821 * doc/standards.texi (C Language): Document C17 support.
822 * doc/cpp.texi (Overview): Mention -std=c17.
823 (Standard Predefined Macros): Document C11 and C17 values of
824 __STDC_VERSION__. Do not refer to C99 support as incomplete.
825 * doc/extend.texi (Inline): Do not list individual options for
826 standards newer than C99.
827 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
829 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
832 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
834 * asan.c (asan_finish_file): Align asan globals array by shadow
837 2017-10-30 Jakub Jelinek <jakub@redhat.com>
840 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
841 (struct store_immediate_info): Add bitregion_start and bitregion_end
843 (store_immediate_info::store_immediate_info): Add brs and bre
844 arguments and initialize bitregion_{start,end} from those.
845 (struct merged_store_group): Add bitregion_start, bitregion_end,
846 align_base and mask fields. Drop unnecessary struct keyword from
847 struct store_immediate_info. Add do_merge method.
848 (clear_bit_region_be): Use memset instead of loop storing zeros.
849 (merged_store_group::do_merge): New method.
850 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
851 stores as long as the surrounding bitregions have no gaps.
852 (merged_store_group::merge_overlapping): Use do_merge.
853 (merged_store_group::apply_stores): Test that bitregion_{start,end}
854 is byte aligned, rather than requiring that start and width are
855 byte aligned. Drop unnecessary struct keyword from
856 struct store_immediate_info. Allocate and populate also mask array.
857 Make start of the arrays relative to bitregion_start rather than
858 start and size them according to bitregion_{end,start} difference.
859 (struct imm_store_chain_info): Drop unnecessary struct keyword from
860 struct store_immediate_info.
861 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
862 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
863 struct keyword from struct store_immediate_info.
864 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
865 between stores as long as the surrounding bitregions have no gaps.
867 (struct split_store): Add orig non-static data member.
868 (split_store::split_store): Initialize orig to false.
869 (find_constituent_stmts): Return store_immediate_info *, non-NULL
870 if there is exactly a single original stmt. Change stmts argument
871 to pointer from reference, if NULL, don't push anything to it. Add
872 first argument, use it to optimize skipping over orig stmts that
873 are known to be before bitpos already. Simplify.
874 (split_group): Return unsigned int count how many stores are or
875 would be needed rather than a bool. Add allow_unaligned argument.
876 Change split_stores argument from reference to pointer, if NULL,
877 only do a dry run computing how many stores would be produced.
878 Rewritten algorithm to use both alignment and misalign if
879 !allow_unaligned and handle bitfield stores with gaps.
880 (imm_store_chain_info::output_merged_store): Set start_byte_pos
881 from bitregion_start instead of start. Compute allow_unaligned
882 here, if true, do 2 split_group dry runs to compute which one
883 produces fewer stores and prefer aligned if equal. Punt if
884 new count is bigger or equal than original before emitting any
885 statements, rather than during that. Remove no longer needed
886 new_ssa_names tracking. Replace num_stmts with
887 split_stores.length (). Use 32-bit stack allocated entries
888 in split_stores auto_vec. Try to reuse original store lhs/rhs1
889 if possible. Handle bitfields with gaps.
890 (pass_store_merging::execute): Ignore bitsize == 0 stores.
891 Compute bitregion_{start,end} for the stores and construct
892 store_immediate_info with that. Formatting fixes.
894 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
897 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
898 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
900 2017-10-29 Jim Wilson <wilson@tuliptree.org>
902 * gimplify.c: Include tm_p.h.
904 * common.opt (gcoff): Re-add as ignored option.
905 (gcoff1, gcoff2, gcoff3): Likewise.
907 * Makefile.in (OBJS): Delete sdbout.o.
908 (GTFILES): Delete $(srcdir)/sdbout.c.
909 * debug.h: Delete sdb_debug_hooks.
910 * final.c: Delete sdbout.h include.
911 (final_scan_insn): Delete SDB_DEBUG check.
912 (rest_of_clean_state): Likewise.
913 * output.h: Delete sdb_begin_function_line.
916 * toplev.c: Delete sdbout.h include.
917 (process_options): Delete SDB_DEBUG check.
918 * tree-core.h (tree_type_common): Delete pointer field of
920 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
922 * tree.h (TYPE_SYMTAB_POINTER): Delete.
923 (TYPE_SYMTAB_IS_POINTER): Delete.
924 (TYPE_SYMTAB_IS_DIE): Renumber.
925 * xcoffout.c: Refer to former sdbout.c file.
926 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
928 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
929 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
930 (Debugging Options): Delete -gcoff.
931 (-gstabs): Delete SDB reference.
933 (-gcoff@var{level}): Delete.
934 * doc/passes.texi (Debugging information output): Delete SDB and
936 * doc/tm.texi: Regenerate.
937 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
938 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
940 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
941 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
942 (SDB_DEBUGGING_INFO): Delete.
943 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
944 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
945 * target.def (output_source_filename): Delete COFF reference.
947 * common.opt (gcoff): Delete.
948 (gxcoff+): Update Negative chain.
949 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
951 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
952 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
953 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
955 (expand_function_start): Change sdb reference to past tense.
956 (expand_function_end): Change sdb reference to past tense.
957 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
958 * opts.c (debug_type_names): Delete coff entry.
959 (common_handle_option): Delete OPT_gcoff case.
960 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
962 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
963 * config/cris/cris.h: Delete SDB reference in comment.
964 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
965 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
966 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
967 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
969 (ix86_expand_prologue): Likewise.
970 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
971 * config/ia64/ia64.h: Likewise.
972 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
973 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
975 * config/mmix/mmix.h: Likewise.
976 * config/nds32/nds32.c: Likewise.
977 * config/stormy/storym16.h: Likewise.
978 * config/visium/visium.h: Likewise.
979 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
981 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
983 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
984 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
985 FRAME_POINTER_REGNUM point at high end of local var area.
987 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
989 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
990 Move comment around. Do not reset best_edge for a copiable
991 destination if the copy would cause a partition change.
992 (better_edge_p): Remove redundant check.
994 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
996 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
998 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
1000 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
1001 for math functions that have _Float<N> and _Float<N>X variants.
1002 (mathfn_built_in_2): Add support for math functions that have
1003 _Float<N> and _Float<N>X variants.
1004 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
1005 (expand_builtin_mathfn_ternary): Add support for fma with
1006 _Float<N> and _Float<N>X variants.
1007 (expand_builtin): Likewise.
1008 (fold_builtin_3): Likewise.
1009 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
1010 create math function _Float<N> and _Float<N>X variants as external
1012 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
1013 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
1014 the __builtin_ prefix and if not strict ansi, without the prefix.
1015 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
1016 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
1017 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
1018 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
1019 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
1020 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
1021 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
1022 function signatures for fma _Float<N> and _Float<N>X variants.
1023 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
1024 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
1025 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
1026 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
1027 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
1028 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
1029 * gencfn-macros.c (print_case_cfn): Add support for math functions
1030 that have _Float<N> and _Float<N>X variants.
1031 (print_define_operator_list): Likewise.
1032 (fltfn_suffixes): Likewise.
1034 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
1035 for math functions that have _Float<N> and _Float<N>X variants.
1036 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
1037 and _Float<N>X variants.
1038 (COPYSIGN): Likewise.
1041 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
1042 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
1044 (integer_valued_read_call_p): Likewise.
1045 * fold-const-call.c (fold_const_call_ss): Likewise.
1046 (fold_const_call_sss): Add support for copysign, fmin, and fmax
1047 _Float<N> and _Float<N>X variants.
1048 (fold_const_call_ssss): Add support for fma _Float<N> and
1049 _Float<N>X variants.
1050 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
1051 support for copysign and fma _Float<N> and _Float<N>X variants.
1052 (backprop::process_builtin_call_use): Likewise.
1053 * tree-call-cdce.c (can_test_argument_range); Add support for
1054 sqrt _Float<N> and _Float<N>X variants.
1055 (edom_only_function): Likewise.
1056 (get_no_error_domain): Likewise.
1057 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
1058 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
1059 copysign _Float<N> and _Float<N>X variants.
1060 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
1061 handled by machine independent code.
1062 (FMAF128): Likewise.
1063 * doc/cpp.texi (Common Predefined Macros): Document defining
1064 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
1065 fma _Float<N> and _Float<N>X variants.
1067 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
1070 * config/i386/i386-modes.def (CCFPU): Remove definition.
1071 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
1072 (ix86_cc_modes_compatible): Ditto.
1073 (ix86_expand_carry_flag_compare): Ditto.
1074 (ix86_expand_int_movcc): Ditto.
1075 (ix86_expand_int_addcc): Ditto.
1076 (ix86_reverse_condition): Ditto.
1077 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
1078 Return true/false for unordered/ordered fp comparisons.
1079 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
1080 (ix86_prepare_fp_compare_args): Update for rename.
1081 (ix86_expand_fp_compare): Update for rename. Generate unordered
1082 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
1083 (ix86_expand_sse_compare_and_jump): Ditto.
1084 * config/i386/predicates.md (fcmov_comparison_operator):
1085 Remove CCFPU mode handling.
1086 (ix86_comparison_operator): Ditto.
1087 (ix86_carry_flag_operator): Ditto.
1088 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
1089 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
1090 (*cmpu<mode>_cc_i387): Ditto.
1091 (FPCMP): Remove mode iterator.
1092 (unord): Remove mode attribute.
1093 (unord_subst): New define_subst transformation
1094 (unord): New define_subst attribute.
1096 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
1097 (*cmpi<unord>xf_i387): Ditto.
1098 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
1099 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
1100 using unord_subst transformation.
1101 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
1102 (round_saeonly): Also handle CCFP mode.
1103 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
1104 Remove UNSPEC_SAHF unspec handling.
1106 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1108 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
1110 2017-10-27 Jeff Law <law@redhat.com>
1112 * gimple-ssa-sprintf.c: Include domwalk.h.
1113 (class sprintf_dom_walker): New class, derived from dom_walker.
1114 (sprintf_dom_walker::before_dom_children): New function.
1115 (struct call_info): Moved into sprintf_dom_walker class
1116 (compute_formath_length, handle_gimple_call): Likewise.
1117 (sprintf_length::execute): Call the dominator walker rather
1118 than walking the statements.
1120 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
1121 gimple statement locations.
1122 (check_array_bounds): Corresponding changes. Get the statement's
1123 location directly from wi->stmt.
1125 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
1128 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
1130 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1132 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
1133 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
1135 2017-10-27 Jakub Jelinek <jakub@redhat.com>
1138 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
1139 * config/i386/i386.c (maybe_get_pool_constant): Removed.
1140 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
1141 maybe_get_pool_constant.
1142 * config/i386/predicates.md (zero_extended_scalar_load_operand):
1145 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1147 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
1148 2.26 caveat. Update gas and gld versions.
1149 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
1152 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1154 * cgraph.h (set_malloc_flag): Declare.
1155 * cgraph.c (set_malloc_flag_1): New function.
1156 (set_malloc_flag): Likewise.
1157 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
1158 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
1160 (read_ipa_call_summary): Add support for reading is_return_callee.
1161 (write_ipa_call_summary): Stream is_return_callee.
1162 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
1163 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
1164 ipa-prop.h, ipa-fnsummary.h.
1165 (pure_const_names): Change to static.
1166 (malloc_state_e): Define.
1167 (malloc_state_names): Define.
1168 (funct_state_d): Add field malloc_state.
1169 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
1170 (check_retval_uses): New function.
1171 (malloc_candidate_p): Likewise.
1172 (analyze_function): Add support for malloc attribute.
1173 (pure_const_write_summary): Stream malloc_state.
1174 (pure_const_read_summary): Add support for reading malloc_state.
1175 (dump_malloc_lattice): New function.
1176 (propagate_malloc): New function.
1177 (warn_function_malloc): New function.
1178 (ipa_pure_const::execute): Call propagate_malloc and
1179 ipa_free_fn_summary.
1180 (pass_local_pure_const::execute): Add support for malloc attribute.
1181 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
1182 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
1184 2017-10-27 Martin Liska <mliska@suse.cz>
1186 PR gcov-profile/82457
1187 * doc/invoke.texi: Document that one needs a non-strict ISO mode
1188 for fork-like functions to be properly instrumented.
1190 2017-10-27 Richard Biener <rguenther@suse.de>
1193 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
1194 info when we redirected EH.
1196 2017-10-26 Michael Collison <michael.collison@arm.com>
1198 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
1200 (<optab>_trunchf<GPI:mode>2: New pattern.
1201 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
1202 * config/aarch64/iterators.md (wv): New mode attribute.
1203 (vf, VF): New mode attributes.
1204 (vgp, VGP): New mode attributes.
1205 (s): Update attribute with SImode and DImode prefixes.
1207 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1209 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
1210 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
1211 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
1212 (nios2_option_overide): Initialize it. Don't allow R0-relative
1213 addressing with PIC.
1214 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
1215 (nios2_symbolic_constant_p): Likewise.
1216 (nios2_legitimate_address_p): Likewise.
1217 (nios2_r0rel_section_name_p): New.
1218 (nios2_symbol_ref_in_r0rel_data_p): New.
1219 (nios2_emit_move_sequence): Handle r0rel_constant_p.
1220 (r0rel_constant_p): New.
1221 (nios2_print_operand_address): Handle r0rel_constant_p.
1222 (nios2_cdx_narrow_form_p): Likewise.
1223 * config/nios2/nios2.opt (mr0rel-sec=): New option.
1224 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
1225 (Nios II Options): Document -mr0rel-sec.
1227 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1229 * config/nios2/nios2.c: Include xregex.h.
1230 (nios2_gprel_sec_regex): New.
1231 (nios2_option_overide): Initialize it. Don't allow GP-relative
1232 addressing with PIC.
1233 (nios2_small_section_name_p): Check for regex match.
1234 * config/nios2/nios2.opt (mgprel-sec=): New option.
1235 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
1236 (Nios II Options): Document -mgprel-sec.
1238 2017-10-26 Jim Wilson <wilson@tuliptree.org>
1240 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
1242 2017-10-26 Tom de Vries <tom@codesourcery.com>
1244 PR tree-optimization/82707
1245 * gimple.c (gimple_copy): Fix unsharing of
1246 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
1248 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
1250 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
1251 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
1252 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
1253 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
1254 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
1255 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
1256 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
1257 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
1258 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
1259 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
1260 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
1261 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
1262 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
1263 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
1264 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
1265 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
1266 _mm512_mask_cmpunord_ps_mask): New intrinsics.
1268 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1270 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
1272 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
1273 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
1274 warning to rs6000.c. Remove the Undocumented flag, since it has
1276 (-mabi=ibmlongdouble): Likewise.
1277 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
1278 already set, set the default format for long double.
1279 (rs6000_debug_reg_global): Print whether long double is IBM or
1281 (rs6000_option_override_internal): Rework setting long double
1282 format. Only warn if the user is changing the long double default
1283 and they did not use -Wno-psabi.
1284 * doc/invoke.texi (PowerPC options): Update the documentation for
1285 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
1287 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1288 Alan Hayward <alan.hayward@arm.com>
1289 David Sherwood <david.sherwood@arm.com>
1291 * rtl.h (wider_subreg_mode): New function.
1292 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
1293 rather than an unsigned int *.
1294 * ira-color.c (regno_max_ref_width): Replace with...
1295 (regno_max_ref_mode): ...this new variable.
1296 (coalesced_pseudo_reg_slot_compare): Update accordingly.
1297 Use wider_subreg_mode.
1298 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
1299 rather than an unsigned int *.
1300 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
1301 (process_alt_operands): Likewise.
1302 (invariant_p): Likewise.
1303 * lra-spills.c (assign_mem_slot): Likewise.
1304 (add_pseudo_to_slot): Likewise.
1305 * lra.c (collect_non_operand_hard_regs): Likewise.
1306 (add_regs_to_insn_regno_info): Likewise.
1307 * reload1.c (regno_max_ref_width): Replace with...
1308 (regno_max_ref_mode): ...this new variable.
1309 (reload): Update accordingly. Update call to
1310 ira_sort_regnos_for_alter_reg.
1311 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
1312 (init_eliminable_invariants): Update to use regno_max_ref_mode.
1313 (scan_paradoxical_subregs): Likewise.
1315 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1317 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
1318 (aarch64_frame): Add emit_frame_chain boolean.
1319 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1320 Move eh_return case to aarch64_layout_frame.
1321 (aarch64_layout_frame): Initialize emit_frame_chain.
1322 (aarch64_expand_prologue): Use emit_frame_chain.
1324 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1326 * config/aarch64/aarch64.c (aarch64_layout_frame):
1327 Ensure LR is always stored at the bottom of the callee-saves.
1328 Remove rarely used frame layout which saves callee-saves at top of
1329 frame, so the store of LR can be used as a valid probe in all cases.
1331 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1333 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
1334 Improve unaligned TImode/TFmode base/offset split.
1336 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1337 Alan Hayward <alan.hayward@arm.com>
1338 David Sherwood <david.sherwood@arm.com>
1340 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
1341 * combine.c (find_single_use_1): Likewise.
1342 (expand_field_assignment): Likewise.
1343 (move_deaths): Likewise.
1344 * lra-constraints.c (simplify_operand_subreg): Likewise.
1345 (curr_insn_transform): Likewise.
1346 * lra.c (collect_non_operand_hard_regs): Likewise.
1347 (add_regs_to_insn_regno_info): Likewise.
1348 * rtlanal.c (reg_referenced_p): Likewise.
1349 (covers_regno_no_parallel_p): Likewise.
1351 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1353 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
1354 Don't print any bits outside the precision of the value.
1355 * wide-int.cc (test_printing): Add some new tests.
1357 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1359 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
1360 supports -xbrace_comment option.
1361 * configure: Regenerate.
1362 * config.in: Regenerate.
1363 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
1364 (ASM_CPU_SPEC): Use it.
1366 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1368 * target.def (static_rtx_alignment): New hook.
1369 * targhooks.h (default_static_rtx_alignment): Declare.
1370 * targhooks.c (default_static_rtx_alignment): New function.
1371 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
1372 * doc/tm.texi: Regenerate.
1373 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
1374 instead of targetm.constant_alignment. Remove call to
1376 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1377 (cris_preferred_mininum_alignment): New function, split out from...
1378 (cris_constant_alignment): ...here.
1379 (cris_static_rtx_alignment): New function.
1380 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
1382 (ix86_constant_alignment): ...here.
1383 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1384 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1385 (mmix_static_rtx_alignment): New function.
1386 * config/spu/spu.c (spu_static_rtx_alignment): New function.
1387 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1389 2017-10-26 Tamar Christina <tamar.christina@arm.com>
1392 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
1393 Add flag_trapping_math and flag_fp_int_builtin_inexact.
1395 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
1397 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
1398 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
1399 mark as a sign-extending load.
1400 (local_pic_load_u): Define.
1402 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
1405 * fold-const.c (operand_equal_for_comparison_p): Also return true
1406 if ARG0 is a simple variant of ARG1 with narrower precision.
1407 (fold_ternary_loc): Always pass unstripped operands to the predicate.
1409 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
1411 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
1413 * i386.h (processor_costs): Add gather_static, gather_per_elt,
1414 scatter_static, scatter_per_elt.
1415 * x86-tune-costs.h: Add new cost entries.
1417 2017-10-25 Richard Biener <rguenther@suse.de>
1419 * tree-ssa-sccvn.h (vn_eliminate): Declare.
1420 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
1421 class pass_fre): Move to ...
1422 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
1423 class pass_fre): ... here and adjust for statistics.
1425 2017-10-25 Jakub Jelinek <jakub@redhat.com>
1428 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
1429 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
1430 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
1431 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
1434 2017-10-25 Richard Biener <rguenther@suse.de>
1436 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
1437 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
1438 eliminate_push_avail, eliminate_insert): Move inside...
1439 (class eliminate_dom_walker): ... this class in preparation
1441 (fini_eliminate): Remove by merging with ...
1442 (eliminate): ... this function. Adjust for class changes.
1443 (pass_pre::execute): Remove fini_eliminate call.
1444 (pass_fre::execute): Likewise.
1446 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1449 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
1450 (VPERMI2, VPERMI2I): New mode iterators.
1451 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
1452 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
1454 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
1455 mode iterator. Remove 3 old define_insn patterns.
1456 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
1457 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
1458 VPERMI2 mode iterator, remove the other two expanders.
1459 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
1460 to use VPERMI2 mode iterator, add another alternative for vpermi2*
1461 instructions, remove the other two patterns.
1462 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
1463 mode iterator, remove the other two patterns.
1464 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
1465 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
1466 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
1467 and adjust argument order accordingly.
1468 (ix86_expand_vec_perm): Adjust caller.
1469 (expand_vec_perm_1): Likewise.
1470 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
1471 (expand_vec_perm_vpermt2_vpshub2): ... this.
1472 (ix86_expand_vec_perm_const_1): Adjust caller.
1473 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
1476 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
1477 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
1478 (vec_shl_<mode>): Remove unused expander.
1479 (avx512bw_<shift_insn><mode>3): New define_insn.
1480 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
1481 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
1483 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
1486 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
1489 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1491 PR rtl-optimization/82396
1492 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
1493 (autopref_multipass_init): Simplify initialization.
1494 (autopref_rank_data): Simplify sort order.
1495 * gcc/sched-int.h (autopref_multipass_data_): Remove
1496 multi_mem_insn_p, min_offset and max_offset.
1498 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1501 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1502 Check special value of flag_omit_frame_pointer.
1503 (aarch64_can_eliminate): Likewise.
1504 (aarch64_override_options_after_change_1): Simplify handling of
1505 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
1507 2017-10-24 Richard Biener <rguenther@suse.de>
1509 PR tree-optimization/82697
1510 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
1511 zero for conditional load and unconditional store.
1513 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1515 * doc/install.texi: Document bootstrap-cet.
1517 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1520 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
1521 ENDBR instruction at function entrance if function is only
1524 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1527 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
1528 patterns to better describe from which operation the CF is computed.
1529 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
1530 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
1531 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
1532 is 0, use _0 suffixed expanders instead of emitting a comparison
1535 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
1537 * config/i386/i386.md(*movsf_internal, *movdf_internal):
1538 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
1540 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
1543 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
1544 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
1545 * loop-iv.c (iv_get_reaching_def): Likewise.
1546 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
1547 variable is promoted and the partition contains undefined values.
1549 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1551 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
1553 (nios2_address_cost): Define.
1554 (nios2_legitimize_address): Recognize (exp + constant) directly.
1555 (TARGET_ADDRESS_COST): Define.
1557 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1559 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
1560 (nios2_symbolic_memory_operand_p): Declare.
1561 (nios2_split_large_constant): Declare.
1562 (nios2_split_symbolic_memory_operand): Declare.
1563 * config/nios2/nios2.c: Adjust includes.
1564 (nios2_symbolic_constant_allowed): New.
1565 (nios2_symbolic_constant_p): New.
1566 (nios2_plus_symbolic_constant_p): New.
1567 (nios2_valid_addr_expr_p): Recognize addresses involving
1569 (nios2_legitimate_address_p): Likewise, also LO_SUM.
1570 (nios2_symbolic_memory_operand_p): New.
1571 (nios2_large_constant_p): New.
1572 (nios2_split_large_constant): New.
1573 (nios2_split_plus_large_constant): New.
1574 (nios2_split_symbolic_memory_operand): New.
1575 (nios2_legitimize_address): Code refactoring. Handle addresses
1576 involving symbolic constants.
1577 (nios2_emit_move_sequence): Likewise.
1578 (nios2_print_operand): Improve error output.
1579 (nios2_print_operand_address): Handle LO_SUM.
1580 (nios2_cdx_narrow_form_p): Likewise.
1581 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
1582 operands involving symbolic constants.
1583 (movhi_internal, movsi_internal): Likewise.
1584 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
1585 (extendhisi2, extendqi<mode>2): Likewise.
1587 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1589 * tree-pass.h (PROP_rtl_split_insns): Define.
1590 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
1592 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1594 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
1596 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1599 * target.def (const_not_ok_for_debug_p): Default to
1600 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
1601 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
1602 * targhooks.c (default_const_not_ok_for_debug_p): New function.
1603 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
1604 which targetm.const_not_ok_for_debug_p returned true.
1605 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
1607 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
1609 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
1610 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
1611 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
1613 (ix86_const_not_ok_for_debug_p): New function.
1614 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
1615 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
1617 2017-10-23 David Malcolm <dmalcolm@redhat.com>
1620 * system.h: Conditionally include "unique-ptr.h" if
1621 INCLUDE_UNIQUE_PTR is defined.
1622 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
1623 of defining INCLUDE_UNIQUE_PTR before including "system.h".
1625 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
1627 * config/rl78/rl78.md: New define_expand "subdi3".
1629 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
1632 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
1633 DF_REF_INSN if DF_REF_INSN_INFO is false.
1635 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
1637 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
1638 xmm_move instead of sse_move.
1639 (sse_store_index): New function.
1640 (ix86_register_move_cost): Be more sensible about mismatch stall;
1641 model AVX moves correctly; make difference between sse->integer and
1643 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
1644 moves; make difference between SSE and AVX.
1645 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
1646 and zmm_move. Increase size of sse load and store tables;
1647 add unaligned load and store tables; add ssemmx_to_integer.
1648 * x86-tune-costs.h: Update all entries according to real
1649 move latencies from Agner Fog's manual and chip documentation.
1651 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1654 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
1655 * config/i386/constraints.md (Wf): New constraint.
1656 * config/i386/i386.md (UNSPEC_SBB): New unspec.
1657 (cmp<dwi>_doubleword): Removed.
1658 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
1659 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
1660 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
1661 expand with cmp<dwi>_doubleword. For LTU and GEU use
1662 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
1664 * common.opt (gcolumn-info): Enable by default.
1665 * doc/invoke.texi (gcolumn-info): Document new default.
1667 2017-10-23 Richard Biener <rguenther@suse.de>
1669 PR tree-optimization/82672
1670 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
1671 Fold the stmt if we propagated into it.
1673 2017-10-23 Richard Biener <rguenther@suse.de>
1675 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
1676 (bitmap_remove_expr_from_set): ... this. All callers call this
1677 for non-constant values.
1678 (bitmap_set_subtract): Rename to...
1679 (bitmap_set_subtract_expressions): ... this. Adjust and
1681 (bitmap_set_contains_value): Remove superfluous check.
1682 (bitmap_set_replace_value): Inline into single caller ...
1683 (bitmap_value_replace_in_set): ... here and simplify.
1684 (dependent_clean): Merge into ...
1685 (clean): ... this using an overload. Adjust.
1686 (prune_clobbered_mems): Adjust.
1687 (compute_antic_aux): Likewise.
1688 (compute_partial_antic_aux): Likewise.
1690 2017-10-23 Richard Biener <rguenther@suse.de>
1692 PR tree-optimization/82129
1694 2017-08-01 Richard Biener <rguenther@suse.de>
1696 PR tree-optimization/81181
1697 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1698 (compute_antic): ... end of iteration here.
1700 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1702 * target.def (starting_frame_offset): New hook.
1703 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
1704 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
1705 * doc/tm.texi.in: Regenerate.
1706 * hooks.h (hook_hwi_void_0): Declare.
1707 * hooks.c (hook_hwi_void_0): New function.
1708 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
1709 STARTING_FRAME_OFFSET.
1710 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
1711 * reload1.c (reload): Likewise.
1712 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
1713 instead of STARTING_FRAME_OFFSET.
1714 * function.c (try_fit_stack_local): Likewise.
1715 (assign_stack_local_1): Likewise
1716 (instantiate_virtual_regs): Likewise.
1717 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
1718 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
1719 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
1720 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
1721 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
1722 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
1723 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
1724 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
1725 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
1726 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
1727 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
1728 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
1729 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
1730 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
1731 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
1732 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
1733 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
1734 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
1735 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
1736 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
1737 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
1738 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
1739 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
1740 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
1741 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
1742 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
1743 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
1744 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
1745 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
1746 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
1747 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
1748 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
1749 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
1750 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
1751 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
1752 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
1753 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
1754 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
1755 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
1756 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
1757 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
1758 * config/avr/avr.c (avr_starting_frame_offset): Make static and
1759 return a HOST_WIDE_INT.
1760 (avr_builtin_setjmp_frame_value): Use it instead of
1761 STARTING_FRAME_OFFSET.
1762 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1763 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
1764 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
1766 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1767 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
1768 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
1769 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1770 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
1771 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
1772 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1773 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
1774 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
1775 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1776 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
1777 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
1779 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1780 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
1781 * config/mips/mips.c (mips_compute_frame_info): Refer to
1782 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
1783 (mips_starting_frame_offset): New function.
1784 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1785 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
1786 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
1787 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
1788 and return a HOST_WIDE_INT.
1789 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1790 (mmix_initial_elimination_offset): Refer to
1791 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
1792 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
1793 * config/pa/pa.c (pa_starting_frame_offset): New function.
1794 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
1795 (pa_expand_prologue): Likewise.
1796 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1797 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
1798 !FRAME_GROWS_DOWNWARD handling to...
1799 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1800 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
1801 !FRAME_GROWS_DOWNWARD handling to...
1802 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1803 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
1804 !FRAME_GROWS_DOWNWARD handling to...
1805 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1806 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
1808 (rs6000_starting_frame_offset): New function.
1809 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
1810 !FRAME_GROWS_DOWNWARD handling to...
1811 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1812 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
1813 !FRAME_GROWS_DOWNWARD handling to...
1814 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1815 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
1816 !FRAME_GROWS_DOWNWARD handling to...
1817 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1818 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
1819 (rs6000_starting_frame_offset): New function.
1820 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
1821 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
1822 * config/vax/vax.c (vax_starting_frame_offset): New function.
1823 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
1824 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1825 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
1826 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
1827 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1828 * system.h (STARTING_FRAME_OFFSET): Poison.
1830 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1832 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
1833 SCALAR_TYPE_MODE instead of TYPE_MODE.
1835 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1836 Alan Hayward <alan.hayward@arm.com>
1837 David Sherwood <david.sherwood@arm.com>
1839 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
1841 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1842 Alan Hayward <alan.hayward@arm.com>
1843 David Sherwood <david.sherwood@arm.com>
1845 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
1847 2017-10-23 Richard Biener <rguenther@suse.de>
1849 PR tree-optimization/82129
1850 * tree-ssa-pre.c (bitmap_set_and): Remove.
1851 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
1852 canonicalizing expressions in the set to those with lowest
1853 ID rather than taking that from the first edge.
1855 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1857 * combine.c (rtx_equal_for_field_assignment_p): Use
1858 byte_lowpart_offset.
1860 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1861 Alan Hayward <alan.hayward@arm.com>
1862 David Sherwood <david.sherwood@arm.com>
1864 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
1865 to a SUBREG_PROMOTED_VAR.
1867 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1868 Alan Hayward <alan.hayward@arm.com>
1869 David Sherwood <david.sherwood@arm.com>
1871 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
1872 (expand_debug_source_expr): Likewise.
1873 * combine.c (combine_simplify_rtx): Likewise.
1874 * cse.c (fold_rtx): Likewise.
1875 * optabs.c (expand_float): Likewise.
1876 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1877 (simplify_binary_operation_1): Likewise.
1879 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1880 Alan Hayward <alan.hayward@arm.com>
1881 David Sherwood <david.sherwood@arm.com>
1883 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
1884 (record_promoted_value): Likewise.
1885 * expr.c (expand_expr_real_2): Likewise.
1886 * ree.c (update_reg_equal_equiv_notes): Likewise.
1887 (combine_set_extension): Likewise.
1888 * rtlanal.c (low_bitmask_len): Likewise.
1889 * simplify-rtx.c (neg_const_int): Likewise.
1890 (simplify_binary_operation_1): Likewise.
1892 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1893 Alan Hayward <alan.hayward@arm.com>
1894 David Sherwood <david.sherwood@arm.com>
1896 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
1897 * regcprop.c (maybe_mode_change): Likewise.
1898 * reload1.c (alter_reg): Likewise.
1900 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1902 * inchash.h (inchash::hash::add_wide_int): New function.
1903 * lto-streamer-out.c (hash_tree): Use it.
1905 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1907 * inchash.h (inchash::hash::add_wide_int): Rename to...
1908 (inchash::hash::add_hwi): ...this.
1909 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
1910 (polymorphic_call_target_hasher::hash): Likewise.
1911 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
1912 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
1913 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
1914 * lto-streamer-out.c (hash_tree): Likewise.
1915 * optc-save-gen.awk: Likewise.
1916 * tree.c (add_expr): Likewise.
1918 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1921 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
1922 for ordered inequality comparisons even with TARGET_IEEE_FP.
1924 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1927 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
1928 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1929 Expand with cmp<dwi>_doubleword.
1931 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1933 * extend.texi: Add x86 specific to 'nocf_check' attribute.
1934 List CET intrinsics.
1935 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
1936 specific to -fcf-protection option.
1938 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1940 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
1941 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
1942 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
1943 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
1944 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
1945 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
1946 (extra_objs): Add cet.o for Linux/x86 targets.
1947 (tmake_file): Add i386/t-cet for Linux/x86 targets.
1948 * config/i386/cet.c: New file.
1949 * config/i386/cetintrin.h: Likewise.
1950 * config/i386/t-cet: Likewise.
1951 * config/i386/cpuid.h (bit_SHSTK): New.
1952 (bit_IBT): Likewise.
1953 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
1954 pass IBT and SHSTK bits.
1955 * config/i386/i386-builtin-types.def
1956 (VOID_FTYPE_UNSIGNED_PVOID): New.
1957 (VOID_FTYPE_UINT64_PVOID): Likewise.
1958 * config/i386/i386-builtin.def: Add CET intrinsics.
1959 * config/i386/i386-c.c (ix86_target_macros_internal): Add
1960 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
1961 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
1962 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
1964 * config/i386/i386.c (rest_of_insert_endbranch): New.
1965 (pass_data_insert_endbranch): Likewise.
1966 (pass_insert_endbranch): Likewise.
1967 (make_pass_insert_endbranch): Likewise.
1968 (ix86_notrack_prefixed_insn_p): Likewise.
1969 (ix86_target_string): Add -mibt, -mshstk flags.
1970 (ix86_option_override_internal): Add flag_cf_protection
1972 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
1973 (ix86_print_operand): Add 'notrack' prefix output.
1974 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
1975 (ix86_expand_builtin): Expand CET intrinsics.
1976 (x86_output_mi_thunk): Add 'endbranch' instruction.
1977 * config/i386/i386.h (TARGET_IBT): New.
1978 (TARGET_IBT_P): Likewise.
1979 (TARGET_SHSTK): Likewise.
1980 (TARGET_SHSTK_P): Likewise.
1981 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
1982 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
1983 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
1984 (builtin_setjmp_setup): New pattern.
1985 (builtin_longjmp): Likewise.
1986 (rdssp<mode>): Likewise.
1987 (incssp<mode>): Likewise.
1988 (saveprevssp): Likewise.
1989 (rstorssp): Likewise.
1990 (wrss<mode>): Likewise.
1991 (wruss<mode>): Likewise.
1992 (setssbsy): Likewise.
1993 (clrssbsy): Likewise.
1994 (nop_endbr): Likewise.
1995 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
1997 * config/i386/immintrin.h: Include <cetintrin.h>.
1998 * config/i386/linux-common.h
1999 (file_end_indicate_exec_stack_and_cet): New prototype.
2000 (TARGET_ASM_FILE_END): New.
2002 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2004 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
2005 latencies instead of having separate table; make difference between
2006 integer and float costs.
2007 * i386.h (processor_costs): Remove scalar_stmt_cost,
2008 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
2009 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
2011 * x86-tune-costs.h: Remove entries which has been removed in
2012 procesor_costs from all tables; make cond_taken_branch_cost
2013 and cond_not_taken_branch_cost COST_N_INSNS based.
2015 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
2017 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
2019 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2021 * config/i386/i386.md (isa): Remove fma_avx512f.
2022 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
2023 <avx512>_fmadd_<mode>_mask3<round_name>,
2024 <avx512>_fmsub_<mode>_mask<round_name>,
2025 <avx512>_fmsub_<mode>_mask3<round_name>,
2026 <avx512>_fnmadd_<mode>_mask<round_name>,
2027 <avx512>_fnmadd_<mode>_mask3<round_name>,
2028 <avx512>_fnmsub_<mode>_mask<round_name>,
2029 <avx512>_fnmsub_<mode>_mask3<round_name>,
2030 <avx512>_fmaddsub_<mode>_mask<round_name>,
2031 <avx512>_fmaddsub_<mode>_mask3<round_name>,
2032 <avx512>_fmsubadd_<mode>_mask<round_name>,
2033 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
2034 (*vec_widen_umult_even_v16si<mask_name>,
2035 *vec_widen_smult_even_v16si<mask_name>): Likewise.
2036 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
2038 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2040 * extend.texi: Add 'nocf_check' documentation.
2041 * gimple.texi: Add second parameter to
2042 gimple_build_call_from_tree.
2043 * invoke.texi: Add -fcf-protection documentation.
2044 * rtl.texi: Add REG_CALL_NOTRACK documenation.
2046 2017-10-20 Richard Biener <rguenther@suse.de>
2048 PR tree-optimization/82473
2049 * tree-vect-loop.c (vectorizable_reduction): Properly get at
2050 the largest input type.
2052 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
2054 * c-attribs.c (handle_nocf_check_attribute): New function.
2055 (c_common_attribute_table): Add 'nocf_check' handling.
2056 * gimple-parser.c: Add second argument NULL to
2057 gimple_build_call_from_tree.
2058 * attrib.c (comp_type_attributes): Check nocf_check attribute.
2059 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
2061 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
2062 * common.opt: Add fcf-protection flag.
2063 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
2064 * flag-types.h: Add enum cf_protection_level.
2065 * gimple.c (gimple_build_call_from_tree): Add second parameter.
2066 Add 'nocf_check' attribute propagation to gimple call.
2067 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
2068 (gimple_build_call_from_tree): Update prototype.
2069 (gimple_call_nocf_check_p): New function.
2070 (gimple_call_set_nocf_check): Likewise.
2071 * gimplify.c: Add second argument to gimple_build_call_from_tree.
2072 * ipa-icf.c: Add nocf_check attribute in statement hash.
2073 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
2074 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
2075 * toplev.c (process_options): Add flag_cf_protection handling.
2077 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2079 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
2081 2017-10-20 Richard Biener <rguenther@suse.de>
2083 PR tree-optimization/82603
2084 * tree-if-conv.c (predicate_mem_writes): Make sure to only
2085 remove false predicated stores.
2087 2017-10-20 Richard Biener <rguenther@suse.de>
2089 * graphite-isl-ast-to-gimple.c
2090 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
2091 Remove return value and simplify, dump copied stmt after lhs
2093 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
2094 Reduce dump verbosity.
2095 (gsi_insert_earliest): Likewise.
2096 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
2097 * graphite.c (print_global_statistics): Adjust dumping.
2098 (print_graphite_scop_statistics): Likewise.
2099 (print_graphite_statistics): Do not dump loops here.
2100 (graphite_transform_loops): But here.
2102 2017-10-20 Nicolas Roche <roche@adacore.com>
2104 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
2105 * configure: Regenerate.
2107 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2110 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
2111 functions when optimizing replace GIMPLE_RETURN stmts with
2112 calls to __builtin_unreachable ().
2115 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
2116 for -fsanitize=thread link of executables.
2117 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
2118 link of executables.
2121 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
2123 (<shift_insn><mode>3<mask_name>): Change the last of the 3
2124 define_insns for logical vector shifts to use VI248_AVX512BW
2125 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
2126 condition, useless isa and prefix attributes. Change the first
2127 2 of these define_insns to ...
2128 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
2129 define_insn for avx512vl.
2130 (<shift_insn><mode>3): ... and this, new define_insn without
2131 masking for non-avx512vl.
2134 * config/i386/sse.md (*andnot<mode>3,
2135 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
2136 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
2137 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
2138 not applied use empty suffix even for TARGET_AVX512VL.
2139 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
2140 is applied, supply evex,evex or evex,evex,evex instead of just
2143 2017-10-20 Julia Koval <julia.koval@intel.com>
2145 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
2146 (OPTION_MASK_ISA_GFNI_UNSET): New.
2147 (ix86_handle_option): Handle OPT_mgfni.
2148 * config/i386/cpuid.h (bit_GFNI): New.
2149 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
2150 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
2151 * config/i386/i386.c (ix86_target_string): Add -mgfni.
2152 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
2153 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
2154 * config/i386/i386.opt: Add mgfni.
2156 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
2158 * config/msp430/msp430.c (msp430_option_override): Disable
2159 -fdelete-null-pointer-checks.
2160 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
2162 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2164 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
2165 of x87 and SSE instructions.
2167 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2169 * asan.c (create_cond_insert_point): Do not update edge count.
2170 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
2171 (afdo_propagate_circuit): Likewise.
2172 (afdo_calculate_branch_prob): Likewise.
2173 (afdo_annotate_cfg): Likewise.
2174 * basic-block.h (struct edge_def): Remove count.
2175 (edge_def::count): New accessor.
2176 * bb-reorder.c (rotate_loop): Update.
2177 (find_traces_1_round): Update.
2178 (connect_traces): Update.
2179 (sanitize_hot_paths): Update.
2180 * cfg.c (unchecked_make_edge): Update.
2181 (make_single_succ_edge): Update.
2182 (check_bb_profile): Update.
2183 (dump_edge_info): Update.
2184 (update_bb_profile_for_threading): Update.
2185 (scale_bbs_frequencies_int): Update.
2186 (scale_bbs_frequencies_gcov_type): Update.
2187 (scale_bbs_frequencies_profile_count): Update.
2188 (scale_bbs_frequencies): Update.
2189 * cfganal.c (connect_infinite_loops_to_exit): Update.
2190 * cfgbuild.c (compute_outgoing_frequencies): Update.
2191 (find_many_sub_basic_blocks): Update.
2192 * cfgcleanup.c (try_forward_edges): Update.
2193 (try_crossjump_to_edge): Update
2194 * cfgexpand.c (expand_gimple_cond): Update
2195 (expand_gimple_tailcall): Update
2196 (construct_exit_block): Update
2197 * cfghooks.c (verify_flow_info): Update
2198 (redirect_edge_succ_nodup): Update
2199 (split_edge): Update
2200 (make_forwarder_block): Update
2201 (duplicate_block): Update
2202 (account_profile_record): Update
2203 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
2204 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
2205 * cfgloopmanip.c (scale_loop_profile): Update.
2207 (lv_adjust_loop_entry_edge): Update.
2208 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
2209 (force_nonfallthru_and_redirect): Update.
2210 (purge_dead_edges): Update.
2211 (rtl_flow_call_edges_add): Update.
2212 * cgraphunit.c (init_lowered_empty_function): Update.
2213 (cgraph_node::expand_thunk): Update.
2214 * gimple-pretty-print.c (dump_probability): Update.
2215 (dump_edge_probability): Update.
2216 * gimple-ssa-isolate-paths.c (isolate_path): Update.
2217 * haifa-sched.c (sched_create_recovery_edges): Update.
2218 * hsa-gen.c (convert_switch_statements): Update.
2219 * ifcvt.c (dead_or_predicable): Update.
2220 * ipa-inline-transform.c (inline_transform): Update.
2221 * ipa-split.c (split_function): Update.
2222 * ipa-utils.c (ipa_merge_profiles): Update.
2223 * loop-doloop.c (add_test): Update.
2224 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
2225 * lto-streamer-in.c (input_cfg): Update.
2226 (input_function): Update.
2227 * lto-streamer-out.c (output_cfg): Update.
2228 * modulo-sched.c (sms_schedule): Update.
2229 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
2230 * predict.c (maybe_hot_edge_p): Update.
2231 (unlikely_executed_edge_p): Update.
2232 (probably_never_executed_edge_p): Update.
2233 (dump_prediction): Update.
2234 (drop_profile): Update.
2235 (propagate_unlikely_bbs_forward): Update.
2236 (determine_unlikely_bbs): Update.
2237 (force_edge_cold): Update.
2238 * profile.c (compute_branch_probabilities): Update.
2239 * reg-stack.c (better_edge): Update.
2240 * shrink-wrap.c (handle_simple_exit): Update.
2241 * tracer.c (better_p): Update.
2242 * trans-mem.c (expand_transaction): Update.
2243 (split_bb_make_tm_edge): Update.
2244 * tree-call-cdce.c: Update.
2245 * tree-cfg.c (gimple_find_sub_bbs): Update.
2246 (gimple_split_edge): Update.
2247 (gimple_duplicate_sese_region): Update.
2248 (gimple_duplicate_sese_tail): Update.
2249 (gimple_flow_call_edges_add): Update.
2250 (insert_cond_bb): Update.
2251 (execute_fixup_cfg): Update.
2252 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
2253 * tree-complex.c (expand_complex_div_wide): Update.
2254 * tree-eh.c (lower_resx): Update.
2255 (unsplit_eh): Update.
2256 (cleanup_empty_eh_move_lp): Update.
2257 * tree-inline.c (copy_edges_for_bb): Update.
2258 (freqs_to_counts): Update.
2259 (copy_cfg_body): Update.
2260 * tree-ssa-dce.c (remove_dead_stmt): Update.
2261 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
2262 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
2263 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
2264 (unloop_loops): Update.
2265 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2266 * tree-ssa-loop-split.c (connect_loops): Update.
2267 (split_loop): Update.
2268 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
2269 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
2270 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
2271 * tree-ssa-reassoc.c (branch_fixup): Update.
2272 * tree-ssa-tail-merge.c (replace_block_by): Update.
2273 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
2274 (compute_path_counts): Update.
2275 (update_profile): Update.
2276 (recompute_probabilities): Update.
2277 (update_joiner_offpath_counts): Update.
2278 (estimated_freqs_path): Update.
2279 (freqs_to_counts_path): Update.
2280 (clear_counts_path): Update.
2281 (ssa_fix_duplicate_block_edges): Update.
2282 (duplicate_thread_path): Update.
2283 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
2284 (case_bit_test_cmp): Update.
2285 (collect_switch_conv_info): Update.
2286 (gen_inbound_check): Update.
2287 (do_jump_if_equal): Update.
2288 (emit_cmp_and_jump_insns): Update.
2289 * tree-tailcall.c (decrease_profile): Update.
2290 (eliminate_tail_call): Update.
2291 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
2292 (vect_do_peeling): Update.
2293 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2294 * ubsan.c (ubsan_expand_null_ifn): Update.
2295 (ubsan_expand_ptr_ifn): Update.
2296 * value-prof.c (gimple_divmod_fixed_value): Update.
2297 (gimple_mod_pow2): Update.
2298 (gimple_mod_subtract): Update.
2299 (gimple_ic): Update.
2300 (gimple_stringop_fixed_value): Update.
2302 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2305 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
2307 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
2309 PR rtl-optimization/82395
2310 * ira-color.c (allocno_priority_compare_func): Fix comparison step
2311 based on non_spilled_static_chain_regno_p.
2313 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2315 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
2316 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
2317 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
2319 2017-10-19 Martin Sebor <msebor@redhat.com>
2321 PR tree-optimization/82596
2322 * tree.c (array_at_struct_end_p): Handle STRING_CST.
2324 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2326 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
2327 (get_mem_refs_of_builtin_call): Likewise.
2328 * builtins.c (expand_builtin_apply): Adjust call to
2329 allocate_dynamic_stack_space.
2330 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
2331 the third argument to allocate_dynamic_stack_space, otherwise -1.
2332 (expand_builtin): Deal with all alloca variants.
2333 (is_inexpensive_builtin): Likewise.
2334 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
2335 * calls.c (special_function_p): Deal with all alloca variants.
2336 (initialize_argument_information): Adjust call to
2337 allocate_dynamic_stack_space.
2338 (expand_call): Likewise.
2339 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
2340 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
2341 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
2342 use it for the stack usage computation.
2343 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
2344 * function.c (gimplify_parameters): Call build_alloca_call_expr.
2345 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
2346 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
2347 (in_loop_p): Remove first argument and useless check.
2348 (pass_walloca::execute): Remove useless test and adjust call to above.
2349 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
2350 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
2351 (gimplify_call_expr): Deal with all alloca variants.
2352 * hsa-gen.c (gen_hsa_alloca): Likewise.
2353 (gen_hsa_insns_for_call): Likewise.
2354 * ipa-pure-const.c (special_builtin_state): Likewise.
2355 * tree-chkp.c (chkp_build_returned_bound): Likewise.
2356 * tree-object-size.c (alloc_object_size): Likewise.
2357 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2358 (call_may_clobber_ref_p_1): Likewise.
2359 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
2360 (ccp_fold_stmt): Likewise.
2361 (optimize_stack_restore): Likewise.
2362 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2363 (mark_all_reaching_defs_necessary_1): Likewise.
2364 (propagate_necessity): Likewise.
2365 (eliminate_unnecessary_stmts): Likewise.
2366 * tree.c (build_common_builtin_nodes): Build
2367 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
2368 (build_alloca_call_expr): New function.
2369 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
2370 (CASE_BUILT_IN_ALLOCA): Likewise.
2371 (build_alloca_call_expr): Declare.
2372 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
2374 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2377 * dwarf2out.c (new_die_raw): New static inline function.
2378 (new_die): Use it to create the DIE.
2379 (add_AT_external_die_ref): Likewise.
2380 (clone_die): Likewise.
2381 (clone_as_declaration): Likewise.
2382 (dwarf2out_vms_debug_main_pointer): Likewise.
2383 (base_type_die): Likewise. Remove early return for corner cases.
2384 Do not call add_pubtype on the DIE here.
2385 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
2386 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
2387 typedefs for base types with DW_AT_endianity. Make sure a DIE with
2388 native order exists for base types, attach the DIE manually and call
2389 add_pubtype on it. Do not equate a reverse order DIE to the type.
2391 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
2393 * config/arm/arm.c (align_ok_ldrd_strd): New function.
2394 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
2396 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
2398 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2400 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
2401 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
2402 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
2403 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
2404 * opts.c (sanitizer_opts): Add builtin.
2405 * ubsan.c (instrument_builtin): New function.
2406 (pass_ubsan::execute): Call it.
2407 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
2408 * doc/invoke.texi: Document -fsanitize=builtin.
2410 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
2411 builtins, store max (log2 (align), 0) into uchar field instead of
2412 align into uptr field.
2413 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
2414 store uchar 0 field instead of uptr 0 field.
2415 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
2416 instead of passing one address of struct with 2 locations pass
2417 two addresses of structs with 1 location each.
2418 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
2419 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
2420 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
2421 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
2422 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
2423 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
2424 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
2425 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
2427 2017-10-19 Martin Liska <mliska@suse.cz>
2430 * file-find.c (remove_prefix): Remove.
2431 * file-find.h (remove_prefix): Likewise.
2432 * gcc-ar.c: Remove smartness of lookup.
2434 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
2436 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
2437 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
2438 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
2440 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2443 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
2444 (setcc + and to xor + setcc): New peephole2.
2446 2017-10-19 Tom de Vries <tom@codesourcery.com>
2448 * doc/sourcebuild.texi (Test Directives, Variants of
2449 dg-require-support): Add dg-require-stack-size.
2451 2017-10-19 Martin Liska <mliska@suse.cz>
2454 * gimplify.c (gimplify_decl_expr): Do not instrument variables
2455 that have a large alignment.
2456 (gimplify_target_expr): Likewise.
2458 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
2460 PR rtl-optimization/82602
2461 * ira.c (rtx_moveable_p): Return false for volatile asm.
2463 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
2466 * config/i386/i386-modes.def (CCGZ): New CC mode.
2467 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
2468 * config/i386/predicates.md (ix86_comparison_operator):
2470 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2471 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
2472 with double-word subtraction.
2473 (put_condition_code): Handle CCGZmode.
2475 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
2477 * wide-int.cc (debug (const wide_int &)): New.
2478 (debug (const wide_int *)): New.
2479 (debug (const widest_int &)): New.
2480 (debug (const widest_int *)): New.
2482 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
2485 * lra-constraints.c (curr_insn_transform): Use non-input operand
2486 instead of output one for matched reload.
2488 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2490 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
2491 (tree-ssa-loop-ivopts.h): New header file.
2492 (struct builtin_info): New fields.
2493 (classify_builtin_1): Compute and record base and offset parts for
2494 memset builtin partition by calling strip_offset.
2495 (offset_cmp, fuse_memset_builtins): New functions.
2496 (finalize_partitions): Fuse adjacent memset partitions by calling
2498 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
2499 Expose the interface.
2500 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
2502 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2504 PR tree-optimization/82574
2505 * tree-loop-distribution.c (find_single_drs): New parameter. Check
2506 that data reference must be executed exactly once per iteration
2507 against the outermost loop in nest.
2508 (classify_partition): Update call to above function.
2510 2017-10-18 Richard Biener <rguenther@suse.de>
2512 PR tree-optimization/82591
2513 * graphite.c (graphite_transform_loops): Move code gen message
2515 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2516 Here. Handle scop_to_isl_ast failing.
2517 (scop_to_isl_ast): Limit the number of ISL operations.
2519 2017-10-18 Richard Biener <rguenther@suse.de>
2521 * graphite-isl-ast-to-gimple.c
2522 (translate_isl_ast_to_gimple::set_rename): Simplify.
2523 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
2524 (graphite_copy_stmts_from_block): ... here.
2525 (copy_bb_and_scalar_dependences): Simplify.
2526 (add_parameters_to_ivs_params): Canonicalize.
2527 (generate_entry_out_of_ssa_copies): Simplify.
2528 * graphite-sese-to-poly.c (extract_affine_name): Simplify
2529 by passing in ISL dimension.
2530 (parameter_index_in_region_1): Rename to ...
2531 (parameter_index_in_region): ... this.
2532 (extract_affine): Adjust assert, pass down parameter index.
2533 (add_param_constraints): Use range-info when available.
2534 (build_scop_context): Adjust.
2535 * sese.c (new_sese_info): Adjust.
2536 (free_sese_info): Likewise.
2537 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
2538 Remove unused typedefs.
2539 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
2541 2017-10-18 Martin Liska <mliska@suse.cz>
2543 * combine.c (simplify_compare_const): Add gcc_fallthrough.
2545 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2547 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
2548 (s390_sched_init): Do not reset s390_sched_state if we entered the
2549 current basic block via a fallthru edge and all others are unlikely.
2551 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2553 * config/s390/s390.c (NUM_SIDES): New variable.
2554 (LONGRUNNING_THRESHOLD): New variable.
2555 (LATENCY_FACTOR): New variable.
2556 (s390_sched_score): Decrease score for long-running instructions on
2558 (s390_sched_variable_issue): Perform bookkeeping for long-running
2561 2017-10-18 Richard Biener <rguenther@suse.de>
2563 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
2564 Simplify with removal of the parameter rename map.
2565 (set_rename): Likewise.
2566 (should_copy_to_new_region): Likewise.
2567 (graphite_copy_stmts_from_block): Likewise.
2568 (copy_bb_and_scalar_dependences): Remove initialization of
2569 unused copied_bb_map.
2571 (copy_internal_parameters): Likewise.
2572 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
2573 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
2574 Use INTEGRAL_TYPE_P.
2575 (parameter_index_in_region_1): Rename to ...
2576 (assign_parameter_index_in_region): ... this. Assert we have
2577 a parameter we handle.
2578 (scan_tree_for_params): Adjust.
2579 * sese.h (parameter_rename_map_t): Remove.
2580 (struct sese_info_t): Remove unused parameter_rename_map and
2581 copied_bb_map members.
2582 * sese.c (new_sese_info): Adjust.
2583 (free_sese_info): Likewise.
2585 2017-10-18 Martin Liska <mliska@suse.cz>
2588 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
2589 on an abnormal edge.
2591 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
2593 * doc/invoke.texi (ffunction-sections and fdata-sections):
2596 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2598 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
2599 the use statement can throw internally.
2601 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2603 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
2604 any RTX present on the RHS of a SET.
2605 * compare-elim.c (try_eliminate_compare): Restore comment.
2607 2017-10-17 Jakub Jelinek <jakub@redhat.com>
2609 * langhooks.h (struct lang_hooks): Document that tree_size langhook
2610 may be also called on tcc_type nodes.
2611 * langhooks.c (lhd_tree_size): Likewise.
2613 2017-10-17 David Malcolm <dmalcolm@redhat.com>
2615 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
2616 format_warning_at_substring.
2617 (maybe_warn): Convert source_range * param to a location_t. Pass
2618 UNKNOWN_LOCATION rather than NULL to fmtwarn.
2619 (format_directive): Remove code to extract source_ranges and
2620 source_range * in favor of just a location_t.
2621 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
2623 * substring-locations.c (format_warning_va): Convert
2624 source_range * param to a location_t.
2625 (format_warning_at_substring): Likewise.
2626 * substring-locations.h (format_warning_va): Likewise.
2627 (format_warning_at_substring): Likewise.
2629 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
2631 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
2633 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
2634 and scatter/gather ops.
2636 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
2637 vec_gather_load and vec_scatter_store.
2638 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
2639 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
2640 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
2641 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
2642 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
2643 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
2645 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
2647 * reg-stack.c (compare_for_stack_reg): Add bool argument.
2648 Detect FTST instruction and handle its register pops. Only pop
2649 second operand if can_pop_second_op is true.
2650 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
2651 set can_pop_second_op to false in the compare_for_stack_reg call.
2653 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
2654 output_fp_compare for stack register operands.
2655 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
2656 instructions here. Do not emit stack register pops here. Assert
2657 that FCOMPP pops next to top stack register. Rewrite function.
2659 2017-10-17 Nathan Sidwell <nathan@acm.org>
2662 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
2663 use DECL_ASSEMBLER_NAME_RAW.
2666 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
2669 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
2670 Wilco Dijkstra <wilco.dijkstra@arm.com>
2672 * builtins.c (expand_builtin_update_setjmp_buf): Add a
2673 converstion to Pmode from the buf_addr.
2675 2017-10-17 Richard Biener <rguenther@suse.de>
2677 * graphite-dependences.c (scop_get_reads_and_writes): Change
2678 output parameters to references.
2680 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
2682 PR 71026/tree-optimization
2683 * fold-const.c (distribute_real_division): Removed.
2684 (fold_binary_loc): Remove calls to distribute_real_divison.
2686 2017-10-17 Richard Biener <rguenther@suse.de>
2688 * graphite-scop-detection.c
2689 (scop_detection::stmt_has_simple_data_refs_p): Always use
2690 the full nest as region.
2691 (try_generate_gimple_bb): Likewise.
2692 * sese.c (scalar_evolution_in_region): Simplify now that
2693 SCEV can handle instantiation in regions.
2694 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
2695 in the non-loop part of a function if requested.
2697 2017-10-17 Richard Biener <rguenther@suse.de>
2699 PR tree-optimization/82563
2700 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
2702 (graphite_regenerate_ast_isl): Call it.
2703 * graphite-scop-detection.c (build_scops): Remove entry edge split.
2705 2017-10-17 Jakub Jelinek <jakub@redhat.com>
2707 PR tree-optimization/82549
2708 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
2709 Formatting fixes. Instead of calling make_bit_field_ref with negative
2712 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
2714 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
2715 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
2716 _mm_maskz_reduce_ss): New.
2717 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
2718 __builtin_ia32_reducess_mask): Ditto..
2719 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
2720 * config/i386/sse.md (reduces<mode>): Renamed to ...
2721 (reduces<mode><mask_scalar_name>): ... this.
2722 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
2724 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
2725 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
2727 2017-10-16 David Malcolm <dmalcolm@redhat.com>
2729 * Makefile.in (OBJS): Add unique-ptr-tests.o.
2730 * selftest-run-tests.c (selftest::run_tests): Call
2731 selftest::unique_ptr_tests_cc_tests.
2732 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
2733 * unique-ptr-tests.cc: New file.
2735 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
2738 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2740 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2741 (make_hard_regno_born, make_hard_regno_dead): Update
2742 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
2744 2017-10-16 Jeff Law <law@redhat.com>
2746 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
2748 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
2750 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
2752 2017-10-16 Olivier Hainque <hainque@adacore.com>
2754 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
2755 with_cpu if we were configured for an e500v2 target cpu name.
2757 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
2759 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
2760 * doc/invoke.texi: Document +nodsp as a valid extension for
2763 2017-10-16 Martin Liska <mliska@suse.cz>
2765 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
2766 (test_set_range): Likewise.
2767 (test_range_functions): Rename to ...
2768 (test_bit_in_range): ... this.
2769 (sbitmap_c_tests): Add new test.
2771 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2773 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
2775 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
2776 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
2778 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2780 * config/aarch64/aarch64-builtins.c
2781 (aarch64_types_quadopu_lane_qualifiers): New.
2782 (TYPES_QUADOPU_LANE): New.
2783 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
2784 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
2785 (aarch64_<sur>dot_laneq<vsi2qi>): New.
2786 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
2787 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
2788 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
2789 (Vdottype, DOTPROD): New.
2790 (sur): Add SDOT and UDOT.
2792 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2794 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
2795 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
2796 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
2798 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
2799 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
2800 Enable TARGET_DOTPROD.
2801 (cortex-a75.cortex-a55): Likewise.
2802 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
2804 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2806 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
2807 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
2809 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
2811 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
2812 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
2813 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
2814 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
2815 * config/arm/types.md (neon_dot, neon_dot_q): New.
2816 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
2818 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2820 * config/arm/arm.h (TARGET_DOTPROD): New.
2821 * config/arm/arm.c (arm_arch_dotprod): New.
2822 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
2823 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
2824 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
2825 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
2826 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
2827 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
2828 * doc/invoke.texi (armv8.2-a): Document dotprod
2830 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2832 * i386.c (ix86_vec_cost): New function.
2833 (ix86_rtx_costs): Handle vector operations better.
2834 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
2835 * x86-tune-costs.h: Add new costs to all tables.
2837 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2839 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
2841 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
2842 divsd, sqrtss and sqrtsd
2843 * x86-tune-costs.h: Add new entries to all costs.
2844 (znver1_cost): Fix to match real instruction latencies.
2846 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2847 Michael Collison <michael.collison@arm.com>
2849 * compare-elim.c: Include emit-rtl.h.
2850 (can_merge_compare_into_arith): New function.
2851 (try_validate_parallel): Likewise.
2852 (try_merge_compare): Likewise.
2853 (try_eliminate_compare): Call the above when no previous clobber
2855 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
2858 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2862 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
2863 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
2867 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
2868 to be any operand_equal_p operands. For & (B - 1) require
2869 B to be power of 2. Recognize
2870 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
2872 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2875 * optabs.c (expand_memory_blockage): Fix call of
2876 targetm.have_memory_blockage.
2878 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2881 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
2882 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
2883 objects to extra_objs instead of overwriting it.
2885 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2887 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
2888 Use any_fp_register_operand as operand[3] predicate. Simplify
2889 equality test for operands[2] and operands[4] memory location.
2890 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
2891 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
2892 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
2893 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
2894 any_fp_register_operand as operand[1] predicate. Simplify
2895 equality test for operands[0] and operands[3] memory location.
2896 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
2897 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
2898 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
2900 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2902 * target-insns.def: Add memory_blockage.
2903 * optabs.c (expand_memory_blockage): New function.
2904 (expand_asm_memory_barrier): Rename ...
2905 (expand_asm_memory_blockage): ... to this.
2906 (expand_mem_thread_fence): Call expand_memory_blockage
2907 instead of expand_asm_memory_barrier.
2908 (expand_mem_singnal_fence): Ditto.
2909 (expand_atomic_load): Ditto.
2910 (expand_atomic_store): Ditto.
2911 * doc/md.texi (Standard Pattern Names For Generation):
2912 Document memory_blockage instruction pattern.
2914 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
2916 * config/rl78/rl78.c (rl78_emit_libcall): New function.
2917 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
2918 * config/rl78/rl78.md: New define_expand "adddi3".
2920 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2922 * cfghooks.c (verify_flow_info): Disable check that all probabilities
2925 2017-10-13 Jeff Law <law@redhat.com>
2927 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
2929 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2932 * internal-fn.c (expand_mul_overflow): If both operands have
2933 the same highpart of -1 or 0 and the topmost bit of lowpart
2934 is different, overflow is if res <= 0 rather than res < 0.
2936 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
2938 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
2939 TARGET_P9_VECTOR code for unaligned_load case.
2941 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2943 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
2945 2017-10-13 Nathan Sidwell <nathan@acm.org>
2947 * tree-core.h (tree_contains_struct): Make bool.
2948 * tree.c (tree_contains_struct): Likewise.
2949 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
2950 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
2951 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
2952 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
2953 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
2955 2017-10-13 Richard Biener <rguenther@suse.de>
2957 * graphite-isl-ast-to-gimple.c
2958 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
2959 parameters and dominance check.
2960 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
2961 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
2962 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
2963 Do not update SSA form here or do intermediate IL verification.
2964 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
2965 (graphite_initialize): Remove check on the number of loops in
2966 the function and inline into graphite_transform_loops.
2967 (graphite_finalize): Inline into graphite_transform_loops.
2968 (graphite_transform_loops): Perform SSA update and IL verification
2970 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
2972 2017-10-13 Richard Biener <rguenther@suse.de>
2974 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
2975 graphite_expression_type_precision): Avoid global constructor
2977 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
2978 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
2979 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
2980 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
2981 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
2982 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
2984 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
2987 * config/i386/i386.h (ix86_red_zone_size): New.
2988 * config/i386/i386.md (push peephole2s): Replace
2989 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
2991 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2992 Alan Hayward <alan.hayward@arm.com>
2993 David Sherwood <david.sherwood@arm.com>
2995 * combine.c (can_change_dest_mode): Reject changes in
2996 REGMODE_NATURAL_SIZE.
2998 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2999 Alan Hayward <alan.hayward@arm.com>
3000 David Sherwood <david.sherwood@arm.com>
3002 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
3003 (expand_debug_source_expr): Likewise.
3004 * combine.c (combine_simplify_rtx): Likewise.
3005 * cse.c (fold_rtx): Likewise.
3006 * fwprop.c (canonicalize_address): Likewise.
3007 * targhooks.c (default_shift_truncation_mask): Likewise.
3009 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3010 Alan Hayward <alan.hayward@arm.com>
3011 David Sherwood <david.sherwood@arm.com>
3013 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
3014 (widened_mode): Likewise.
3015 (expand_unop): Likewise.
3016 * ree.c (transform_ifelse): Likewise.
3017 (merge_def_and_ext): Likewise.
3018 (combine_reaching_defs): Likewise.
3019 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
3021 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3022 Alan Hayward <alan.hayward@arm.com>
3023 David Sherwood <david.sherwood@arm.com>
3025 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
3026 * combine.c (gen_lowpart_for_combine): Likewise.
3027 * dwarf2out.c (rtl_for_decl_location): Likewise.
3028 * final.c (alter_subreg): Likewise.
3029 * rtlhooks.c (gen_lowpart_general): Likewise.
3030 (gen_lowpart_if_possible): Likewise.
3032 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
3033 Alan Hayward <alan.hayward@arm.com>
3034 David Sherwood <david.sherwood@arm.com>
3036 * calls.c (expand_call): Use subreg_lowpart_offset.
3037 * cse.c (cse_insn): Likewise.
3038 * regcprop.c (copy_value): Likewise.
3039 (copyprop_hardreg_forward_1): Likewise.
3041 2017-10-13 Jakub Jelinek <jakub@redhat.com>
3044 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
3045 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
3046 =Q constraints to +Q and into insn condition add check
3047 that operands[0] and operands[1] are equal.
3048 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
3049 =Q constraints to +Q and into insn condition add check
3050 that operands[0] is equal to either operands[1] or operands[2].
3053 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
3054 instead of handling MINUS_EXPR twice (once for each argument),
3055 canonicalize operand order and handle just once, use rtype where
3056 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
3059 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
3060 any values of __C while still being pattern recognizable as a simple
3063 2017-10-13 Richard Biener <rguenther@suse.de>
3065 PR tree-optimization/82451
3067 2017-10-02 Richard Biener <rguenther@suse.de>
3069 PR tree-optimization/82355
3070 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
3071 a mapping for the enclosing loop but avoid generating one for
3073 (copy_bb_and_scalar_dependences): Remove premature codegen
3074 error on PHIs in blocks duplicated into multiple places.
3075 * graphite-scop-detection.c
3076 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
3077 in the region use it as loop and nest to analyze the DR in.
3078 (try_generate_gimple_bb): Likewise.
3079 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
3080 (add_loop_constraints): For blocks in a loop not in the region
3081 create a dimension with a single iteration.
3082 * sese.h (gbb_loop_at_index): Remove assert.
3084 * cfgloop.c (loop_preheader_edge): For the loop tree root
3085 return the single successor of the entry block.
3086 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
3087 Reset the SCEV hashtable and niters.
3088 * graphite-scop-detection.c
3089 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
3090 assert that we only have POLYNOMIAL_CHREC that vary in loops
3091 contained in the region.
3092 (scop_detection::graphite_can_represent_expr): Adjust.
3093 (scop_detection::stmt_has_simple_data_refs_p): For loops
3094 not in the region set loop to NULL. The nest is now the
3095 entry edge to the region.
3096 (try_generate_gimple_bb): Likewise.
3097 * sese.c (scalar_evolution_in_region): Adjust for
3098 instantiate_scev change.
3099 * tree-data-ref.h (graphite_find_data_references_in_stmt):
3100 Make nest parameter the edge into the region.
3101 (create_data_ref): Likewise.
3102 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
3103 entry edge into a region and adjust instantiate_scev calls.
3104 (create_data_ref): Likewise.
3105 (graphite_find_data_references_in_stmt): Likewise.
3106 (find_data_references_in_stmt): Pass the loop preheader edge
3107 from the nest argument.
3108 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
3109 parameter the edge into the region.
3110 (instantiate_parameters): Use the loop preheader edge as entry.
3111 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
3113 (get_instantiated_value_entry): Make instantiate_below parameter
3114 the edge into the region.
3115 (instantiate_scev_name): Likewise. Adjust dominance checks,
3116 when we cannot use loop-based instantiation instantiate by
3117 walking use-def chains.
3118 (instantiate_scev_poly): Adjust.
3119 (instantiate_scev_binary): Likewise.
3120 (instantiate_scev_convert): Likewise.
3121 (instantiate_scev_not): Likewise.
3122 (instantiate_array_ref): Remove.
3123 (instantiate_scev_3): Likewise.
3124 (instantiate_scev_2): Likewise.
3125 (instantiate_scev_1): Likewise.
3126 (instantiate_scev_r): Do not blindly handle N-operand trees.
3127 Do not instantiate array-refs. Handle all constants and invariants.
3128 (instantiate_scev): Make instantiate_below parameter
3129 the edge into the region.
3130 (resolve_mixers): Use the loop preheader edge for the region
3131 parameter to instantiate_scev_r.
3132 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
3134 2017-10-13 Richard Biener <rguenther@suse.de>
3136 PR tree-optimization/82525
3137 * graphite-isl-ast-to-gimple.c
3138 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
3140 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
3141 Fail code generation when we cannot represent the isl integer.
3142 (binary_op_to_tree): Elide modulo operations that are no-ops
3143 in the type we code generate. Remove now superfluous code
3146 2017-10-13 Richard Biener <rguenther@suse.de>
3148 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
3149 (scop_detection::harmful_loop_in_region): Remove premature
3150 IV type restriction.
3151 (scop_detection::graphite_can_represent_scev): We can handle
3152 pointer IVs just fine.
3154 2017-10-13 Alan Modra <amodra@gmail.com>
3156 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
3157 "Clobbers and Scratch Registers". Add paragraph on
3158 alternative to clobbers for scratch registers and OpenBLAS
3161 2017-10-13 Alan Modra <amodra@gmail.com>
3163 * doc/extend.texi (Clobbers): Correct vax example. Delete old
3164 example of a memory input for a string of known length. Move
3165 commentary out of table. Add a number of new examples
3166 covering array memory inputs.
3168 2017-10-12 Martin Liska <mliska@suse.cz>
3170 PR tree-optimization/82493
3171 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
3172 (test_range_functions): New function.
3173 (sbitmap_c_tests): Likewise.
3174 * selftest-run-tests.c (selftest::run_tests): Run new tests.
3175 * selftest.h (sbitmap_c_tests): New function.
3177 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3179 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
3181 * config/rs6000/amo.h: Fix spacing issue.
3183 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3186 * config/i386/i386.md (*ashl<mode>3_mask_1,
3187 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
3188 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
3191 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3193 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
3194 (profile_probability): Set max_probability
3195 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
3197 * profile-count.c (profile_probability::differs_from_p): Do not
3198 rely on max_probaiblity == 10000
3200 2017-10-12 Jeff Law <law@redhat.com>
3202 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
3205 2017-10-12 Martin Sebor <msebor@redhat.com>
3209 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
3210 (handle_alias_pairs): Call it.
3211 * common.opt (-Wattribute-alias): New option.
3212 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
3213 * doc/invoke.texi (-Wattribute-alias): Document.
3215 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
3218 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3220 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3222 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3223 (make_hard_regno_born, make_hard_regno_dead): Update
3224 bb_killed_pseudos and bb_gen_pseudos.
3226 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3228 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
3230 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
3232 * config/alpha/alpha.c (alpha_split_conditional_move):
3233 Use std::swap instead of manually swapping.
3234 (alpha_stdarg_optimize_hook): Ditto.
3235 (alpha_canonicalize_comparison): Ditto.
3237 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3239 * tree-loop-distribution.c (struct builtin_info): New struct.
3240 (struct partition): Refactor fields into struct builtin_info.
3241 (partition_free): Free struct builtin_info.
3242 (build_size_arg_loc, build_addr_arg_loc): Delete.
3243 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
3244 information from struct builtin_info.
3245 (find_single_drs): New function refactored from classify_partition.
3246 Also moved builtin validity checks to this function.
3247 (compute_access_range, alloc_builtin): New functions.
3248 (classify_builtin_st, classify_builtin_ldst): New functions.
3249 (classify_partition): Refactor code into functions find_single_drs,
3250 classify_builtin_st and classify_builtin_ldst.
3251 (distribute_loop): Don't do runtime alias check when distributing
3253 (find_seed_stmts_for_distribution): New function.
3254 (pass_loop_distribution::execute): Refactor code finding seed
3255 stmts into above function. Support distribution for the innermost
3256 two-level loop nest. Adjust dump information.
3258 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3260 * tree-loop-distribution.c: Adjust the general comment.
3261 (NUM_PARTITION_THRESHOLD): New macro.
3262 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
3263 (classify_partition): Skip builtin pattern of loop nest's inner loop.
3264 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
3265 in call to build_partition_graph.
3266 (finalize_partitions): New parameter. Make loop distribution more
3267 conservative by fusing more partitions.
3268 (distribute_loop): Don't do runtime alias check in case of loop nest
3270 (find_seed_stmts_for_distribution): New function.
3271 (prepare_perfect_loop_nest): New function.
3272 (pass_loop_distribution::execute): Refactor code finding seed stmts
3273 and loop nest into above functions. Support loop nest distribution.
3274 Adjust dump information accordingly.
3276 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3278 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
3279 and set PTYPE_SEQUENTIAL for merged partition.
3281 2017-10-12 Richard Biener <rguenther@suse.de>
3283 PR tree-optimization/69728
3285 2017-09-19 Richard Biener <rguenther@suse.de>
3287 PR tree-optimization/69728
3288 * graphite-sese-to-poly.c (schedule_error): New global.
3289 (add_loop_schedule): Handle empty domain by failing the
3291 (build_original_schedule): Handle schedule_error.
3293 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
3294 domain by returning an unchanged schedule.
3296 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3298 * genrecog.c (validate_pattern): For VEC_SELECT verify that
3299 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
3301 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
3303 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
3306 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3309 * expr.c (store_field): Don't optimize away bitsize == 0 store
3310 from CALL_EXPR with addressable return type.
3312 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3314 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
3315 * config/rs6000/rs6000.md (sel): Delete mode attribute.
3316 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
3317 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
3318 TARGET_ISEL instead of TARGET_ISEL<sel>.
3320 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
3322 * config/rs6000/rs6000.c
3323 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
3325 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3327 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
3329 (reg_or_cint_operand): Fix comment.
3330 (reg_or_zero_operand): New predicate.
3331 * config/rs6000/rs6000-protos.h (output_isel): Delete.
3332 * config/rs6000/rs6000.c (output_isel): Delete.
3333 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
3334 instead of reg_or_cint_operand. Output instruction directly (not via
3336 (isel_unsigned_<mode>): Ditto.
3337 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
3338 gpc_reg_operand. Add an instruction alternative for this. Output
3339 instruction directly.
3340 (*isel_reversed_unsigned_<mode>): Ditto.
3342 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3344 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
3345 (TARGET_CANONICALIZE_COMPARISON): Define.
3347 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
3350 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
3351 Check whether the dest is REG before adding REG_EQUIV note.
3353 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3356 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3358 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3359 (make_hard_regno_born, make_hard_regno_dead): Update
3360 bb_killed_pseudos and bb_gen_pseudos.
3362 2017-10-11 Nathan Sidwell <nathan@acm.org>
3364 * incpath.h (enum incpath_kind): Name enum, prefix values.
3365 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
3366 * incpath.c (heads, tails): Use INC_MAX.
3367 (add_env_var_paths, add_standard_paths): Use incpath_kind.
3368 (merge_include_chains, split_quote_chain,
3369 register_include_chains): Update incpath_kind names.
3370 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
3371 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
3373 (add_framework_path, darwin_register_objc_includes): Likewise.
3374 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
3376 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3378 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
3379 Do not use float_operator operator predicate.
3380 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
3381 * config/i386/predicates.md (float_operator): Remove predicate.
3383 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3385 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
3386 (*jccxf_i387): Ditto.
3387 (*jcc<mode>_i387): Ditto.
3388 (*jccu<mode>_i387): Ditto.
3389 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
3390 (*jcc_*_i387 splitters): Remove.
3391 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
3392 * config/i386/i386.c (ix86_split_fp_branch): Remove.
3393 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
3396 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3398 * profile-count.h (slow_safe_scale_64bit): New function.
3399 (safe_scale_64bit): New inline.
3400 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
3401 * profile-count.c: Include wide-int.h
3402 (slow_safe_scale_64bit): New.
3404 2017-10-11 Nathan Sidwell <nathan@acm.org>
3406 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
3407 HAS_DECL_ASSEMBLER_NAME_P.
3408 * gimple-expr.c (gimple_decl_printable_name: Check
3409 HAS_DECL_ASSEMBLER_NAME_P too.
3410 * ipa-utils.h (type_in_anonymous_namespace_p): Check
3411 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
3412 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
3413 * passes.c (rest_of_decl_compilation): Check
3414 HAS_DECL_ASSEMBLER_NAME_P too.
3415 * recog.c (verify_changes): Likewise.
3416 * tree-pretty-print.c (dump_decl_name): Likewise.
3417 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
3419 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
3420 (SET_DECL_ASSEMBLER_NAME): Use it.
3421 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
3422 (COPY_DECL_ASSEMBLER_NAME): Likewise.
3423 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
3425 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3427 * config.gcc (i386, x86_64): Add extra objects.
3428 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
3429 (ix86_min_insn_size): Declare.
3430 (ix86_issue_rate): Declare.
3431 (ix86_adjust_cost): Declare.
3432 (ia32_multipass_dfa_lookahead): Declare.
3433 (ix86_macro_fusion_p): Declare.
3434 (ix86_macro_fusion_pair_p): Declare.
3435 (ix86_bd_has_dispatch): Declare.
3436 (ix86_bd_do_dispatch): Declare.
3437 (ix86_core2i7_init_hooks): Declare.
3438 (ix86_atom_sched_reorder): Declare.
3439 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
3440 (COSTS_N_BYTES): Move to x86-tune-costs.h.
3441 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
3442 (rip_relative_addr_p): Rename to ...
3443 (ix86_rip_relative_addr_p): ... this one; export.
3444 (memory_address_length): Update.
3445 (ix86_issue_rate): Move to x86-tune-sched.c.
3446 (ix86_flags_dependent): Move to x86-tune-sched.c.
3447 (ix86_agi_dependent): Move to x86-tune-sched.c.
3448 (exact_dependency_1): Move to x86-tune-sched.c.
3449 (exact_store_load_dependency): Move to x86-tune-sched.c.
3450 (ix86_adjust_cost): Move to x86-tune-sched.c.
3451 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
3452 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
3453 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
3454 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
3455 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
3456 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
3457 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
3458 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
3459 (min_insn_size): Rename to ...
3460 (ix86_min_insn_size): ... this one; export.
3461 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
3462 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
3463 (core2i7_first_cycle_multipass_backtrack): Move to
3464 x86-tune-sched-core.c.
3465 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
3466 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
3467 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
3468 (ix86_avoid_jump_mispredicts): Update.
3469 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
3470 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
3471 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
3472 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
3473 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
3474 (MAX_INSN): Move to ix86-tune-sched-bd.c.
3475 (MAX_IMM): Move to ix86-tune-sched-bd.c.
3476 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
3477 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
3478 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
3479 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
3480 (MAX_STORE): Move to ix86-tune-sched-bd.c.
3481 (BIG): Move to ix86-tune-sched-bd.c.
3482 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
3483 (enum insn_path): Move to ix86-tune-sched-bd.c.
3484 (get_mem_group): Move to ix86-tune-sched-bd.c.
3485 (is_cmp): Move to ix86-tune-sched-bd.c.
3486 (dispatch_violation): Move to ix86-tune-sched-bd.c.
3487 (is_branch): Move to ix86-tune-sched-bd.c.
3488 (is_prefetch): Move to ix86-tune-sched-bd.c.
3489 (init_window): Move to ix86-tune-sched-bd.c.
3490 (allocate_window): Move to ix86-tune-sched-bd.c.
3491 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
3492 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
3493 (process_end_window): Move to ix86-tune-sched-bd.c.
3494 (allocate_next_window): Move to ix86-tune-sched-bd.c.
3495 (find_constant): Move to ix86-tune-sched-bd.c.
3496 (get_num_immediates): Move to ix86-tune-sched-bd.c.
3497 (has_immediate): Move to ix86-tune-sched-bd.c.
3498 (get_insn_path): Move to ix86-tune-sched-bd.c.
3499 (get_insn_group): Move to ix86-tune-sched-bd.c.
3500 (count_num_restricted): Move to ix86-tune-sched-bd.c.
3501 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
3502 (add_insn_window): Move to ix86-tune-sched-bd.c.
3503 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
3504 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
3505 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
3506 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
3507 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
3508 (do_dispatch): Move to ix86-tune-sched-bd.c.
3509 (has_dispatch): Move to ix86-tune-sched-bd.c.
3510 * config/i386/t-i386: Add new object files.
3511 * config/i386/x86-tune-costs.h: New file.
3512 * config/i386/x86-tune-sched-atom.c: New file.
3513 * config/i386/x86-tune-sched-bd.c: New file.
3514 * config/i386/x86-tune-sched-core.c: New file.
3515 * config/i386/x86-tune-sched.c: New file.
3517 2017-10-11 Liu Hao <lh_mouse@126.com>
3519 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
3520 the generic version below instead.
3521 (should_colorize): Recognize Windows consoles as terminals
3523 * pretty-print.c [__MINGW32__] (write_all): New function.
3524 [__MINGW32__] (find_esc_head): Likewise.
3525 [__MINGW32__] (find_esc_terminator): Likewise.
3526 [__MINGW32__] (eat_esc_sequence): Likewise.
3527 [__MINGW32__] (mingw_ansi_fputs): New function that handles
3529 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
3532 2017-10-11 Richard Biener <rguenther@suse.de>
3534 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
3535 Properly call analyze_scalar_evolution with the loop of the stmt.
3537 2017-10-11 Richard Biener <rguenther@suse.de>
3539 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
3540 * tree-core.h (tree_base): Add chrec_var union member.
3541 * tree.h (CHREC_VAR): Remove.
3542 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
3543 * tree-chrec.h (build_polynomial_chrec): Adjust.
3544 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
3545 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
3547 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
3549 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
3550 * match.pd: ... here.
3551 ((T) X == (T) Y): Relax condition.
3553 2017-10-11 Bin Cheng <bin.cheng@arm.com>
3555 PR tree-optimization/82472
3556 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
3558 (break_alias_scc_partitions): Update postorder number.
3560 2017-10-11 Martin Liska <mliska@suse.cz>
3563 * opts.c (parse_no_sanitize_attribute): Do not use error_value
3565 * opts.h (parse_no_sanitize_attribute): Remove last argument.
3567 2017-10-11 Martin Liska <mliska@suse.cz>
3569 * print-rtl.c (print_insn): Move declaration of idbuf
3570 to same scope as name.
3572 2017-10-11 Martin Liska <mliska@suse.cz>
3577 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3580 2017-10-11 Martin Liska <mliska@suse.cz>
3583 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3586 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
3588 * config.gcc (default_gnu_indirect_function): Default to yes for
3589 arm*-*-linux* with glibc.
3591 2017-10-11 Richard Biener <rguenther@suse.de>
3593 * tree-scalar-evolution.c (get_scalar_evolution): Handle
3594 default-defs and types we do not want to analyze.
3595 (interpret_loop_phi): Replace unreachable code with an assert.
3596 (compute_scalar_evolution_in_loop): Remove and inline ...
3597 (analyze_scalar_evolution_1): ... here, replacing condition with
3598 what makes the intent clearer. Remove handling of cases
3599 get_scalar_evolution now handles.
3601 2017-10-10 Jim Wilson <wilson@tuliptree.org>
3603 PR rtl-optimization/81434
3604 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
3605 comment for main loop. In sched_group_found if, also add checks for
3606 pass and min_cost_group.
3608 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
3610 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
3611 (rs6000_insn_cost): New function.
3612 * config/rs6000/rs6000.md (cost): New attribute.
3614 2017-10-10 Jakub Jelinek <jakub@redhat.com>
3615 H.J. Lu <hongjiu.lu@intel.com>
3619 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
3620 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
3621 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
3622 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
3623 to OPTION_MASK_ISA_AVX512VL - builtins that have both
3624 OPTION_MASK_ISA_MMX and some other bit set require both
3625 mmx and the ISAs without the mmx bit.
3626 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
3627 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
3628 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
3629 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
3630 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
3631 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
3632 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
3633 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
3634 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
3635 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
3636 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
3637 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
3638 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
3639 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
3640 Add OPTION_MASK_ISA_MMX.
3642 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
3644 * config.gcc (armv7*-*-freebsd*): New target.
3645 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
3647 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
3649 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
3650 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
3653 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
3655 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
3656 a template parameter.
3657 (WIDE_INT_REF_FOR): Update accordingly.
3658 * tree.h (wi::int_traits <const_tree>): Delete.
3659 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
3660 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
3661 (wi::tree_to_wide_ref): New typedef.
3662 (wi::to_wide): New function.
3663 * calls.c (get_size_range): Use wi::to_wide when operating on
3665 * cgraph.c (cgraph_node::create_thunk): Likewise.
3666 * config/i386/i386.c (ix86_data_alignment): Likewise.
3667 (ix86_local_alignment): Likewise.
3668 * dbxout.c (stabstr_O): Likewise.
3669 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
3670 * expr.c (const_vector_from_tree): Likewise.
3671 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
3672 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
3673 (fold_negate_expr_1, int_const_binop_1, const_binop)
3674 (fold_convert_const_int_from_real, optimize_bit_field_compare)
3675 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
3676 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
3677 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
3678 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
3679 (fold_not_const, round_up_loc): Likewise.
3680 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
3681 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
3682 (alloca_call_type): Likewise.
3683 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
3684 * godump.c (go_output_typedef): Likewise.
3685 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
3686 * internal-fn.c (get_min_precision): Likewise.
3687 * ipa-cp.c (ipcp_store_vr_results): Likewise.
3688 * ipa-polymorphic-call.c
3689 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
3690 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
3691 (ipa_modify_call_arguments): Likewise.
3692 * match.pd: Likewise.
3693 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
3694 * print-tree.c (print_node_brief, print_node): Likewise.
3695 * stmt.c (expand_case): Likewise.
3696 * stor-layout.c (layout_type): Likewise.
3697 * tree-affine.c (tree_to_aff_combination): Likewise.
3698 * tree-cfg.c (group_case_labels_stmt): Likewise.
3699 * tree-data-ref.c (dr_analyze_indices): Likewise.
3700 (prune_runtime_alias_test_list): Likewise.
3701 * tree-dump.c (dequeue_and_dump): Likewise.
3702 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
3703 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
3704 * tree-pretty-print.c (dump_generic_node): Likewise.
3705 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
3706 (simple_iv_with_niters): Likewise.
3707 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
3708 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
3709 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
3710 * tree-ssa-loop-niter.c (split_to_var_and_offset)
3711 (refine_value_range_using_guard, number_of_iterations_ne_max)
3712 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
3713 (get_cst_init_from_scev, record_nonwrapping_iv)
3714 (scev_var_range_cant_overflow): Likewise.
3715 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
3716 * tree-ssa-pre.c (compute_avail): Likewise.
3717 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
3718 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
3719 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
3720 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
3721 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
3722 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
3723 (dump_case_nodes, try_switch_expansion): Likewise.
3724 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
3725 (vect_do_peeling): Likewise.
3726 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
3727 * tree-vect-stmts.c (vectorizable_load): Likewise.
3728 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
3729 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
3730 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
3731 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
3732 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
3733 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
3734 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
3735 (evrp_dom_walker::before_dom_children): Likewise.
3736 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
3737 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
3738 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
3739 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
3740 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
3741 (get_range_pos_neg): Likewise.
3742 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
3743 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
3744 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
3745 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
3746 * config/avr/avr.c (avr_fold_builtin): Likewise.
3747 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
3748 * config/msp430/msp430.c (msp430_attr): Likewise.
3749 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
3750 * config/powerpcspe/powerpcspe-c.c
3751 (altivec_resolve_overloaded_builtin): Likewise.
3752 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
3753 (rs6000_expand_ternop_builtin): Likewise.
3754 * config/rs6000/rs6000-c.c
3755 (altivec_resolve_overloaded_builtin): Likewise.
3756 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
3757 (rs6000_expand_ternop_builtin): Likewise.
3758 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
3760 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3762 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
3763 when copying loop nest with only one inner loop.
3765 2017-10-10 Richard Biener <rguenther@suse.de>
3767 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
3768 blocks if SCEV is active.
3769 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
3771 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
3772 (scev_initialize): Assert we are not yet initialized.
3774 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3776 * tree-loop-distribution.c (generate_loops_for_partition): Remove
3777 inner loop's exit stmt by making it always exit the loop, otherwise
3778 we would generate an infinite empty loop.
3780 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3782 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
3783 renaming variables in new preheader if it's deleted.
3785 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3787 * tree-loop-distribution.c (struct partition): Remove unused field
3788 loops of the structure.
3789 (partition_alloc, partition_free): Ditto.
3790 (build_rdg_partition_for_vertex): Ditto.
3792 2017-10-09 Jeff Law <law@redhat.com>
3794 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
3795 return type to match prototype and documentation.
3797 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3799 * config/rs6000/rs6000.c (processor_costs): Move to ...
3800 * config/rs6000/rs6000.h: ... here.
3801 (rs6000_cost): Declare.
3803 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
3805 * except.c (setjmp_fn): New global variable.
3806 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
3807 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
3808 if DONT_USE_BUILTIN_SETJMP is defined.
3810 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3812 * target.def (insn_cost): New hook.
3813 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
3814 * doc/tm.texi: Regenerate.
3815 * rtlanal.c (insn_cost): Use the new hook.
3817 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3819 * combine.c (combine_validate_cost): Compute the new insn_cost,
3820 not just pattern_cost.
3821 (try_combine): Adjust comment.
3823 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3825 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
3827 * combine.c (uid_insn_cost): Adjust comment.
3828 (combine_validate_cost): Adjust comment. Use pattern_cost instead
3830 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
3831 * dse.c (find_shift_sequence): Ditto.
3832 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
3833 (bb_valid_for_noce_process_p): Use pattern_cost.
3834 * rtl.h (insn_rtx_cost): Delete.
3835 (pattern_cost): New prototype.
3836 (insn_cost): New prototype.
3837 * rtlanal.c (insn_rtx_cost): Rename to...
3838 (pattern_cost): ... this.
3841 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
3843 * config/i386/i386.md (*jcc_2): Remove insn pattern.
3844 (*jcc<mode>_0_r_i387): Ditto.
3845 (*jccxf_r_i387): Ditto.
3846 (*jcc<mode>_r_i387): Ditto.
3847 (*jccu<mode>_r_i387): Ditto.
3848 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
3849 (*jcc): Rename from *jcc_1.
3851 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3853 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
3854 deferred rescans after the lvx/stvx recombination pre-pass.
3856 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
3858 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
3859 memory operation instruction support.
3860 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
3861 (rs6000-ibm-aix[789]*): Likewise.
3862 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
3863 Document new functions.
3865 2017-10-09 Richard Biener <rguenther@suse.de>
3867 PR tree-optimization/82397
3868 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
3869 equality only for semantically equal trees.
3871 2017-10-09 Richard Biener <rguenther@suse.de>
3873 PR tree-optimization/82449
3874 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
3875 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
3876 allow constant addresses.
3877 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
3880 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3882 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
3885 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3888 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
3891 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3894 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
3896 2017-10-09 Jakub Jelinek <jakub@redhat.com>
3899 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
3900 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
3902 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
3904 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
3905 (WI_BINARY_PREDICATE_RESULT): Likewise.
3906 (wi::binary_traits::operator_result): New type.
3907 (wi::binary_traits::predicate_result): Likewise.
3908 (generic_wide_int::operator~, unary generic_wide_int::operator-)
3909 (generic_wide_int::operator==, generic_wide_int::operator!=)
3910 (generic_wide_int::operator&, generic_wide_int::and_not)
3911 (generic_wide_int::operator|, generic_wide_int::or_not)
3912 (generic_wide_int::operator^, generic_wide_int::operator+
3913 (binary generic_wide_int::operator-, generic_wide_int::operator*):
3915 (operator~, unary operator-, operator==, operator!=, operator&)
3916 (operator|, operator^, operator+, binary operator-, operator*): New
3918 * expr.c (get_inner_reference): Use wi::bit_and_not.
3919 * fold-const.c (fold_binary_loc): Likewise.
3920 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
3921 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
3922 (bit_value_binop): Likewise.
3923 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
3924 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
3925 (extract_range_from_binary_expr_1): Likewise.
3926 (masked_increment): Likewise.
3927 (simplify_bit_ops_using_ranges): Likewise.
3929 2017-10-09 Martin Jambor <mjambor@suse.cz>
3932 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
3933 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
3934 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
3936 (hsa_fixup_mov_insn_type): New function.
3937 (hsa_op_with_type::get_in_type): Use it.
3938 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
3939 immediates in an assert.
3940 (hsa_op_with_type::extend_int_to_32bit): New method.
3941 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
3942 types. Convert to dest type if necessary.
3943 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
3944 (reg_for_gimple_ssa): Pass false as min32int to
3945 hsa_type_for_scalar_tree_type.
3946 (gen_hsa_addr): Fixup type when creating addresable temporary.
3947 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
3948 (gen_hsa_unary_operation): Extend operands and convert to dest type if
3949 necessary. Call hsa_fixup_mov_insn_type.
3950 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
3951 extend operands and convert to dest type if necessary.
3952 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
3953 to dest type if necessary.
3954 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
3955 if conversion nt necessary and size matches.
3956 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
3957 to dest type if necessary.
3958 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
3959 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
3961 (gen_hsa_clrsb): Likewise.
3962 (gen_hsa_ffs): Likewise.
3963 (gen_hsa_divmod): Extend operands and convert to dest type if
3965 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
3967 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3969 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
3970 Remove empty default arguments. Use a brace block as output
3972 (conditional return): Ditto.
3974 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
3975 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
3977 (group_ending_nop): Ditto.
3978 (doloop_end): Ditto.
3979 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
3980 (splitters for those): Ditto.
3982 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3984 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
3985 a conditional jump (and the compare for it) so that pc_rtx is the
3987 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
3988 for the deleted and renamed ctr<mode>_internal[234] patterns.
3989 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
3990 Delete second conditional return pattern.
3991 (ctr<mode>_internal2): Delete this second bdnz pattern.
3992 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
3993 (ctr<mode>_internal4): Delete this second bdz pattern.
3995 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3997 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
3998 (always_initialized_rtx_for_ssa_name_p): New predicate.
3999 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
4000 (finish_out_of_ssa): Free new field of SA.
4001 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
4002 * tree-ssa-coalesce.c: Include tree-ssa.h.
4003 (get_parm_default_def_partitions): Remove extern keyword.
4004 (get_undefined_value_partitions): New function.
4005 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
4006 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
4008 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
4010 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
4012 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
4014 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4016 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
4017 for targets that preffer 128bit.
4019 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
4021 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
4023 2017-10-08 Olivier Hainque <hainque@adacore.com>
4025 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
4026 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
4028 (thumb_set_return_address): Likewise.
4030 2017-10-08 Olivier Hainque <hainque@adacore.com>
4032 * common/config/arm/arm-common.c (arm_except_unwind_info):
4033 Handle DWARF2_UNWIND_INFO.
4035 2017-10-07 Michael Collison <michael.collison@arm.com>
4037 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
4040 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
4042 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
4043 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
4044 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
4045 defined, force the creation of a new block for a dispatch label.
4047 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
4049 * invoke.texi (Wsuggest-attribute=cold): Document.
4050 * common.opt (Wsuggest-attribute=cold): New
4051 * ipa-pure-const.c (warn_function_cold): New function.
4052 * predict.c (compute_function_frequency): Use it.
4053 * predict.h (warn_function_cold): Declare.
4055 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
4057 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
4060 2017-10-06 Martin Liska <mliska@suse.cz>
4062 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
4063 keyword for member functions.
4064 (struct sanopt_tree_couple): New struct.
4065 (struct sanopt_tree_couple_hash): New function.
4066 (struct sanopt_ctx): Add new hash_map.
4067 (has_dominating_ubsan_ptr_check): New function.
4068 (record_ubsan_ptr_check_stmt): Likewise.
4069 (maybe_optimize_ubsan_ptr_ifn): Likewise.
4070 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
4071 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
4073 2017-10-06 Sudakshina Das <sudi.das@arm.com>
4076 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
4077 aarch64_simd_valid_immediate on CONST_VECTORs.
4078 (aarch64_reg_or_bic_imm): Likewise.
4080 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
4082 PR rtl-optimization/82396
4083 * haifa-sched.c (ready_sort_real): Disable qsort checking.
4085 2017-10-06 Sebastian Pop <sebpop@gmail.com>
4087 * graphite-dependences.c (scop_get_reads): Move code to...
4088 (scop_get_must_writes): Move code to...
4089 (scop_get_may_writes): Move code to...
4090 (scop_get_reads_and_writes): ... here.
4091 (scop_get_dependences): Call scop_get_reads_and_writes.
4093 2017-10-06 Jakub Jelinek <jakub@redhat.com>
4095 PR tree-optimization/82434
4096 * fold-const.h (can_native_encode_type_p,
4097 can_native_encode_string_p): Remove.
4098 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
4099 don't encode anything, just return what would be otherwise returned.
4100 (native_encode_fixed, native_encode_complex, native_encode_vector):
4102 (native_encode_string): Likewise. Inline by hand
4103 can_native_encode_string_p.
4104 (can_native_encode_type_p): Remove.
4105 (can_native_encode_string_p): Remove.
4106 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
4107 STRING_CSTs using can_native_encode_string_p, test all
4108 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
4109 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
4110 argument from native_encode_expr.
4111 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
4112 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
4115 2017-10-06 Richard Biener <rguenther@suse.de>
4117 PR tree-optimization/82397
4118 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
4119 operand_equal_p but rely on data_ref_compare_tree for detecting
4121 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
4122 to match up with dr_group_sort_cmp.
4124 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4127 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
4129 * config/s390/s390-builtin-types.def: Regenerate.
4131 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4134 * config/s390/s390-builtin-types.def: Regenerate.
4135 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
4136 Change flag from B_VXE to B_VX.
4137 (s390_vec_min_dbl): Remove B_VXE flag.
4139 2017-10-06 Richard Biener <rguenther@suse.de>
4141 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
4142 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
4143 translate_isl_ast_to_gimple::is_valid_rename,
4144 translate_isl_ast_to_gimple::get_rename,
4145 translate_isl_ast_to_gimple::get_def_bb_for_const,
4146 translate_isl_ast_to_gimple::get_new_name,
4147 translate_isl_ast_to_gimple::collect_all_ssa_names,
4148 translate_isl_ast_to_gimple::copy_loop_phi_args,
4149 translate_isl_ast_to_gimple::collect_all_ssa_names,
4150 translate_isl_ast_to_gimple::copy_loop_phi_args,
4151 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
4152 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
4153 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
4154 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
4155 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
4156 translate_isl_ast_to_gimple::copy_cond_phi_args,
4157 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
4158 translate_isl_ast_to_gimple::edge_for_new_close_phis,
4159 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
4160 translate_isl_ast_to_gimple::rename_uses,
4161 translate_isl_ast_to_gimple::rename_all_uses): Remove.
4162 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
4163 (set_rename_for_each_def): Likewise.
4164 (graphite_copy_stmts_from_block): Handle debug stmt resetting
4165 here. Handle rewriting SCEV analyzable uses here.
4166 (copy_bb_and_scalar_dependences): Generate code for PHI
4168 (graphite_regenerate_ast_isl): Adjust.
4169 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
4170 (add_write, add_read): New functions.
4171 (build_cross_bb_scalars_def): Use it and simplify.
4172 (build_cross_bb_scalars_use): Likewise.
4173 (graphite_find_cross_bb_scalar_vars): Inline into...
4174 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
4175 simulating out-of-SSA. Compute liveout and add dependencies.
4176 (build_scops): Force an empty entry block.
4177 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
4179 (sese_build_liveouts): Declare.
4180 (sese_trivially_empty_bb_p): Likewise.
4181 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
4182 compute liveout and debug_liveout.
4183 (sese_bad_liveouts_use): Remove.
4184 (sese_reset_debug_liveouts_bb): Likewise.
4185 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
4186 (sese_build_liveouts): Build liveout and debug_liveout and store
4188 (new_sese_info): Adjust.
4189 (free_sese_info): Likewise.
4190 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
4191 do not build liveout here.
4192 (move_sese_in_condition): Adjust region entry.
4193 (scev_analyzable_p): Match up with chrec_apply requirements.
4194 (sese_trivially_empty_bb_p): New.
4195 * tree-into-ssa.c (get_reaching_def): Properly support generating
4196 default-defs for incremental rewrite of anonymous names.
4198 2017-10-06 Richard Biener <rguenther@suse.de>
4200 * graphite-sese-to-poly.c (extract_affine): For casts increasing
4201 precision do not perform modulo reduction.
4203 2017-10-06 Richard Biener <rguenther@suse.de>
4205 PR tree-optimization/82436
4206 * tree-vect-slp.c (vect_supported_load_permutation_p): More
4207 conservatively choose the vectorization factor when checking
4208 whether we can perform the required load permutation.
4209 (vect_transform_slp_perm_load): Assert when we may not fail.
4211 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4213 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
4214 message for incompatible -msdata=* and -mcall-* options.
4216 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4218 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
4219 rate for post-reload scheduling.
4221 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4223 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
4225 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4227 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
4228 to improve monte carlo in scimark.
4230 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4232 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
4233 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
4234 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
4235 pentium4_cost, nocona_cost): Set reassociation width to 1.
4236 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
4237 width to 2 for fp operations and 1 otherwise.
4238 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
4240 (atom_cost): Set reassociation width to 2.
4241 (slm_cost, generic_cost): Set fp reassociation width
4242 to 2 and 1 otherwise.
4243 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
4244 (core_cost): Set fp reassociation width to 4 and vector to 2.
4245 (ix86_reassociation_width): Rewrite using cost table; special case
4246 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
4247 and TARGET_AVX128_OPTIMAL.
4248 * config/i386/i386.h (processor_costs): Add
4249 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
4250 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
4251 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
4252 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
4253 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
4254 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
4256 2017-10-05 Nathan Sidwell <nathan@acm.org>
4258 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
4260 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4262 * config/arm/arm.c (arm_test_fpu_data): New.
4263 (arm_run_selftests): Call arm_test_fpu_data.
4265 2017-10-04 Nathan Sidwell <nathan@acm.org>
4267 * toplev.c (toplev::main): Remove excess parens on pretty_printer
4269 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
4271 2017-10-04 Sudakshina Das <sudi.das@arm.com>
4273 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
4274 check type for aarch64_simd_valid_immediate.
4275 (aarch64_output_simd_mov_immediate): Update prototype.
4276 (aarch64_simd_valid_immediate): Update prototype.
4277 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
4278 support for ORR-immediate.
4279 (and<mode>3): modified pattern to add support for BIC-immediate.
4280 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
4281 now checks for valid immediate for BIC and ORR based on new enum
4283 (aarch64_output_simd_mov_immediate): Function now used to output
4284 BIC/ORR imm as well based on new enum argument.
4285 * config/aarch64/constraints.md (Do): New vector immediate constraint.
4287 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
4288 (aarch64_reg_or_bic_imm): Likewise.
4290 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4292 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
4293 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
4295 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4299 PR rtl-optimization/82396
4300 * haifa-sched.c (autopref_multipass_init): Simplify
4302 (autopref_rank_data): Simplify sort order.
4303 * sched-int.h (autopref_multipass_data_): Remove
4304 multi_mem_insn_p, min_offset and max_offset.
4306 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4308 * doc/sourcebuild.texi: Document vect_peeling_profitable.
4310 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4312 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
4315 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4317 * doc/sourcebuild.texi: Document vect_long_mult.
4319 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
4321 PR tree-optimization/82413
4322 * fold-const.c (build_range_check): Use widest_int when comparing
4323 the maximum ETYPE value with HIGH.
4325 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4327 PR rtl-optimization/82396
4328 * haifa-sched.c (autopref_multipass_init): Simplify
4330 (autopref_rank_data): Simplify sort order.
4331 * sched-int.h (autopref_multipass_data_): Remove
4332 multi_mem_insn_p, min_offset and max_offset.
4334 2017-10-04 Jakub Jelinek <jakub@redhat.com>
4336 PR tree-optimization/82381
4337 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
4338 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
4341 PR tree-optimization/82374
4342 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
4343 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
4344 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
4345 current_function_decl to the new decl.
4347 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
4349 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
4350 helper macro for IEEE float128 hardware built-in functions.
4351 (SQRTF128_ODD): Add built-in functions with the round-to-odd
4353 (TRUNCF128_ODD): Likewise.
4354 (ADDF128_ODD): Likewise.
4355 (SUBF128_ODD): Likewise.
4356 (MULF128_ODD): Likewise.
4357 (DIVF128_ODD): Likewise.
4358 (FMAF128_ODD): Likewise.
4359 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
4360 UNSPEC_TRUNC_ROUND_TO_ODD.
4361 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
4362 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
4363 floating point round to odd instructions.
4364 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
4365 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
4366 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
4367 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
4368 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
4369 (trunc<mode>sf2_hw): Change the truncate with round to odd
4370 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
4371 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
4372 to odd hardware instructions.
4373 (sub<mode>3_odd): Likewise.
4374 (mul<mode>3_odd): Likewise.
4375 (div<mode>3_odd): Likewise.
4376 (sqrt<mode>2_odd): Likewise.
4377 (fma<mode>4_odd): Likewise.
4378 (fms<mode>4_odd): Likewise.
4379 (nfma<mode>4_odd): Likewise.
4380 (nfms<mode>4_odd): Likewise.
4381 (trunc<mode>df2_odd): Change the truncate with round to odd
4382 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
4384 * doc/extend.texi (PowerPC built-in functions): Update documentation
4385 for existing IEEE float128-bit built-in functions. Add built-in
4386 functions that generate the IEEE 128-bit floating point round to
4389 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
4391 PR rtl-optimization/77729
4392 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
4393 to (X&(C1&~C2))|C2 transformations.
4395 2017-10-03 Martin Jambor <mjambor@suse.cz>
4397 PR tree-optimization/82363
4398 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
4399 mismatch, mark lacc written regardless of racc.
4401 2017-10-03 Jakub Jelinek <jakub@redhat.com>
4403 PR tree-optimization/82381
4404 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
4405 stmt_to_insert nor wheather SSA_NAMEs are default defs.
4406 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
4407 fallthrough into reassoc_stmt_dominates_stmt_p.
4410 * combine.c (combine_instructions): Don't combine in unreachable
4413 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4416 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
4417 function to not use the have_cpu variable. Do not set cpu_index,
4418 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
4420 (rs6000_valid_attribute_p): Remove duplicate initializations of
4421 old_optimize and func_optimize.
4422 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
4423 (rs6000_activate_target_options): Make global.
4424 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
4427 2017-10-02 Jakub Jelinek <jakub@redhat.com>
4429 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
4430 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
4431 Set *poffset to 0 and *psize and *pmax_size to -1 if
4432 *poffset + *psize overflows in HOST_WIDE_INT.
4434 PR tree-optimization/82387
4435 PR tree-optimization/82388
4436 PR tree-optimization/82389
4437 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
4438 instead of live_bytes non-NULL.
4440 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
4443 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
4446 2017-10-02 Richard Biener <rguenther@suse.de>
4448 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
4449 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
4450 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
4452 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4454 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
4455 requested precision matches the type's.
4456 * calls.c (alloc_max_size): Calculate the new candidate size as
4457 a widest_int and use wi::to_widest when comparing it with the
4458 current candidate size.
4459 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
4460 zero rather than integer_zero_node.
4461 * match.pd: Check for a no-op conversion before using wi::add
4462 rather than after. Use tree_to_uhwi when summing small shift
4463 counts into an unsigned int.
4465 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4466 Alan Hayward <alan.hayward@arm.com>
4467 David Sherwood <david.sherwood@arm.com>
4470 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
4471 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
4472 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
4473 POINTER_AND_FP_REGS.
4475 2017-10-02 Richard Biener <rguenther@suse.de>
4477 PR tree-optimization/82355
4478 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4479 a mapping for the enclosing loop but avoid generating one for
4481 (copy_bb_and_scalar_dependences): Remove premature codegen
4482 error on PHIs in blocks duplicated into multiple places.
4483 * graphite-scop-detection.c
4484 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4485 in the region use it as loop and nest to analyze the DR in.
4486 (try_generate_gimple_bb): Likewise.
4487 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4488 (add_loop_constraints): For blocks in a loop not in the region
4489 create a dimension with a single iteration.
4490 * sese.h (gbb_loop_at_index): Remove assert.
4492 2017-10-01 Kevin Buettner <kevinb@redhat.com>
4494 * omp-expand.c (adjust_context_scope): New function.
4495 (expand_parallel_call): Call adjust_context_scope.
4497 2017-10-01 Jeff Law <law@redhat.com>
4499 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
4500 dom_opt_dom_walker class with direct access to private members.
4501 Add comments. Call test_for_singularity.
4502 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
4503 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
4504 m_dummy_cond anymore.
4505 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
4507 (pass_dominator:execute): Build the dummy_cond here and pass it
4508 to the dom_opt_dom_walker ctor.
4509 (test_for_singularity): New function.
4511 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
4512 Maya Rashish <coypu@sdf.org>
4514 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
4515 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
4516 (alpha*-*-netbsd*) Use nbsd_tm_file.
4517 (arm*-*-netbsdelf*) Likewise.
4518 (i[34567]86-*-netbsdelf*) Likewise.
4519 (x86_64-*-netbsd*) Likewise.
4520 (mips*-*-netbsd*) Likewise.
4521 (powerpc-*-netbsd*) Likewise.
4522 (sh*-*-netbsd*) Likewise.
4523 (sparc-*-netbsdelf*) Likewise.
4524 (sparc64-*-netbsd*) Likewise.
4525 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
4527 (vax-*-netbsdelf*) Likewise.
4528 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
4529 (UINT_FAST8_TYPE) Likewise.
4530 (INT_FAST16_TYPE) Check CHAR_FAST16.
4531 (UINT_FAST16_TYPE) Likewise.
4533 2017-09-30 Jakub Jelinek <jakub@redhat.com>
4536 * config/i386/i386.md
4537 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
4538 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
4539 *divmodsi4_zext_2): New define_insn_and_split.
4540 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
4541 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
4542 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
4543 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
4544 New define_insn_and_split.
4545 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
4546 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
4547 operands[1] having DImode when mode is SImode.
4549 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
4550 always SImode for DIV and MOD in REG_EQUAL notes.
4552 2017-09-29 Yury Gribov <tetra2005@gmail.com>
4555 * match.pd: Fix handling of NaNs in pattern.
4557 2017-09-29 Jeff Law <law@redhat.com>
4559 * sbitmap.c (bitmap_bit_in_range_p): New function.
4560 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
4561 * tree-ssa-dse.c (live_bytes_read): New function.
4562 (dse_classify_store): Ignore reads of dead bytes.
4564 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
4565 typos and whitespace errors.
4566 * config/i386/predicates.md (address_no_seg_operand): Likewise.
4567 * config/s390/s390.c (s390_emit_prologue): Likewise.
4569 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4572 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
4573 with a symbol for LRA.
4575 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4577 PR rtl-optimization/82338
4578 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
4580 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
4582 * genmodes.c (calc_wider_mode): Suppress qsort macro.
4583 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
4584 (qsort_chk): Declare.
4585 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
4586 (qsort_chk): New function.
4588 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4590 PR tree-optimization/82337
4591 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
4592 phi definition if the PHI result appears in an abnormal PHI.
4593 (find_basis_for_base_expr): Don't record a basis if the LHS of the
4594 basis appears in an abnormal PHI.
4596 2017-09-29 Richard Biener <rguenther@suse.de>
4598 * graphite-isl-ast-to-gimple.c
4599 (translate_isl_ast_to_gimple::set_codegen_error): New function.
4600 (binary_op_to_tree): Use it.
4601 (get_rename_from_scev): Likewise.
4602 (copy_loop_phi_nodes): Likewise.
4603 (copy_bb_and_scalar_dependences): Likewise.
4604 (translate_pending_phi_nodes): Likewise.
4606 2017-09-29 Jakub Jelinek <jakub@redhat.com>
4609 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
4610 for movabsq $(i32 << shift), r64.
4612 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
4614 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
4615 index when encoding %esp as %rsp to avoid 0x67 prefix.
4617 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
4619 * config/i386/i386.md (*movsf_internal, *movdf_internal):
4620 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
4622 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
4624 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
4625 Extensions with more than 16 double VFP registers.
4626 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
4627 to_clear_mask and all code related to it. Replace the remaining
4628 entry by a sbitmap and adapt code accordingly.
4630 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
4632 * brig-builtins.def: Change pure attributes to const.
4634 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4636 * config.gcc (default_gnu_indirect_function): Default to yes for
4637 sparc*-*-linux* with glibc.
4639 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4641 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
4642 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
4643 when creating .init_array and .fini_array sections with priority
4646 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
4649 * config/aarch64/aarch64.c
4650 (aarch64_builtin_support_vector_misalignment): Always return false
4651 when misalignment is unknown.
4653 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
4655 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
4656 this function to return false if the definition used by the swap
4657 instruction is artificial, or if the memory address from which the
4658 constant value is loaded is not represented by a base address held
4659 in a register or if the base address register is a frame or stack
4660 pointer. Additionally, return false if the base address of the
4661 loaded constant is a SYMBOL_REF but is not considered to be a
4663 (replace_swapped_load_constant): New function.
4664 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
4665 loaded constant vector with a load of a swapped constant vector.
4667 2017-09-27 Carl Love <cel@us.ibm.com>
4669 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
4670 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
4671 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
4674 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
4676 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
4677 first, always call autopref_rank_data otherwise.
4679 2017-09-27 Richard Biener <rguenther@suse.de>
4681 * graphite-scop-detection.c (find_scop_parameters): Move
4682 loop bound handling ...
4683 (gather_bbs::before_dom_children): ... here, avoiding the need
4684 to build scop_info->loop_nest.
4685 (record_loop_in_sese): Remove.
4686 * sese.h (sese_info_t::loop_nest): Remove.
4687 * sese.c (new_sese_info): Do not allocate loop_nest.
4688 (free_sese_info): Do not free loop_nest.
4690 2017-09-27 Jakub Jelinek <jakub@redhat.com>
4693 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
4694 lhs from calls if the lhs has addressable type.
4696 2017-09-27 Richard Biener <rguenther@suse.de>
4698 * graphite.h (scop::max_alias_set): New member.
4699 * graphite-scop-detection.c: Remove references to non-existing
4700 --param in comments.
4701 (build_alias_sets): Record the maximum alias set used for drs.
4702 (build_scops): Support zero as unlimited for
4703 --param graphite-max-arrays-per-scop.
4704 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
4706 (build_poly_sr_1): ... here. Compute alias set based on the
4707 maximum alias set used for drs rather than
4708 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
4710 2017-09-27 Richard Biener <rguenther@suse.de>
4712 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
4713 --param loop-block-tile-size=0 to disable tiling.
4715 2017-09-27 Richard Biener <rguenther@suse.de>
4717 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
4718 (graphite-max-nb-scop-params): Document special value zero.
4719 * domwalk.h (dom_walker::STOP): New symbolical constant.
4720 (dom_walker::dom_walker): Add optional parameter for bb to
4722 (dom_walker::~dom_walker): Declare.
4723 (dom_walker::before_dom_children): Document STOP return value.
4724 (dom_walker::m_user_bb_to_rpo): New member.
4725 (dom_walker::m_bb_to_rpo): Likewise.
4726 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
4727 mapping here if not provided by the user.
4728 (dom_walker::~dom_walker): Free bb to RPO mapping if not
4729 provided by the user.
4730 (dom_walker::STOP): Define.
4731 (dom_walker::walk): Do not compute bb to RPO mapping here.
4732 Support STOP return value from before_dom_children to stop
4734 * graphite-optimize-isl.c (optimize_isl): If the schedule
4735 is the same still generate code if -fgraphite-identity
4736 or -floop-parallelize-all are given.
4737 * graphite-scop-detection.c: Include cfganal.h.
4738 (gather_bbs::gather_bbs): Get and pass through bb to RPO
4740 (gather_bbs::before_dom_children): Return STOP for BBs
4742 (build_scops): Compute bb to RPO mapping and pass it to
4743 the domwalk. Treat --param graphite-max-nb-scop-params=0
4744 as not limiting the number of params.
4745 * graphite.c (graphite_initialize): Remove limit on the
4746 number of basic-blocks in a function.
4747 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
4748 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
4749 default value of 10.
4751 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4753 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
4754 Adjust code to eliminate needing to do the shift right 32-bits
4755 operation after XSCVDPSPN.
4757 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4759 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
4760 ((X / Y) != 0 -> X >= Y): Likewise.
4762 2017-09-26 Carl Love <cel@us.ibm.com>
4764 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
4765 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
4766 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
4767 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
4768 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
4769 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
4770 definitions and overloading.
4771 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
4772 statement for P9V_BUILTIN_XST_LEN_R.
4773 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
4774 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
4775 define_expand and define_insn for the instructions and builtins.
4776 * doc/extend.texi: Update the built-in documentation file for the new
4778 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
4779 define_insn for the instructions
4781 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
4784 * gcc/config/netbsd-protos.h: New file.
4785 * gcc/config/netbsd.c: New file.
4786 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
4787 * gcc/config/t-netbsd: New file.
4788 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
4789 (tmake_file) Add t-netbsd.
4790 (extra_objs) Add netbsd.o.
4792 2017-09-26 Janus Weil <janus@gcc.gnu.org>
4796 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
4798 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4800 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
4801 sign extension from a vector register to a GPR by doing a 32-bit
4802 direct move and then an EXTSW.
4803 (extendsi<mode>2 splitter): Likewise.
4804 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
4805 right or vector extract after doing XSCVDPSPN. Use
4806 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
4808 (movdi_from_sf_zero_ext): Likewise.
4809 (reload_gpr_from_vsxsf): Likewise.
4810 (p8_mfvsrd_4_disf): Delete, no longer used.
4811 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
4812 then needing to move the SFmode to a GPR to use the XSCVDPSP
4813 instruction instead of FRSP and XSCVDPSPN.
4814 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
4815 it is adjacent to the other XSCVSPDP insns.
4816 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
4817 SFmode to be in traditional Altivec registers.
4818 (vsx_xscvdpspn): Eliminate useless alternative constraint.
4819 (vsx_xscvspdpn): Likewise.
4820 (vsx_xscvspdpn_scalar): Likewise.
4822 2017-09-26 Martin Jambor <mjambor@suse.cz>
4824 * tree-sra.c (compare_access_positions): Put integral types first,
4825 stabilize sorting of integral types, remove conditions putting
4826 non-full-precision integers last.
4827 (sort_and_splice_var_accesses): Disable scalarization if a
4828 non-integert would be represented by a non-full-precision integer.
4830 2017-09-26 Joseph Myers <joseph@codesourcery.com>
4832 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
4833 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
4834 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
4835 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
4836 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
4837 conditionals inside the function instead of around it. Call
4838 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
4839 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
4841 2017-09-26 Richard Biener <rguenther@suse.de>
4843 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
4845 (scop_detection::build_scop_breadth): ... this. Removed.
4846 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
4847 (scop_detection::harmful_stmt_in_bb): Likewise.
4848 (scop_detection::graphite_can_represent_stmt): Likewise.
4849 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
4850 (scop_detection::can_represent_loop): Remove recursion, fold in ...
4851 (scop_detection::can_represent_loop_1): ... this. Removed.
4852 (scop_detection::harmful_loop_in_region): Simplify after inlining
4853 the above and remove more quadraticness.
4854 (build_scops): Adjust.
4855 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
4858 2017-09-26 Jakub Jelinek <jakub@redhat.com>
4861 * config/i386/i386.c (ix86_print_operand_address_as): Only test
4862 REGNO (base) == SP_REG if base is a REG.
4865 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
4866 if it is different SSA_NAME.
4867 (optimize_range_tests_cmp_bitwise): New function.
4868 (optimize_range_tests): Call it.
4870 2017-09-26 Richard Biener <rguenther@suse.de>
4872 PR tree-optimization/82321
4873 * graphite.c (canonicalize_loop_closed_ssa): Properly check
4874 for the def being inside the loop.
4876 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4878 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
4880 * config/s390/s390-builtins.def: Fix constraint on op4.
4882 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4884 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
4885 independent expanders.
4886 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
4887 ("vec_ordered", "vec_unordered"): New expanders.
4889 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4891 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
4894 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4896 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
4897 vec_unpacks_lo_v16qi.
4898 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
4900 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4902 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
4903 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
4904 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
4906 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4908 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
4910 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
4912 ("*vec_slb<mode>"): New insn pattern.
4913 ("vec_shr_<mode>"): New expander.
4914 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
4915 and force the shift count operand to V16QImode.
4916 ("vec_srb<mode>"): Set shift count mode to V16QI.
4918 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4920 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
4921 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
4922 ("vec_widen_smult_hi_<mode>"): New expander definitions.
4924 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
4927 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
4929 2017-09-26 Richard Biener <rguenther@suse.de>
4931 PR tree-optimization/82320
4932 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
4935 2017-09-25 Jeff Law <law@redhat.com>
4937 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
4938 prototype for new argument.
4939 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
4940 mostly extracted from rs6000_emit_allocate_stack.
4941 (rs6000_emit_probe_stack_range_stack_clash): New function.
4942 (rs6000_emit_allocate_stack): Call
4943 rs6000_emit_probe_stack_range_stack_clash as needed.
4944 (rs6000_emit_probe_stack_range): Add additional argument
4945 to call to gen_probe_stack_range{si,di}.
4946 (output_probe_stack_range): New.
4947 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
4948 (output_probe_stack_range_stack_clash): New.
4949 (rs6000_emit_prologue): Emit notes into dump file as requested.
4950 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
4951 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
4952 Add additional operand and pass it to output_probe_stack_range.
4954 2017-09-25 Bin Cheng <bin.cheng@arm.com>
4956 PR tree-optimization/82163
4957 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
4958 (checking_verify_loop_closed_ssa): New parameter.
4959 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
4960 (check_loop_closed_ssa_stmt): Delete.
4961 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
4962 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
4963 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
4966 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
4968 * brig-builtins.def: Treat HSAIL barrier builtins as
4969 setjmp/longjump style functions.
4971 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4973 * target.def (constant_alignment): New hook.
4974 * defaults.h (CONSTANT_ALIGNMENT): Delete.
4975 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
4976 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
4977 * doc/tm.texi: Regenerate.
4978 * targhooks.h (default_constant_alignment): Declare.
4979 (constant_alignment_word_strings): Likewise.
4980 * targhooks.c (default_constant_alignment): New function.
4981 (constant_alignment_word_strings): Likewise.
4982 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
4983 instead of CONSTANT_ALIGNMENT.
4984 * varasm.c (align_variable, get_variable_align, build_constant_desc)
4985 (force_const_mem): Likewise.
4986 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
4987 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
4988 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
4989 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4990 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
4992 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
4993 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4994 constant_alignment_word_strings.
4995 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
4996 (CONSTANT_ALIGNMENT): Likewise.
4997 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4998 (arm_constant_alignment): New function.
4999 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
5000 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5001 constant_alignment_word_strings.
5002 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
5003 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5004 constant_alignment_word_strings.
5005 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
5006 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5007 (cris_constant_alignment): New function.
5008 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
5009 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5010 (epiphany_constant_alignment): New function.
5011 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
5012 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5013 constant_alignment_word_strings.
5014 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
5015 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5016 constant_alignment_word_strings.
5017 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
5018 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5019 constant_alignment_word_strings.
5020 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
5021 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
5022 * config/i386/i386.c (ix86_constant_alignment): Make static.
5023 Use the same interface as the target hook.
5024 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5025 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
5026 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5027 constant_alignment_word_strings.
5028 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
5029 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
5030 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5031 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
5032 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5033 constant_alignment_word_strings.
5034 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
5035 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5036 constant_alignment_word_strings.
5037 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
5038 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5039 constant_alignment_word_strings.
5040 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
5041 * config/microblaze/microblaze.c (microblaze_constant_alignment):
5043 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5044 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
5045 * config/mips/mips.c (mips_constant_alignment): New function.
5046 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5047 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
5048 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
5049 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5050 (mmix_constant_alignment): Make static. Use the same interface
5052 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
5053 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5054 constant_alignment_word_strings.
5055 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
5056 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5057 constant_alignment_word_strings.
5058 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
5059 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5060 constant_alignment_word_strings.
5061 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
5062 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5063 (rs6000_constant_alignment): New function.
5064 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
5065 * config/riscv/riscv.c (riscv_constant_alignment): New function.
5066 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5067 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
5068 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5069 (rs6000_constant_alignment): New function.
5070 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
5071 * config/s390/s390.c (s390_constant_alignment): New function.
5072 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5073 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
5074 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5075 constant_alignment_word_strings.
5076 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
5077 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5078 (sparc_constant_alignment): New function.
5079 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
5080 * config/spu/spu.c (spu_constant_alignment): New function.
5081 (TARGET_CONSTANT_ALIGNMENT): Redefine.
5082 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
5083 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5084 constant_alignment_word_strings.
5085 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
5086 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5087 constant_alignment_word_strings.
5088 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
5089 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
5090 constant_alignment_word_strings.
5091 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
5092 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5093 (visium_constant_alignment): New function.
5094 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
5095 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5096 (xtensa_constant_alignment): New function.
5097 * system.h (CONSTANT_ALIGNMENT): Poison.
5099 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
5101 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5102 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
5103 (rs6000_builtin_valid_without_lhs): New helper function.
5104 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5105 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
5107 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5109 * target.h (vec_perm_indices): Use unsigned short rather than
5111 (auto_vec_perm_indices): Likewise.
5112 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
5113 Use unsigned int rather than unsigned char.
5114 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
5116 2017-09-25 Richard Biener <rguenther@suse.de>
5118 * cfgloop.h (sort_sibling_loops): Declare.
5119 * cfgloop.c (sort_sibling_loops_cmp): New helper.
5120 (sort_sibling_loops): New function sorting the sibling loop list
5122 * graphite.c (graphite_transform_loops): Sort sibling loops.
5124 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
5126 * target.def (vec_perm_const_ok): Change sel parameter to
5128 * optabs-query.c (can_vec_perm_p): Update accordingly.
5129 * doc/tm.texi: Regenerate.
5130 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
5131 auto_vec_perm_indices and remove separate nelt field.
5132 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
5133 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
5134 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
5135 (aarch64_expand_vec_perm_const): Update accordingly.
5136 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
5137 to vec_perm_indices.
5138 * config/arm/arm.c (expand_vec_perm_d): Change perm to
5139 auto_vec_perm_indices and remove separate nelt field.
5140 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
5141 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
5142 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
5144 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
5145 to vec_perm_indices.
5146 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
5147 sel to vec_perm_indices.
5148 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
5149 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
5150 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
5152 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
5155 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
5158 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
5159 on the FUNCTION_DECL function context if it has a DIE that is a
5162 2017-09-25 Richard Biener <rguenther@suse.de>
5164 PR tree-optimization/82285
5165 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
5168 2017-09-25 Tom de Vries <tom@codesourcery.com>
5172 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
5175 2017-09-25 Richard Biener <rguenther@suse.de>
5177 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
5178 ISL errors other than isl_error_quota happen. Dump if the
5179 schedule is the same.
5180 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
5181 errors instead of aborting inside ISL.
5183 2017-09-25 Iain Sandoe <iain@codesourcery.com>
5186 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
5187 of libgcc_eh for m64.
5188 * config/i386/darwin64.h: Likewise.
5190 2017-09-25 Richard Biener <rguenther@suse.de>
5193 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
5194 attribute for incomplete types nor twice for complete ones.
5196 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
5199 * config/i386/i386.c (ix86_print_operand_address_as): Encode
5200 %esp as %rsp to avoid 0x67 prefix if there is no index or base
5203 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
5206 * config/i386/i386.opt (mprefer-avx256): Use
5207 ix86_target_flags variable.
5208 * config/i386/i386.c (ix86_target_string): Move
5209 -mprefer-avx256 to flag2_opts.
5211 2017-09-22 Jakub Jelinek <jakub@redhat.com>
5214 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
5215 and x != -1 | y != -1 into (x & y) != -1.
5217 2017-09-22 Steve Ellcey <sellcey@cavium.com>
5219 * config.gcc: Add new case statement to set
5220 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
5221 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
5222 s390x-*-linux* case statements. Added aarch64 to the list of
5223 supported architectures.
5225 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5227 PR tree-optimization/82289
5228 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
5229 STMT_VINFO_RELEVANT_P.
5231 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5232 Alan Hayward <alan.hayward@arm.com>
5233 David Sherwood <david.sherwood@arm.com>
5235 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
5236 for VR_RANGE only; don't allow VR_ANTI_RANGE.
5237 (extract_range_from_binary_expr_1): Don't call
5238 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
5240 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5241 Alan Hayward <alan.hayward@arm.com>
5242 David Sherwood <david.sherwood@arm.com>
5244 * target.def (preferred_vector_alignment): New hook.
5245 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
5247 * doc/tm.texi: Regenerate.
5248 * targhooks.h (default_preferred_vector_alignment): Declare.
5249 * targhooks.c (default_preferred_vector_alignment): New function.
5250 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
5252 (DR_TARGET_ALIGNMENT): New macro.
5253 (aligned_access_p): Update commentary.
5254 (vect_known_alignment_in_bytes): New function.
5255 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
5257 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
5258 Calculate the misalignment based on the target alignment rather than
5260 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
5261 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
5262 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
5263 the target alignment, rather than masking the element misalignment
5264 with the number of elements in a vector. Also use the target
5265 alignment when calculating the maximum number of peels.
5266 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
5267 instead of TYPE_ALIGN_UNIT.
5268 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
5269 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
5270 (vect_create_addr_base_for_vector_ref): Update call accordingly.
5271 (vect_create_data_ref_ptr): Likewise.
5272 (vect_setup_realignment): Realign by ANDing with
5273 -DR_TARGET_MISALIGNMENT.
5274 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
5275 the number of peels based on DR_TARGET_ALIGNMENT.
5276 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
5277 with the guaranteed alignment boundary when deciding whether
5279 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
5280 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
5281 (ensure_base_align): Remove stmt_info parameter. Get the
5282 target base alignment from DR_TARGET_ALIGNMENT.
5283 (vectorizable_store): Update call accordingly. Interpret
5284 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
5286 (vectorizable_load): Likewise.
5288 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5289 Alan Hayward <alan.hayward@arm.com>
5290 David Sherwood <david.sherwood@arm.com>
5292 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
5293 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
5294 (vect_enhance_data_refs_alignment): Likewise.
5296 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5298 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5299 error. Only quit immediately if parsing is complete.
5300 (BEGIN): Initialize fatal_err and parse_done.
5301 (begin fpu, end fpu): Check number of arguments.
5302 (begin arch, end arch): Likewise.
5303 (begin cpu, end cpu): Likewise.
5304 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5305 (optalias): Likewise.
5307 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5309 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5310 * config/arm/arm-isa.h: Delete. Move definitions to ...
5311 * arm-cpus.in: ... here. Use new feature and fgroup values.
5312 * config/arm/arm.c (arm_option_override): Use lower case for feature
5314 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5315 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5316 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5318 (print_isa_bits_for): New function.
5319 (gen_isa): New function.
5320 (gen_comm_data): Use print_isa_bits_for.
5321 (define feature): New keyword.
5322 (define fgroup): New keyword.
5323 * config/arm/t-arm (TM_H): Remove.
5324 (GTM_H): Add arm-isa.h.
5325 (arm-isa.h): Add rule to generate file.
5326 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5327 case for feature bit names.
5329 2017-09-22 Richard Biener <rguenther@suse.de>
5331 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
5333 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
5334 print of no dependency loops ...
5335 * graphite.c (graphite_transform_loops): ... here.
5336 (canonicalize_loop_closed_ssa_form): Work from inner to outer
5338 (same_close_phi_node, remove_duplicate_close_phi,
5339 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
5340 (canonicalize_loop_closed_ssa): ... here and simplify.
5341 * graphite-optimize-isl.c: Include tree-vectorizer.h.
5342 (optimize_isl): Use dump_printf_loc to tell when we stopped
5343 optimizing because of an ISL timeout.
5345 2017-09-22 Richard Biener <rguenther@suse.de>
5347 PR tree-optimization/82291
5348 * tree-if-conv.c (predicate_mem_writes): Make sure to
5349 remove writes in blocks predicated with false.
5351 2017-09-22 Richard Biener <rguenther@suse.de>
5353 * sese.c: Include cfganal.h.
5354 (if_region_set_false_region): Remove.
5355 (create_if_region_on_edge): Likewise.
5356 (move_sese_in_condition): Re-implement without destroying
5359 2017-09-22 Richard Biener <rguenther@suse.de>
5361 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
5362 Verify both BBs contain loop PHI nodes before dispatching to
5364 (graphite_regenerate_ast_isl): Do not recompute dominators,
5365 do not verify three times. Restructure for clarity.
5366 * graphite-scop-detection.c (same_close_phi_node,
5367 remove_duplicate_close_phi, make_close_phi_nodes_unique,
5368 defined_in_loop_p, canonicalize_loop_closed_ssa,
5369 canonicalize_loop_closed_ssa_form): Simplify, remove excess
5370 checking and SSA rewrite, move to ...
5371 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
5372 (graphite_initialize): Do not pass in ctx, do not reset the
5373 SCEV cache, compute only dominators.
5374 (graphite_transform_loops): Allocate ISL ctx after
5375 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
5376 Maintain post-dominators only around build_scops.
5377 * sese.c (if_region_set_false_region): Make static. Free
5378 and recompute dominators.
5379 (move_sese_in_condition): Assert we don't get called with
5380 post-dominators computed.
5381 * sese.h (if_region_set_false_region): Remove.
5383 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
5385 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
5386 mode attribute for TARGET_AVX512VL.
5388 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
5390 * config/i386/i386.opt (mprefer-avx256): New option.
5391 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
5393 (ix86_preferred_simd_mode): Return 256-bit AVX modes
5394 for TARGET_PREFER_AVX256.
5395 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
5397 2017-09-21 Jeff Law <law@redhat.com>
5399 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
5400 Fix dump output if the only stack space is for pushed registers.
5402 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5404 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
5407 2017-09-21 Martin Sebor <msebor@redhat.com>
5410 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
5411 code (in C++) or code that triggers warnings.
5413 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
5415 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
5417 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
5419 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
5420 * sched-rgn.c: Ditto.
5421 * sel-sched-ir.c: Ditto.
5423 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
5425 * toplev.h (set_random_seed): Adjust return type.
5426 * toplev.c (init_local_tick): Move eager initialization of random_seed
5427 to get_random_seed. Adjust comment.
5428 (init_random_seed): Inline to get_random_seed, delete.
5429 (get_random_seed): Initialize random_seed lazily.
5430 (set_random_seed): Do not return previous value.
5431 (print_switch_value): Do not call get_random_seed.
5433 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
5435 * cgraph.c (delete_function_version): New, broken out from...
5436 (cgraph_node::delete_function_version): ...here. Rename to
5437 cgraph_node::delete_function_version_by_decl. Update all uses.
5438 (cgraph_node::remove): Call delete_function_version.
5440 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5443 * tree-inline.c (expand_call_inline): Emit clobber stmts for
5444 VAR_DECLs to which addressable non-volatile parameters are mapped
5445 and for id->retvar after the return value assignment. Clear
5446 id->retval and id->retbnd after inlining.
5448 2017-09-21 Richard Biener <rguenther@suse.de>
5450 PR tree-optimization/82276
5451 PR tree-optimization/82244
5452 * tree-vrp.c (build_assert_expr_for): Set
5453 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
5455 (remove_range_assertions): Revert earlier change.
5457 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
5460 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
5462 2017-09-21 Richard Biener <rguenther@suse.de>
5464 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5465 Restore valid IL after code generation errors.
5466 * graphite.c (graphite_transform_loops): Diagnose code
5467 generation issues as MSG_MISSED_OPTIMIZATION and continue
5468 with processing SCOPs.
5470 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5471 Alan Hayward <alan.hayward@arm.com>
5472 David Sherwood <david.sherwood@arm.com>
5474 * calls.c (compute_argument_addresses): Use simplify_gen_binary
5475 rather than choosing between plus_constant and gen_rtx_<CODE>.
5476 * expr.c (emit_push_insn): Likewise.
5477 (expand_expr_real_2): Likewise.
5479 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5480 Alan Hayward <alan.hayward@arm.com>
5481 David Sherwood <david.sherwood@arm.com>
5483 * loop-unroll.c (split_iv): Call copy_rtx on the step.
5485 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5486 Alan Hayward <alan.hayward@arm.com>
5487 David Sherwood <david.sherwood@arm.com>
5489 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
5490 calling tree_to_uhwi.
5492 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5493 Alan Hayward <alan.hayward@arm.com>
5494 David Sherwood <david.sherwood@arm.com>
5496 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
5497 INTEGER_CST rather than a negative test for ADDR_EXPR.
5499 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5500 Alan Hayward <alan.hayward@arm.com>
5501 David Sherwood <david.sherwood@arm.com>
5503 * tree-vrp.c (extract_range_from_binary_expr_1): Check
5504 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
5506 2017-09-21 Richard Biener <rguenther@suse.de>
5508 PR tree-optimization/71351
5509 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
5510 graphite_create_new_loop_guard): Remove, fold remaining parts
5512 (translate_isl_ast_node_for): ... here and simplify.
5514 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5517 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
5518 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
5519 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
5520 alternative always use QI mode, for -Os imov (=R,R) alternative
5521 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
5524 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5525 Jeff Law <law@redhat.com>
5527 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
5528 (allocate_stack_space): New function, partially extracted from
5530 (s390_emit_prologue): Track offset to most recent stack probe.
5531 Code to allocate space moved into allocate_stack_space.
5532 Dump actions when no stack is allocated.
5533 (s390_prologue_plus_offset): New function.
5534 (s390_emit_stack_probe): Likewise.
5536 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
5538 * common.opt (Wa, Wl, Wp, g, gz=): Add
5540 (gno-column-info): Remove.
5541 (gcolumn-info): Drop RejectNegative.
5543 (gno-record-gcc-switches): Remove.
5544 (grecord-gcc-switches): Drop RejectNegative.
5545 (gno-split-dwarf): Remove.
5546 (gsplit-dwarf): Drop RejectNegative.
5547 (gno-strict-dwarf): Remove.
5548 (gstrict-dwarf): Drop RejectNegative.
5549 * config/darwin.opt (gfull, gused): Add RejectNegative.
5550 * dwarf2out.c (gen_producer_string): Drop
5551 gno-record-gcc-switches handler.
5552 * optc-gen.awk: Add g to prefixes with negative forms.
5553 * opts-common.c (remapping_prefix_p): New.
5554 (find_opt): Check it.
5555 (generate_canonical_option): Test g prefix.
5556 (option_map): Add -gno- mapping.
5557 (add_misspelling_candidates): Check remapping_prefix_p.
5559 2017-09-20 Jeff Law <law@redhat.com>
5561 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
5562 thinko in stack clash protection support.
5564 * explow.c (compute_stack_clash_protection_loop_data): Use
5565 CONST_INT_P instead of explicit test. Verify object is a
5566 CONST_INT_P before looking at INTVAL.
5567 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
5568 instead of explicit test.
5570 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
5573 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
5574 address instead of to r1 and r11.
5576 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
5578 * config.gcc: Support "knm".
5579 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
5580 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5582 * config/i386/i386.c (m_KNM): Define.
5583 (processor_target_table): Add "knm".
5585 (ix86_option_override_internal): Add "knm".
5586 (ix86_issue_rate): Add PROCESSOR_KNM.
5587 (ix86_adjust_cost): Ditto.
5588 (ia32_multipass_dfa_lookahead): Ditto.
5589 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
5590 (fold_builtin_cpu): Add M_INTEL_KNM.
5591 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
5592 (processor_type): Add PROCESSOR_KNM.
5593 * config/i386/x86-tune.def: Add m_KNM.
5594 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
5596 2017-09-20 Richard Biener <rguenther@suse.de>
5598 PR tree-optimization/80213
5599 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
5600 are allowed in empty BBs as well.
5601 (canonicalize_loop_closed_ssa): Also look for other complex
5603 (scop_detection::get_sese): Include the loop-closed PHI block
5605 (scop_detection::merge_sese): Remove code adding extra blocks.
5606 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
5607 (build_scops): Assert the final returned scop is invalid.
5609 2017-09-20 Richard Biener <rguenther@suse.de>
5611 PR tree-optimization/82264
5612 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
5614 (vn_phi_lookup): Likewise.
5615 (vn_phi_insert): Likewise.
5617 2017-09-20 Jakub Jelinek <jakub@redhat.com>
5619 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
5620 that fits into uhwi or shwi, add DW_AT_const_value regardless
5621 of early_dwarf without going through RTL, using add_AT_unsigned
5624 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
5625 (DEBUG_LTO_ABBREV_SECTION): Likewise.
5626 (DEBUG_LTO_MACINFO_SECTION): Likewise.
5627 (DEBUG_MACRO_SECTION): Likewise.
5628 (DEBUG_LTO_MACRO_SECTION): Likewise.
5629 (DEBUG_STR_DWO_SECTION): Likewise.
5630 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
5631 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
5632 (DEBUG_LTO_DWO_LINE_SECTION): Define.
5633 (DEBUG_LTO_LINE_STR_SECTION): Define.
5634 (init_sections_and_labels): Initialize debug_line_str_section
5635 variable. Initialize debug_loc_section for -gdwarf-5 to
5636 DEBUG_LOCLISTS_SECTION. Formatting fixes.
5638 2017-09-20 Richard Biener <rguenther@suse.de>
5640 * graphite-sese-to-poly.c (extract_affine): Properly handle
5641 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
5643 2017-09-20 Richard Biener <rguenther@suse.de>
5645 PR tree-optimization/81373
5646 * graphite-scop-detection.c (build_cross_bb_scalars_def):
5647 Force SESE live-out defs to be handled even if they are
5650 2017-09-19 Jeff Law <law@redhat.com>
5652 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
5653 nothing for stack adjustments with REG_STACK_CHECK.
5654 * sched-deps.c (parse_add_or_inc): Reject insns with
5655 REG_STACK_CHECK from dependency breaking.
5656 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
5657 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
5658 * reg-notes.def (STACK_CHECK): New note.
5660 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
5661 (ix86_expand_prologue): Dump stack clash info as needed.
5662 Call ix86_adjust_stack_and_probe_stack_clash as needed.
5664 * function.c (dump_stack_clash_frame_info): New function.
5665 * function.h (dump_stack_clash_frame_info): Prototype.
5666 (enum stack_clash_probes): New enum.
5668 * config/alpha/alpha.c (alpha_expand_prologue): Also check
5669 flag_stack_clash_protection.
5670 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
5671 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
5672 (arm_frame_pointer_required): Likewise.
5673 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
5674 (ia64_expand_prologue): Likewise.
5675 * config/mips/mips.c (mips_expand_prologue): Likewise.
5676 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
5677 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
5678 (sparc_flat_expand_prologue): Likewise.
5679 * config/spu/spu.c (spu_expand_prologue): Likewise.
5681 * explow.c: Include "params.h".
5682 (anti_adjust_stack_and_probe_stack_clash): New function.
5683 (get_stack_check_protect): Likewise.
5684 (compute_stack_clash_protection_loop_data): Likewise.
5685 (emit_stack_clash_protection_loop_start): Likewise.
5686 (emit_stack_clash_protection_loop_end): Likewise.
5687 (allocate_dynamic_stack_space): Use get_stack_check_protect.
5688 Use anti_adjust_stack_and_probe_stack_clash.
5689 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
5690 (emit_stack_clash_protection_loop_start): Likewise.
5691 (emit_stack_clash_protection_loop_end): Likewise.
5692 * rtl.h (get_stack_check_protect): Prototype.
5693 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
5694 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
5695 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
5697 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
5699 * doc/tm.texi: Rebuilt.
5700 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
5701 get_stack_check_protect.
5702 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
5703 * config/arm/arm.c (arm_expand_prologue): Likewise.
5704 (arm_frame_pointer_required): Likewise.
5705 * config/i386/i386.c (ix86_expand_prologue): Likewise.
5706 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
5707 * config/mips/mips.c (mips_expand_prologue): Likewise.
5708 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
5709 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
5710 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
5711 (sparc_flat_expand_prologue): Likewise.
5713 * common.opt (-fstack-clash-protection): New option.
5714 * flag-types.h (enum stack_check_type): Note difference between
5715 -fstack-check= and -fstack-clash-protection.
5716 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
5717 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
5718 * toplev.c (process_options): Issue warnings/errors for cases
5719 not handled with -fstack-clash-protection.
5720 * doc/invoke.texi (-fstack-clash-protection): Document new option.
5721 (-fstack-check): Note additional problem with -fstack-check=generic.
5722 Note that -fstack-check is primarily for Ada and refer users
5723 to -fstack-clash-protection for stack-clash-protection.
5724 Document new params for stack clash protection.
5726 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
5728 * config/i386/i386.c (ix86_split_long_move): Do not handle
5729 address used for LEA in a special way.
5731 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
5733 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
5735 2017-09-19 Martin Sebor <msebor@redhat.com>
5738 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
5739 of incompatible types.
5741 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
5743 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5744 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
5745 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5746 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
5748 2017-09-19 Richard Biener <rguenther@suse.de>
5750 PR tree-optimization/82244
5751 * tree-vrp.c (remove_range_assertions): Do not propagate
5752 a constant to abnormals but replace the assert with a copy.
5754 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
5756 PR rtl-optimization/57878
5757 PR rtl-optimization/68988
5758 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
5759 avoidance test involving non_reload_pseudos. Move frequency test
5760 below the general fragmentation avoidance test.
5762 2017-09-19 Richard Biener <rguenther@suse.de>
5764 PR tree-optimization/69728
5765 * graphite-sese-to-poly.c (schedule_error): New global.
5766 (add_loop_schedule): Handle empty domain by failing the
5768 (build_original_schedule): Handle schedule_error.
5770 2017-09-19 Richard Biener <rguenther@suse.de>
5772 * graphite-scop-detection.c (scop_detection::can_represent_loop):
5773 Do not iterate to sibling loops but only to siblings of inner
5776 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
5779 * config/m68k/m68k.md (moveq feeding equality comparison): Check
5780 that the registers are different.
5782 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
5784 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
5785 to processor_model and "amdfam17h" to arch_names_table.
5786 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
5788 2017-09-18 Jakub Jelinek <jakub@redhat.com>
5791 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
5793 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5794 Alan Hayward <alan.hayward@arm.com>
5795 David Sherwood <david.sherwood@arm.com>
5797 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
5799 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
5801 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
5802 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
5803 vect_schedule_slp_instance.
5804 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
5805 Update call to vect_slp_analyze_node_operations. Simplify return
5807 (vect_slp_analyze_bb_1): Update call accordingly.
5808 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
5809 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
5810 (vect_schedule_slp): Update call accordingly.
5812 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5813 Alan Hayward <alan.hayward@arm.com>
5814 David Sherwood <david.sherwood@arm.com>
5816 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
5817 with types that aren't in fact scalar.
5819 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5821 * tree-vect-slp.c (vect_record_max_nunits): New function,
5823 (vect_build_slp_tree_1): ...here.
5824 (vect_build_slp_tree_2): Call it for phis too.
5826 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5828 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
5829 to vect_get_vec_def_for_operand when getting the mask operand.
5831 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5832 Alan Hayward <alan.hayward@arm.com>
5833 David Sherwood <david.sherwood@arm.com>
5835 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
5838 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5839 Alan Hayward <alan.hayward@arm.com>
5840 David Sherwood <david.sherwood@arm.com>
5842 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
5843 calculation for vector booleans.
5845 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5846 Alan Hayward <alan.hayward@arm.com>
5847 David Sherwood <david.sherwood@arm.com>
5849 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
5851 (vect_transform_stmt): ...here.
5852 (vect_analyze_stmt): Use it instead of calling
5853 vectorizable_live_operation directly.
5855 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
5857 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
5858 non-SIMT targets in acc vector loops.
5860 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
5862 * configure.ac: Add arc and check if assembler supports gdwarf2.
5863 * configure: Regenerate.
5865 2017-09-18 Richard Biener <rguenther@suse.de>
5867 PR tree-optimization/82220
5868 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
5869 epilogue niters from the min_profitable_iters compute.
5871 2017-09-18 Jakub Jelinek <jakub@redhat.com>
5874 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
5875 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
5876 (ix86_init_pic_reg): Revert 2017-09-01 changes.
5878 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
5881 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
5882 switching to a new text section.
5884 2017-09-18 Richard Biener <rguenther@suse.de>
5886 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
5888 (build_alias_set): Reject aliases with no access function.
5890 2017-09-18 Richard Biener <rguenther@suse.de>
5892 PR tree-optimization/79622
5893 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
5895 (build_cross_bb_scalars_use): Likewise.
5897 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
5899 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
5901 2017-09-18 Alan Modra <amodra@gmail.com>
5904 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
5905 stack_pointer_rtx for count 0. Update comments. Break up
5906 large rtl expression.
5908 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
5910 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
5911 Increase to 20 bytes.
5912 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
5913 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
5914 or avx version of the stub.
5916 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
5919 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
5920 compute the minimum stack alignment. Also update preferred stack
5921 boundary for leaf functions.
5923 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
5925 PR tree-optimization/82228
5926 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
5929 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5931 * common/config/nds32/nds32-common.c
5932 (nds32_option_optimization_table): Refine formatting.
5933 (nds32_option_optimization_table): Use -fsched-pressure and
5934 -fomit-frame-pointer for specific optimization level.
5936 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5938 * config/nds32/nds32.c: Refine formatting and comments.
5939 * config/nds32/nds32.h: Likewise.
5940 * config/nds32/nds32.md: Likewise.
5941 * config/nds32/nds32-cost.c: Likewise.
5942 * config/nds32/nds32-isr.c: Likewise.
5943 * config/nds32/nds32-md-auxiliary.c: Likewise.
5944 * config/nds32/nds32-multiple.md: Likewise.
5945 * config/nds32/nds32-predicates.c: Likewise.
5947 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
5948 Jakub Jelinek <jakub@redhat.com>
5950 Add support for -std=c++2a.
5951 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
5953 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
5955 2017-09-15 Steve Ellcey <sellcey@cavium.com>
5958 * doc/extend.texi (Common Function Attributes): Add
5959 references to ARM, AArch64, and S/390 specific attributes.
5960 (Function Specific Option Pragmas): Add AArch64 and S/390
5961 to list of back ends that support the target pragma.
5963 2017-09-15 Nathan Sidwell <nathan@acm.org>
5965 * doc/standards.texi: Fix C++17 description. Update URLs for
5968 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
5970 * common.opt (Wcast-align=strict): New warning option.
5971 * doc/invoke.texi: Document -Wcast-align=strict.
5973 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
5975 * cgraph.h (cgraph_thunk_info): Add comments.
5976 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
5977 assert for VIRTUAL_* arguments stricter.
5979 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
5981 PR tree-optimization/71026
5982 * match.pd: Move RDIV patterns from fold-const.c
5983 * fold-const.c (distribute_real_division): Removed.
5984 (fold_binary_loc): Remove calls to distribute_real_divison.
5986 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5988 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
5989 c++1z and gnu++1z as deprecated. Change other references to
5990 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
5991 Change -Wc++1z-compat to -Wc++17-compat.
5992 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
5993 * dwarf2out.c (highest_c_language): Handle C++17.
5994 (gen_compile_unit_die): Likewise.
5996 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5998 PR rtl-optimization/82192
5999 * combine.c (make_extraction): Don't look through non-paradoxical
6000 SUBREGs or TRUNCATE if pos + len is or might be bigger than
6003 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6004 Alan Hayward <alan.hayward@arm.com>
6005 David Sherwood <david.sherwood@arm.com>
6007 * target.def (function_arg_offset): New hook.
6008 * targhooks.h (default_function_arg_offset): Declare.
6009 * targhooks.c (default_function_arg_offset): New function.
6010 * function.c (locate_and_pad_parm): Use
6011 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
6012 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
6013 (TARGET_FUNCTION_ARG_OFFSET): ...this.
6014 * doc/tm.texi: Regenerate.
6015 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
6016 * config/spu/spu.c (spu_function_arg_offset): New function.
6017 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
6018 * system.h (FUNCTION_ARG_OFFSET): Poison.
6020 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6021 Alan Hayard <alan.hayward@arm.com>
6022 David Sherwood <david.sherwood@arm.com>
6024 * target.def (truly_noop_truncation): New hook.
6025 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
6026 than TRULY_NOOP_TRUNCATION.
6027 * hooks.h (hook_bool_uint_uint_true): Declare.
6028 * hooks.c (hook_bool_uint_uint_true): New function.
6029 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
6030 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
6031 * doc/tm.texi: Regenerate.
6032 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
6033 rather than TRULY_NOOP_TRUNCATION in comments.
6034 (simplify_comparison): Likewise.
6035 (record_truncated_value): Likewise.
6036 * expmed.c (extract_bit_field_1): Likewise.
6037 (extract_split_bit_field): Likewise.
6038 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
6039 instead of TRULY_NOOP_TRUNCATION.
6040 * function.c (assign_parm_setup_block): Likewise.
6041 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
6042 * rtlhooks.c: Include target.h.
6043 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
6044 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
6045 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
6046 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
6047 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
6048 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
6049 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
6050 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
6051 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
6052 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
6053 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
6054 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
6055 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
6056 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
6057 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
6058 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
6059 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
6060 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
6061 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
6062 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
6063 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
6064 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
6065 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
6066 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
6067 * config/mips/mips.c (mips_truly_noop_truncation): New function.
6068 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6069 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6070 rather than TRULY_NOOP_TRUNCATION in comments.
6071 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
6072 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
6073 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
6074 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
6075 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
6076 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
6077 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
6078 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
6079 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
6080 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
6081 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
6082 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6083 rather than TRULY_NOOP_TRUNCATION in comments.
6084 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
6085 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
6086 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
6087 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
6088 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
6089 TRULY_NOOP_TRUNCATION condition.
6090 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
6091 (TRULY_NOOP_TRUNCATION): Delete.
6092 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
6093 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
6094 * config/spu/spu.c (spu_truly_noop_truncation): New function.
6095 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6096 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
6097 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
6098 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
6099 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6100 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6101 rather than TRULY_NOOP_TRUNCATION in comments.
6102 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
6103 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
6104 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
6105 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
6106 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
6107 * system.h (TRULY_NOOP_TRUNCATION): Poison.
6109 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
6112 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
6113 (*cmp_ior): Likewise.
6114 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
6115 (*ior_scc_scc_cmp): Likewise.
6116 (*and_scc_scc): Likewise.
6117 (*and_scc_scc_cmp): Likewise.
6119 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6120 Alan Hayard <alan.hayward@arm.com>
6121 David Sherwood <david.sherwood@arm.com>
6123 * target.def (can_change_mode_class): New hook.
6124 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
6125 (hard_regno_nregs): Likewise.
6126 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
6127 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
6128 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
6129 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
6130 (LOAD_EXTEND_OP): Update accordingly.
6131 * doc/tm.texi: Regenerate.
6132 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
6133 CANNOT_CHANGE_MODE_CLASS.
6134 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
6135 (REG_CAN_CHANGE_MODE_P): ...this new macro.
6136 * combine.c (simplify_set): Update accordingly.
6137 * emit-rtl.c (validate_subreg): Likewise.
6138 * recog.c (general_operand): Likewise.
6139 * regcprop.c (mode_change_ok): Likewise.
6140 * reload1.c (choose_reload_regs): Likewise.
6141 (inherit_piecemeal_p): Likewise.
6142 * rtlanal.c (simplify_subreg_regno): Likewise.
6143 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
6144 instead of CANNOT_CHANGE_MODE_CLASS.
6145 (reload_cse_simplify_operands): Likewise.
6146 * reload.c (push_reload): Use targetm.can_change_mode_class
6147 instead of CANNOT_CHANGE_MODE_CLASS.
6148 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
6149 REG_CANNOT_CHANGE_MODE_P.
6150 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6151 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
6152 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6153 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6154 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6155 (arm_can_change_mode_class): New function.
6156 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6157 than CANNOT_CHANGE_MODE_CLASS in comments.
6158 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6159 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
6160 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
6161 (ix86_can_change_mode_class): ...this new function, inverting the
6162 sense of the return value.
6163 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6164 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6165 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6166 (ia64_can_change_mode_class): New function.
6167 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6168 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
6169 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
6170 (m32c_can_change_mode_class): ...this new function, inverting the
6171 sense of the return value.
6172 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6173 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6174 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
6175 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
6176 (mips_can_change_mode_class): ...this new function, inverting the
6177 sense of the return value.
6178 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6179 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6180 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6181 (msp430_can_change_mode_class): New function.
6182 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6183 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
6184 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6185 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6186 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6187 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
6188 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6189 (pa_cannot_change_mode_class): Replace with...
6190 (pa_can_change_mode_class): ...this new function, inverting the
6191 sense of the return value.
6192 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6193 than CANNOT_CHANGE_MODE_CLASS in comments.
6194 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6195 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
6196 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6197 (pdp11_cannot_change_mode_class): Replace with...
6198 (pdp11_can_change_mode_class): ...this new function, inverting the
6199 sense of the return value.
6200 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6201 * config/powerpcspe/powerpcspe-protos.h
6202 (rs6000_cannot_change_mode_class_ptr): Delete.
6203 * config/powerpcspe/powerpcspe.c
6204 (rs6000_cannot_change_mode_class_ptr): Delete.
6205 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6206 (rs6000_option_override_internal): Assign to
6207 targetm.can_change_mode_class instead of
6208 rs6000_cannot_change_mode_class_ptr.
6209 (rs6000_cannot_change_mode_class): Replace with...
6210 (rs6000_can_change_mode_class): ...this new function, inverting the
6211 sense of the return value.
6212 (rs6000_debug_cannot_change_mode_class): Replace with...
6213 (rs6000_debug_can_change_mode_class): ...this new function.
6214 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6215 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
6216 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6217 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6218 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
6220 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
6221 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6222 (rs6000_option_override_internal): Assign to
6223 targetm.can_change_mode_class instead of
6224 rs6000_cannot_change_mode_class_ptr.
6225 (rs6000_cannot_change_mode_class): Replace with...
6226 (rs6000_can_change_mode_class): ...this new function, inverting the
6227 sense of the return value.
6228 (rs6000_debug_cannot_change_mode_class): Replace with...
6229 (rs6000_debug_can_change_mode_class): ...this new function.
6230 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6231 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
6232 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
6233 (s390_can_change_mode_class): ...this new function, inverting the
6234 sense of the return value.
6235 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6236 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6237 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
6238 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6239 (sh_cannot_change_mode_class): Replace with...
6240 (sh_can_change_mode_class): ...this new function, inverting the
6241 sense of the return value.
6242 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6243 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6244 (sparc_can_change_mode_class): New function.
6245 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6246 * config/spu/spu.c (spu_can_change_mode_class): New function.
6247 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6248 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6249 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6250 (visium_can_change_mode_class): New function.
6251 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
6253 2017-09-15 Richard Biener <rguenther@suse.de>
6255 PR tree-optimization/82217
6256 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
6257 but not undefined case.
6259 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6262 * postreload.c (reload_cse_simplify_operands): Skip
6263 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
6265 2017-09-15 Richard Biener <rguenther@suse.de>
6267 PR tree-optimization/68823
6268 * graphite-scop-detection.c (build_alias_set): If we have a
6269 possible dependence check whether we can handle them by just
6270 looking at the DRs DR_ACCESS_FNs.
6271 (build_scops): If build_alias_set fails, fail the SCOP.
6273 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6275 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
6276 to support float128 built-in functions that require the ISA 3.0
6278 (BU_FLOAT128_3_HW): Likewise.
6279 (SQRTF128): Add support for the IEEE 128-bit square root and fma
6281 (FMAF128): Likewise.
6283 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
6284 support for built-in functions that need the ISA 3.0 IEEE 128-bit
6285 floating point instructions.
6286 (rs6000_invalid_builtin): Likewise.
6287 (rs6000_builtin_mask_names): Likewise.
6288 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
6289 (RS6000_BTM_FLOAT128_HW): Likewise.
6290 (RS6000_BTM_COMMON): Likewise.
6291 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
6293 * doc/extend.texi (RS/6000 built-in functions): Document the
6294 IEEE 128-bit floating point square root and fused multiply-add
6297 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
6299 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
6300 reg (r2) isn't in the set of registers defined in the prologue.
6302 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6303 Alan Hayward <alan.hayward@arm.com>
6304 David Sherwood <david.sherwood@arm.com>
6306 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
6307 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
6308 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
6309 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
6310 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
6312 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
6313 max_vectorization_factor.
6314 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
6316 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6317 Alan Hayward <alan.hayward@arm.com>
6318 David Sherwood <david.sherwood@arm.com>
6320 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
6321 (vect_worthwhile_without_simd_p): Declare.
6322 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
6323 (vectorizable_reduction): Use it.
6324 * tree-vect-stmts.c (vectorizable_shift): Likewise.
6325 (vectorizable_operation): Likewise.
6327 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6328 Alan Hayward <alan.hayward@arm.com>
6329 David Sherwood <david.sherwood@arm.com>
6331 * tree-vectorizer.h (vect_get_num_copies): New function.
6332 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
6333 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6334 (vectorizable_induction): Likewise.
6335 (vectorizable_live_operation): Likewise.
6336 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6337 (vectorizable_bswap): Likewise.
6338 (vectorizable_call): Likewise.
6339 (vectorizable_conversion): Likewise.
6340 (vectorizable_assignment): Likewise.
6341 (vectorizable_shift): Likewise.
6342 (vectorizable_operation): Likewise.
6343 (vectorizable_store): Likewise.
6344 (vectorizable_load): Likewise.
6345 (vectorizable_condition): Likewise.
6346 (vectorizable_comparison): Likewise.
6347 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
6349 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6350 Alan Hayward <alan.hayward@arm.com>
6351 David Sherwood <david.sherwood@arm.com>
6353 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
6354 of vect_init_vector.
6356 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6357 Alan Hayward <alan.hayward@arm.com>
6358 David Sherwood <david.sherwood@arm.com>
6360 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
6361 an inline wrapper that provides a location.
6362 (gimple_build_vector): Likewise.
6363 * gimple-fold.c (gimple_build_vector_from_val): New function.
6364 (gimple_build_vector): Likewise.
6365 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
6366 functions to build the initial value. Always return a gimple value.
6367 (get_initial_defs_for_reduction): Likewise. Only compute
6369 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
6370 vect_init_vector on the results from get_initial_def(s)_for_reduction.
6371 (vectorizable_induction): Use gimple_build_vector rather than
6374 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6375 Alan Hayward <alan.hayward@arm.com>
6376 David Sherwood <david.sherwood@arm.com>
6378 * target.h (vec_perm_indices): New typedef.
6379 (auto_vec_perm_indices): Likewise.
6380 * optabs-query.h: Include target.h
6381 (can_vec_perm_p): Take a vec_perm_indices *.
6382 * optabs-query.c (can_vec_perm_p): Likewise.
6383 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
6384 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6385 * tree-vect-generic.c (lower_vec_perm): Likewise.
6386 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
6387 (vect_grouped_load_supported): Likewise.
6388 (vect_shift_permute_load_chain): Likewise.
6389 (vect_permute_store_chain): Use auto_vec_perm_indices.
6390 (vect_permute_load_chain): Likewise.
6391 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
6392 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
6393 Update uses of can_vec_perm_p.
6394 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
6395 mode with a number of elements. Take a vec_perm_indices *.
6396 (vect_create_epilog_for_reduction): Update accordingly.
6397 Use auto_vec_perm_indices.
6398 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
6399 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
6400 (vect_transform_slp_perm_load): Likewise.
6401 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
6402 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
6403 (vect_gen_perm_mask_checked): Likewise.
6404 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
6405 (vect_gen_perm_mask_checked): Likewise.
6406 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
6407 (vectorizable_store): Likewise.
6408 (vectorizable_load): Likewise.
6409 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
6410 (vectorizable_bswap): Likewise.
6412 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6413 Alan Hayward <alan.hayward@arm.com>
6414 David Sherwood <david.sherwood@arm.com>
6416 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
6417 * tree.c (build_vector): Likewise.
6418 (build_vector_from_ctor): Update accordingly.
6419 (build_vector_from_val): Likewise.
6420 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
6421 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6422 * tree-vect-generic.c (add_rshift): Likewise.
6423 (expand_vector_divmod): Likewise.
6424 (optimize_vector_constructor): Likewise.
6425 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
6426 (vect_transform_slp_perm_load): Likewise.
6427 (vect_schedule_slp_instance): Likewise.
6428 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
6429 (vectorizable_call): Likewise.
6430 (vect_gen_perm_mask_any): Likewise. Add elements in order.
6431 * expmed.c (make_tree): Likewise.
6432 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
6433 a vector passed to build_vector.
6434 (fold_convert_const): Likewise.
6435 (exact_inverse): Likewise.
6436 (fold_ternary_loc): Likewise.
6437 (fold_relational_const): Likewise.
6438 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
6439 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
6440 (const_unop): Likewise. Store the reduction accumulator in a
6441 variable rather than an array.
6442 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
6443 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
6444 the new vector, rather than constructing it after the input arrays.
6445 (native_interpret_vector): Use auto_vec<tree> when building
6446 a vector passed to build_vector. Add elements in order.
6447 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
6448 auto_vec<tree> when building a vector passed to build_vector.
6449 (vect_create_epilog_for_reduction): Likewise.
6450 (vectorizable_induction): Likewise.
6451 (get_initial_def_for_reduction): Likewise. Fix indentation of
6453 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
6455 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
6456 passed to build_vector.
6458 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6459 Alan Hayward <alan.hayward@arm.com>
6460 David Sherwood <david.sherwood@arm.com>
6462 * tree-core.h (tree_base::u): Add an "nelts" field.
6463 (tree_vector): Use VECTOR_CST_NELTS as the length.
6464 * tree.c (tree_size): Likewise.
6465 (make_vector): Initialize VECTOR_CST_NELTS.
6466 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
6467 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
6468 TYPE_VECTOR_SUBPARTS.
6469 * expr.c (const_vector_mask_from_tree): Consistently use "units"
6470 as the number of units, setting it from VECTOR_CST_NELTS.
6471 (const_vector_from_tree): Likewise.
6472 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
6473 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
6474 (fold_negate_expr_1): Likewise.
6475 (fold_convert_const): Likewise.
6476 (const_binop): Likewise. Differentiate the number of output and
6478 (const_unop): Likewise.
6479 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
6480 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
6481 in cases that did the opposite.
6483 2017-09-14 Richard Biener <rguenther@suse.de>
6485 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
6488 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
6490 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
6492 2017-09-14 Jakub Jelinek <jakub@redhat.com>
6495 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
6496 if and where to split a bb, except for splitting before debug insn
6497 sequences followed by non-label real insn. Delete debug insns
6498 in between basic blocks.
6500 * combine.c (make_compound_operation_int): Formatting fixes.
6502 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
6503 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6504 * config/netbsd.h (LINK_EH_SPEC): Likewise.
6505 * config/sol2.h (LINK_EH_SPEC): Likewise.
6506 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6507 * config/s390/linux.h (LINK_SPEC): Likewise.
6508 * config/freebsd.h (LINK_EH_SPEC): Likewise.
6509 * config/openbsd.h (LINK_EH_SPEC): Likewise.
6510 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6511 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
6512 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
6513 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6514 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
6515 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
6517 2017-09-13 Jakub Jelinek <jakub@redhat.com>
6519 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
6521 (ENDFILE_LINUX_SPEC): Likewise.
6522 (LINK_EH_SPEC): Likewise.
6523 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
6524 (LINK_OS_LINUX_SPEC32): Likewise.
6525 (LINK_OS_LINUX_SPEC64): Likewise.
6526 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
6527 (LINK_OS_LINUX_SPEC): Likewise.
6529 2017-09-13 Martin Liska <mliska@suse.cz>
6532 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
6533 CASE_HIGH is NULL_TREE.
6535 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6536 Alan Hayward <alan.hayward@arm.com>
6537 David Sherwood <david.sherwood@arm.com>
6539 * target.def (secondary_memory_needed): New hook.
6540 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
6541 instead of SECONDARY_MEMORY_NEEDED.
6542 (secondary_memory_needed_mode): Likewise.
6543 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
6544 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
6545 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
6546 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
6547 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
6548 * doc/tm.texi: Regenerate.
6549 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
6550 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
6551 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6552 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
6553 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
6554 * config/i386/i386.c (inline_secondary_memory_needed): Put the
6555 mode argument first and change the reg_class arguments to reg_class_t.
6556 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
6557 Make static. Update the call to inline_secondary_memory_needed.
6558 (ix86_register_move_cost): Update the call to
6559 inline_secondary_memory_needed.
6560 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6561 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
6563 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
6564 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6566 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
6567 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
6568 * config/mips/mips.c (mips_secondary_memory_needed): Make static
6569 and match hook interface. Add comment from mips.h.
6570 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6571 * config/mmix/mmix.md (truncdfsf2): Refer to
6572 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6574 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
6575 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
6576 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6577 (pa_secondary_memory_needed): New function.
6578 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
6579 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
6580 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6581 (pdp11_secondary_memory_needed): Make static and match hook interface.
6582 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
6583 * config/powerpcspe/powerpcspe-protos.h
6584 (rs6000_secondary_memory_needed_ptr): Delete.
6585 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
6587 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6588 (rs6000_option_override_internal): Assign to
6589 targetm.secondary_memory_needed rather than
6590 rs6000_secondary_memory_needed_ptr.
6591 (rs6000_secondary_memory_needed): Match hook interface.
6592 (rs6000_debug_secondary_memory_needed): Likewise.
6593 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
6594 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
6595 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
6596 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6597 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
6598 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
6600 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
6601 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6602 (rs6000_option_override_internal): Assign to
6603 targetm.secondary_memory_needed rather than
6604 rs6000_secondary_memory_needed_ptr.
6605 (rs6000_secondary_memory_needed): Match hook interface.
6606 (rs6000_debug_secondary_memory_needed): Likewise.
6607 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
6608 * config/s390/s390.c (s390_secondary_memory_needed): New function.
6609 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6610 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
6611 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6612 (sparc_secondary_memory_needed): New function.
6613 * lra-constraints.c (check_and_process_move): Refer to
6614 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6616 (curr_insn_transform): Likewise.
6617 (process_alt_operands): Use targetm.secondary_memory_needed
6618 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6619 (check_secondary_memory_needed_p): Likewise.
6620 (choose_split_class): Likewise.
6621 * reload.c: Unconditionally include code that was previously
6622 conditional on SECONDARY_MEMORY_NEEDED.
6623 (push_secondary_reload): Use targetm.secondary_memory_needed
6624 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6625 (push_reload): Likewise.
6626 * reload1.c: Unconditionally include code that was previously
6627 conditional on SECONDARY_MEMORY_NEEDED.
6628 (choose_reload_regs): Use targetm.secondary_memory_needed
6629 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6630 (gen_reload): Likewise.
6631 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
6633 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6634 Alan Hayward <alan.hayward@arm.com>
6635 David Sherwood <david.sherwood@arm.com>
6637 * target.def (secondary_memory_needed_mode): New hook:
6638 * targhooks.c (default_secondary_memory_needed_mode): Declare.
6639 * targhooks.h (default_secondary_memory_needed_mode): New function.
6640 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
6641 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
6642 * doc/tm.texi: Regenerate.
6643 * lra-constraints.c (check_and_process_move): Use
6644 targetm.secondary_memory_needed_mode instead of
6645 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
6646 (curr_insn_transform): Likewise.
6647 * reload.c (get_secondary_mem): Likewise.
6648 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6649 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
6651 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6652 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6653 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
6654 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6655 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
6657 * config/powerpcspe/powerpcspe-protos.h
6658 (rs6000_secondary_memory_needed_mode): Delete.
6659 * config/powerpcspe/powerpcspe.c
6660 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6661 (rs6000_secondary_memory_needed_mode): Make static.
6662 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6663 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
6665 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6667 (rs6000_secondary_memory_needed_mode): Make static.
6668 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6669 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
6670 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6671 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6672 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6674 (sparc_secondary_memory_needed_mode): New function.
6675 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
6677 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
6679 * config/aarch64/constraints.md (Umq): New constraint.
6680 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6682 (mov<mode>): Update condition.
6684 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6686 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
6687 when bitposition is the same.
6689 2017-09-13 Richard Biener <rguenther@suse.de>
6691 * dwarf2out.c (output_die_symbol): Remove.
6692 (output_die): Do not output a DIEs symbol.
6694 2017-09-13 Richard Biener <rguenther@suse.de>
6697 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
6698 default-def to avoid breaking iterator update with the weird
6699 interaction with cgraph_update_edges_for_call_stmt_node.
6701 2017-09-13 Richard Biener <rguenther@suse.de>
6703 * tree-cfg.c (verify_gimple_assign_binary): Add verification
6704 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
6705 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
6706 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
6708 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
6710 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
6711 Disable pc relative literal load irrespective of
6712 TARGET_FIX_ERR_A53_84341 for default.
6714 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
6716 * config/sparc/sparc.c (output_return): Output the source location of
6717 the insn in the delay slot, if any.
6718 (output_sibcall): Likewise.
6720 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
6723 * common.opt (-static-pie): New alias.
6724 (shared): Negate static-pie.
6725 (-no-pie): Update help text.
6727 (static-pie): New option.
6728 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
6729 -static-pie support.
6730 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
6731 (LINK_EH_SPEC): Likewise.
6732 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6733 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
6734 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
6735 * gcc.c (LINK_COMMAND_SPEC): Likewise.
6736 (init_gcc_specs): Likewise.
6737 (init_spec): Likewise.
6738 (display_help): Update help message for -pie.
6739 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
6742 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
6744 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
6745 (movdi_aarch64): Likewise.
6746 (movti_aarch64): Likewise.
6748 2017-09-12 Simon Wright <simon@pushface.org>
6751 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
6752 calculation of the minor version, always output as 0.
6754 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6757 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6758 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
6759 on it early, rather than manual conversion late. For
6760 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
6761 instead of performing manual conversion.
6763 2017-09-12 Carl Love <cel@us.ibm.com>
6765 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
6766 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
6768 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6769 VMULOSW): Add definitions.
6770 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6771 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6772 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6773 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6774 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6776 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
6778 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
6780 (movti_aarch64): Likewise.
6781 (movdf_aarch64): Likewise.
6782 (movtf_aarch64): Likewise.
6783 (load_pairdi): Likewise.
6784 (store_pairdi): Likewise.
6785 (load_pairdf): Likewise.
6786 (store_pairdf): Likewise.
6787 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
6788 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
6789 (ldr_got_small_<mode>): Likewise.
6790 (ldr_got_small_28k_<mode>): Likewise.
6791 (ldr_got_tiny): Likewise.
6792 * config/aarch64/iterators.md (ldst_sz): New.
6793 (ldpstp_sz): Likewise.
6794 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
6796 (thunderx_load): Split load_8 to load_16.
6797 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
6799 (thunderx2t99_storepair_basic): Split store_8 to store_16.
6800 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
6801 (xgene1_store_pair): Split store_8 to store_16.
6802 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
6803 (falkor_st_0_st_sd): Split store_8 to store_16.
6805 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
6807 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
6808 and store1/2/3/4 to store_4/8/12/16.
6809 * config/aarch64/aarch64.md: Update for rename.
6810 * config/arm/arm.md: Likewise.: Likewise.
6811 * config/arm/arm.c: Likewise.
6812 * config/arm/thumb1.md: Likewise.
6813 * config/arm/thumb2.md: Likewise.
6814 * config/arm/vfp.md: Likewise.
6815 * config/arm/arm-generic.md: Likewise.
6816 * config/arm/arm1020e.md: Likewise.
6817 * config/arm/arm1026ejs.md: Likewise.
6818 * config/arm/arm1136jfs.md: Likewise.
6819 * config/arm/arm926ejs.md: Likewise.
6820 * config/arm/cortex-a15.md: Likewise.
6821 * config/arm/cortex-a17.md: Likewise.
6822 * config/arm/cortex-a5.md: Likewise.
6823 * config/arm/cortex-a53.md: Likewise.
6824 * config/arm/cortex-a57.md: Likewise.
6825 * config/arm/cortex-a7.md: Likewise.
6826 * config/arm/cortex-a8.md: Likewise.
6827 * config/arm/cortex-a9.md: Likewise.
6828 * config/arm/cortex-m4.md: Likewise.
6829 * config/arm/cortex-m7.md: Likewise.
6830 * config/arm/cortex-r4.md: Likewise.
6831 * config/arm/exynos-m1.md: Likewise.
6832 * config/arm/fa526.md: Likewise.
6833 * config/arm/fa606te.md: Likewise.
6834 * config/arm/fa626te.md: Likewise.
6835 * config/arm/fa726te.md: Likewise.
6836 * config/arm/fmp626.md: Likewise.
6837 * config/arm/iwmmxt.md: Likewise.
6838 * config/arm/ldmstm.md: Likewise.
6839 * config/arm/marvell-pj4.md: Likewise.
6840 * config/arm/xgene1.md: Likewise.
6841 * config/aarch64/thunderx.md: Likewise.
6842 * config/aarch64/thunderx2t99.md: Likewise.
6843 * config/aarch64/falkor.md: Likewise.
6845 2017-09-12 Martin Liska <mliska@suse.cz>
6847 * attribs.c (private_lookup_attribute): New function.
6848 * attribs.h (private_lookup_attribute): Declared here.
6849 (lookup_attribute): Called from this place.
6851 2017-09-12 Richard Biener <rguenther@suse.de>
6853 PR tree-optimization/82157
6854 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
6855 stmts with side-effects.
6857 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6858 Alan Hayward <alan.hayward@arm.com>
6859 David Sherwood <david.sherwood@arm.com>
6861 * target.def (hard_regno_nregs): New hook.
6862 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
6863 * targhooks.h (default_hard_regno_nregs): Declare.
6864 * targhooks.c (default_hard_regno_nregs): New function.
6865 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
6866 (TARGET_HARD_REGNO_NREGS): ...this hook.
6867 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
6868 (CLASS_MAX_NREGS): Likewise.
6869 * doc/tm.texi: Regenerate.
6870 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
6871 instead of HARD_REGNO_NREGS.
6872 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
6873 HARD_REGNO_NREGS in the comment.
6874 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
6875 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
6876 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
6877 Return an unsigned int.
6878 (TARGET_HARD_REGNO_NREGS): Redefine.
6879 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
6880 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
6881 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6882 (arc_hard_regno_nregs): New function.
6883 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
6884 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
6885 (arm_hard_regno_nregs): New function.
6886 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
6887 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
6888 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
6889 (TARGET_HARD_REGNO_NREGS): Redefine.
6890 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
6891 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
6892 (HARD_REGNO_NREGS): Delete.
6893 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
6894 (cr16_hard_regno_nregs): New function.
6895 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
6896 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
6897 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
6898 (cris_hard_regno_nregs): New function.
6899 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
6900 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
6901 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6902 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
6903 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
6904 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
6905 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
6906 (frv_hard_regno_nregs): Make static. Take and return an
6908 (frv_class_max_nregs): Remove outdated copy of documentation.
6909 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
6910 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
6911 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
6912 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
6913 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
6914 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
6915 (TARGET_HARD_REGNO_NREGS): Redefine.
6916 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
6917 (CLASS_MAX_NREGS): Update comment.
6918 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
6919 (ia64_hard_regno_nregs): New function.
6920 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
6921 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
6922 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
6923 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
6924 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
6926 (m32c_hard_regno_nregs): Likewise. Make static.
6927 (TARGET_HARD_REGNO_NREGS): Redefine.
6928 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
6929 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
6930 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
6931 (m68k_hard_regno_nregs): New function.
6932 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
6933 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
6934 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
6935 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
6936 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
6937 Take and return an unsigned int.
6938 (TARGET_HARD_REGNO_NREGS): Redefine.
6939 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
6940 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6941 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
6942 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
6943 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
6944 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
6945 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
6946 (msp430_hard_regno_nregs): Make static. Take and return an
6948 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
6949 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
6950 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
6951 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
6952 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
6953 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
6954 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
6955 (TARGET_HARD_REGNO_NREGS): Redefine.
6956 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
6957 (PA_HARD_REGNO_NREGS): ...this.
6958 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
6959 (PA_HARD_REGNO_NREGS): ...this.
6960 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6961 (pa_hard_regno_nregs): New function.
6962 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
6963 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
6964 (pdp11_hard_regno_nregs): New function.
6965 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
6966 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
6967 (rs6000_hard_regno_nregs_hook): New function.
6968 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
6969 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6970 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
6971 Take and return an unsigned int. Move earlier in file.
6972 (TARGET_HARD_REGNO_NREGS): Redefine.
6973 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
6974 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
6975 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
6976 (rl78_hard_regno_nregs): Make static. Take and return an
6978 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
6979 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
6980 (rs6000_hard_regno_nregs_hook): New function.
6981 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
6982 * config/rx/rx.c (rx_hard_regno_nregs): New function.
6983 (TARGET_HARD_REGNO_NREGS): Redefine.
6984 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
6985 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
6986 instead of HARD_REGNO_NREGS.
6987 (s390_hard_regno_nregs): New function.
6988 (s390_hard_regno_mode_ok): Add comment from s390.h.
6989 (TARGET_HARD_REGNO_NREGS): Redefine.
6990 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
6991 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
6992 (sh_hard_regno_nregs): New function.
6993 (sh_pass_in_reg_p): Use it.
6994 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
6995 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6996 (sparc_hard_regno_nregs): New function.
6997 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
6998 * config/spu/spu.c (spu_hard_regno_nregs): New function.
6999 (spu_function_arg_advance): Use it, supplying a valid register number.
7000 (TARGET_HARD_REGNO_NREGS): Redefine.
7001 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
7002 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
7003 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
7004 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
7005 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
7006 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
7007 (CLASS_MAX_NREGS): Remove copy of old documentation.
7008 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
7009 (visium_hard_regno_nregs): New function.
7010 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
7011 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
7012 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
7013 xtensa_hard_regno_nregs): New function.
7014 * system.h (HARD_REGNO_NREGS): Poison.
7016 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7018 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
7019 hard_regno_nregs instead of HARD_REGNO_NREGS.
7020 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
7021 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
7022 (c6x_expand_epilogue): Likewise.
7023 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
7024 (frv_read_iacc_argument): Likewise.
7025 * config/sh/sh.c: Include regs.h.
7026 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7027 (regs_used): Likewise.
7028 (output_stack_adjust): Likewise.
7029 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
7030 * expmed.c: Include regs.h.
7031 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
7032 * ree.c: Include regs.h.
7033 (combine_reaching_defs): Use hard_regno_nregs instead of
7035 (add_removable_extension): Likewise.
7037 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7039 * regs.h (hard_regno_nregs): Turn into a function.
7040 (end_hard_regno): Update accordingly.
7041 * caller-save.c (setup_save_areas): Likewise.
7042 (save_call_clobbered_regs): Likewise.
7043 (replace_reg_with_saved_mem): Likewise.
7044 (insert_restore): Likewise.
7045 (insert_save): Likewise.
7046 * combine.c (can_change_dest_mode): Likewise.
7047 (move_deaths): Likewise.
7048 (distribute_notes): Likewise.
7049 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
7050 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
7051 (rs6000_split_multireg_move): Likewise.
7052 (rs6000_register_move_cost): Likewise.
7053 (rs6000_memory_move_cost): Likewise.
7054 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
7055 (rs6000_split_multireg_move): Likewise.
7056 (rs6000_register_move_cost): Likewise.
7057 (rs6000_memory_move_cost): Likewise.
7058 * cselib.c (cselib_reset_table): Likewise.
7059 (cselib_lookup_1): Likewise.
7060 * emit-rtl.c (set_mode_and_regno): Likewise.
7061 * function.c (aggregate_value_p): Likewise.
7062 * ira-color.c (setup_profitable_hard_regs): Likewise.
7063 (check_hard_reg_p): Likewise.
7064 (calculate_saved_nregs): Likewise.
7065 (assign_hard_reg): Likewise.
7066 (improve_allocation): Likewise.
7067 (calculate_spill_cost): Likewise.
7068 * ira-emit.c (modify_move_list): Likewise.
7069 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
7070 (ira_hard_reg_in_set_p): Likewise.
7071 * ira.c (setup_reg_mode_hard_regset): Likewise.
7072 (clarify_prohibited_class_mode_regs): Likewise.
7073 (check_allocation): Likewise.
7074 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7075 (lra_setup_reg_renumber): Likewise.
7076 (setup_try_hard_regno_pseudos): Likewise.
7077 (spill_for): Likewise.
7078 (assign_hard_regno): Likewise.
7079 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
7080 * lra-constraints.c (in_class_p): Likewise.
7081 (lra_constraint_offset): Likewise.
7082 (simplify_operand_subreg): Likewise.
7083 (lra_constraints): Likewise.
7084 (split_reg): Likewise.
7085 (split_if_necessary): Likewise.
7086 (invariant_p): Likewise.
7087 (inherit_in_ebb): Likewise.
7088 * lra-lives.c (process_bb_lives): Likewise.
7089 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
7090 (get_hard_regs): Likewise.
7091 (do_remat): Likewise.
7092 * lra-spills.c (assign_spill_hard_regs): Likewise.
7093 * mode-switching.c (create_pre_exit): Likewise.
7094 * postreload.c (reload_combine_recognize_pattern): Likewise.
7095 * recog.c (peep2_find_free_register): Likewise.
7096 * regcprop.c (kill_value_regno): Likewise.
7097 (set_value_regno): Likewise.
7098 (copy_value): Likewise.
7099 (maybe_mode_change): Likewise.
7100 (find_oldest_value_reg): Likewise.
7101 (copyprop_hardreg_forward_1): Likewise.
7102 * regrename.c (check_new_reg_p): Likewise.
7103 (regrename_do_replace): Likewise.
7104 * reload.c (push_reload): Likewise.
7105 (combine_reloads): Likewise.
7106 (find_dummy_reload): Likewise.
7107 (operands_match_p): Likewise.
7108 (find_reloads): Likewise.
7109 (find_equiv_reg): Likewise.
7110 (reload_adjust_reg_for_mode): Likewise.
7111 * reload1.c (count_pseudo): Likewise.
7112 (count_spilled_pseudo): Likewise.
7113 (find_reg): Likewise.
7114 (clear_reload_reg_in_use): Likewise.
7115 (free_for_value_p): Likewise.
7116 (allocate_reload_reg): Likewise.
7117 (choose_reload_regs): Likewise.
7118 (reload_adjust_reg_for_temp): Likewise.
7119 (emit_reload_insns): Likewise.
7120 (delete_output_reload): Likewise.
7121 * rtlanal.c (subreg_get_info): Likewise.
7122 * sched-deps.c (sched_analyze_reg): Likewise.
7123 * sel-sched.c (init_regs_for_mode): Likewise.
7124 (mark_unavailable_hard_regs): Likewise.
7125 (choose_best_reg_1): Likewise.
7126 (verify_target_availability): Likewise.
7127 * valtrack.c (dead_debug_insert_temp): Likewise.
7128 * var-tracking.c (track_loc_p): Likewise.
7129 (emit_note_insn_var_location): Likewise.
7130 * varasm.c (make_decl_rtl): Likewise.
7131 * reginfo.c (choose_hard_reg_mode): Likewise.
7132 (init_reg_modes_target): Refer directly to
7133 this_target_regs->x_hard_regno_nregs.
7135 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7137 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
7138 instead of hard_regno_nregs.
7140 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7142 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
7143 end_hard_regno instead of hard_regno_nregs.
7144 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
7145 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
7146 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
7147 * ira-color.c (improve_allocation): Likewise.
7148 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7149 * lra-lives.c (mark_regno_live): Likewise.
7150 (mark_regno_dead): Likewise.
7151 * lra-remat.c (operand_to_remat): Likewise.
7152 * lra.c (collect_non_operand_hard_regs): Likewise.
7153 * postreload.c (reload_combine_note_store): Likewise.
7154 (move2add_valid_value_p): Likewise.
7155 * reload.c (regno_clobbered_p): Likewise.
7157 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7159 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
7161 * config/v850/v850.c (v850_reorg): Likewise.
7162 * reload.c (refers_to_regno_for_reload_p): Likewise.
7163 (find_equiv_reg): Likewise.
7164 * reload1.c (reload_reg_reaches_end_p): Likewise.
7166 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7168 * caller-save.c (add_used_regs): Use REG_NREGS instead of
7170 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
7171 * config/arm/arm.c (output_move_neon): Likewise.
7172 (arm_attr_length_move_neon): Likewise.
7173 (neon_split_vcombine): Likewise.
7174 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
7175 (c6x_mark_reg_written): Likewise.
7176 (c6x_dwarf_register_span): Likewise.
7177 * config/i386/i386.c (ix86_save_reg): Likewise.
7178 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
7179 (rws_access_reg): Likewise.
7180 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
7181 * mode-switching.c (create_pre_exit): Likewise.
7182 * ree.c (combine_reaching_defs): Likewise.
7183 (add_removable_extension): Likewise.
7184 * regcprop.c (find_oldest_value_reg): Likewise.
7185 (copyprop_hardreg_forward_1): Likewise.
7186 * reload.c (reload_inner_reg_of_subreg): Likewise.
7187 (push_reload): Likewise.
7188 (combine_reloads): Likewise.
7189 (find_dummy_reload): Likewise.
7190 (reload_adjust_reg_for_mode): Likewise.
7191 * reload1.c (find_reload_regs): Likewise.
7192 (forget_old_reloads_1): Likewise.
7193 (reload_reg_free_for_value_p): Likewise.
7194 (reload_adjust_reg_for_temp): Likewise.
7195 (emit_reload_insns): Likewise.
7196 (delete_output_reload): Likewise.
7197 * sel-sched.c (choose_best_reg_1): Likewise.
7198 (choose_best_pseudo_reg): Likewise.
7200 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7201 Alan Hayward <alan.hayward@arm.com>
7202 David Sherwood <david.sherwood@arm.com>
7204 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
7205 * target.def (slow_unaligned_access): New hook.
7206 * targhooks.h (default_slow_unaligned_access): Declare.
7207 * targhooks.c (default_slow_unaligned_access): New function.
7208 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
7209 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
7210 * doc/tm.texi: Regenerate.
7211 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
7212 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
7213 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
7215 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
7216 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
7218 (rs6000_slow_unaligned_access): New function.
7219 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7220 (expand_block_compare): Likewise.
7221 (expand_strn_compare): Likewise.
7222 (rs6000_rtx_costs): Likewise.
7223 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
7224 (riscv_slow_unaligned_access): Likewise.
7225 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
7226 (riscv_slow_unaligned_access_p): ...this and make static.
7227 (riscv_option_override): Update accordingly.
7228 (riscv_slow_unaligned_access): New function.
7229 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7230 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
7231 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7232 (rs6000_slow_unaligned_access): New function.
7233 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7234 (rs6000_rtx_costs): Likewise.
7235 * config/rs6000/rs6000-string.c (expand_block_compare)
7236 (expand_strn_compare): Use targetm.slow_unaligned_access instead
7237 of SLOW_UNALIGNED_ACCESS.
7238 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
7239 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
7240 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
7241 of SLOW_UNALIGNED_ACCESS.
7242 * expmed.c (simple_mem_bitfield_p): Likewise.
7243 * expr.c (alignment_for_piecewise_move): Likewise.
7244 (emit_group_load_1): Likewise.
7245 (emit_group_store): Likewise.
7246 (copy_blkmode_from_reg): Likewise.
7247 (emit_push_insn): Likewise.
7248 (expand_assignment): Likewise.
7249 (store_field): Likewise.
7250 (expand_expr_real_1): Likewise.
7251 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
7252 * lra-constraints.c (simplify_operand_subreg): Likewise.
7253 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7254 * gimple-ssa-store-merging.c: Likewise in block comment at start
7256 * tree-ssa-strlen.c: Include target.h.
7257 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
7258 of SLOW_UNALIGNED_ACCESS.
7259 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
7261 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7263 PR rtl-optimization/82185
7264 * expmed.c (emit_store_flag_int): Only test tem if it has been
7267 2017-09-12 Richard Biener <rguenther@suse.de>
7270 * match.pd ((FTYPE) N CMP CST): Fix typo.
7272 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
7274 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
7276 (mips_near_type_p): Add 'short_call' attribute as a synonym
7278 * doc/extend.texi (short_call): Document new function attribute.
7280 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7283 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
7284 assertion check that in the condition.
7285 (get_atomic_generic_size): Likewise. Before testing if parameter
7286 has pointer type, if it has array type, call for C++
7287 default_conversion to perform array-to-pointer conversion.
7289 2017-09-12 Richard Biener <rguenther@suse.de>
7291 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
7292 for operations we cannot scalarize.
7294 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
7296 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
7297 vectors heap vectors. Clean up comments.
7298 Make visited_bbs a reference.
7299 (profitable_jump_thread_path): Make GC
7300 vectors heap vectors. Clean up comments.
7302 (convert_and_register_jump_thread_path): Make GC vectors heap
7304 (check_subpath_and_update_thread_path): Same. Clean up comments.
7305 Make visited_bbs a reference.
7306 (handle_phi): Abstract common code to to
7307 register_jump_thread_path_if_profitable.
7308 Rename VAR_BB to DEF_BB.
7310 Make GC vectors heap vectors.
7311 Make visited_bbs a reference.
7312 (handle_assignment): Same.
7313 (register_jump_thread_path_if_profitable): New.
7314 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
7316 Make GC vectors heap vectors. Clean up comments.
7317 Make visited_bbs a reference.
7318 (find_jump_threads_backwards): Make visited_bbs live in the stack.
7319 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
7322 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
7325 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
7326 words of E_DImode object are reachable by xtensa_uimm8x4 access.
7328 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
7330 Revert r251800 and r251799.
7332 2017-09-11 Martin Jambor <mjambor@suse.cz>
7335 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
7336 arguments in advance.
7337 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
7338 use it to find predecessor edges.
7339 (naive_outof_ssa): Collect vector of predecessors.
7341 2017-09-08 Jason Merrill <jason@redhat.com>
7343 PR c++/70029 - ICE with ref-qualifier and -flto
7344 * langhooks.h (struct lang_hooks_for_types): Add
7345 copy_lang_qualifiers.
7346 * attribs.c (build_type_attribute_qual_variant): Use it.
7347 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
7349 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
7350 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
7352 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
7355 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
7356 (*mulsi3_sp64): New instruction.
7357 (mulsi3): New expander.
7359 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
7361 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
7363 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7365 * sancov.c: Include memmodel.h.
7367 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
7370 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
7373 2017-09-07 Carl Love <cel@us.ibm.com>
7375 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
7376 the sldi instruction.
7378 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
7380 * sancov.c: Include tm_p.h.
7382 2017-09-07 Jakub Jelinek <jakub@redhat.com>
7385 * output.h (switch_to_other_text_partition): New declaration.
7386 * varasm.c (switch_to_other_text_partition): New function.
7387 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
7388 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
7389 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
7390 to the other text partition before emitting LCL label and switch back
7391 after emitting the word after it.
7393 2017-09-07 Richard Biener <rguenther@suse.de>
7395 * passes.def (pass_split_crit_edges): Remove instance before PRE.
7396 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
7397 critical edges here, after loop init.
7398 (pass_data_pre): Remove PROP_no_crit_edges flags.
7399 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
7400 for valueization of call args to avoid leaking VN_TOP.
7401 (visit_use): Assert we do not visit default defs.
7402 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
7403 Use error_mark_node to more easily detect leaking VN_TOP.
7404 All default-defs are varying, not VN_TOP. Mark them visited.
7405 (run_scc_vn): Make code match comment.
7407 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7409 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
7410 OPTION_MASK_FLOAT128_KEYWORD.
7411 (POWERPC_MASKS): Likewise.
7412 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
7413 support for the -mfloat128-type option, and make -mfloat128
7414 default on PowerPC Linux systems. Define or undefine
7415 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
7416 Define __float128 to be __ieee128 if IEEE 128-bit support is
7417 enabled, or undefine it.
7418 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
7419 Delete defining __FLOAT128_TYPE__.
7420 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
7421 -mfloat128-type option and make -mfloat128 default on PowerPC
7423 (TARGET_FLOAT128_TYPE): Likewise.
7424 (-mfloat128-type): Likewise.
7425 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7426 Delete the -mfloat128-type option and make -mfloat128 default on
7427 PowerPC Linux systems. Always use __ieee128 to be the keyword for
7428 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
7429 128-bit floating point is enabled. Change tests from using
7430 -mfloat128-type to -mfloat128.
7431 (rs6000_mangle_type): Use the correct mangling for the __float128
7432 type even if normal long double is restricted to 64-bits.
7433 (floatn_mode): Enable the _Float128 type by default on VSX Linux
7435 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
7436 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
7437 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
7439 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
7440 documentation for -mfloat128.
7442 2017-09-06 Olivier Hainque <hainque@adacore.com>
7444 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
7446 2017-09-06 Wish Wu <wishwu007@gmail.com>
7447 Jakub Jelinek <jakub@redhat.com>
7449 * asan.c (initialize_sanitizer_builtins): Add
7450 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
7451 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
7452 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
7453 BT_FN_VOID_UINT64_PTR variables.
7454 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
7455 (BT_FN_VOID_UINT16_UINT16): Likewise.
7456 (BT_FN_VOID_UINT32_UINT32): Likewise.
7457 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
7458 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
7459 (BT_FN_VOID_UINT64_PTR): Likewise.
7460 * common.opt (flag_sanitize_coverage): New variable.
7461 (fsanitize-coverage=trace-pc): Remove.
7462 (fsanitize-coverage=): Add.
7463 * flag-types.h (enum sanitize_coverage_code): New enum.
7464 * fold-const.c (fold_range_test): Disable non-short-circuit
7465 optimization if flag_sanitize_coverage.
7466 (fold_truth_andor): Likewise.
7467 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7468 * opts.c (COVERAGE_SANITIZER_OPT): Define.
7469 (coverage_sanitizer_opts): New array.
7470 (get_closest_sanitizer_option): Add OPTS argument, handle also
7471 OPT_fsanitize_coverage_.
7472 (parse_sanitizer_options): Adjusted to also handle
7473 OPT_fsanitize_coverage_.
7474 (common_handle_option): Add OPT_fsanitize_coverage_.
7475 * sancov.c (instrument_comparison, instrument_switch): New function.
7476 (sancov_pass): Add trace-cmp support.
7477 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
7478 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
7479 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
7480 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
7481 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
7482 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
7483 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
7484 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
7485 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
7486 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
7488 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7490 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7491 error. Only quit immediately if parsing is complete.
7492 (BEGIN): Initialize fatal_err and parse_done.
7493 (begin fpu, end fpu): Check number of arguments.
7494 (begin arch, end arch): Likewise.
7495 (begin cpu, end cpu): Likewise.
7496 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7497 (optalias): Likewise.
7499 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7501 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7502 * config/arm/arm-isa.h: Delete. Move definitions to ...
7503 * arm-cpus.in: ... here. Use new feature and fgroup values.
7504 * config/arm/arm.c (arm_option_override): Use lower case for feature
7506 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7507 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7508 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7510 (print_isa_bits_for): New function.
7511 (gen_isa): New function.
7512 (gen_comm_data): Use print_isa_bits_for.
7513 (define feature): New keyword.
7514 (define fgroup): New keyword.
7515 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
7516 (arm-isa.h): Add rule to generate file.
7517 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7518 case for feature bit names.
7520 2017-09-06 Richard Biener <rguenther@suse.de>
7522 * tree-ssa-pre.c (NECESSARY): Remove.
7523 (create_expression_by_pieces): Do not touch pass-local flags.
7524 (insert_into_preds_of_block): Likewise.
7525 (do_pre_regular_insertion): Likewise.
7526 (eliminate_insert): Likewise.
7527 (eliminate_dom_walker::before_dom_children): Likewise.
7528 (fini_eliminate): Do not look at inserted_exprs.
7529 (mark_operand_necessary): Remove.
7530 (remove_dead_inserted_code): Replace with simple work-list
7531 algorithm based on inserted_exprs and SSA uses.
7532 (pass_pre::execute): Re-order fini_eliminate and
7533 remove_dead_inserted_code.
7535 2017-09-06 Olivier Hainque <hainque@adacore.com>
7537 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
7538 for VxWorks 7. Adjust surrounding comments.
7540 2017-09-06 Richard Biener <rguenther@suse.de>
7542 * gimple-ssa-strength-reduction.c
7543 (find_candidates_dom_walker::before_dom_children): Also allow
7546 2017-09-06 Richard Biener <rguenther@suse.de>
7548 PR tree-optimization/82108
7549 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
7550 for gap in the non-permutation SLP case.
7552 2017-09-06 Martin Jambor <mjambor@suse.cz>
7554 PR tree-optimization/82078
7555 * tree-sra.c (sort_and_splice_var_accesses): Move call to
7556 add_access_to_work_queue...
7557 (build_accesses_from_assign): ...here.
7558 (propagate_all_subaccesses): Make sure racc is the group
7559 representative, if there is one.
7561 2017-09-06 Jakub Jelinek <jakub@redhat.com>
7564 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
7567 2017-09-06 Richard Biener <rguenther@suse.de>
7569 * gimple-ssa-strength-reduction.c
7570 (find_candidates_dom_walker::before_doom_children): Use a
7571 type and not a mode check.
7573 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
7576 * config/arm/predicates.md (arm_general_adddi_operand): Create new
7578 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
7580 2017-09-05 Jeff Law <law@redhat.com>
7582 PR tree-optimization/64910
7583 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
7584 cases where we have 3 or more operands.
7586 2017-09-05 Jakub Jelinek <jakub@redhat.com>
7589 * omp-low.c (lower_omp_for): Recompute tree invariant if
7590 gimple_omp_for_initial/final is ADDR_EXPR.
7593 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
7594 into gimple val before gimplification fo the COND_EXPR.
7596 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
7598 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
7599 REGION_COPY argument.
7600 (thread_through_all_blocks): Remove unused argument to
7601 duplicate_thread_path.
7603 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7604 Alan Hayward <alan.hayward@arm.com>
7605 David Sherwood <david.sherwood@arm.com>
7607 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
7608 Take a scalar_mode rather than a machine_mode.
7609 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7610 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
7611 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7612 (aarch64_gen_adjusted_ldpstp): Likewise.
7613 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
7615 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7616 Alan Hayward <alan.hayward@arm.com>
7617 David Sherwood <david.sherwood@arm.com>
7619 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
7620 Take a scalar_int_mode instead of a machine_mode.
7621 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7622 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7623 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7624 (aarch64_simd_attr_length_rglist): Delete.
7625 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
7626 a scalar_int_mode instead of a machine_mode.
7627 (aarch64_add_offset): Likewise.
7628 (aarch64_internal_mov_immediate): Likewise
7629 (aarch64_add_constant_internal): Likewise.
7630 (aarch64_add_constant): Likewise.
7631 (aarch64_movw_imm): Likewise.
7632 (aarch64_rtx_arith_op_extract_p): Likewise.
7633 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7634 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7635 Remove assert that the mode isn't a vector.
7636 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7637 (aarch64_expand_mov_immediate): Update calls after above changes.
7638 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
7639 (aarch64_and_bitmask_imm): Check for scalar integer modes.
7640 (aarch64_move_imm): Likewise.
7641 (aarch64_can_const_movi_rtx_p): Likewise.
7642 (aarch64_strip_extend): Likewise.
7643 (aarch64_extr_rtx_p): Likewise.
7644 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
7645 a CONST_INT when the mode parameter is VOIDmode.
7646 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
7648 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7650 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
7651 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7652 (bitwise_type_for_mode): Update accordingly.
7654 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7656 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
7657 * stor-layout.c (mode_for_size_tree): Likewise.
7658 (mode_for_array): Update accordingly.
7659 (layout_decl): Likewise.
7660 (compute_record_mode): Likewise. Only set the mode once.
7662 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7664 * target.def (get_mask_mode): Change return type to opt_mode.
7666 * doc/tm.texi: Regenerate.
7667 * targhooks.h (default_get_mask_mode): Return an opt_mode.
7668 * targhooks.c (default_get_mask_mode): Likewise.
7669 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7670 * optabs-query.c (can_vec_mask_load_store_p): Update use of
7671 targetm.get_mask_mode.
7672 * tree.c (build_truth_vector_type): Likewise.
7674 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7676 * machmode.h (mode_for_vector): Return an opt_mode.
7677 * stor-layout.c (mode_for_vector): Likewise.
7678 (mode_for_int_vector): Update accordingly.
7679 (layout_type): Likewise.
7680 * config/i386/i386.c (emit_memmov): Likewise.
7681 (ix86_expand_set_or_movmem): Likewise.
7682 (ix86_expand_vector_init): Likewise.
7683 (ix86_get_mask_mode): Likewise.
7684 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
7686 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
7687 * expmed.c (extract_bit_field_1): Likewise.
7688 * expr.c (expand_expr_real_2): Likewise.
7689 * optabs-query.c (can_vec_perm_p): Likewise.
7690 (can_vec_mask_load_store_p): Likewise.
7691 * optabs.c (expand_vec_perm): Likewise.
7692 * targhooks.c (default_get_mask_mode): Likewise.
7693 * tree-vect-stmts.c (vectorizable_store): Likewise.
7694 (vectorizable_load): Likewise.
7695 (get_vectype_for_scalar_type_and_size): Likewise.
7697 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7699 * machmode.h (mode_for_int_vector): New function.
7700 * stor-layout.c (mode_for_int_vector): Likewise.
7701 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
7702 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
7703 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
7704 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
7705 (s390_expand_vcond): Likewise.
7707 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7709 * machmode.h (opt_machine_mode): New type.
7710 (opt_mode<T>): Allow construction from anything that can be
7712 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
7713 (mode_for_size): Return an opt_machine_mode.
7714 * stor-layout.c (mode_for_size): Likewise.
7715 (mode_for_size_tree): Update call accordingly.
7716 (bitwise_mode_for_mode): Likewise.
7717 (make_fract_type): Likewise.
7718 (make_accum_type): Likewise.
7719 * caller-save.c (replace_reg_with_saved_mem): Update call
7721 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7722 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7723 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7724 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7725 * expmed.c (extract_bit_field_1): Likewise.
7726 * reload.c (get_secondary_mem): Likewise.
7727 * varasm.c (assemble_integer): Likewise.
7728 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
7731 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7733 * machmode.h (decimal_float_mode_for_size): New function.
7734 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
7735 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
7736 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
7737 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
7738 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
7739 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
7741 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7743 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
7744 (get_builtin_sync_mode): Likewise.
7745 (expand_ifn_atomic_compare_exchange): Likewise.
7746 (expand_builtin_atomic_clear): Likewise.
7747 (expand_builtin_atomic_test_and_set): Likewise.
7748 (fold_builtin_atomic_always_lock_free): Likewise.
7749 * calls.c (compute_argument_addresses): Likewise.
7750 (emit_library_call_value_1): Likewise.
7751 (store_one_arg): Likewise.
7752 * combine.c (combine_instructions): Likewise.
7753 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
7754 * config/arm/arm.c (arm_function_value): Likewise.
7755 (aapcs_allocate_return_reg): Likewise.
7756 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
7757 * config/i386/i386.c (construct_container): Likewise.
7758 (ix86_gimplify_va_arg): Likewise.
7759 (ix86_expand_sse_cmp): Likewise.
7760 (emit_memmov): Likewise.
7761 (emit_memset): Likewise.
7762 (expand_small_movmem_or_setmem): Likewise.
7763 (ix86_expand_pextr): Likewise.
7764 (ix86_expand_pinsr): Likewise.
7765 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
7766 * config/microblaze/microblaze.c (microblaze_block_move_straight):
7768 * config/mips/mips.c (mips_function_value_1) Likewise.
7769 (mips_block_move_straight): Likewise.
7770 (mips_expand_ins_as_unaligned_store): Likewise.
7771 * config/powerpcspe/powerpcspe.c
7772 (rs6000_darwin64_record_arg_advance_flush): Likewise.
7773 (rs6000_darwin64_record_arg_flush): Likewise.
7774 * config/rs6000/rs6000.c
7775 (rs6000_darwin64_record_arg_advance_flush): Likewise.
7776 (rs6000_darwin64_record_arg_flush): Likewise.
7777 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
7778 (sparc_function_value_1): Likewise.
7779 * config/spu/spu.c (adjust_operand): Likewise.
7780 (spu_emit_branch_or_set): Likewise.
7781 (arith_immediate_p): Likewise.
7782 * emit-rtl.c (gen_lowpart_common): Likewise.
7783 * expr.c (expand_expr_real_1): Likewise.
7784 * function.c (assign_parm_setup_block): Likewise.
7785 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7786 * reload1.c (alter_reg): Likewise.
7787 * stor-layout.c (mode_for_vector): Likewise.
7788 (layout_type): Likewise.
7790 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7792 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
7793 (spu_convert_move): Likewise.
7794 * lower-subreg.c (resolve_simple_move): Likewise.
7796 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7799 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
7800 define_insn to a define_expand.
7801 (altivec_vsum2sws_direct): New define_insn.
7802 (altivec_vsumsws): Convert from a define_insn to a define_expand.
7804 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7806 * config/arm/arm.c (arm_option_params_internal): Improve setting of
7809 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
7814 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
7815 realign stack if stack alignment needed is less than incoming
7818 2017-09-05 Marek Polacek <polacek@redhat.com>
7821 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
7824 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7826 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
7828 2017-09-05 Richard Biener <rguenther@suse.de>
7830 PR tree-optimization/82084
7831 * fold-const.c (can_native_encode_string_p): Handle wide characters.
7833 2017-09-05 Richard Biener <rguenther@suse.de>
7835 PR tree-optimization/82102
7836 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
7838 2017-09-05 Martin Liska <mliska@suse.cz>
7840 PR tree-optimization/82032
7841 * tree-cfg.c (generate_range_test): New function.
7842 * tree-cfg.h (generate_range_test): Declared here.
7843 * tree-cfgcleanup.c (convert_single_case_switch): New function.
7844 (cleanup_control_expr_graph): Use it.
7845 * tree-switch-conversion.c (try_switch_expansion): Remove
7847 (emit_case_nodes): Use generate_range_test.
7849 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7852 * config/i386/i386.md (*<btsc><mode>_mask): Add
7853 TARGET_USE_BT to insn constraint.
7854 (*btr<mode>_mask): Ditto.
7856 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
7858 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
7859 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
7861 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
7864 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
7865 TARGET_NEON and TARGET_IWMMXT.
7866 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
7867 TARGET_NEON and TARGET_IWMMXT.
7868 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
7870 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7872 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
7873 (ix86_rewrite_tls_address): Ditto.
7874 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
7875 (ix86_rewrite_tls_address_1): Ditto.
7876 (ix86_rewrite_tls_address): Ditto.
7877 * config/i386/predicates.md (tls_address_pattern): New predicate.
7878 * config/i386/i386.md (TLS address splitter): New splitter.
7880 2017-09-04 Richard Biener <rguenther@suse.de>
7882 PR tree-optimization/82084
7883 * fold-const.h (can_native_encode_string_p): Declare.
7884 * fold-const.c (can_native_encode_string_p): Factor out from ...
7885 (native_encode_string): ... here.
7886 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
7887 vectorizing stores from constants we later cannot handle.
7889 2017-09-04 Marek Polacek <polacek@redhat.com>
7892 * doc/invoke.texi: Update -Wtautological-compare documentation.
7894 2017-09-04 Jeff Law <law@redhat.com>
7896 PR tree-optimization/64910
7897 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
7898 swap the first and last operand if the last is a constant.
7900 2017-09-04 Marek Polacek <polacek@redhat.com>
7903 * convert.c (do_narrow): When sanitizing signed integer overflows,
7904 bail out for signed types.
7905 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
7907 2017-09-04 Richard Biener <rguenther@suse.de>
7909 PR tree-optimization/82060
7910 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
7911 Move devirtualization after stmt folding and before EH/AB/noreturn
7912 cleanup to get the stmt refs canonicalized. Use a bool instead
7913 of gimple_modified_p since that doesn't work for NOPs. Schedule
7914 NOPs generated by folding for removal.
7916 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7917 Alan Hayward <alan.hayward@arm.com>
7918 David Sherwood <david.sherwood@arm.com>
7920 * coretypes.h (pad_direction): New enum.
7921 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
7922 (FUNCTION_ARG_PADDING): Likewise.
7923 * target.def (function_arg_padding): New hook.
7924 * targhooks.h (default_function_arg_padding): Declare.
7925 * targhooks.c (default_function_arg_padding): New function.
7926 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
7927 (TARGET_FUNCTION_ARG_PADDING): ...this.
7928 * doc/tm.texi: Regenerate.
7929 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
7930 instead of direction.
7931 (compute_argument_addresses): Likewise.
7932 (load_register_parameters): Likewise.
7933 (emit_library_call_value_1): Likewise.
7934 (store_one_arg): Use targetm.calls.function_arg_padding instead
7935 of FUNCTION_ARG_PADDING.
7936 (must_pass_in_stack_var_size_or_pad): Likewise.
7937 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
7938 (emit_group_store): Likewise.
7939 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
7940 instead of FUNCTION_ARG_PADDING.
7941 (emit_push_insn): Likewise, and propagate enum change throughout
7943 * function.h (direction): Delete.
7944 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
7946 * function.c (assign_parm_find_stack_rtl): Likewise.
7947 (assign_parm_setup_block_p): Likewise.
7948 (assign_parm_setup_block): Likewise.
7949 (gimplify_parameters): Likewise.
7950 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
7951 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
7953 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
7954 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7955 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
7956 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
7957 (aarch64_function_arg_padding): ...this new function.
7958 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
7959 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7960 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
7961 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7962 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
7963 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7964 (arm_pad_arg_upward): Replace with...
7965 (arm_function_arg_padding): ...this new function.
7966 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
7968 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
7969 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
7970 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7971 (ia64_hpux_function_arg_padding): Replace with...
7972 (ia64_function_arg_padding): ...this new function. Use pad_direction
7973 instead of direction. Check for TARGET_HPUX.
7974 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
7975 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7976 (iq2000_function_arg_padding): New function.
7977 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
7978 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
7979 (mips_function_arg_padding): ...this new function.
7980 (mips_pad_reg_upward): Update accordingly.
7981 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7982 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
7983 targetm.calls.function_arg_padding.
7984 (FUNCTION_ARG_PADDING): Delete.
7985 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7986 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
7987 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7988 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
7989 (nios2_block_reg_padding): Return pad_direction instead of direction.
7990 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
7991 instead of direction.
7992 (nios2_function_arg_padding): Likewise. Make static.
7993 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7994 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
7995 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
7996 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
7997 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7998 (pa_function_arg_padding): Make static. Return pad_direction instead
8000 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
8001 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8002 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
8003 instead of direction. Use targetm.calls.function_arg_padding.
8004 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
8005 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8006 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
8007 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
8008 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
8010 (function_arg_padding): Rename to...
8011 (rs6000_function_arg_padding): ...this. Make static. Return
8012 pad_direction instead of direction.
8013 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8014 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
8015 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
8016 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
8017 instead of direction. Use targetm.calls.function_arg_padding.
8018 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
8019 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
8020 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
8021 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
8022 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8023 (function_arg_padding): Rename to...
8024 (rs6000_function_arg_padding): ...this. Make static. Return
8025 pad_direction instead of direction.
8026 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
8027 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
8028 * config/s390/s390.c (s390_function_arg_padding): New function.
8029 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8030 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
8031 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
8032 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
8033 (function_arg_padding): Rename to...
8034 (sparc_function_arg_padding): ...this. Make static. Return
8035 pad_direction instead of direction.
8036 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
8037 * config/spu/spu.c (spu_function_arg_padding): New function.
8038 (TARGET_FUNCTION_ARG_PADDING): Redefine.
8039 * system.h (FUNCTION_ARG_PADDING): Poison.
8041 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8042 Alan Hayward <alan.hayward@arm.com>
8043 David Sherwood <david.sherwood@arm.com>
8045 * target.def (modes_tieable_p): New hook.
8046 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
8047 (TARGET_MODES_TIEABLE_P): ...this.
8048 * doc/tm.texi.in: Regenerate.
8049 * hooks.h (hook_bool_mode_mode_true): Declare.
8050 * hooks.c (hook_bool_mode_mode_true): New function.
8051 * combine.c (subst): Use targetm.modes_tieable_p instead of
8053 * dse.c (find_shift_sequence): Likewise.
8054 * expmed.c (extract_low_bits): Likewise.
8055 * lower-subreg.c: Include target.h.
8056 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
8058 * rtlanal.c (rtx_cost): Likewise.
8059 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
8060 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
8061 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
8062 (TARGET_MODES_TIEABLE_P): Redefine.
8063 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
8064 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
8065 (TARGET_MODES_TIEABLE_P): Redefine.
8066 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
8067 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
8068 (arc_modes_tieable_p): New function.
8069 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
8070 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
8071 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
8072 (arm_modes_tieable_p): Make static.
8073 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
8074 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
8075 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
8076 (TARGET_MODES_TIEABLE_P): Redefine.
8077 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
8078 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
8079 (TARGET_MODES_TIEABLE_P): Redefine.
8080 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
8081 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
8082 (cr16_modes_tieable_p): New function.
8083 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
8084 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
8085 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
8086 (TRULY_NOOP_TRUNCATION): Update comment.
8087 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
8088 (TRULY_NOOP_TRUNCATION): Update comment.
8089 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
8090 (frv_modes_tieable_p): New function.
8091 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
8092 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
8093 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
8094 (TARGET_MODES_TIEABLE_P): Redefine.
8095 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
8096 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
8097 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
8098 (TARGET_MODES_TIEABLE_P): Redefine.
8099 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
8100 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
8101 (ia64_modes_tieable_p): New function.
8102 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
8103 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
8104 (iq2000_modes_tieable_p): New function.
8105 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
8106 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
8107 (lm32_modes_tieable_p): New function.
8108 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
8109 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
8110 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
8111 (TARGET_MODES_TIEABLE_P): Redefine.
8112 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
8113 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
8114 (m32r_modes_tieable_p): New function.
8115 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
8116 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
8117 (m68k_modes_tieable_p): New function.
8118 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
8119 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
8120 (mcore_modes_tieable_p): New function.
8121 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
8122 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
8124 (TARGET_MODES_TIEABLE_P): Redefine.
8125 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
8126 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
8127 * config/mips/mips.c (mips_modes_tieable_p): Make static.
8128 (TARGET_MODES_TIEABLE_P): Redefine.
8129 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
8130 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
8131 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
8132 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
8133 (mn10300_modes_tieable_p): ...this and make static.
8134 (TARGET_MODES_TIEABLE_P): Redefine.
8135 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
8136 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
8137 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
8138 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
8139 (msp430_modes_tieable_p): Make static.
8140 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
8141 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
8142 (TARGET_MODES_TIEABLE_P): Redefine.
8143 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
8144 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
8145 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
8146 (TARGET_MODES_TIEABLE_P): Redefine.
8147 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
8148 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
8149 * config/pa/pa.c (pa_modes_tieable_p): Make static.
8150 (TARGET_MODES_TIEABLE_P): Redefine.
8151 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
8152 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
8153 (pdp11_modes_tieable_p): New function.
8154 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
8155 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
8156 (rs6000_modes_tieable_p): New function.
8157 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8158 * config/powerpcspe/powerpcspe.md: Update comment.
8159 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
8160 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
8161 (TARGET_MODES_TIEABLE_P): Redefine.
8162 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
8163 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
8164 (rl78_modes_tieable_p): New function.
8165 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
8166 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
8167 (rs6000_modes_tieable_p): New function.
8168 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8169 * config/rs6000/rs6000.md: Update comment.
8170 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
8171 * config/rx/rx.c (rx_modes_tieable_p): New function.
8172 (TARGET_MODES_TIEABLE_P): Redefine.
8173 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
8174 * config/s390/s390.c (s390_modes_tieable_p): New function.
8175 (TARGET_MODES_TIEABLE_P): Redefine.
8176 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
8177 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
8178 (sh_modes_tieable_p): New function.
8179 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
8180 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
8181 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
8182 (sparc_modes_tieable_p): Make static.
8183 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
8184 * config/spu/spu.c (spu_modes_tieable_p): New function.
8185 (TARGET_MODES_TIEABLE_P): Redefine.
8186 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
8187 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
8188 (TARGET_MODES_TIEABLE_P): Redefine.
8189 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
8190 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
8191 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
8192 * config/v850/v850.c (v850_modes_tieable_p): New function.
8193 (TARGET_MODES_TIEABLE_P): Redefine.
8194 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
8195 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
8196 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
8197 (visium_modes_tieable_p): New function.
8198 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
8199 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
8200 (xtensa_modes_tieable_p): New function.
8201 * system.h (MODES_TIEABLE_P): Poison.
8203 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8204 Alan Hayward <alan.hayward@arm.com>
8205 David Sherwood <david.sherwood@arm.com>
8207 * target.def (hard_regno_mode_ok): New hook.
8208 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
8209 (TARGET_HARD_REGNO_MODE_OK): ...this.
8210 * doc/tm.texi.in: Regenerate.
8211 * hooks.h (hook_bool_uint_mode_true): Declare.
8212 * hooks.c (hook_bool_uint_mode_true): New function.
8213 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
8215 * genpreds.c (write_insn_preds_c): Add an include of target.h.
8216 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
8217 instead of HARD_REGNO_MODE_OK.
8218 * caller-save.c: Include target.h.
8219 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
8221 * combine.c (can_combine_p): Likewise.
8222 (combinable_i3pat): Likewise.
8223 (can_change_dest_mode): Likewise.
8224 * expr.c (init_expr_target): Likewise.
8225 (convert_move): Likewise.
8226 (convert_modes): Likewise.
8227 * ira.c (setup_prohibited_class_mode_regs): Likewise.
8228 (setup_prohibited_mode_move_regs): Likewise.
8229 * ira.h (target_ira): Likewise.
8230 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8231 * lra-constraints.c (process_alt_operands): Likewise.
8232 (split_reg): Likewise.
8233 * recog.c (peep2_find_free_register): Likewise.
8234 * ree.c (combine_reaching_defs): Likewise.
8235 * regcprop.c (maybe_mode_change): Likewise.
8236 * reginfo.c (init_reg_sets_1): Likewise.
8237 (choose_hard_reg_mode): Likewise.
8238 (simplifiable_subregs): Likewise.
8239 * regrename.c (check_new_reg_p): Likewise.
8240 * reload.c (find_valid_class): Likewise.
8241 (find_valid_class_1): Likewise.
8242 (reload_inner_reg_of_subreg): Likewise.
8243 (push_reload): Likewise.
8244 (combine_reloads): Likewise.
8245 (find_dummy_reload): Likewise.
8246 (find_reloads): Likewise.
8247 * reload1.c (find_reg): Likewise.
8248 (set_reload_reg): Likewise.
8249 (allocate_reload_reg): Likewise.
8250 (choose_reload_regs): Likewise.
8251 (reload_adjust_reg_for_temp): Likewise.
8252 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
8253 (simplify_subreg_regno): Likewise.
8254 * sel-sched.c (init_regs_for_mode): Likewise.
8255 * varasm.c (make_decl_rtl): Likewise.
8256 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
8257 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
8259 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
8260 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
8261 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8262 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
8263 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
8264 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8265 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
8266 (arc_mode_class): Delete.
8267 (HARD_REGNO_MODE_OK): Delete.
8268 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8269 (arc_hard_regno_mode_ok): Rename old array to...
8270 (arc_hard_regno_mode_ok_modes): ...this.
8271 (arc_conditional_register_usage): Update accordingly.
8272 (arc_mode_class): Make static.
8273 (arc_hard_regno_mode_ok): New function.
8274 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
8275 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
8276 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8277 (arm_hard_regno_mode_ok): Make static.
8278 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
8280 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
8281 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
8282 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
8284 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8285 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
8286 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
8287 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
8288 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
8289 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8290 * config/bfin/predicates.md (valid_reg_operand): Use
8291 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
8292 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
8293 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
8294 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8295 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
8296 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
8297 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8298 (cr16_hard_regno_mode_ok): Make static and return a bool.
8299 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
8300 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8301 (cris_hard_regno_mode_ok): New function.
8302 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
8303 (epiphany_mode_class): Delete.
8304 (HARD_REGNO_MODE_OK): Delete.
8305 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
8306 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8307 (hard_regno_mode_ok): Rename to...
8308 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
8309 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
8310 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
8312 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
8313 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
8314 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8315 (frv_hard_regno_mode_ok): Make static and return a bool.
8316 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
8318 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
8319 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
8320 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
8321 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
8323 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8324 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
8325 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
8326 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
8328 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8329 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
8330 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8331 (ia64_hard_regno_mode_ok): New function.
8332 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
8333 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8334 (iq2000_hard_regno_mode_ok): New function.
8335 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
8336 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8337 (lm32_hard_regno_mode_ok): New function.
8338 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
8339 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
8340 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
8341 instead of HARD_REGNO_MODE_OK.
8342 (m32c_hard_regno_ok): Rename to...
8343 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
8344 (m32c_cannot_change_mode_class): Update accordingly.
8345 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8346 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
8347 (m32r_mode_class): Delete.
8348 (HARD_REGNO_MODE_OK): Delete.
8349 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8350 (m32r_hard_regno_mode_ok): Rename to...
8351 (m32r_hard_regno_modes): ...this.
8352 (m32r_mode_class): Make static.
8353 (m32r_hard_regno_mode_ok): New function.
8354 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
8355 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
8356 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8357 (m68k_hard_regno_mode_ok): Make static.
8358 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
8359 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8360 (mcore_hard_regno_mode_ok): New function.
8361 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
8362 (HARD_REGNO_MODE_OK): Delete.
8363 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
8365 (microblaze_hard_regno_mode_ok_p): ...this and make static.
8366 (microblaze_hard_regno_mode_ok): New function.
8367 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8368 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
8369 (mips_hard_regno_mode_ok): Delete.
8370 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
8371 (mips_hard_regno_mode_ok_p): ...this and make static.
8372 (mips_hard_regno_mode_ok_p): Rename to...
8373 (mips_hard_regno_mode_ok_uncached): ...this.
8374 (mips_hard_regno_mode_ok): New function.
8375 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
8376 of HARD_REGNO_MODE_OK.
8377 (mips_option_override): Update after above name changes.
8378 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8379 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
8380 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
8381 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
8382 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
8383 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8384 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
8385 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
8386 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
8387 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8388 (msp430_hard_regno_mode_ok): Make static and return a bool.
8389 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
8390 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
8391 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
8393 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8394 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
8395 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
8396 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
8397 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
8398 (PA_HARD_REGNO_MODE_OK): ...this
8399 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
8400 (PA_HARD_REGNO_MODE_OK): ...this.
8401 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8402 (pa_hard_regno_mode_ok): New function.
8403 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
8404 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8405 (pdp11_hard_regno_mode_ok): New function.
8406 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
8407 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
8409 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
8411 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8412 (rs6000_hard_regno_mode_ok): Rename to...
8413 (rs6000_hard_regno_mode_ok_uncached): ...this.
8414 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8415 (rs6000_hard_regno_mode_ok): New function.
8416 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
8417 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
8418 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
8419 (riscv_hard_regno_mode_ok): ...this and make static.
8420 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8421 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
8422 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
8423 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8424 (rl78_hard_regno_mode_ok): Make static and return bool.
8425 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
8426 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
8428 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
8429 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8430 (rs6000_hard_regno_mode_ok): Rename to...
8431 (rs6000_hard_regno_mode_ok_uncached): ...this.
8432 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8433 (rs6000_hard_regno_mode_ok): New function.
8434 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
8435 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
8436 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8437 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
8438 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
8439 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
8440 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8441 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
8442 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
8443 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8444 (sh_hard_regno_mode_ok): Make static.
8445 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
8446 instead of HARD_REGNO_MODE_OK.
8447 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
8448 (sparc_mode_class): Delete.
8449 (HARD_REGNO_MODE_OK): Delete.
8450 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8451 (hard_regno_mode_classes): Make static.
8452 (sparc_mode_class): Likewise.
8453 (sparc_hard_regno_mode_ok): New function.
8454 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
8455 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
8456 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
8458 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8459 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
8460 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
8461 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
8462 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
8463 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8464 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
8465 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
8466 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8467 (visium_hard_regno_mode_ok): New function.
8468 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
8469 instead of HARD_REGNO_MODE_OK.
8470 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
8471 (HARD_REGNO_MODE_OK): Delete.
8472 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
8473 (xtensa_hard_regno_mode_ok_p): ...this and make static.
8474 (xtensa_option_override): Update accordingly.
8475 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8476 (xtensa_hard_regno_mode_ok): New function.
8477 * system.h (HARD_REGNO_MODE_OK): Poison.
8479 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8480 Alan Hayward <alan.hayward@arm.com>
8481 David Sherwood <david.sherwood@arm.com>
8483 * target.def (hard_regno_call_part_clobbered): New hook.
8484 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
8485 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
8486 * doc/tm.texi: Regenerate.
8487 * hooks.h (hook_bool_uint_mode_false): Declare.
8488 * hooks.c (hook_bool_uint_mode_false): New function.
8489 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8490 * cselib.c (cselib_process_insn): Use
8491 targetm.hard_regno_call_part_clobbered instead of
8492 HARD_REGNO_CALL_PART_CLOBBERED.
8493 * ira-conflicts.c (ira_build_conflicts): Likewise.
8494 * ira-costs.c (ira_tune_allocno_costs): Likewise.
8495 * lra-constraints.c (need_for_call_save_p): Likewise.
8496 * lra-lives.c: Include target.h.
8497 (check_pseudos_live_through_calls): Use
8498 targetm.hard_regno_call_part_clobbered instead of
8499 HARD_REGNO_CALL_PART_CLOBBERED.
8500 * regcprop.c: Include target.h.
8501 (copyprop_hardreg_forward_1): Use
8502 targetm.hard_regno_call_part_clobbered instead of
8503 HARD_REGNO_CALL_PART_CLOBBERED.
8504 * reginfo.c (choose_hard_reg_mode): Likewise.
8505 * regrename.c (check_new_reg_p): Likewise.
8506 * reload.c (find_equiv_reg): Likewise.
8507 * reload1.c (emit_reload_insns): Likewise.
8508 * sched-deps.c (deps_analyze_insn): Likewise.
8509 * sel-sched.c (init_regs_for_mode): Likewise.
8510 (mark_unavailable_hard_regs): Likewise.
8511 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
8512 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8513 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
8515 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8516 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8517 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
8519 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
8521 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8522 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8523 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
8525 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8526 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8527 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
8529 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8530 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
8532 * config/powerpcspe/powerpcspe.c
8533 (rs6000_hard_regno_call_part_clobbered): New function.
8534 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8535 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8536 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
8538 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8539 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8540 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
8542 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8543 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8544 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
8546 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8547 Alan Hayward <alan.hayward@arm.com>
8548 David Sherwood <david.sherwood@arm.com>
8550 * rtl.h (subreg_memory_offset): Declare.
8551 * emit-rtl.c (subreg_memory_offset): New function.
8552 * expmed.c (store_bit_field_1): Use it.
8553 * expr.c (undefined_operand_subword_p): Likewise.
8554 * simplify-rtx.c (simplify_subreg): Likewise.
8556 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
8558 PR rtl-optimization/57448
8561 * optabs.c (expand_atomic_load): Place compiler memory barriers if
8562 using atomic_load pattern.
8563 (expand_atomic_store): Likewise.
8565 2017-09-04 Jakub Jelinek <jakub@redhat.com>
8568 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
8569 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
8570 handling. Use replace_call_with_value with NULL instead of
8571 gsi_replace, unlink_stmt_vdef and release_defs.
8573 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
8576 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
8578 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8581 * rtl.h (emit_library_call_value_1): Declare.
8582 (emit_library_call): Replace declaration with a series of overloads.
8583 Remove the parameter count argument.
8584 (emit_library_call_value): Likewise.
8585 * calls.c (emit_library_call_value_1): Make global. Replace varargs
8586 with an "rtx_mode_t *".
8587 (emit_library_call_value): Delete.
8588 (emit_library_call): Likewise.
8589 * asan.c (asan_emit_stack_protection): Update calls accordingly.
8590 (asan_emit_allocas_unpoison): Likewise.
8591 * builtins.c (expand_builtin_powi): Likewise.
8592 (expand_asan_emit_allocas_unpoison): Likewise.
8593 * cfgexpand.c (expand_main_function): Likewise.
8594 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
8595 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
8596 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
8597 * config/arm/arm.c (arm_trampoline_init): Likewise.
8598 (arm_call_tls_get_addr): Likewise.
8599 (arm_expand_divmod_libfunc): Likewise.
8600 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
8601 (smulsi3_highpart): Likewise.
8602 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
8603 (c6x_expand_compare): Likewise.
8604 (c6x_expand_movmem): Likewise.
8605 * config/frv/frv.c (frv_trampoline_init): Likewise.
8606 * config/i386/i386.c (ix86_trampoline_init): Likewise.
8607 (ix86_expand_divmod_libfunc): Likewise.
8608 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
8609 (ia64_expand_compare): Likewise.
8610 (ia64_profile_hook): Likewise.
8611 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
8612 (nonlocal_goto): Likewise.
8613 (restore_stack_nonlocal): Likewise.
8614 * config/m32r/m32r.c (block_move_call): Likewise.
8615 (m32r_trampoline_init): Likewise.
8616 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
8617 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
8618 (m68k_call_m68k_read_tp): Likewise.
8619 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
8620 (microblaze_expand_divide): Likewise.
8621 * config/mips/mips.h (mips_args): Likewise.
8622 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
8623 (MIPS_ICACHE_SYNC): Likewise.
8624 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
8625 (nios2_trampoline_init): Likewise.
8626 * config/pa/pa.c (hppa_tls_call): Likewise.
8627 (pa_trampoline_init): Likewise.
8628 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
8629 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
8630 (expand_strn_compare): Likewise.
8631 (rs6000_generate_compare): Likewise.
8632 (rs6000_expand_float128_convert): Likewise.
8633 (output_profile_hook): Likewise.
8634 (rs6000_trampoline_init): Likewise.
8635 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
8636 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
8637 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
8638 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
8639 (rs6000_generate_compare): Likewise.
8640 (rs6000_expand_float128_convert): Likewise.
8641 (output_profile_hook): Likewise.
8642 (rs6000_trampoline_init): Likewise.
8643 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
8644 * config/sh/sh.c (sh_trampoline_init): Likewise.
8645 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
8646 (sparc_emit_float_lib_cmp): Likewise.
8647 (sparc32_initialize_trampoline): Likewise.
8648 (sparc64_initialize_trampoline): Likewise.
8649 (sparc_profile_hook): Likewise.
8650 * config/spu/spu.c (ea_load_store): Likewise.
8651 * config/spu/spu.md (floatunssidf2): Likewise.
8652 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
8653 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
8654 * config/visium/visium.c (expand_block_move_4): Likewise.
8655 (expand_block_move_2): Likewise.
8656 (expand_block_move_1): Likewise.
8657 (expand_block_set_4): Likewise.
8658 (expand_block_set_2): Likewise.
8659 (expand_block_set_1): Likewise.
8660 (visium_trampoline_init): Likewise.
8661 (visium_profile_hook): Likewise.
8662 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
8663 (xtensa_setup_frame_addresses): Likewise.
8664 (xtensa_trampoline_init): Likewise.
8665 * except.c (sjlj_emit_function_enter): Likewise.
8666 (sjlj_emit_function_exit): Likewise.
8667 * explow.c (allocate_dynamic_stack_space): Likewise.
8668 (probe_stack_range): Likewise.
8669 * expr.c (convert_mode_scalar): Likewise.
8670 * optabs.c (expand_binop): Likewise.
8671 (expand_twoval_binop_libfunc): Likewise.
8672 (expand_unop): Likewise.
8673 (prepare_cmp_insn): Likewise.
8674 (prepare_float_lib_cmp): Likewise.
8675 (expand_float): Likewise.
8676 (expand_fix): Likewise.
8677 (expand_fixed_convert): Likewise.
8678 (maybe_emit_sync_lock_test_and_set): Likewise.
8679 (expand_atomic_compare_and_swap): Likewise.
8680 (expand_mem_thread_fence): Likewise.
8681 (expand_atomic_fetch_op): Likewise.
8683 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
8685 * doc/generic.texi (OpenACC): Adjust URL.
8686 * doc/invoke.texi (C Dialect Options): Ditto.
8688 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
8690 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
8691 predicate for operand 1. Add (m,<S>) constraint.
8692 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
8693 Prevent memory operand 1 with register operand 2.
8695 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
8697 PR rtl-optimization/82024
8698 * combine.c (try_combine): If the combination result is a PARALLEL,
8699 and we only need to retain the SET in there that would be placed
8700 at I2, check that we can place that at I3 instead, before doing so.
8702 2017-09-01 Jakub Jelinek <jakub@redhat.com>
8705 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
8707 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
8708 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
8711 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
8712 Jeff Law <law@redhat.com>
8714 * varasm.c (bss_initializer_p): Do not put constants into .bss
8715 (categorize_decl_for_section): Handle bss_initializer_p returning
8716 false when DECL_INITIAL is NULL.
8718 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8721 * config/s390/s390.c (s390_can_inline_p): New function.
8723 2017-09-01 Jeff Law <law@redhat.com>
8725 PR tree-optimization/82052
8726 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8727 Always initialize the returned slot after a hash table miss
8728 when INSERT is true.
8730 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
8732 * config/s390/s390.md (mem_signal_fence): Remove.
8733 * doc/md.texi (mem_signal_fence): Remove.
8734 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
8736 * target-insns.def (mem_signal_fence): Remove.
8738 2017-09-01 Jakub Jelinek <jakub@redhat.com>
8741 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
8744 * asan.c (create_odr_indicator): Strip name encoding from assembler
8745 name before appending it after __odr_asan_.
8747 2017-09-01 Martin Liska <mliska@suse.cz>
8749 PR tree-optimization/82059
8750 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
8751 frequency only when an edge is redirected.
8753 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8755 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
8756 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
8757 (arc_conditional_register_usage): Remove ARC600 lp_count
8759 (arc_file_start): Emit Tag_ARC_CPU_variation.
8760 (arc_can_use_doloop_p): New conditions to use ZOLs.
8761 (hwloop_fail): New function.
8762 (hwloop_optimize): Likewise.
8763 (hwloop_pattern_reg): Likewise.
8764 (arc_doloop_hooks): New struct, to be used with reorg_loops.
8765 (arc_reorg_loops): New function, calls reorg_loops.
8766 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
8767 (arc600_corereg_hazard): Remove ZOL checking, case handled by
8769 (arc_loop_hazard): Remove function, functionality moved into
8771 (arc_hazard): Remove arc_loop_hazard call.
8772 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
8773 into hwloop_optimize.
8774 (arc_label_align): Remove ZOL handling.
8775 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
8776 * config/arc/arc.md (doloop_begin): Remove pattern.
8777 (doloop_begin_i): Likewise.
8778 (doloop_end_i): Likewise.
8779 (doloop_fallback): Likewise.
8780 (doloop_fallback_m): Likewise.
8781 (doloop_end): Reimplement expand.
8782 (arc_lp): New pattern for LP instruction.
8783 (loop_end): New pattern.
8784 (loop_fail): Likewise.
8785 (decrement_and_branch_until_zero): Likewise.
8786 * config/arc/arc.opt (mlpc-width): New option.
8787 * doc/invoke.texi (mlpc-width): Document option.
8789 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8791 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
8792 (arc_ccfsm_advance): Fix checking for delay slots.
8793 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
8795 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8797 * config/arc/arc.md (movqi_insn): Add stores to save constant long
8799 (movhi_insn): Update store instruction constraint which are saving
8800 6-bit short immediates.
8801 (movsi_insn): Consider also short scaled load operations.
8802 (zero_extendhisi2_i): Use Usd constraint instead of T.
8803 (extendhisi2_i): Add q constraint.
8804 (arc_clzsi2): Add type and length attributes.
8805 (arc_ctzsi2): Likewise.
8806 * config/arc/constraints.md (Usc): Update constraint, the
8807 assembler can parse two relocations for a single instruction.
8809 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8811 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
8812 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
8814 2017-08-31 Olivier Hainque <hainque@adacore.com>
8816 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
8817 match as powerpc-wrs-vxworks*.
8819 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
8821 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
8822 register constraint for by-element operand.
8823 (aarch64_mls_elt_merge<mode>): Likewise.
8825 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8827 * config/arc/arc.c (arc_can_follow_jump): Check for short
8830 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8832 * config.gcc: Use g.opt for arc.
8833 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
8834 functionality moved to ...
8835 (legitimate_scaled_address_p): New function, ...here.
8836 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
8837 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
8838 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
8840 (arc_override_options): Handle G option.
8841 (arc_output_pic_addr_const): Correct function definition.
8842 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
8843 (arc_decl_anon_ns_mem_p): Delete.
8844 (arc_in_small_data_p): Overhaul this function to take into
8845 consideration the value given via G option.
8846 (arc_rewrite_small_data_1): Renamed and corrected old
8847 arc_rewrite_small_data function.
8848 (arc_rewrite_small_data): New function.
8849 (small_data_pattern): Don't use pic_offset_table_rtx.
8850 * config/arc/arc.h (CC1_SPEC): Recognize G option.
8851 * config/arc/simdext.md (movmisalignv2hi): Use
8852 prepare_move_operands function.
8854 (movmisalign*): Likewise.
8855 * doc/invoke.texi (ARC options): Document -G option.
8857 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8859 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
8861 * config/arc/arc.c (arc_print_operand): Output scalled address for
8862 sdata whenever is possible.
8863 (arc_in_small_data_p): Allow sdata for 64bit datum when double
8864 load/stores are available.
8865 (compact_sda_memory_operand): Check for the alignment required by
8866 code density instructions.
8867 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
8869 * config/arc/constraints.md (Usd): Update constraint.
8870 (Us0): New constraint.
8871 (Usc): Update constraint.
8873 2017-08-31 Richard Biener <rguenther@suse.de>
8876 * dwarf2out.c (dwarf2out_early_global_decl): Process each
8879 2017-08-31 Tamar Christina <tamar.christina@arm.com>
8881 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
8882 Resize type_signature.
8884 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8885 Alan Hayward <alan.hayward@arm.com>
8886 David Sherwood <david.sherwood@arm.com>
8888 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
8889 subregs whose inner modes can be stored in GPRs.
8890 (aarch64_classify_index): Likewise.
8892 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8893 Alan Hayward <alan.hayward@arm.com>
8894 David Sherwood <david.sherwood@arm.com>
8896 * config/aarch64/iterators.md (V_cmp_result): Rename to...
8897 (V_INT_EQUIV): ...this.
8898 (v_cmp_result): Rename to...
8899 (v_int_equiv): ...this.
8900 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
8901 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
8902 (copysign<mode>3): Likewise.
8903 (aarch64_simd_bsl<mode>_internal): Likewise.
8904 (aarch64_simd_bsl<mode>): Likewise.
8905 (vec_cmp<mode><mode>): Likewise.
8906 (vcond<mode><mode>): Likewise.
8907 (vcond<v_cmp_mixed><mode>): Likewise.
8908 (vcondu<mode><v_cmp_mixed>): Likewise.
8909 (aarch64_cm<optab><mode>): Likewise.
8910 (aarch64_cmtst<mode>): Likewise.
8911 (aarch64_fac<optab><mode>): Likewise.
8912 (vec_perm_const<mode>): Likewise.
8913 (vcond_mask_<mode><v_cmp_result>): Rename to...
8914 (vcond_mask_<mode><v_int_equiv>): ...this.
8915 (vec_cmp<mode><v_cmp_result>): Rename to...
8916 (vec_cmp<mode><v_int_equiv>): ...this.
8918 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8919 Alan Hayward <alan.hayward@arm.com>
8920 David Sherwood <david.sherwood@arm.com>
8922 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
8924 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
8925 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
8926 (UNSPEC_LD4_DREG): New unspecs.
8927 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
8928 (aarch64_ld2<mode>_dreg_be): Replace with...
8929 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
8931 (aarch64_ld3<mode>_dreg_le)
8932 (aarch64_ld3<mode>_dreg_be): Replace with...
8933 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
8935 (aarch64_ld4<mode>_dreg_le)
8936 (aarch64_ld4<mode>_dreg_be): Replace with...
8937 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
8940 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8942 PR tree-optimization/81987
8943 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
8944 insert an initializer in a location not dominated by the stride
8947 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8949 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
8950 on the entire header of the finally block in the fallthru case.
8952 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8954 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
8956 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
8958 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
8959 rs6000_emit_move_from_cr and call renamed function.
8960 (rs6000_emit_prologue): Call renamed functions.
8961 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
8962 movesi_from_cr, remove volatile CRs.
8964 2017-08-30 Jon Beniston <jon@beniston.com>
8965 Richard Biener <rguenther@suse.de>
8967 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
8968 of VECTOR_MODE_P check.
8969 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
8970 element vector types.
8972 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8974 * df.h (df_read_modify_subreg_p): Remove in favor of...
8975 * rtl.h (read_modify_subreg_p): ...this new function. Take a
8976 const_rtx instead of an rtx.
8977 * cprop.c (local_cprop_find_used_regs): Update accordingly.
8978 * df-problems.c (df_word_lr_mark_ref): Likewise.
8979 * ira-lives.c (mark_pseudo_reg_live): Likewise.
8980 (mark_pseudo_reg_dead): Likewise.
8981 (mark_ref_dead): Likewise.
8982 * reginfo.c (init_subregs_of_mode): Likewise.
8983 * sched-deps.c (sched_analyze_1): Likewise.
8984 * df-scan.c (df_def_record_1): Likewise.
8985 (df_uses_record): Likewise.
8986 (df_read_modify_subreg_p): Remove in favor of...
8987 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
8988 const_rtx instead of an rtx.
8990 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8991 Alan Hayward <alan.hayward@arm.com>
8992 David Sherwood <david.sherwood@arm.com>
8994 * rtl.h (partial_subreg_p): New function.
8995 * caller-save.c (save_call_clobbered_regs): Use it.
8996 * calls.c (expand_call): Likewise.
8997 * combine.c (combinable_i3pat): Likewise.
8998 (simplify_set): Likewise.
8999 (make_extraction): Likewise.
9000 (make_compound_operation_int): Likewise.
9001 (gen_lowpart_or_truncate): Likewise.
9002 (force_to_mode): Likewise.
9003 (make_field_assignment): Likewise.
9004 (reg_truncated_to_mode): Likewise.
9005 (record_truncated_value): Likewise.
9006 (move_deaths): Likewise.
9007 * cse.c (record_jump_cond): Likewise.
9008 (cse_insn): Likewise.
9009 * cselib.c (cselib_lookup_1): Likewise.
9010 * expmed.c (extract_bit_field_using_extv): Likewise.
9011 * function.c (assign_parm_setup_reg): Likewise.
9012 * ifcvt.c (noce_convert_multiple_sets): Likewise.
9013 * ira-build.c (create_insn_allocnos): Likewise.
9014 * lra-coalesce.c (merge_pseudos): Likewise.
9015 * lra-constraints.c (match_reload): Likewise.
9016 (simplify_operand_subreg): Likewise.
9017 (curr_insn_transform): Likewise.
9018 * lra-lives.c (process_bb_lives): Likewise.
9019 * lra.c (new_insn_reg): Likewise.
9020 (lra_substitute_pseudo): Likewise.
9021 * regcprop.c (mode_change_ok): Likewise.
9022 (maybe_mode_change): Likewise.
9023 (copyprop_hardreg_forward_1): Likewise.
9024 * reload.c (push_reload): Likewise.
9025 (find_reloads): Likewise.
9026 (find_reloads_subreg_address): Likewise.
9027 * reload1.c (alter_reg): Likewise.
9028 (eliminate_regs_1): Likewise.
9029 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9031 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
9033 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
9034 back to if statements, including unpack.
9036 2017-08-30 Martin Liska <mliska@suse.cz>
9039 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
9041 (func_checker::compare_asm_inputs_outputs): ... this function.
9042 (func_checker::compare_gimple_asm): Use the function to compare
9043 also ASM constrains.
9044 * ipa-icf-gimple.h: Rename the function.
9046 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9047 Alan Hayward <alan.hayward@arm.com>
9048 David Sherwood <david.sherwood@arm.com>
9050 * coretypes.h (complex_mode): New type.
9051 * gdbhooks.py (build_pretty_printer): Handle it.
9052 * machmode.h (complex_mode): New class.
9053 (complex_mode::includes_p): New function.
9054 (is_complex_int_mode): Likewise.
9055 (is_complex_float_mode): Likewise.
9056 * genmodes.c (get_mode_class): Handle complex mode classes.
9057 * function.c (expand_function_end): Use is_complex_int_mode.
9059 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9060 Alan Hayward <alan.hayward@arm.com>
9061 David Sherwood <david.sherwood@arm.com>
9063 * coretypes.h (scalar_mode_pod): New typedef.
9064 * gdbhooks.py (build_pretty_printer): Handle it.
9065 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
9066 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
9067 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
9068 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
9071 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9072 Alan Hayward <alan.hayward@arm.com>
9073 David Sherwood <david.sherwood@arm.com>
9075 * machmode.h (mode_for_vector): Take a scalar_mode instead
9077 * stor-layout.c (mode_for_vector): Likewise.
9078 * explow.c (promote_mode): Use as_a <scalar_mode>.
9079 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
9081 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9082 Alan Hayward <alan.hayward@arm.com>
9083 David Sherwood <david.sherwood@arm.com>
9085 * target.def (preferred_simd_mode): Take a scalar_mode
9086 instead of a machine_mode.
9087 * targhooks.h (default_preferred_simd_mode): Likewise.
9088 * targhooks.c (default_preferred_simd_mode): Likewise.
9089 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
9090 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
9091 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
9092 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
9093 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
9094 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
9095 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
9096 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
9098 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
9099 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
9100 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
9101 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
9102 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
9103 * doc/tm.texi: Regenerate.
9104 * optabs-query.c (can_vec_mask_load_store_p): Return false for
9107 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9108 Alan Hayward <alan.hayward@arm.com>
9109 David Sherwood <david.sherwood@arm.com>
9111 * target.def (scalar_mode_supported_p): Take a scalar_mode
9112 instead of a machine_mode.
9113 * targhooks.h (default_scalar_mode_supported_p): Likewise.
9114 * targhooks.c (default_scalar_mode_supported_p): Likewise.
9115 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
9116 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9117 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
9118 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
9119 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
9120 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
9121 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
9122 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
9123 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
9124 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
9125 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
9126 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
9128 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
9129 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
9130 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9131 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9132 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
9134 * doc/tm.texi: Regenerate.
9136 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9137 Alan Hayward <alan.hayward@arm.com>
9138 David Sherwood <david.sherwood@arm.com>
9140 * coretypes.h (opt_scalar_mode): New typedef.
9141 * gdbhooks.py (build_pretty_printers): Handle it.
9142 * machmode.h (mode_iterator::get_2xwider): Add overload for
9144 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
9146 * expr.c (convert_mode_scalar): Likewise.
9147 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9148 * optabs.c (expand_float): Likewise.
9149 (expand_fix): Likewise.
9150 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9152 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9153 Alan Hayward <alan.hayward@arm.com>
9154 David Sherwood <david.sherwood@arm.com>
9156 * optabs.c (expand_float): Explicitly check for scalars before
9157 using a branching expansion.
9158 (expand_fix): Likewise.
9160 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9161 Alan Hayward <alan.hayward@arm.com>
9162 David Sherwood <david.sherwood@arm.com>
9164 * expr.c (convert_mode): Split scalar handling out into...
9165 (convert_mode_scalar): ...this new function. Treat the modes
9168 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9169 Alan Hayward <alan.hayward@arm.com>
9170 David Sherwood <david.sherwood@arm.com>
9172 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
9174 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
9176 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9177 Alan Hayward <alan.hayward@arm.com>
9178 David Sherwood <david.sherwood@arm.com>
9180 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
9181 rather than a machine_mode.
9182 (fixed_from_string): Likewise.
9183 (fixed_convert): Likewise.
9184 (fixed_convert_from_int): Likewise.
9185 (fixed_convert_from_real): Likewise.
9186 (real_convert_from_fixed): Likewise.
9187 * fixed-value.c (fixed_from_double_int): Likewise.
9188 (fixed_from_string): Likewise.
9189 (fixed_convert): Likewise.
9190 (fixed_convert_from_int): Likewise.
9191 (fixed_convert_from_real): Likewise.
9192 (real_convert_from_fixed): Likewise.
9193 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
9195 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9196 Alan Hayward <alan.hayward@arm.com>
9197 David Sherwood <david.sherwood@arm.com>
9199 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
9200 of separate mode class checks. Do not allow vector modes here.
9201 (immed_wide_int_const): Use as_a <scalar_mode>.
9202 * explow.c (trunc_int_for_mode): Likewise.
9203 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
9204 (wi::shwi): Likewise.
9205 (wi::min_value): Likewise.
9206 (wi::max_value): Likewise.
9207 * dwarf2out.c (loc_descriptor): Likewise.
9208 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
9211 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9212 Alan Hayward <alan.hayward@arm.com>
9213 David Sherwood <david.sherwood@arm.com>
9215 * tree.h (SCALAR_TYPE_MODE): New macro.
9216 * expr.c (expand_expr_addr_expr_1): Use it.
9217 (expand_expr_real_2): Likewise.
9218 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
9219 (fold_convert_const_fixed_from_int): Likewise.
9220 (fold_convert_const_fixed_from_real): Likewise.
9221 (native_encode_fixed): Likewise
9222 (native_encode_complex): Likewise
9223 (native_encode_vector): Likewise.
9224 (native_interpret_fixed): Likewise.
9225 (native_interpret_real): Likewise.
9226 (native_interpret_complex): Likewise.
9227 (native_interpret_vector): Likewise.
9228 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
9229 (simd_clone_adjust_argument_types): Likewise.
9230 (simd_clone_init_simd_arrays): Likewise.
9231 (simd_clone_adjust): Likewise.
9232 * stor-layout.c (layout_type): Likewise.
9233 * tree.c (build_minus_one_cst): Likewise.
9234 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9235 * tree-inline.c (estimate_move_cost): Likewise.
9236 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
9237 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
9238 (vectorizable_reduction): Likewise.
9239 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
9240 (vect_recog_mixed_size_cond_pattern): Likewise.
9241 (check_bool_pattern): Likewise.
9242 (adjust_bool_pattern): Likewise.
9243 (search_type_for_mask_1): Likewise.
9244 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
9245 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9246 (vectorizable_load): Likewise.
9247 (vectorizable_store): Likewise.
9248 * ubsan.c (ubsan_encode_value): Likewise.
9249 * varasm.c (output_constant): Likewise.
9251 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9252 Alan Hayward <alan.hayward@arm.com>
9253 David Sherwood <david.sherwood@arm.com>
9255 * coretypes.h (scalar_mode): New class.
9256 * machmode.h (scalar_mode): Likewise.
9257 (scalar_mode::includes_p): New function.
9258 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
9259 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
9260 * genmodes.c (get_mode_class): Handle remaining scalar modes.
9261 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
9262 * expmed.c (store_bit_field_1): Likewise.
9263 (extract_bit_field_1): Likewise.
9264 * expr.c (write_complex_part): Likewise.
9265 (read_complex_part): Likewise.
9266 (emit_move_complex_push): Likewise.
9267 (expand_expr_real_2): Likewise.
9268 * function.c (assign_parm_setup_reg): Likewise.
9269 (assign_parms_unsplit_complex): Likewise.
9270 * optabs.c (expand_binop): Likewise.
9271 * rtlanal.c (subreg_get_info): Likewise.
9272 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9273 * varasm.c (output_constant_pool_2): Likewise.
9275 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9276 Alan Hayward <alan.hayward@arm.com>
9277 David Sherwood <david.sherwood@arm.com>
9279 * expmed.c (extract_high_half): Use scalar_int_mode and remove
9281 (expmed_mult_highpart_optab): Likewise.
9282 (expmed_mult_highpart): Likewise.
9284 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9285 Alan Hayward <alan.hayward@arm.com>
9286 David Sherwood <david.sherwood@arm.com>
9288 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
9289 instead of a machine_mode.
9290 (builtin_memset_read_str): Likewise.
9291 * builtins.c (c_readstr): Likewise.
9292 (builtin_memcpy_read_str): Likewise.
9293 (builtin_strncpy_read_str): Likewise.
9294 (builtin_memset_read_str): Likewise.
9295 (builtin_memset_gen_str): Likewise.
9296 (expand_builtin_signbit): Use scalar_int_mode for local variables.
9297 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
9298 instead of a machine_mode.
9299 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
9301 (make_extraction): Likewise.
9302 (try_widen_shift_mode): Take and return scalar_int_modes instead
9304 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
9305 a scalar_int_mode instead of a machine_mode.
9306 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
9307 (avr_addr_space_pointer_mode): Likewise.
9308 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
9309 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
9310 (msp430_unwind_word_mode): Likewise.
9311 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
9312 (spu_addr_space_pointer_mode): Likewise.
9313 (spu_addr_space_address_mode): Likewise.
9314 (spu_libgcc_cmp_return_mode): Likewise.
9315 (spu_libgcc_shift_count_mode): Likewise.
9316 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
9317 (rl78_addr_space_pointer_mode): Likewise.
9318 (fl78_unwind_word_mode): Likewise.
9319 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
9321 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
9322 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
9323 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
9324 (mips_valid_pointer_mode): Likewise.
9325 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
9326 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
9327 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
9329 (ft32_addr_space_address_mode): Likewise.
9330 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
9331 scalar_int_mode instead of a machine_mode.
9332 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
9334 (m32c_addr_space_address_mode): Likewise.
9335 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
9336 (rs6000_eh_return_filter_mode): Likewise.
9337 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
9338 (rs6000_eh_return_filter_mode): Likewise.
9339 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
9340 (s390_libgcc_shift_count_mode): Likewise.
9341 (s390_unwind_word_mode): Likewise.
9342 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
9344 * target.def (mode_rep_extended): Likewise.
9345 (valid_pointer_mode): Likewise.
9346 (addr_space.valid_pointer_mode): Likewise.
9347 (eh_return_filter_mode): Return a scalar_int_mode rather than
9349 (libgcc_cmp_return_mode): Likewise.
9350 (libgcc_shift_count_mode): Likewise.
9351 (unwind_word_mode): Likewise.
9352 (addr_space.pointer_mode): Likewise.
9353 (addr_space.address_mode): Likewise.
9354 * doc/tm.texi: Regenerate.
9355 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
9357 (do_jump): Use scalar_int_mode for local variables.
9358 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
9359 rather than a machine_mode.
9360 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
9361 (scompare_loc_descriptor_wide): Likewise.
9362 (scompare_loc_descriptor_narrow): Likewise.
9363 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
9365 * except.c (sjlj_emit_dispatch_table): Likewise.
9366 (expand_builtin_eh_copy_values): Likewise.
9367 * explow.c (convert_memory_address_addr_space_1): Likewise.
9368 Take a scalar_int_mode rather than a machine_mode.
9369 (convert_memory_address_addr_space): Take a scalar_int_mode rather
9370 than a machine_mode.
9371 (memory_address_addr_space): Use scalar_int_mode for local variables.
9372 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
9373 rather than a machine_mode.
9374 * expmed.c (mask_rtx): Likewise.
9375 (init_expmed_one_conv): Likewise.
9376 (expand_mult_highpart_adjust): Likewise.
9377 (extract_high_half): Likewise.
9378 (expmed_mult_highpart_optab): Likewise.
9379 (expmed_mult_highpart): Likewise.
9380 (expand_smod_pow2): Likewise.
9381 (expand_sdiv_pow2): Likewise.
9382 (emit_store_flag_int): Likewise.
9383 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
9385 (extract_low_bits): Likewise.
9386 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
9388 * expr.c (pieces_addr::adjust): Likewise.
9389 (can_store_by_pieces): Likewise.
9390 (store_by_pieces): Likewise.
9391 (clear_by_pieces_1): Likewise.
9392 (expand_expr_addr_expr_1): Likewise.
9393 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
9394 (expand_expr_real_1): Likewise.
9395 (try_casesi): Likewise.
9396 * final.c (shorten_branches): Likewise.
9397 * fold-const.c (fold_convert_const_int_from_fixed): Change the
9398 type of "mode" to machine_mode.
9399 * internal-fn.c (expand_arith_overflow_result_store): Take a
9400 scalar_int_mode rather than a machine_mode.
9401 (expand_mul_overflow): Use scalar_int_mode for local variables.
9402 * loop-doloop.c (doloop_modify): Likewise.
9403 (doloop_optimize): Likewise.
9404 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
9405 than a machine_mode.
9406 (expand_doubleword_shift_condmove): Likewise.
9407 (expand_doubleword_shift): Likewise.
9408 (expand_doubleword_clz): Likewise.
9409 (expand_doubleword_popcount): Likewise.
9410 (expand_doubleword_parity): Likewise.
9411 (expand_absneg_bit): Use scalar_int_mode for local variables.
9412 (prepare_float_lib_cmp): Likewise.
9413 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
9414 rather than a machine_mode.
9415 (convert_memory_address_addr_space): Likewise.
9416 (get_mode_bounds): Likewise.
9417 (get_address_mode): Return a scalar_int_mode rather than a
9419 * rtlanal.c (get_address_mode): Likewise.
9420 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
9421 than a machine_mode.
9422 * targhooks.c (default_mode_rep_extended): Likewise.
9423 (default_valid_pointer_mode): Likewise.
9424 (default_addr_space_valid_pointer_mode): Likewise.
9425 (default_eh_return_filter_mode): Return a scalar_int_mode rather
9426 than a machine_mode.
9427 (default_libgcc_cmp_return_mode): Likewise.
9428 (default_libgcc_shift_count_mode): Likewise.
9429 (default_unwind_word_mode): Likewise.
9430 (default_addr_space_pointer_mode): Likewise.
9431 (default_addr_space_address_mode): Likewise.
9432 * targhooks.h (default_eh_return_filter_mode): Likewise.
9433 (default_libgcc_cmp_return_mode): Likewise.
9434 (default_libgcc_shift_count_mode): Likewise.
9435 (default_unwind_word_mode): Likewise.
9436 (default_addr_space_pointer_mode): Likewise.
9437 (default_addr_space_address_mode): Likewise.
9438 (default_mode_rep_extended): Take a scalar_int_mode rather than
9440 (default_valid_pointer_mode): Likewise.
9441 (default_addr_space_valid_pointer_mode): Likewise.
9442 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
9444 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
9445 rather than a machine_mode.
9446 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
9447 for local variables.
9448 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9449 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
9450 than a machine_mode.
9452 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9453 Alan Hayward <alan.hayward@arm.com>
9454 David Sherwood <david.sherwood@arm.com>
9456 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
9457 the mode argument to scalar_int_mode.
9458 (do_jump_by_parts_zero_rtx): Likewise.
9459 (do_jump_by_parts_equality_rtx): Likewise.
9460 (do_jump_by_parts_greater): Take a mode argument.
9461 (do_jump_by_parts_equality): Likewise.
9462 (do_jump_1): Update calls accordingly.
9464 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9465 Alan Hayward <alan.hayward@arm.com>
9466 David Sherwood <david.sherwood@arm.com>
9468 * is-a.h (safe_dyn_cast): New function.
9469 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
9470 (jump_table_for_label): Likewise.
9471 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
9472 instead of an rtx_insn *.
9473 (shorten_branches): Use dyn_cast instead of LABEL_P and
9474 JUMP_TABLE_DATA_P. Use jump_table_for_label and
9475 rtx_jump_table_data::get_data_mode.
9476 (final_scan_insn): Likewise.
9478 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9479 Alan Hayward <alan.hayward@arm.com>
9480 David Sherwood <david.sherwood@arm.com>
9482 * combine.c (try_combine): Use is_a <scalar_int_mode> when
9483 trying to combine a full-register integer set with a subreg
9486 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9487 Alan Hayward <alan.hayward@arm.com>
9488 David Sherwood <david.sherwood@arm.com>
9490 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
9491 that is always either address_mode or pointer_mode.
9493 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9494 Alan Hayward <alan.hayward@arm.com>
9495 David Sherwood <david.sherwood@arm.com>
9497 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
9498 when the two are known to be equal.
9500 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9501 Alan Hayward <alan.hayward@arm.com>
9502 David Sherwood <david.sherwood@arm.com>
9504 * simplify-rtx.c (simplify_const_unary_operation): Use
9505 is_a <scalar_int_mode> instead of checking for a nonzero
9506 precision. Forcibly convert op_mode to a scalar_int_mode
9507 in that case. More clearly differentiate the operand and
9508 result modes and use the former when deciding what the value
9509 of a count-bits operation should be. Use is_int_mode instead
9510 of checking for a MODE_INT. Remove redundant check for whether
9511 this mode has a zero precision.
9513 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9514 Alan Hayward <alan.hayward@arm.com>
9515 David Sherwood <david.sherwood@arm.com>
9517 * optabs.c (widen_leading): Change the type of the mode argument
9518 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
9519 (widen_bswap): Likewise.
9520 (expand_parity): Likewise.
9521 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
9522 (expand_ffs): Likewise.
9523 (epand_unop): Check for scalar integer modes before calling the
9526 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9527 Alan Hayward <alan.hayward@arm.com>
9528 David Sherwood <david.sherwood@arm.com>
9530 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
9532 (expand_expr_real_1): Update call accordingly.
9534 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9535 Alan Hayward <alan.hayward@arm.com>
9536 David Sherwood <david.sherwood@arm.com>
9538 * expmed.c (store_bit_field_using_insv): Add op0_mode and
9539 value_mode arguments. Use scalar_int_mode internally.
9540 (store_bit_field_1): Rename the new integer mode from imode
9541 to op0_mode and use it instead of GET_MODE (op0). Update calls
9542 to store_split_bit_field, store_bit_field_using_insv and
9543 store_fixed_bit_field.
9544 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
9545 Use scalar_int_mode internally. Use a bit count rather than a mode
9546 when calculating the largest bit size for get_best_mode.
9547 Update calls to store_split_bit_field and store_fixed_bit_field_1.
9548 (store_fixed_bit_field_1): Add mode and value_mode arguments.
9549 Remove assertion that OP0 has a scalar integer mode.
9550 (store_split_bit_field): Add op0_mode and value_mode arguments.
9551 Update calls to extract_fixed_bit_field.
9552 (extract_bit_field_using_extv): Add an op0_mode argument.
9553 Use scalar_int_mode internally.
9554 (extract_bit_field_1): Rename the new integer mode from imode to
9555 op0_mode and use it instead of GET_MODE (op0). Update calls to
9556 extract_split_bit_field, extract_bit_field_using_extv and
9557 extract_fixed_bit_field.
9558 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
9559 to extract_split_bit_field and extract_fixed_bit_field_1.
9560 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
9561 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
9563 (extract_split_bit_field): Add an op0_mode argument. Update call
9564 to extract_fixed_bit_field.
9566 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9567 Alan Hayward <alan.hayward@arm.com>
9568 David Sherwood <david.sherwood@arm.com>
9570 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
9571 * explow.c (hard_function_value): Likewise.
9572 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
9573 convert_to_mode call outside the loop.
9574 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
9575 for the mode iterator. Require the mode specified by max_pieces
9577 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
9579 (copy_blkmode_to_reg): Likewise.
9580 (set_storage_via_setmem): Likewise.
9581 * optabs.c (prepare_cmp_insn): Likewise.
9582 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9583 * stor-layout.c (finish_bitfield_representative): Likewise.
9585 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9586 Alan Hayward <alan.hayward@arm.com>
9587 David Sherwood <david.sherwood@arm.com>
9589 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
9590 * expr.c (convert_move): Use them.
9591 (convert_modes): Likewise.
9592 (store_expr_with_bounds): Likewise.
9594 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9595 Alan Hayward <alan.hayward@arm.com>
9596 David Sherwood <david.sherwood@arm.com>
9598 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
9599 parameter for the mode of "x". Remove the "known_x", "known_mode"
9600 and "known_ret" arguments. Change the type of the mode argument
9602 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
9603 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
9604 (reg_num_sign_bit_copies_for_combine): Likewise.
9605 * rtlanal.c (nonzero_bits1): Likewise.
9606 (num_sign_bit_copies1): Likewise.
9607 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
9608 (reg_num_sign_bit_copies_general): Likewise.
9609 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
9610 (reg_nonzero_bits_general): Likewise.
9612 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9613 Alan Hayward <alan.hayward@arm.com>
9614 David Sherwood <david.sherwood@arm.com>
9616 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
9617 than in subroutines. Return 1 for non-integer modes.
9618 (cached_num_sign_bit_copies): Change the type of the mode parameter
9620 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
9621 classes. Handle CONST_INT_P first and then check whether X also
9622 has a scalar integer mode. Check the same thing for inner registers
9623 of a SUBREG and for values that are being extended or truncated.
9625 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9626 Alan Hayward <alan.hayward@arm.com>
9627 David Sherwood <david.sherwood@arm.com>
9629 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
9630 in subroutines. Return the mode mask for non-integer modes.
9631 (cached_nonzero_bits): Change the type of the mode parameter
9633 (nonzero_bits1): Likewise. Remove early exit for other mode
9634 classes. Handle CONST_INT_P first and then check whether X
9635 also has a scalar integer mode.
9637 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9638 Alan Hayward <alan.hayward@arm.com>
9639 David Sherwood <david.sherwood@arm.com>
9641 * expr.c (widest_int_mode_for_size): Make the comment match the code.
9642 Return a scalar_int_mode and assert that the size is greater than
9644 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
9645 (op_by_pieces_d::op_by_pieces_d): Likewise.
9646 (op_by_pieces_d::run): Likewise.
9647 (can_store_by_pieces): Likewise.
9649 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9650 Alan Hayward <alan.hayward@arm.com>
9651 David Sherwood <david.sherwood@arm.com>
9653 * combine.c (extract_left_shift): Add a mode argument and update
9655 (make_compound_operation_int): Change the type of the mode parameter
9656 to scalar_int_mode and update the call to extract_left_shift.
9658 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9659 Alan Hayward <alan.hayward@arm.com>
9660 David Sherwood <david.sherwood@arm.com>
9662 * combine.c (simplify_and_const_int): Change the type of the mode
9663 parameter to scalar_int_mode.
9664 (simplify_and_const_int_1): Likewise. Update recursive call.
9666 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9667 Alan Hayward <alan.hayward@arm.com>
9668 David Sherwood <david.sherwood@arm.com>
9670 * combine.c (simplify_compare_const): Check that the mode is a
9671 scalar_int_mode (rather than VOIDmode) before testing its
9673 (simplify_comparison): Move COMPARISON_P handling out of the
9674 loop and restrict the latter part of the loop to scalar_int_modes.
9675 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
9676 and when considering SUBREG_REGs. Use is_int_mode instead of
9677 checking GET_MODE_CLASS against MODE_INT.
9679 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9680 Alan Hayward <alan.hayward@arm.com>
9681 David Sherwood <david.sherwood@arm.com>
9683 * combine.c (try_widen_shift_mode): Move check for equal modes to...
9684 (simplify_shift_const_1): ...here. Use scalar_int_mode for
9685 shift_unit_mode and for modes involved in scalar shifts.
9687 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9688 Alan Hayward <alan.hayward@arm.com>
9689 David Sherwood <david.sherwood@arm.com>
9691 * combine.c (force_int_to_mode): New function, split out from...
9692 (force_to_mode): ...here. Keep xmode up-to-date and use it
9693 instead of GET_MODE (x).
9695 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9696 Alan Hayward <alan.hayward@arm.com>
9697 David Sherwood <david.sherwood@arm.com>
9699 * optabs-query.h (extraction_insn::struct_mode): Change type to
9700 opt_scalar_int_mode and update comment.
9701 (extraction_insn::field_mode): Change type to scalar_int_mode.
9702 (extraction_insn::pos_mode): Likewise.
9703 * combine.c (make_extraction): Update accordingly.
9704 * optabs-query.c (get_traditional_extraction_insn): Likewise.
9705 (get_optab_extraction_insn): Likewise.
9706 * recog.c (simplify_while_replacing): Likewise.
9707 * expmed.c (narrow_bit_field_mem): Change the type of the mode
9708 parameter to opt_scalar_int_mode.
9710 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9711 Alan Hayward <alan.hayward@arm.com>
9712 David Sherwood <david.sherwood@arm.com>
9714 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
9715 to a scalar_int_mode instead of a machine_mode.
9716 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
9717 (get_best_mode): Return a boolean and use a pointer argument to store
9718 the selected mode. Replace the limit mode parameter with a bit limit.
9719 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
9720 for the values returned by bit_field_mode_iterator::next_mode.
9721 (store_bit_field): Update call to get_best_mode.
9722 (store_fixed_bit_field): Likewise.
9723 (extract_fixed_bit_field): Likewise.
9724 * expr.c (optimize_bitfield_assignment_op): Likewise.
9725 * fold-const.c (optimize_bit_field_compare): Likewise.
9726 (fold_truth_andor_1): Likewise.
9727 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
9728 Update for new type of m_mode.
9729 (get_best_mode): As above.
9731 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9732 Alan Hayward <alan.hayward@arm.com>
9733 David Sherwood <david.sherwood@arm.com>
9735 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
9736 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
9737 (store_bit_field): Check is_a <scalar_int_mode> before calling
9738 strict_volatile_bitfield_p.
9739 (extract_bit_field): Likewise.
9741 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9742 Alan Hayward <alan.hayward@arm.com>
9743 David Sherwood <david.sherwood@arm.com>
9745 * target.def (cstore_mode): Return a scalar_int_mode.
9746 * doc/tm.texi: Regenerate.
9747 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
9748 * targhooks.h (default_cstore_mode): Likewise.
9749 * targhooks.c (default_cstore_mode): Likewise, using a forced
9751 * expmed.c (emit_cstore): Expect the target of the cstore to be
9754 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9755 Alan Hayward <alan.hayward@arm.com>
9756 David Sherwood <david.sherwood@arm.com>
9758 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
9760 (niter_desc): Likewise mode.
9761 (iv_analyze): Add a mode parameter.
9763 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
9764 and change its type to scalar_int_mode.
9765 * loop-iv.c: Update commentary at head of file.
9766 (iv_constant): Pass the mode paraeter before the rtx it describes
9767 and change its type to scalar_int_mode. Remove VOIDmode handling.
9768 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
9769 (iv_extend): Likewise.
9770 (shorten_into_mode): Likewise.
9771 (iv_add): Use scalar_int_mode.
9772 (iv_mult): Likewise.
9773 (iv_shift): Likewise.
9774 (canonicalize_iv_subregs): Likewise.
9775 (get_biv_step_1): Pass the outer_mode parameter before the rtx
9776 it describes and change its mode to scalar_int_mode. Also change
9777 the type of the returned inner_mode to scalar_int_mode.
9778 (get_biv_step): Likewise, turning outer_mode from a pointer
9779 into a direct parameter. Update call to get_biv_step_1.
9780 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
9781 iv_constant and get_biv_step.
9782 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
9783 and change its type to scalar_int_mode. Don't initialise iv->mode
9784 to VOIDmode and remove later checks for its still being VOIDmode.
9785 Update calls to iv_analyze_op and iv_analyze_expr. Check
9786 is_a <scalar_int_mode> when changing the mode under consideration.
9787 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
9788 Update call to iv_analyze_expr.
9789 (iv_analyze_op): Add a mode parameter. Reject subregs whose
9790 inner register is not also a scalar_int_mode. Update call to
9792 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
9793 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
9794 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
9795 separate mode class checks. Update calls to iv_analyze. Remove
9796 fix-up of VOIDmodes after iv_analyze_biv.
9797 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
9798 don't have a scalar_int_mode. Update call to biv_p.
9800 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9801 Alan Hayward <alan.hayward@arm.com>
9802 David Sherwood <david.sherwood@arm.com>
9804 * cfgexpand.c (convert_debug_memory_address): Use
9805 as_a <scalar_int_mode>.
9806 * combine.c (expand_compound_operation): Likewise.
9807 (make_extraction): Likewise.
9808 (change_zero_ext): Likewise.
9809 (simplify_comparison): Likewise.
9810 * cse.c (cse_insn): Likewise.
9811 * dwarf2out.c (minmax_loc_descriptor): Likewise.
9812 (mem_loc_descriptor): Likewise.
9813 (loc_descriptor): Likewise.
9814 * expmed.c (init_expmed_one_mode): Likewise.
9815 (synth_mult): Likewise.
9816 (emit_store_flag_1): Likewise.
9817 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
9818 of a comparison with size.
9819 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
9820 (reduce_to_bit_field_precision): Likewise.
9821 * function.c (expand_function_end): Likewise.
9822 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
9823 * loop-doloop.c (doloop_modify): Likewise.
9824 * optabs.c (expand_binop): Likewise.
9825 (expand_unop): Likewise.
9826 (expand_copysign_absneg): Likewise.
9827 (prepare_cmp_insn): Likewise.
9828 (maybe_legitimize_operand): Likewise.
9829 * recog.c (const_scalar_int_operand): Likewise.
9830 * rtlanal.c (get_address_mode): Likewise.
9831 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9832 (simplify_cond_clz_ctz): Likewise.
9833 * tree-nested.c (get_nl_goto_field): Likewise.
9834 * tree.c (build_vector_type_for_mode): Likewise.
9835 * var-tracking.c (use_narrower_mode): Likewise.
9837 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9838 Alan Hayward <alan.hayward@arm.com>
9839 David Sherwood <david.sherwood@arm.com>
9841 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
9842 * builtins.c (expand_builtin_signbit): Use it.
9843 * cfgexpand.c (expand_debug_expr): Likewise.
9844 * dojump.c (do_jump): Likewise.
9845 (do_compare_and_jump): Likewise.
9846 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
9847 * expmed.c (make_tree): Likewise.
9848 * expr.c (expand_expr_real_2): Likewise.
9849 (expand_expr_real_1): Likewise.
9850 (try_casesi): Likewise.
9851 * fold-const-call.c (fold_const_call_ss): Likewise.
9852 * fold-const.c (unextend): Likewise.
9853 (extract_muldiv_1): Likewise.
9854 (fold_single_bit_test): Likewise.
9855 (native_encode_int): Likewise.
9856 (native_encode_string): Likewise.
9857 (native_interpret_int): Likewise.
9858 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
9859 * internal-fn.c (expand_addsub_overflow): Likewise.
9860 (expand_neg_overflow): Likewise.
9861 (expand_mul_overflow): Likewise.
9862 (expand_arith_overflow): Likewise.
9863 * match.pd: Likewise.
9864 * stor-layout.c (layout_type): Likewise.
9865 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9866 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
9867 * tree-ssanames.c (get_range_info): Likewise.
9868 * tree-switch-conversion.c (array_value_type) Likewise.
9869 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
9870 (vect_recog_divmod_pattern): Likewise.
9871 (vect_recog_mixed_size_cond_pattern): Likewise.
9872 * tree-vrp.c (extract_range_basic): Likewise.
9873 (simplify_float_conversion_using_ranges): Likewise.
9874 * tree.c (int_fits_type_p): Likewise.
9875 * ubsan.c (instrument_bool_enum_load): Likewise.
9876 * varasm.c (mergeable_string_section): Likewise.
9877 (narrowing_initializer_constant_valid_p): Likewise.
9878 (output_constant): Likewise.
9880 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9881 Alan Hayward <alan.hayward@arm.com>
9882 David Sherwood <david.sherwood@arm.com>
9884 * machmode.h (NARROWEST_INT_MODE): New macro.
9885 * expr.c (alignment_for_piecewise_move): Use it instead of
9886 GET_CLASS_NARROWEST_MODE (MODE_INT).
9887 (push_block): Likewise.
9888 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
9890 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9892 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9893 Alan Hayward <alan.hayward@arm.com>
9894 David Sherwood <david.sherwood@arm.com>
9896 * postreload.c (move2add_valid_value_p): Change the type of the
9897 mode parameter to scalar_int_mode.
9898 (move2add_use_add2_insn): Add a mode parameter and use it instead
9900 (move2add_use_add3_insn): Likewise.
9901 (reload_cse_move2add): Update accordingly.
9903 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9904 Alan Hayward <alan.hayward@arm.com>
9905 David Sherwood <david.sherwood@arm.com>
9907 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
9909 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
9910 * optabs.c (expand_unop): Likewise.
9912 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9913 Alan Hayward <alan.hayward@arm.com>
9914 David Sherwood <david.sherwood@arm.com>
9916 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
9917 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
9918 (popcount_loc_descriptor): Likewise.
9919 (bswap_loc_descriptor): Likewise.
9920 (rotate_loc_descriptor): Likewise.
9921 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
9922 calling the functions above.
9924 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9925 Alan Hayward <alan.hayward@arm.com>
9926 David Sherwood <david.sherwood@arm.com>
9928 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
9930 (try_combine): Likewise.
9931 (simplify_if_then_else): Likewise.
9932 * cse.c (cse_insn): Likewise.
9933 * dwarf2out.c (mem_loc_descriptor): Likewise.
9934 * emit-rtl.c (gen_lowpart_common): Likewise.
9935 * simplify-rtx.c (simplify_truncation): Likewise.
9936 (simplify_binary_operation_1): Likewise.
9937 (simplify_const_relational_operation): Likewise.
9938 (simplify_ternary_operation): Likewise.
9939 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
9941 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9942 Alan Hayward <alan.hayward@arm.com>
9943 David Sherwood <david.sherwood@arm.com>
9945 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
9946 * reload.c (push_reload): Likewise.
9947 (find_reloads): Likewise.
9949 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9950 Alan Hayward <alan.hayward@arm.com>
9951 David Sherwood <david.sherwood@arm.com>
9953 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
9954 (make_compound_operation_int): Likewise.
9955 (change_zero_ext): Likewise.
9956 * expr.c (convert_move): Likewise.
9957 (convert_modes): Likewise.
9958 * fwprop.c (forward_propagate_subreg): Likewise.
9959 * loop-iv.c (get_biv_step_1): Likewise.
9960 * optabs.c (widen_operand): Likewise.
9961 * postreload.c (move2add_valid_value_p): Likewise.
9962 * recog.c (simplify_while_replacing): Likewise.
9963 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9964 (simplify_binary_operation_1): Likewise. Remove redundant
9965 mode equality check.
9967 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9968 Alan Hayward <alan.hayward@arm.com>
9969 David Sherwood <david.sherwood@arm.com>
9971 * combine.c (combine_simplify_rtx): Add checks for
9972 is_a <scalar_int_mode>.
9973 (simplify_if_then_else): Likewise.
9974 (make_field_assignment): Likewise.
9975 (simplify_comparison): Likewise.
9976 * ifcvt.c (noce_try_bitop): Likewise.
9977 * loop-invariant.c (canonicalize_address_mult): Likewise.
9978 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9980 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9981 Alan Hayward <alan.hayward@arm.com>
9982 David Sherwood <david.sherwood@arm.com>
9984 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
9985 is_a <scalar_int_mode> instead of != BLKmode.
9987 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9988 Alan Hayward <alan.hayward@arm.com>
9989 David Sherwood <david.sherwood@arm.com>
9991 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
9992 instead of != VOIDmode.
9993 * combine.c (if_then_else_cond): Likewise.
9994 (change_zero_ext): Likewise.
9995 * dwarf2out.c (mem_loc_descriptor): Likewise.
9996 (loc_descriptor): Likewise.
9997 * rtlanal.c (canonicalize_condition): Likewise.
9998 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
10000 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10001 Alan Hayward <alan.hayward@arm.com>
10002 David Sherwood <david.sherwood@arm.com>
10004 * simplify-rtx.c (simplify_binary_operation_1): Use
10005 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
10007 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10008 Alan Hayward <alan.hayward@arm.com>
10009 David Sherwood <david.sherwood@arm.com>
10011 * wide-int.h (int_traits<unsigned char>) New class.
10012 (int_traits<unsigned short>) Likewise.
10013 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
10014 Use GET_MODE_UNIT_PRECISION and remove redundant test for
10016 * combine.c (set_nonzero_bits_and_sign_copies): Use
10017 is_a <scalar_int_mode>.
10018 (find_split_point): Likewise.
10019 (combine_simplify_rtx): Likewise.
10020 (simplify_logical): Likewise.
10021 (expand_compound_operation): Likewise.
10022 (expand_field_assignment): Likewise.
10023 (make_compound_operation): Likewise.
10024 (extended_count): Likewise.
10025 (change_zero_ext): Likewise.
10026 (simplify_comparison): Likewise.
10027 * dwarf2out.c (scompare_loc_descriptor): Likewise.
10028 (ucompare_loc_descriptor): Likewise.
10029 (minmax_loc_descriptor): Likewise.
10030 (mem_loc_descriptor): Likewise.
10031 (loc_descriptor): Likewise.
10032 * expmed.c (init_expmed_one_mode): Likewise.
10033 * lra-constraints.c (lra_constraint_offset): Likewise.
10034 * optabs.c (prepare_libcall_arg): Likewise.
10035 * postreload.c (move2add_note_store): Likewise.
10036 * reload.c (operands_match_p): Likewise.
10037 * rtl.h (load_extend_op): Likewise.
10038 * rtlhooks.c (gen_lowpart_general): Likewise.
10039 * simplify-rtx.c (simplify_truncation): Likewise.
10040 (simplify_unary_operation_1): Likewise.
10041 (simplify_binary_operation_1): Likewise.
10042 (simplify_const_binary_operation): Likewise.
10043 (simplify_const_relational_operation): Likewise.
10044 (simplify_subreg): Likewise.
10045 * stor-layout.c (bitwise_mode_for_mode): Likewise.
10046 * var-tracking.c (adjust_mems): Likewise.
10047 (prepare_call_arguments): Likewise.
10049 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10050 Alan Hayward <alan.hayward@arm.com>
10051 David Sherwood <david.sherwood@arm.com>
10053 * machmode.h (is_int_mode): New fuction.
10054 * combine.c (find_split_point): Use it.
10055 (combine_simplify_rtx): Likewise.
10056 (simplify_if_then_else): Likewise.
10057 (simplify_set): Likewise.
10058 (simplify_shift_const_1): Likewise.
10059 (simplify_comparison): Likewise.
10060 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
10061 * cse.c (notreg_cost): Likewise.
10062 (cse_insn): Likewise.
10063 * cselib.c (cselib_lookup_1): Likewise.
10064 * dojump.c (do_jump_1): Likewise.
10065 (do_compare_rtx_and_jump): Likewise.
10066 * dse.c (get_call_args): Likewise.
10067 * dwarf2out.c (rtl_for_decl_init): Likewise.
10068 (native_encode_initializer): Likewise.
10069 * expmed.c (emit_store_flag_1): Likewise.
10070 (emit_store_flag): Likewise.
10071 * expr.c (convert_modes): Likewise.
10072 (store_field): Likewise.
10073 (expand_expr_real_1): Likewise.
10074 * fold-const.c (fold_read_from_constant_string): Likewise.
10075 * gimple-ssa-sprintf.c (get_format_string): Likewise.
10076 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
10077 * optabs.c (expand_binop): Likewise.
10078 (expand_unop): Likewise.
10079 (expand_abs_nojump): Likewise.
10080 (expand_one_cmpl_abs_nojump): Likewise.
10081 * simplify-rtx.c (mode_signbit_p): Likewise.
10082 (val_signbit_p): Likewise.
10083 (val_signbit_known_set_p): Likewise.
10084 (val_signbit_known_clear_p): Likewise.
10085 (simplify_relational_operation_1): Likewise.
10086 * tree.c (vector_type_mode): Likewise.
10088 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10089 Alan Hayward <alan.hayward@arm.com>
10090 David Sherwood <david.sherwood@arm.com>
10092 * machmode.h (smallest_mode_for_size): Fix formatting.
10093 (smallest_int_mode_for_size): New function.
10094 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
10095 instead of smallest_mode_for_size.
10096 * combine.c (make_extraction): Likewise.
10097 * config/arc/arc.c (arc_expand_movmem): Likewise.
10098 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
10099 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10100 * config/s390/s390.c (s390_expand_insv): Likewise.
10101 * config/sparc/sparc.c (assign_int_registers): Likewise.
10102 * config/spu/spu.c (spu_function_value): Likewise.
10103 (spu_function_arg): Likewise.
10104 * coverage.c (get_gcov_type): Likewise.
10105 (get_gcov_unsigned_t): Likewise.
10106 * dse.c (find_shift_sequence): Likewise.
10107 * expmed.c (store_bit_field_1): Likewise.
10108 * expr.c (convert_move): Likewise.
10109 (store_field): Likewise.
10110 * internal-fn.c (expand_arith_overflow): Likewise.
10111 * optabs-query.c (get_best_extraction_insn): Likewise.
10112 * optabs.c (expand_twoval_binop_libfunc): Likewise.
10113 * stor-layout.c (layout_type): Likewise.
10114 (initialize_sizetypes): Likewise.
10115 * targhooks.c (default_get_mask_mode): Likewise.
10116 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
10118 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10119 Alan Hayward <alan.hayward@arm.com>
10120 David Sherwood <david.sherwood@arm.com>
10122 * machmode.h (opt_mode::else_blk): New function.
10123 (int_mode_for_mode): Declare.
10124 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
10125 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
10127 * cfgexpand.c (expand_debug_expr): Likewise.
10128 * combine.c (gen_lowpart_or_truncate): Likewise.
10129 (gen_lowpart_for_combine): Likewise.
10130 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
10131 * config/avr/avr.c (avr_to_int_mode): Likewise.
10132 (avr_out_plus_1): Likewise.
10133 (avr_out_plus): Likewise.
10134 (avr_out_round): Likewise.
10135 * config/i386/i386.c (ix86_split_to_parts): Likewise.
10136 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10137 (s390_expand_vcond): Likewise.
10138 * config/spu/spu.c (spu_split_immediate): Likewise.
10139 (spu_expand_mov): Likewise.
10140 * dse.c (get_stored_val): Likewise.
10141 * expmed.c (store_bit_field_1): Likewise.
10142 (convert_extracted_bit_field): Use int_mode_for_mode instead of
10144 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
10145 (extract_low_bits): Likewise.
10146 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
10147 handling rather than repeating the check.
10148 (emit_group_store): Likewise.
10149 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
10150 * optabs.c (expand_absneg_bit): Likewise.
10151 (expand_copysign_absneg): Likewise.
10152 (expand_copysign_bit): Likewise.
10153 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
10154 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
10155 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
10156 * var-tracking.c (prepare_call_arguments): Likewise.
10157 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
10158 int_mode_for_mode instead of mode_for_size.
10159 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10161 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10162 Alan Hayward <alan.hayward@arm.com>
10163 David Sherwood <david.sherwood@arm.com>
10165 * machmode.h (int_mode_for_size): New function.
10166 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
10167 instead of mode_for_size.
10168 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
10170 * combine.c (expand_field_assignment): Use int_mode_for_size
10171 instead of mode_for_size.
10172 (make_extraction): Likewise.
10173 (simplify_shift_const_1): Likewise.
10174 (simplify_comparison): Likewise.
10175 * dojump.c (do_jump): Likewise.
10176 * dwarf2out.c (mem_loc_descriptor): Likewise.
10177 * emit-rtl.c (init_derived_machine_modes): Likewise.
10178 * expmed.c (flip_storage_order): Likewise.
10179 (convert_extracted_bit_field): Likewise.
10180 * expr.c (copy_blkmode_from_reg): Likewise.
10181 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
10182 * internal-fn.c (expand_mul_overflow): Likewise.
10183 * lower-subreg.c (simple_move): Likewise.
10184 * optabs-libfuncs.c (init_optabs): Likewise.
10185 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10186 * tree.c (vector_type_mode): Likewise.
10187 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
10188 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
10189 * tree-vect-generic.c (expand_vector_parallel): Likewise.
10190 * tree-vect-stmts.c (vectorizable_load): Likewise.
10191 (vectorizable_store): Likewise.
10193 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10194 Alan Hayward <alan.hayward@arm.com>
10195 David Sherwood <david.sherwood@arm.com>
10197 * coretypes.h (pod_mode): New type.
10198 (scalar_int_mode_pod): New typedef.
10199 * machmode.h (pod_mode): New class.
10200 (int_n_data_t::m): Change type to scalar_int_mode_pod.
10201 * genmodes.c (emit_mode_int_n): Update accordingly.
10202 * lower-subreg.h (target_lower_subreg): Change type to
10203 scalar_int_mode_pod.
10204 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
10205 scalar_int_mode_pod.
10207 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10208 Alan Hayward <alan.hayward@arm.com>
10209 David Sherwood <david.sherwood@arm.com>
10211 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
10212 machine_mode to scalar_int_mode.
10213 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
10214 (rs6000_option_override_internal): Remove cast to int.
10215 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
10216 machine_mode to scalar_int_mode.
10217 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
10218 (rs6000_option_override_internal): Remove cast to int.
10219 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
10220 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
10222 * config/s390/s390.c (s390_expand_builtin): Likewise.
10223 * coretypes.h (scalar_int_mode): New type.
10224 (opt_scalar_int_mode): New typedef.
10225 * machmode.h (scalar_int_mode): New class.
10226 (scalar_int_mode::includes_p): New function.
10227 (byte_mode): Change type to scalar_int_mode.
10228 (word_mode): Likewise.
10229 (ptr_mode): Likewise.
10230 * emit-rtl.c (byte_mode): Likewise.
10231 (word_mode): Likewise.
10232 (ptr_mode): Likewise.
10233 (init_derived_machine_modes): Update accordingly.
10234 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
10235 and MODE_PARTIAL_INT.
10236 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
10237 opt_scalar_int_mode.
10239 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10240 Alan Hayward <alan.hayward@arm.com>
10241 David Sherwood <david.sherwood@arm.com>
10243 * target.def (libgcc_floating_mode_supported_p): Take a
10245 * doc/tm.texi: Regenerate.
10246 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
10248 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10249 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
10252 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10253 Alan Hayward <alan.hayward@arm.com>
10254 David Sherwood <david.sherwood@arm.com>
10256 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
10257 * doc/tm.texi: Regenerate.
10258 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
10259 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
10260 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
10261 * targhooks.h (default_floatn_mode): Likewise.
10262 * targhooks.c (default_floatn_mode): Likewise.
10263 * tree.c (build_common_tree_nodes): Update accordingly.
10265 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10266 Alan Hayward <alan.hayward@arm.com>
10267 David Sherwood <david.sherwood@arm.com>
10269 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
10270 (mode_iterator::iterate_p): Likewise.
10271 (mode_iterator::get_wider): Likewise.
10272 * expr.c (init_expr_target): Use opt_scalar_float_mode.
10274 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10275 Alan Hayward <alan.hayward@arm.com>
10276 David Sherwood <david.sherwood@arm.com>
10278 * coretypes.h (opt_scalar_float_mode): New typedef.
10279 * machmode.h (float_mode_for_size): New function.
10280 * emit-rtl.c (double_mode): Delete.
10281 (init_emit_once): Use float_mode_for_size.
10282 * stor-layout.c (layout_type): Likewise.
10283 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
10285 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10286 Alan Hayward <alan.hayward@arm.com>
10287 David Sherwood <david.sherwood@arm.com>
10289 * output.h (assemble_real): Take a scalar_float_mode.
10290 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
10291 * config/arm/arm.md (consttable_4): Likewise.
10292 (consttable_8): Likewise.
10293 (consttable_16): Likewise.
10294 * config/mips/mips.md (consttable_float): Likewise.
10295 * config/s390/s390.c (s390_output_pool_entry): Likewise.
10296 * varasm.c (assemble_real): Take a scalar_float_mode.
10297 (output_constant_pool_2): Update accordingly.
10298 (output_constant): Likewise.
10300 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10301 Alan Hayward <alan.hayward@arm.com>
10302 David Sherwood <david.sherwood@arm.com>
10304 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
10305 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
10306 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
10307 (native_encode_real): Likewise.
10308 (native_interpret_real): Likewise.
10309 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
10310 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10312 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10313 Alan Hayward <alan.hayward@arm.com>
10314 David Sherwood <david.sherwood@arm.com>
10316 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
10317 <scalar_float_mode>. Simplify.
10318 (gen_extend_conv_libfunc): Likewise.
10320 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10321 Alan Hayward <alan.hayward@arm.com>
10322 David Sherwood <david.sherwood@arm.com>
10324 * coretypes.h (scalar_float_mode): New type.
10325 * machmode.h (mode_traits::from_int): Use machine_mode if
10326 USE_ENUM_MODES is defined.
10327 (is_a): New function.
10329 (dyn_cast): Likewise.
10330 (scalar_float_mode): New class.
10331 (scalar_float_mode::includes_p): New function.
10332 (is_float_mode): Likewise.
10333 * gdbhooks.py (MachineModePrinter): New class.
10334 (build_pretty_printer): Use it for scalar_float_mode.
10335 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
10336 (format_helper::format_helper): Turn into a template.
10337 * genmodes.c (get_mode_class): New function.
10338 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
10339 or machine_mode if none.
10340 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
10341 as_a <scalar_float_mode>.
10342 * dwarf2out.c (mem_loc_descriptor): Likewise.
10343 (insert_float): Likewise.
10344 (add_const_value_attribute): Likewise.
10345 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10346 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
10347 (expand_unop): Update accordingly.
10348 (expand_abs_nojump): Likewise.
10349 (expand_copysign_absneg): Take a scalar_float_mode.
10350 (expand_copysign_bit): Likewise.
10351 (expand_copysign): Update accordingly.
10353 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10354 Alan Hayward <alan.hayward@arm.com>
10355 David Sherwood <david.sherwood@arm.com>
10357 * coretypes.h (opt_mode): New class.
10358 * machmode.h (opt_mode): Likewise.
10359 (opt_mode::else_void): New function.
10360 (opt_mode::require): Likewise.
10361 (opt_mode::exists): Likewise.
10362 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
10363 (GET_MODE_2XWIDER_MODE): Likewise.
10364 (mode_iterator::get_wider): Update accordingly.
10365 (mode_iterator::get_2xwider): Likewise.
10366 (mode_iterator::get_known_wider): Likewise, turning into a template.
10367 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
10368 forcing a wider mode to exist.
10369 * config/cr16/cr16.h (LONG_REG_P): Likewise.
10370 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10371 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
10372 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
10373 * lower-subreg.c (init_lower_subreg): Likewise.
10374 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
10375 on the final iteration.
10376 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
10377 a wider mode exists before asking for a move pattern.
10378 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
10379 forcing a wider mode to exist.
10380 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
10381 returning false if no such mode exists.
10382 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
10383 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
10384 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
10385 Avoid checking for a MODE_INT if we already know the mode is not a
10387 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
10388 forcing a wider mode to exist.
10389 (expmed_mult_highpart_optab): Likewise.
10390 (expmed_mult_highpart): Likewise.
10391 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
10393 * lto-streamer-in.c (lto_input_mode_table): Likewise.
10394 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
10395 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10396 * internal-fn.c (expand_mul_overflow): Update use of
10397 GET_MODE_2XWIDER_MODE.
10398 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10399 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
10400 GET_MODE_WIDER_MODE.
10401 (convert_plusminus_to_widen): Likewise.
10402 * tree-switch-conversion.c (array_value_type): Likewise.
10403 * var-tracking.c (emit_note_insn_var_location): Likewise.
10404 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10405 Return false inside rather than outside the loop if no wider mode
10407 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
10408 and GET_MODE_2XWIDER_MODE
10409 (can_compare_p): Use else_void.
10410 * gdbhooks.py (OptMachineModePrinter): New class.
10411 (build_pretty_printer): Use it for opt_mode.
10413 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10414 Alan Hayward <alan.hayward@arm.com>
10415 David Sherwood <david.sherwood@arm.com>
10417 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
10418 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
10420 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10421 Alan Hayward <alan.hayward@arm.com>
10422 David Sherwood <david.sherwood@arm.com>
10424 * machmode.h (mode_traits): New structure.
10425 (get_narrowest_mode): New function.
10426 (mode_iterator::start): Likewise.
10427 (mode_iterator::iterate_p): Likewise.
10428 (mode_iterator::get_wider): Likewise.
10429 (mode_iterator::get_known_wider): Likewise.
10430 (mode_iterator::get_2xwider): Likewise.
10431 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
10432 (FOR_EACH_MODE): Likewise.
10433 (FOR_EACH_MODE_FROM): Likewise.
10434 (FOR_EACH_MODE_UNTIL): Likewise.
10435 (FOR_EACH_WIDER_MODE): Likewise.
10436 (FOR_EACH_2XWIDER_MODE): Likewise.
10437 * builtins.c (expand_builtin_strlen): Use new mode iterators.
10438 * combine.c (simplify_comparison): Likewise
10439 * config/i386/i386.c (type_natural_mode): Likewise.
10440 * cse.c (cse_insn): Likewise.
10441 * dse.c (find_shift_sequence): Likewise.
10442 * emit-rtl.c (init_derived_machine_modes): Likewise.
10443 (init_emit_once): Likewise.
10444 * explow.c (hard_function_value): Likewise.
10445 * expmed.c (extract_fixed_bit_field_1): Likewise.
10446 (extract_bit_field_1): Likewise.
10447 (expand_divmod): Likewise.
10448 (emit_store_flag_1): Likewise.
10449 * expr.c (init_expr_target): Likewise.
10450 (convert_move): Likewise.
10451 (alignment_for_piecewise_move): Likewise.
10452 (widest_int_mode_for_size): Likewise.
10453 (emit_block_move_via_movmem): Likewise.
10454 (copy_blkmode_to_reg): Likewise.
10455 (set_storage_via_setmem): Likewise.
10456 (compress_float_constant): Likewise.
10457 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10458 * optabs-query.c (get_best_extraction_insn): Likewise.
10459 * optabs.c (expand_binop): Likewise.
10460 (expand_twoval_unop): Likewise.
10461 (expand_twoval_binop): Likewise.
10462 (widen_leading): Likewise.
10463 (widen_bswap): Likewise.
10464 (expand_parity): Likewise.
10465 (expand_unop): Likewise.
10466 (prepare_cmp_insn): Likewise.
10467 (prepare_float_lib_cmp): Likewise.
10468 (expand_float): Likewise.
10469 (expand_fix): Likewise.
10470 (expand_sfix_optab): Likewise.
10471 * postreload.c (move2add_use_add2_insn): Likewise.
10472 * reg-stack.c (reg_to_stack): Likewise.
10473 * reginfo.c (choose_hard_reg_mode): Likewise.
10474 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10475 * stor-layout.c (mode_for_size): Likewise.
10476 (smallest_mode_for_size): Likewise.
10477 (mode_for_vector): Likewise.
10478 (finish_bitfield_representative): Likewise.
10479 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
10480 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
10481 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10482 * var-tracking.c (prepare_call_arguments): Likewise.
10484 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10485 Alan Hayward <alan.hayward@arm.com>
10486 David Sherwood <david.sherwood@arm.com>
10488 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
10489 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
10490 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
10491 * machmode.h (mode_size): Move earlier in file.
10492 (mode_precision): Likewise.
10493 (mode_inner): Likewise.
10494 (mode_nunits): Likewise.
10495 (mode_unit_size): Likewise.
10496 (unit_unit_precision): Likewise.
10497 (mode_wider): Likewise.
10498 (mode_2xwider): Likewise.
10499 (machine_mode): New class.
10500 (mode_to_bytes): New function.
10501 (mode_to_bits): Likewise.
10502 (mode_to_precision): Likewise.
10503 (mode_to_inner): Likewise.
10504 (mode_to_unit_size): Likewise.
10505 (mode_to_unit_precision): Likewise.
10506 (mode_to_nunits): Likewise.
10507 (GET_MODE_SIZE): Use mode_to_bytes.
10508 (GET_MODE_BITSIZE): Use mode_to_bits.
10509 (GET_MODE_PRECISION): Use mode_to_precision.
10510 (GET_MODE_INNER): Use mode_to_inner.
10511 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
10512 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
10513 (GET_MODE_NUNITS): Use mode_to_nunits.
10514 * system.h (ALWAYS_INLINE): New macro.
10515 * config/powerpcspe/powerpcspe-c.c
10516 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
10517 int for arg1_mode and arg2_mode.
10519 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10520 Alan Hayward <alan.hayward@arm.com>
10521 David Sherwood <david.sherwood@arm.com>
10523 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10524 Prefix mode names with E_ in case statements.
10525 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10526 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
10527 (aarch64_split_simd_move): Likewise.
10528 (aarch64_gen_storewb_pair): Likewise.
10529 (aarch64_gen_loadwb_pair): Likewise.
10530 (aarch64_gen_store_pair): Likewise.
10531 (aarch64_gen_load_pair): Likewise.
10532 (aarch64_get_condition_code_1): Likewise.
10533 (aarch64_constant_pool_reload_icode): Likewise.
10534 (get_rsqrte_type): Likewise.
10535 (get_rsqrts_type): Likewise.
10536 (get_recpe_type): Likewise.
10537 (get_recps_type): Likewise.
10538 (aarch64_gimplify_va_arg_expr): Likewise.
10539 (aarch64_simd_container_mode): Likewise.
10540 (aarch64_emit_load_exclusive): Likewise.
10541 (aarch64_emit_store_exclusive): Likewise.
10542 (aarch64_expand_compare_and_swap): Likewise.
10543 (aarch64_gen_atomic_cas): Likewise.
10544 (aarch64_emit_bic): Likewise.
10545 (aarch64_emit_atomic_swap): Likewise.
10546 (aarch64_emit_atomic_load_op): Likewise.
10547 (aarch64_evpc_trn): Likewise.
10548 (aarch64_evpc_uzp): Likewise.
10549 (aarch64_evpc_zip): Likewise.
10550 (aarch64_evpc_ext): Likewise.
10551 (aarch64_evpc_rev): Likewise.
10552 (aarch64_evpc_dup): Likewise.
10553 (aarch64_gen_ccmp_first): Likewise.
10554 (aarch64_gen_ccmp_next): Likewise.
10555 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10556 (alpha_emit_xfloating_libcall): Likewise.
10557 (emit_insxl): Likewise.
10558 (alpha_arg_type): Likewise.
10559 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
10560 (arc_preferred_simd_mode): Likewise.
10561 (arc_secondary_reload): Likewise.
10562 (get_arc_condition_code): Likewise.
10563 (arc_print_operand): Likewise.
10564 (arc_legitimate_constant_p): Likewise.
10565 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10566 * config/arc/arc.md (casesi_load): Likewise.
10567 (casesi_compact_jump): Likewise.
10568 * config/arc/predicates.md (proper_comparison_operator): Likewise.
10569 (cc_use_register): Likewise.
10570 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10571 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10572 (arm_init_iwmmxt_builtins): Likewise.
10573 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
10574 (neon_expand_vector_init): Likewise.
10575 (arm_attr_length_move_neon): Likewise.
10576 (maybe_get_arm_condition_code): Likewise.
10577 (arm_emit_vector_const): Likewise.
10578 (arm_preferred_simd_mode): Likewise.
10579 (arm_output_iwmmxt_tinsr): Likewise.
10580 (thumb1_output_casesi): Likewise.
10581 (thumb2_output_casesi): Likewise.
10582 (arm_emit_load_exclusive): Likewise.
10583 (arm_emit_store_exclusive): Likewise.
10584 (arm_expand_compare_and_swap): Likewise.
10585 (arm_evpc_neon_vuzp): Likewise.
10586 (arm_evpc_neon_vzip): Likewise.
10587 (arm_evpc_neon_vrev): Likewise.
10588 (arm_evpc_neon_vtrn): Likewise.
10589 (arm_evpc_neon_vext): Likewise.
10590 (arm_validize_comparison): Likewise.
10591 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
10592 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
10593 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
10594 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
10595 (c6x_preferred_simd_mode): Likewise.
10596 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
10597 (epiphany_rtx_costs): Likewise.
10598 * config/epiphany/predicates.md (proper_comparison_operator):
10600 * config/frv/frv.c (condexec_memory_operand): Likewise.
10601 (frv_emit_move): Likewise.
10602 (output_move_single): Likewise.
10603 (output_condmove_single): Likewise.
10604 (frv_hard_regno_mode_ok): Likewise.
10605 (frv_matching_accg_mode): Likewise.
10606 * config/h8300/h8300.c (split_adds_subs): Likewise.
10607 (h8300_rtx_costs): Likewise.
10608 (h8300_print_operand): Likewise.
10609 (compute_mov_length): Likewise.
10610 (output_logical_op): Likewise.
10611 (compute_logical_op_length): Likewise.
10612 (compute_logical_op_cc): Likewise.
10613 (h8300_shift_needs_scratch_p): Likewise.
10614 (output_a_shift): Likewise.
10615 (compute_a_shift_length): Likewise.
10616 (compute_a_shift_cc): Likewise.
10617 (expand_a_rotate): Likewise.
10618 (output_a_rotate): Likewise.
10619 * config/i386/i386.c (classify_argument): Likewise.
10620 (function_arg_advance_32): Likewise.
10621 (function_arg_32): Likewise.
10622 (function_arg_64): Likewise.
10623 (function_value_64): Likewise.
10624 (ix86_gimplify_va_arg): Likewise.
10625 (ix86_legitimate_constant_p): Likewise.
10626 (put_condition_code): Likewise.
10627 (split_double_mode): Likewise.
10628 (ix86_avx256_split_vector_move_misalign): Likewise.
10629 (ix86_expand_vector_logical_operator): Likewise.
10630 (ix86_split_idivmod): Likewise.
10631 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
10632 (ix86_build_const_vector): Likewise.
10633 (ix86_build_signbit_mask): Likewise.
10634 (ix86_match_ccmode): Likewise.
10635 (ix86_cc_modes_compatible): Likewise.
10636 (ix86_expand_branch): Likewise.
10637 (ix86_expand_sse_cmp): Likewise.
10638 (ix86_expand_sse_movcc): Likewise.
10639 (ix86_expand_int_sse_cmp): Likewise.
10640 (ix86_expand_vec_perm_vpermi2): Likewise.
10641 (ix86_expand_vec_perm): Likewise.
10642 (ix86_expand_sse_unpack): Likewise.
10643 (ix86_expand_int_addcc): Likewise.
10644 (ix86_split_to_parts): Likewise.
10645 (ix86_vectorize_builtin_gather): Likewise.
10646 (ix86_vectorize_builtin_scatter): Likewise.
10647 (avx_vpermilp_parallel): Likewise.
10648 (inline_memory_move_cost): Likewise.
10649 (ix86_tieable_integer_mode_p): Likewise.
10650 (x86_maybe_negate_const_int): Likewise.
10651 (ix86_expand_vector_init_duplicate): Likewise.
10652 (ix86_expand_vector_init_one_nonzero): Likewise.
10653 (ix86_expand_vector_init_one_var): Likewise.
10654 (ix86_expand_vector_init_concat): Likewise.
10655 (ix86_expand_vector_init_interleave): Likewise.
10656 (ix86_expand_vector_init_general): Likewise.
10657 (ix86_expand_vector_set): Likewise.
10658 (ix86_expand_vector_extract): Likewise.
10659 (emit_reduc_half): Likewise.
10660 (ix86_emit_i387_round): Likewise.
10661 (ix86_mangle_type): Likewise.
10662 (ix86_expand_round_sse4): Likewise.
10663 (expand_vec_perm_blend): Likewise.
10664 (canonicalize_vector_int_perm): Likewise.
10665 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
10666 (expand_vec_perm_1): Likewise.
10667 (expand_vec_perm_interleave3): Likewise.
10668 (expand_vec_perm_even_odd_pack): Likewise.
10669 (expand_vec_perm_even_odd_1): Likewise.
10670 (expand_vec_perm_broadcast_1): Likewise.
10671 (ix86_vectorize_vec_perm_const_ok): Likewise.
10672 (ix86_expand_vecop_qihi): Likewise.
10673 (ix86_expand_mul_widen_hilo): Likewise.
10674 (ix86_expand_sse2_abs): Likewise.
10675 (ix86_expand_pextr): Likewise.
10676 (ix86_expand_pinsr): Likewise.
10677 (ix86_preferred_simd_mode): Likewise.
10678 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
10679 * config/i386/sse.md (*andnot<mode>3): Likewise.
10680 (<mask_codefor><code><mode>3<mask_name>): Likewise.
10681 (*<code><mode>3): Likewise.
10682 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
10683 (ia64_expand_atomic_op): Likewise.
10684 (ia64_arg_type): Likewise.
10685 (ia64_mode_to_int): Likewise.
10686 (ia64_scalar_mode_supported_p): Likewise.
10687 (ia64_vector_mode_supported_p): Likewise.
10688 (expand_vec_perm_broadcast): Likewise.
10689 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
10690 (iq2000_function_arg_advance): Likewise.
10691 (iq2000_function_arg): Likewise.
10692 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
10693 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
10694 (m68k_libcall_value): Likewise.
10695 (m68k_function_value): Likewise.
10696 (sched_attr_op_type): Likewise.
10697 * config/mcore/mcore.c (mcore_output_move): Likewise.
10698 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
10700 (microblaze_function_arg): Likewise.
10701 * config/mips/mips.c (mips16_build_call_stub): Likewise.
10702 (mips_print_operand): Likewise.
10703 (mips_mode_ok_for_mov_fmt_p): Likewise.
10704 (mips_vector_mode_supported_p): Likewise.
10705 (mips_preferred_simd_mode): Likewise.
10706 (mips_expand_vpc_loongson_even_odd): Likewise.
10707 (mips_expand_vec_unpack): Likewise.
10708 (mips_expand_vi_broadcast): Likewise.
10709 (mips_expand_vector_init): Likewise.
10710 (mips_expand_vec_reduc): Likewise.
10711 (mips_expand_msa_cmp): Likewise.
10712 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
10713 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
10714 (cc_flags_for_mode): Likewise.
10715 * config/msp430/msp430.c (msp430_print_operand): Likewise.
10716 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
10717 (nds32_output_casesi_pc_relative): Likewise.
10718 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10719 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
10720 (nvptx_gen_unpack): Likewise.
10721 (nvptx_gen_pack): Likewise.
10722 (nvptx_gen_shuffle): Likewise.
10723 (nvptx_gen_wcast): Likewise.
10724 (nvptx_preferred_simd_mode): Likewise.
10725 * config/pa/pa.c (pa_secondary_reload): Likewise.
10726 * config/pa/predicates.md (base14_operand): Likewise.
10727 * config/powerpcspe/powerpcspe-c.c
10728 (altivec_resolve_overloaded_builtin): Likewise.
10729 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
10731 (rs6000_preferred_simd_mode): Likewise.
10732 (output_vec_const_move): Likewise.
10733 (rs6000_expand_vector_extract): Likewise.
10734 (rs6000_split_vec_extract_var): Likewise.
10735 (reg_offset_addressing_ok_p): Likewise.
10736 (rs6000_legitimate_offset_address_p): Likewise.
10737 (rs6000_legitimize_address): Likewise.
10738 (rs6000_emit_set_const): Likewise.
10739 (rs6000_const_vec): Likewise.
10740 (rs6000_emit_move): Likewise.
10741 (spe_build_register_parallel): Likewise.
10742 (rs6000_darwin64_record_arg_recurse): Likewise.
10743 (swap_selector_for_mode): Likewise.
10744 (spe_init_builtins): Likewise.
10745 (paired_init_builtins): Likewise.
10746 (altivec_init_builtins): Likewise.
10747 (do_load_for_compare): Likewise.
10748 (rs6000_generate_compare): Likewise.
10749 (rs6000_expand_float128_convert): Likewise.
10750 (emit_load_locked): Likewise.
10751 (emit_store_conditional): Likewise.
10752 (rs6000_output_function_epilogue): Likewise.
10753 (rs6000_handle_altivec_attribute): Likewise.
10754 (rs6000_function_value): Likewise.
10755 (emit_fusion_gpr_load): Likewise.
10756 (emit_fusion_p9_load): Likewise.
10757 (emit_fusion_p9_store): Likewise.
10758 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
10759 (fusion_gpr_mem_load): Likewise.
10760 (fusion_addis_mem_combo_load): Likewise.
10761 (fusion_addis_mem_combo_store): Likewise.
10762 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
10763 (fusion_gpr_mem_load): Likewise.
10764 (fusion_addis_mem_combo_load): Likewise.
10765 (fusion_addis_mem_combo_store): Likewise.
10766 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10768 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
10769 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10770 (rs6000_preferred_simd_mode): Likewise.
10771 (output_vec_const_move): Likewise.
10772 (rs6000_expand_vector_extract): Likewise.
10773 (rs6000_split_vec_extract_var): Likewise.
10774 (reg_offset_addressing_ok_p): Likewise.
10775 (rs6000_legitimate_offset_address_p): Likewise.
10776 (rs6000_legitimize_address): Likewise.
10777 (rs6000_emit_set_const): Likewise.
10778 (rs6000_const_vec): Likewise.
10779 (rs6000_emit_move): Likewise.
10780 (rs6000_darwin64_record_arg_recurse): Likewise.
10781 (swap_selector_for_mode): Likewise.
10782 (paired_init_builtins): Likewise.
10783 (altivec_init_builtins): Likewise.
10784 (rs6000_expand_float128_convert): Likewise.
10785 (emit_load_locked): Likewise.
10786 (emit_store_conditional): Likewise.
10787 (rs6000_output_function_epilogue): Likewise.
10788 (rs6000_handle_altivec_attribute): Likewise.
10789 (rs6000_function_value): Likewise.
10790 (emit_fusion_gpr_load): Likewise.
10791 (emit_fusion_p9_load): Likewise.
10792 (emit_fusion_p9_store): Likewise.
10793 * config/rx/rx.c (rx_gen_move_template): Likewise.
10794 (flags_from_mode): Likewise.
10795 * config/s390/predicates.md (s390_alc_comparison): Likewise.
10796 (s390_slb_comparison): Likewise.
10797 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
10798 (s390_vector_mode_supported_p): Likewise.
10799 (s390_cc_modes_compatible): Likewise.
10800 (s390_match_ccmode_set): Likewise.
10801 (s390_canonicalize_comparison): Likewise.
10802 (s390_emit_compare_and_swap): Likewise.
10803 (s390_branch_condition_mask): Likewise.
10804 (s390_rtx_costs): Likewise.
10805 (s390_secondary_reload): Likewise.
10806 (__SECONDARY_RELOAD_CASE): Likewise.
10807 (s390_expand_cs): Likewise.
10808 (s390_preferred_simd_mode): Likewise.
10809 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
10810 * config/sh/sh.c (sh_print_operand): Likewise.
10811 (dump_table): Likewise.
10812 (sh_secondary_reload): Likewise.
10813 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10814 * config/sh/sh.md (casesi_worker_1): Likewise.
10815 (casesi_worker_2): Likewise.
10816 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
10817 (fcc_comparison_operator): Likewise.
10818 * config/sparc/sparc.c (sparc_expand_move): Likewise.
10819 (emit_soft_tfmode_cvt): Likewise.
10820 (sparc_preferred_simd_mode): Likewise.
10821 (output_cbranch): Likewise.
10822 (sparc_print_operand): Likewise.
10823 (sparc_expand_vec_perm_bmask): Likewise.
10824 (vector_init_bshuffle): Likewise.
10825 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10826 (spu_vector_mode_supported_p): Likewise.
10827 (spu_expand_insv): Likewise.
10828 (spu_emit_branch_or_set): Likewise.
10829 (spu_handle_vector_attribute): Likewise.
10830 (spu_builtin_splats): Likewise.
10831 (spu_builtin_extract): Likewise.
10832 (spu_builtin_promote): Likewise.
10833 (spu_expand_sign_extend): Likewise.
10834 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10835 (tilegx_simd_int): Likewise.
10836 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
10837 (tilepro_simd_int): Likewise.
10838 * config/v850/v850.c (const_double_split): Likewise.
10839 (v850_print_operand): Likewise.
10840 (ep_memory_offset): Likewise.
10841 * config/vax/vax.c (vax_rtx_costs): Likewise.
10842 (vax_output_int_move): Likewise.
10843 (vax_output_int_add): Likewise.
10844 (vax_output_int_subtract): Likewise.
10845 * config/visium/predicates.md (visium_branch_operator): Likewise.
10846 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
10847 (visium_print_operand_address): Likewise.
10848 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10849 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
10850 (xtensa_expand_conditional_branch): Likewise.
10851 (xtensa_copy_incoming_a7): Likewise.
10852 (xtensa_output_literal): Likewise.
10853 * dfp.c (decimal_real_maxval): Likewise.
10854 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10856 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10857 Alan Hayward <alan.hayward@arm.com>
10858 David Sherwood <david.sherwood@arm.com>
10860 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
10861 (mode_nunits_inline): Likewise.
10862 (mode_inner_inline): Likewise.
10863 (mode_unit_size_inline): Likewise.
10864 (mode_unit_precision_inline): Likewise.
10865 (emit_insn_modes_h): Likewise. Also emit a #define of the
10867 (emit_mode_wider): Add an E_ prefix to mode names.
10868 (emit_mode_complex): Likewise.
10869 (emit_mode_inner): Likewise.
10870 (emit_mode_adjustments): Likewise.
10871 (emit_mode_int_n): Likewise.
10872 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
10873 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
10874 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
10875 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
10876 (CRC32_BUILTIN, ENTRY): Likewise.
10877 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
10878 (aarch64_pop_regs): Likewise.
10879 (aarch64_process_components): Likewise.
10880 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
10881 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
10882 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
10883 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
10884 * config/arm/arm.c (arm_init_libfuncs): Likewise.
10885 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
10887 * config/i386/i386-builtin.def (pcmpestr): Likewise.
10888 (pcmpistr): Likewise.
10889 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
10890 * config/mmix/mmix.c (mmix_output_condition): Likewise.
10891 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10893 * config/rl78/rl78.c (mduc_regs): Likewise.
10894 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10895 (htm_expand_builtin): Likewise.
10896 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
10897 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
10898 * config/xtensa/xtensa.c (print_operand): Likewise.
10899 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
10900 (NUM_MODE_VECTOR_INT): Likewise.
10901 * genoutput.c (null_operand): Likewise.
10902 (output_operand_data): Likewise.
10903 * genrecog.c (print_parameter_value): Likewise.
10904 * lra.c (debug_operand_data): Likewise.
10906 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10908 * dwarf2out.c (scompare_loc_descriptor_wide)
10909 (scompare_loc_descriptor_narrow): New functions, split out from...
10910 (scompare_loc_descriptor): ...here.
10911 * expmed.c (emit_store_flag_int): New function, split out from...
10912 (emit_store_flag): ...here.
10914 2017-08-30 Richard Biener <rguenther@suse.de>
10916 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
10917 (dwarf2out_early_finish): Move setting of AT_pubnames from
10918 early debug output to early finish.
10920 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10922 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
10923 and -mdata-region to the assembler.
10925 2017-08-30 Richard Biener <rguenther@suse.de>
10927 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
10929 (gen_subprogram_die): Add DW_AT_object_pointer only early.
10930 (dwarf2out_early_global_decl): Only generate a DIE for the
10931 abstract origin if it doesn't already exist or is a declaration DIE.
10932 (resolve_addr): Do not add the linkage name twice when
10933 generating a stub DIE for the DW_TAG_GNU_call_site target.
10935 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10937 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10938 Use machine_mode rather than int for arg1_mode.
10940 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
10943 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
10944 that the second argument of the built-in functions to unpack
10945 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
10946 switch statement instead a lot of if statements.
10947 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
10948 Allow 64-bit values to be in Altivec registers as well as
10949 traditional floating point registers.
10950 (pack<mode>, FMOVE128_VSX iterator): Likewise.
10952 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
10954 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
10955 MAX_REGS_PER_ADDRESS == 1.
10957 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
10959 * config/i386/i386.opt (flag_fentry): Do not init to -1.
10960 * config/i386/i386.c (ix86_option_override_internal): Simplify
10961 setting of opts->x_flag_entry.
10963 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10964 Jakub Jelinek <jakub@redhat.com>
10965 Richard Biener <rguenther@suse.de>
10967 PR tree-optimization/81503
10968 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
10969 folded constant fits in the target type; reorder tests for clarity.
10971 2017-08-29 Martin Liska <mliska@suse.cz>
10973 * passes.def: Include pass_lower_switch.
10974 * stmt.c (dump_case_nodes): Remove and move to
10975 tree-switch-conversion.
10976 (case_values_threshold): Likewise.
10977 (expand_switch_as_decision_tree_p): Likewise.
10978 (emit_case_decision_tree): Likewise.
10979 (expand_case): Likewise.
10980 (balance_case_nodes): Likewise.
10981 (node_has_low_bound): Likewise.
10982 (node_has_high_bound): Likewise.
10983 (node_is_bounded): Likewise.
10984 (emit_case_nodes): Likewise.
10985 (struct simple_case_node): New struct.
10986 (add_case_node): Remove.
10987 (emit_case_dispatch_table): Use vector instead of case_list.
10988 (reset_out_edges_aux): Remove.
10989 (compute_cases_per_edge): Likewise.
10990 (expand_case): Build list of simple_case_node.
10991 (expand_sjlj_dispatch_table): Use it.
10992 * tree-switch-conversion.c (struct case_node): Moved from
10993 stmt.c and adjusted.
10994 (emit_case_nodes): Likewise.
10995 (node_has_low_bound): Likewise.
10996 (node_has_high_bound): Likewise.
10997 (node_is_bounded): Likewise.
10998 (case_values_threshold): Likewise.
10999 (reset_out_edges_aux): Likewise.
11000 (compute_cases_per_edge): Likewise.
11001 (add_case_node): Likewise.
11002 (dump_case_nodes): Likewise.
11003 (balance_case_nodes): Likewise.
11004 (expand_switch_as_decision_tree_p): Likewise.
11005 (emit_jump): Likewise.
11006 (emit_case_decision_tree): Likewise.
11007 (try_switch_expansion): Likewise.
11008 (do_jump_if_equal): Likewise.
11009 (emit_cmp_and_jump_insns): Likewise.
11010 (fix_phi_operands_for_edge): New function.
11011 (record_phi_operand_mapping): Likewise.
11012 (class pass_lower_switch): New pass.
11013 (pass_lower_switch::execute): New function.
11014 (make_pass_lower_switch): Likewise.
11015 (conditional_probability):
11016 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
11017 * tree-pass.h: Add make_pass_lower_switch.
11019 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11022 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
11025 2017-08-29 Richard Biener <rguenther@suse.de>
11027 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
11028 we do not add a DW_AT_inline attribute twice.
11029 (gen_subprogram_die): Remove code setting DW_AT_inline on
11030 DECL_ABSTRACT_P nodes.
11032 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
11034 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
11035 calls to internal functions.
11036 (gimplify_modify_expr): Likewise.
11037 * tree-call-cdce.c (use_internal_fn): Likewise.
11038 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
11039 (convert_to_divmod): Set the nothrow flag.
11040 * tree-if-conv.c (predicate_mem_writes): Likewise.
11041 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
11042 (vectorizable_call): Likewise.
11043 (vectorizable_store): Likewise.
11044 (vectorizable_load): Likewise.
11045 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
11046 (vect_recog_mask_conversion_pattern): Likewise.
11048 2017-08-29 Martin Liska <mliska@suse.cz>
11051 * gcc.c (driver_handle_option): Add new argument.
11052 * opts-common.c (handle_option): Pass
11053 target_option_override_hook.
11054 * opts-global.c (lang_handle_option): Add new option.
11055 (set_default_handlers): Add new argument.
11056 (decode_options): Likewise.
11057 * opts.c (target_handle_option): Likewise.
11058 (common_handle_option): Call target_option_override_hook.
11059 * opts.h (struct cl_option_handler_func): Add hook for
11060 target option override.
11061 (struct cl_option_handlers): Likewise.
11062 (set_default_handlers): Add new argument.
11063 (decode_options): Likewise.
11064 (common_handle_option): Likewise.
11065 (target_handle_option): Likewise.
11066 * toplev.c (toplev::main): Pass targetm.target_option.override
11069 2017-08-29 Richard Biener <rguenther@suse.de>
11070 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
11072 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
11073 life to the active subtree.
11075 2017-08-28 Jeff Law <law@redhat.com>
11077 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
11078 derive_equivalences.
11079 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
11080 Code moved into....
11081 (edge_info::derive_equivalences): New private member function
11083 * tree-ssa-dom.c (class edge_info): Changed from a struct
11084 to a class. Add ctor/dtor, methods and data members.
11085 (edge_info::edge_info): Renamed from allocate_edge_info.
11086 Initialize additional members.
11087 (edge_info::~edge_info): New.
11088 (free_dom_edge_info): Delete the edge info.
11089 (record_edge_info): Use new class & associated member functions.
11090 Tighten forms for testing for edge equivalences.
11091 (record_temporary_equivalences): Iterate over the simple
11092 equivalences rather than assuming there's only one per edge.
11093 (cprop_into_successor_phis): Iterate over the simple
11094 equivalences rather than assuming there's only one per edge.
11095 (optimize_stmt): Use operand_equal_p rather than pointer
11096 equality for mini-DSE code.
11098 2017-08-28 Nathan Sidwell <nathan@acm.org>
11100 * gcc.c (execute): Fold SIGPIPE handling into switch
11101 statement. Adjust internal error message.
11103 2017-08-28 Richard Biener <rguenther@suse.de>
11106 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
11107 Do nothing for removed DIEs.
11109 2017-08-28 Richard Biener <rguenther@suse.de>
11111 PR tree-optimization/81977
11112 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
11115 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
11118 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
11119 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
11120 using targetm.gen_mem_thread_fence.
11122 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
11125 * config/i386/i386.md (*<btsc><mode>): Change operand 2
11126 predicate to register_operand. Reorder operands.
11127 (*btr<mode>): Ditto.
11128 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
11129 (*btr<mode>_mask): Ditto.
11131 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
11133 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
11134 * config/rs6000/xmmintrin.h: New file.
11135 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
11137 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11140 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
11141 parameter and_insn and return it.
11142 (recombine_lvx_pattern): Insert a copy to ensure availability of
11143 the base register of the copied masking operation at the point of
11144 the instruction replacement.
11145 (recombine_stvx_pattern): Likewise.
11147 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
11149 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
11150 undocumented switches.
11151 (-mpower9-dform-vector): Likewise.
11152 (-mpower9-dform): Likewise.
11153 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
11154 comments to delete references to -mpower9-dform* switches.
11155 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
11156 Delete reference to -mpower9-dform* switches, test for
11157 -mpower9-vector instead.
11158 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
11159 (OTHER_P9_VECTOR_MASKS): Likewise.
11160 (POWERPC_MASKS): Likewise.
11161 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
11162 tests against -mpower9-dform* to -mpower9-vector. Delete code
11163 that checked for -mpower9-dform* consistancy with other options.
11164 Add test for -mpower9-misc to enable other power9 switches.
11165 (rs6000_init_hard_regno_mode_ok): Likewise.
11166 (rs6000_option_override_internal): Likewise.
11167 (rs6000_emit_prologue): Likewise.
11168 (rs6000_emit_epilogue): Likewise.
11169 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
11170 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
11171 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
11173 (emit_fusion_p9_store): Likewise.
11174 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
11175 resetting these macros if the assembler does not support ISA 3.0
11177 (TARGET_P9_DFORM_VECTOR): Likewise.
11178 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
11179 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
11181 2017-08-25 Alan Modra <amodra@gmail.com>
11183 PR rtl-optimization/81747
11184 * cse.c (cse_extended_basic_block): Don't attempt to record
11185 equivalences for degenerate conditional branches that branch
11186 to their fall-through.
11188 2017-08-24 Martin Sebor <msebor@redhat.com>
11190 PR middle-end/81908
11191 * gimple-fold.c (size_must_be_zero_p): New function.
11192 (gimple_fold_builtin_memory_op): Call it.
11194 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
11196 * config/rs6000/mm_malloc.h: New file.
11198 2017-08-24 Bin Cheng <bin.cheng@arm.com>
11200 PR tree-optimization/81913
11201 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
11202 analysis when either IVs in condition can wrap.
11204 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
11206 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
11207 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
11209 2017-08-24 Richard Biener <rguenther@suse.de>
11212 * targhooks.c (default_target_can_inline_p): Properly
11213 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
11214 is present and always compare.
11215 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
11216 infer -mfpmath=sse from TARGET_SSE_P.
11217 (ix86_can_inline_p): Properly use target_option_default_node when
11218 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
11220 2017-08-24 Richard Biener <rguenther@suse.de>
11223 * dwarf2out.c (output_die): Handle flag_generate_offload like
11225 (output_comp_unit): Likewise.
11226 (gen_array_type_die): Likewise.
11227 (dwarf2out_early_finish): Likewise.
11228 (note_variable_value_in_expr): Likewise.
11229 (dwarf2out_finish): Likewise. Adjust assert.
11230 * cgraphunit.c (symbol_table::compile): Move setting of
11231 flag_generate_offload earlier ...
11232 (symbol_table::finalize_compilation_unit): ... here, before
11233 early debug finalization.
11235 2017-08-24 Richard Biener <rguenther@suse.de>
11237 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
11238 and ipa-fnsummary.h.
11239 (ix86_can_inline_p): When ix86_fpmath flags do not match
11240 check whether the callee uses FP math at all.
11242 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
11244 PR middle-end/81931
11245 * tree-ssanames.c (get_nonzero_bits): Use element_precision
11246 instead of TYPE_PRECISION.
11248 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
11249 Alan Hayward <alan.hayward@arm.com>
11250 David Sherwood <david.sherwood@arm.com>
11252 * combine.c (make_extraction): Use subreg_offset_from_lsb.
11254 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
11256 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
11258 (ix86_frame::stack_realign_allocate): New field.
11259 (struct machine_frame_state): Modify comments.
11260 (machine_frame_state::sp_realigned_fp_end): New field.
11261 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
11262 layout calculation.
11263 (sp_valid_at): Add assertion to assure no attempt to access invalid
11264 offset of a realigned stack.
11265 (fp_valid_at): Likewise.
11266 (choose_baseaddr): Modify comments.
11267 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
11268 ix86_expand_prologue.
11269 (ix86_expand_prologue): Modify stack realignment and allocation.
11270 (ix86_expand_epilogue): Modify comments.
11271 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
11272 avx2_runtime, avx512f, and avx512f_runtime.
11274 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
11276 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
11277 (mstackrealign): Do not init to -1.
11278 * config/i386/i386.c (ix86_option_override_internal):
11279 Check opts_set, not opts when setting default value of
11280 opts->x_ix86_force_align_arg_pointer.
11282 2017-08-23 Richard Biener <rguenther@suse.de>
11284 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
11285 lang_hooks.decl_printable_name.
11286 * print-rtl-function.c (print_rtx_function): Likewise.
11287 * tree-pretty-print.c (dump_function_header): Likewise.
11289 2017-08-23 Richard Biener <rguenther@suse.de>
11292 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
11293 -g0 at compile-time.
11295 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11297 PR middle-end/19706
11298 * doc/sourcebuild.texi (Other hardware attributes):
11301 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11303 PR middle-end/19706
11304 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
11305 Fix single-use check.
11307 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
11309 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
11311 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11313 * doc/install.texi: Modify to add more details on running selected
11316 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11318 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
11319 is combined with -mabi=ms.
11320 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
11323 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11325 PR tree-optimization/81488
11326 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
11327 and cached_basis fields.
11328 (MAX_SPREAD): New constant.
11329 (alloc_cand_and_find_basis): Initialize new fields.
11330 (clear_visited): New function.
11331 (create_phi_basis_1): Rename from create_phi_basis, set visited
11332 and cached_basis fields.
11333 (create_phi_basis): New wrapper function.
11334 (phi_add_costs_1): Rename from phi_add_costs, add spread
11335 parameter, set visited field, short-circuit when limits reached.
11336 (phi_add_costs): New wrapper function.
11337 (record_phi_increments_1): Rename from record_phi_increments, set
11339 (record_phi_increments): New wrapper function.
11340 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
11341 (phi_incr_cost): New wrapper function.
11342 (all_phi_incrs_profitable_1): Rename from
11343 all_phi_incrs_profitable, set visited field.
11344 (all_phi_incrs_profitable): New wrapper function.
11346 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11347 Alan Hayward <alan.hayward@arm.com>
11348 David Sherwood <david.sherwood@arm.com>
11350 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
11351 that takes the outer and inner modes.
11352 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
11353 comparison as the canonical test for a paradoxical subreg.
11354 * combine.c (simplify_set): Use paradoxical_subreg_p.
11355 (make_extraction): Likewise.
11356 (force_to_mode): Likewise.
11357 (rtx_equal_for_field_assignment_p): Likewise.
11358 (gen_lowpart_for_combine): Likewise.
11359 (simplify_comparison): Likewise.
11360 * cse.c (equiv_constant): Likewise.
11361 * expmed.c (store_bit_field_1): Likewise.
11362 * final.c (alter_subreg): Likewise.
11363 * fwprop.c (propagate_rtx): Likewise.
11364 (forward_propagate_subreg): Likewise.
11365 * ira-conflicts.c (ira_build_conflicts): Likewise.
11366 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
11367 * lra-constraints.c (curr_insn_transform): Likewise.
11368 (split_reg): Likewise.
11369 * lra-eliminations.c (move_plus_up): Likewise.
11370 (lra_eliminate_regs_1): Likewise.
11371 * recog.c (general_operand): Likewise.
11372 * ree.c (combine_reaching_defs): Likewise.
11373 * reload.c (push_reload): Likewise.
11374 (find_reloads): Likewise.
11375 * reload1.c (elimination_effects): Likewise.
11376 (compute_reload_subreg_offset): Likewise.
11377 (choose_reload_regs): Likewise.
11378 * rtlanal.c (subreg_lsb_1): Likewise.
11379 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11380 (simplify_subreg): Likewise.
11381 * var-tracking.c (track_loc_p): Likewise.
11382 * emit-rtl.c (byte_lowpart_offset): Likewise.
11383 (paradoxical_subreg_p): Delete out-of-line definition.
11385 2017-08-22 Jeff Law <law@redhat.com>
11387 PR tree-optimization/81741
11388 PR tree-optimization/71947
11389 * tree-ssa-dom.c: Include tree-inline.h.
11390 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
11391 equivalences if one is more expensive to compute than the other.
11392 * tree-ssa-scopedtables.h (class const_or_copies): Make
11393 record_const_or_copy_raw method private.
11394 (class avail_exprs_stack): New method simplify_binary_operation.
11395 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
11396 avail_exprs_stack::simplify_binary_operation as needed.
11397 (avail_exprs_stack::simplify_binary_operation): New function.
11399 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
11401 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
11402 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
11403 (DOT_SYMBOLS): Likewise.
11404 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
11405 (RELOCATABLE_NEEDS_FIXUP): Likewise.
11406 (RS6000_ABI_NAME): Likewise.
11407 (TARGET_CMODEL): Likewise.
11408 (TOC_SECTION_ASM_OP): Likewise.
11409 (SET_CMODEL): New macro.
11410 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
11412 2017-08-22 Richard Biener <rguenther@suse.de>
11414 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
11415 to allow for free-lang-data replacements similar to verify_type_variant.
11417 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11418 Alan Hayward <alan.hayward@arm.com>
11419 David Sherwood <david.sherwood@arm.com>
11421 * config/aarch64/aarch64.md (casesi): Use DImode rather than
11422 VOIDmode for the LABEL_REF.
11424 2017-08-22 Richard Biener <rguenther@suse.de>
11426 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
11428 2017-08-22 Richard Biener <rguenther@suse.de>
11430 * common.opt (feliminate-dwarf2-dups): Ignore.
11431 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
11432 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
11433 same_die_p_wrap, compute_section_prefix,
11434 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
11435 (comdat_symbol_id, comdat_symbol_number): Likewise.
11436 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
11438 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
11439 (output_die): Mark unreachable path unreachable.
11440 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
11441 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
11442 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
11443 (dwarf2out_early_finish): Likewise.
11445 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
11447 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
11449 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
11452 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
11453 not VAR_P. Filter attribute warnings with OPT_Wattributes.
11454 (avr_attribute_table) <io, io_low, address>: Initialize
11455 .decl_required with true.
11457 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
11459 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
11460 undocumented debugging options.
11461 (-mvsx-scalar-double): Likewise.
11462 (-mallow-df-permute): Likewise.
11463 (-mvectorize-builtins): Likewise.
11464 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11465 (rs6000_builtin_vectorized_function): Likewise.
11466 (rs6000_builtin_md_vectorized_function): Likewise.
11467 (rs6000_opt_vars): Likewise.
11469 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
11472 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
11473 (*btrq_imm): Rename from *btrq.
11474 (*btcq_imm): Rename from *btcq.
11475 (btsc): New code attribute.
11476 (*<btsc><mode>): New insn pattern.
11477 (*btr<mode>): Ditto.
11478 (*<btsc><mode>_mask): New insn_and_split pattern.
11479 (*btr<mode>_mask): Ditto.
11481 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11482 Alan Hayward <alan.hayward@arm.com>
11483 David Sherwood <david.sherwood@arm.com>
11485 * function.c (pad_below): Simplify padding calculation.
11487 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11488 Alan Hayward <alan.hayward@arm.com>
11489 David Sherwood <david.sherwood@arm.com>
11491 * target.def (function_prologue): Remove frame size argument.
11492 (function_epilogue): Likewise.
11493 * doc/tm.texi: Regenerate.
11494 * final.c (final_start_function): Update call to function_prologue.
11495 (final_end_function): Update call to function_epilogue.
11496 (default_function_pro_epilogue): Remove frame size argument.
11497 * output.h (default_function_pro_epilogue): Likewise.
11498 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
11499 (arm_output_function_prologue): Likewise.
11500 * config/frv/frv.c (frv_function_prologue): Likewise.
11501 (frv_function_epilogue): Likewise.
11502 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
11503 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
11504 (ia64_output_function_epilogue): Likewise.
11505 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
11506 (m32r_output_function_epilogue): Likewise.
11507 * config/microblaze/microblaze.c (microblaze_function_prologue)
11508 (microblaze_function_epilogue): Likewise.
11509 * config/mips/mips.c (mips_output_function_prologue): Likewise.
11510 (mips_output_function_epilogue): Likewise.
11511 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
11512 (mmix_target_asm_function_epilogue): Likewise.
11513 * config/msp430/msp430.c (msp430_start_function): Likewise.
11514 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
11515 (nds32_asm_function_epilogue): Likewise.
11516 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
11517 * config/pa/pa.c (pa_output_function_prologue): Likewise.
11518 (pa_output_function_epilogue): Likewise.
11519 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
11520 (rs6000_output_function_epilogue): Likewise.
11521 * config/rl78/rl78.c (rl78_start_function): Likewise.
11522 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
11523 (rs6000_output_function_epilogue): Likewise.
11524 * config/rx/rx.c (rx_output_function_prologue): Likewise.
11525 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
11526 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
11527 (sparc_asm_function_epilogue): Likewise.
11529 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11531 * tree.h (type_has_mode_precision_p): New function.
11532 * convert.c (convert_to_integer_1): Use it.
11533 * expr.c (expand_expr_real_2): Likewise.
11534 (expand_expr_real_1): Likewise.
11535 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
11536 * match.pd: Likewise.
11537 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
11538 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
11539 * tree-tailcall.c (process_assignment): Likewise.
11540 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11541 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
11542 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
11543 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11544 (vectorizable_assignment): Likewise.
11545 (vectorizable_shift): Likewise.
11546 (vectorizable_operation): Likewise.
11547 * tree-vrp.c (register_edge_assert_for_2): Likewise.
11549 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
11551 * match.pd: Add pow (C, x) simplification.
11553 2017-08-21 Richard Biener <rguenther@suse.de>
11555 PR tree-optimization/81900
11556 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
11557 for blocks with abnormal predecessors.
11558 (compute_antic): Do not set visited flag prematurely.
11560 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
11563 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
11565 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11567 * stor-layout.h (vector_type_mode): Move to...
11568 * tree.h (vector_type_mode): ...here.
11569 * stor-layout.c (vector_type_mode): Move to...
11570 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
11572 2017-08-21 Richard Biener <rguenther@suse.de>
11574 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
11575 register_external_die hooks.
11576 (debug_false_tree_charstarstar_uhwistar): Declare.
11577 (debug_nothing_tree_charstar_uhwi): Likewise.
11578 * debug.c (do_nothing_debug_hooks): Adjust.
11579 (debug_false_tree_charstarstar_uhwistar): New do nothing.
11580 (debug_nothing_tree_charstar_uhwi): Likewise.
11581 * dbxout.c (dbx_debug_hooks): Adjust.
11582 (xcoff_debug_hooks): Likewise.
11583 * sdbout.c (sdb_debug_hooks): Likewise.
11584 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11585 * dwarf2out.c (macinfo_label_base): New global.
11586 (dwarf2out_register_external_die): New function for the
11587 register_external_die hook.
11588 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
11589 (dwarf2_debug_hooks): Use them.
11590 (dwarf2_lineno_debug_hooks): Adjust.
11591 (struct die_struct): Add with_offset flag.
11592 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
11593 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
11594 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
11595 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
11596 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
11597 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
11598 defining section names for the early LTO debug variants.
11599 (reset_indirect_string): New helper.
11600 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
11601 (print_dw_val): Add support for offsetted symbol references.
11602 (get_ultimate_context): Split out from is_cxx.
11603 (is_cxx): Use get_ultimate_context.
11604 (is_fortran): Add decl overload.
11605 (compute_comp_unit_symbol): Split out worker from
11606 compute_section_prefix.
11607 (compute_section_prefix): Call compute_comp_unit_symbol and
11608 set comdat_type_p here.
11609 (output_die): Skip DIE symbol output for the LTO added one.
11610 Handle DIE symbol references with offset.
11611 (output_comp_unit): Guard section name mangling properly.
11612 For LTO debug sections emit a symbol at the section beginning
11613 which we use to refer to its DIEs.
11614 (add_abstract_origin_attribute): For DIEs registered via
11615 dwarf2out_register_external_die directly refer to the early
11616 DIE rather than indirectly through the shadow one we created.
11617 Remove obsolete call to dwarf2out_abstract_function for
11618 non-function/block origins.
11619 (gen_array_type_die): When generating early LTO debug do
11620 not emit DW_AT_string_length.
11621 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
11622 late when in LTO. As suggested place a gcc_unreachable for
11623 the DECL_ABSTRACT_P case.
11624 (gen_subprogram_die): Avoid another specification DIE
11625 for early built declarations/definitions for the late LTO case.
11626 (gen_variable_die): Add type references for late duplicated VLA dies
11628 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
11629 we have the abstract instance already.
11630 (process_scope_var): Adjust decl DIE contexts in LTO which
11631 first puts them in limbo.
11632 (gen_decl_die): Do not generate type DIEs late apart from
11633 types for VLAs or for decls we do not yet have a DIE. Do not
11634 call dwarf2out_abstract_function late.
11635 (dwarf2out_early_global_decl): Make sure to create DIEs
11636 for abstract instances of a decl first.
11637 (dwarf2out_late_global_decl): Adjust comment.
11638 (output_macinfo_op): With multiple macro sections use
11639 macinfo_label_base to distinguish labels.
11640 (output_macinfo): Likewise. Update macinfo_label_base.
11641 Pass in the line info label.
11642 (note_variable_value_in_expr): When generating LTO resolve
11643 all variable values here by generating DIEs as needed.
11644 (init_sections_and_labels): Add early LTO debug flag parameter
11645 and generate different sections and names if set. Add generation
11646 counter for the labels so we can have multiple of them.
11647 (reset_dies): Helper to allow DIEs to be output multiple times.
11648 (dwarf2out_finish): When outputting DIEs to the fat part of an
11649 LTO object first reset DIEs.
11650 (dwarf2out_early_finish): Output early DIEs when generating LTO.
11651 (modified_type_die): Check for decl_ultimate_origin being self
11653 (gen_type_die_with_usage): Likewise.
11654 (gen_typedef_die): Allow decl_ultimate_origin being self.
11655 (set_decl_abstract_flags): Remove.
11656 (set_block_abstract_flags): Likewise.
11657 (dwarf2out_abstract_function): Treat the early generated DIEs
11658 as the abstract copy and only add DW_AT_inline and
11659 DW_AT_artificial here and call set_decl_origin_self.
11660 If the DIE has an abstract origin don't do anything.
11661 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
11662 if we have none yet (Go fails to build one, PR78628).
11663 (variably_modified_type_p): Prevent endless recursion for Ada
11664 cyclic pointer types.
11665 * lto-streamer-in.c: Include debug.h.
11666 (dref_queue): New global.
11667 (lto_read_tree_1): Stream in DIE references.
11668 (lto_input_tree): Register DIE references.
11669 (input_function): Stream DECL_DEBUG_ARGS.
11670 * lto-streamer-out.c: Include debug.h.
11671 (lto_write_tree_1): Output DIE references.
11672 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
11673 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
11674 (output_function): Stream DECL_DEBUG_ARGS.
11675 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
11676 Stream DECL_ABSTRACT_ORIGIN.
11677 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
11678 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
11679 DECL_CONTEXT for file-scope decls.
11680 * lto-streamer.h (struct dref_entry): Declare.
11681 (dref_queue): Likewise.
11682 * cfgexpand.c (pass_expand::execute): Do not call the
11683 outlining_inline_function hook here.
11684 * lto-wrapper.c (debug_obj): New global.
11685 (tool_cleanup): Unlink it if required.
11686 (debug_objcopy): New function.
11687 (run_gcc): Handle early debug sections in the IL files by
11688 extracting them to separate files, partially linkin them and
11689 feeding the result back as result to the linker.
11690 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
11691 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
11692 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
11693 sections into a separate segment.
11694 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
11696 (darwin_asm_dwarf_section): Likewise.
11697 (darwin_asm_output_dwarf_offset): Likewise.
11698 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
11700 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11701 Alan Hayward <alan.hayward@arm.com>
11702 David Sherwood <david.sherwood@arm.com>
11704 * read-md.h (md_reader::record_potential_iterator_use): Replace
11705 pointer argument with an rtx and an index.
11706 * read-rtl.c (iterator_group::apply_iterator): Likewise.
11707 (apply_mode_iterator): Likewise.
11708 (apply_code_iterator): Likewise.
11709 (apply_int_iterator): Likewise.
11710 (apply_subst_iterator): Likewise.
11711 (record_iterator_use): Likewise.
11712 (record_attribute_use): Likewise.
11713 (md_reader::record_potential_iterator_use): Likewise. Update calls
11714 to record_iterator_use and apply_iterator.
11715 (iterator_use): Replace ptr with x and index.
11716 (attribute_use): Likewise.
11717 (apply_attribute_uses): Update calls to apply_iterator.
11718 (apply_iterators): Likewise. Update initialization of iterator_use.
11719 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
11720 and record_potential_iterator_use.
11721 (rtx_reader::read_rtx_operand): Likewise.
11723 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11724 Alan Hayward <alan.hayward@arm.com>
11725 David Sherwood <david.sherwood@arm.com>
11727 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
11730 2017-08-21 Richard Biener <rguenther@suse.de>
11732 PR middle-end/81884
11733 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
11734 at struct end conservatively when comparing common bases.
11736 2017-08-21 Richard Biener <rguenther@suse.de>
11738 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
11739 (mem_ref_in_stmt): Remove.
11740 (determine_max_movement): Use ref index to get at the reference.
11741 (invariantness_dom_walker::before_dom_children): Deal with
11742 lim data already initialized.
11743 (gather_mem_refs_stmt): Initialize lim data and record ref index.
11745 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
11747 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
11748 (TARGET_ISA_ROUND): Ditto.
11749 (TARGET_ROUND): Ditto.
11750 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
11751 * config/i386/i386.md: Ditto.
11752 * config/i386/sse.md: Ditto.
11753 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
11754 with OPTION_MASK_ISA_SSE4_1.
11756 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
11759 * doc/extend.texi (x86 Built-in Functions): Correct the name of
11760 __builtin_ia32_lzcnt_u16.
11762 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
11765 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
11766 (rs6000_set_current_function): Rewrite function to use it.
11768 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
11771 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
11772 and TYPE_WARN_IF_NOT_ALIGN.
11773 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
11774 (handle_warn_if_not_align): New.
11775 (place_union_field): Call handle_warn_if_not_align.
11776 (place_field): Call handle_warn_if_not_align.
11777 Copy TYPE_WARN_IF_NOT_ALIGN.
11778 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
11779 (layout_type): Likewise.
11780 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
11782 (tree_decl_common): Add warn_if_not_align.
11783 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
11784 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
11785 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
11786 (DECL_WARN_IF_NOT_ALIGN): Likewise.
11787 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
11788 * doc/extend.texi: Document warn_if_not_aligned attribute.
11789 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
11791 2017-08-17 Martin Liska <mliska@suse.cz>
11794 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
11795 (get_data_dependence): Use it as pointer type.
11796 (distribute_loop): Likewise.
11798 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11800 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
11801 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
11802 (p8_vmrgow_<mode>_direct): New define_insn.
11803 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
11804 handle endianness for vmrgew and vmrgow permute patterns.
11806 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11808 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
11809 * config/rs6000/rs6000-cpus.def: Remove comment.
11810 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
11811 (POWERPC_MASKS): Likewise.
11812 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
11813 use of TARGET_VSX_TIMODE.
11814 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
11815 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
11816 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
11817 (rs6000_option_override_internal): Remove dead code.
11818 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
11819 (rs6000_legitimize_reload_address): Likewise.
11820 (rs6000_legitimate_address_p): Likewise.
11821 (rs6000_opt_masks): Delete "vsx-timode".
11822 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
11823 from function comment.
11824 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
11825 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
11826 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
11828 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
11829 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
11830 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
11831 (VSX_TI): Likewise.
11833 (define_peephole2): Likewise.
11835 2017-08-17 Martin Sebor <msebor@redhat.com>
11838 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
11839 past the end of an array.
11840 (test_pp_format): Add test cases.
11842 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
11844 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
11845 missing ECF_NOTHROW flags.
11847 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11850 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
11851 operands residing in integer registers.
11852 (*vsx_le_perm_load_<mode>): Likewise.
11853 (*vsx_le_perm_store_<mode>): Likewise.
11854 (define_peephole2): Add peepholes to optimize the above.
11856 2017-08-17 Marek Polacek <polacek@redhat.com>
11858 PR middle-end/81814
11859 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
11860 to mimic what shorten_compare did. Change the return type to bool.
11861 (fold_cond_expr_with_comparison): Update call to
11862 operand_equal_for_comparison_p.
11863 (fold_ternary_loc): Likewise.
11865 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
11867 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
11869 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
11871 2017-08-17 Richard Biener <rguenther@suse.de>
11873 * tree-ssa-structalias.c (solve_graph): When propagating
11874 to successors update the graphs succ edges and avoid duplicate work.
11876 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
11879 * config/i386/i386.c (ix86_option_override_internal): Save target
11880 specific options after ix86_stack_protector_guard_reg was changed.
11882 2017-08-17 Richard Biener <rguenther@suse.de>
11884 PR tree-optimization/81827
11885 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
11886 (new_var_info): Initialize it conservatively.
11887 (get_call_vi): Mark register vars.
11888 (new_scalar_tmp_constraint_exp): Likewise.
11889 (handle_rhs_call): Likewise.
11890 (handle_const_call): Likewise.
11891 (create_function_info_for): Likewise.
11892 (solve_constraints): Sort varinfos to separate register from
11893 non-register vars to pack points-to solution bitmaps during
11896 2017-08-17 Marek Polacek <polacek@redhat.com>
11898 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
11900 2017-08-17 Richard Biener <rguenther@suse.de>
11902 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
11903 to true when overflow is undefined and we saturated the result.
11905 2017-08-17 Alan Modra <amodra@gmail.com>
11908 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
11909 Don't use store multiple if only one reg needs saving.
11910 (interesting_frame_related_regno): New function.
11911 (rs6000_frame_related): Don't emit frame info for regs that
11913 (rs6000_emit_epilogue): Likewise.
11915 2017-08-16 Nathan Sidwell <nathan@acm.org>
11917 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
11918 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
11919 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
11920 * tree.c (free_lang_data_in_type): Use else-if chain. Always
11921 clear TYPE_LANG_1. Remove obsolete member-function stripping.
11922 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
11923 (verify_type): Adjust for TYPE_BINFO move.
11924 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
11925 process TYPE_BINFO directly.
11926 (hash_tree): Likewise.
11927 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11929 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11932 2017-08-16 David Malcolm <dmalcolm@redhat.com>
11934 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
11936 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
11939 * config/i386/i386.md (*anddi_1_btr): Change predicates of
11940 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
11941 Add ix86_binary_operator_ok to insn constraint.
11942 (*iordi_1_bts): Ditto.
11943 (*xordi_1_btc): Ditto.
11944 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
11945 Update corresponding peephole2 pattern.
11949 2017-08-16 Bin Cheng <bin.cheng@arm.com>
11951 PR tree-optimization/81832
11952 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
11953 copy loop header which has IFN_LOOP_DIST_ALIAS call.
11955 2017-08-16 Marek Polacek <polacek@redhat.com>
11958 * fold-const.c (fold_indirect_ref_1): Restore original behavior
11959 regarding size_zero_node.
11961 2017-08-16 Martin Liska <mliska@suse.cz>
11964 * config.gcc: Respect previously set extra_objs in case
11967 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
11969 PR tree-optimization/81835
11970 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
11971 the phi SSA_NAME. Check that the condition in a COND_EXPR does
11972 not depend on the phi.
11974 2017-08-16 Alan Modra <amodra@gmail.com>
11976 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
11979 2017-08-16 Alan Modra <amodra@gmail.com>
11981 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
11982 (save_reg_p): ..into this. Update all callers.
11983 (first_reg_to_save): Simplify.
11985 2017-08-16 Alan Modra <amodra@gmail.com>
11987 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
11990 2017-08-15 Joseph Myers <joseph@codesourcery.com>
11994 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
11995 constant count if that count is less than 32.
11997 2017-08-15 Nathan Sidwell <nathan@acm.org>
11999 * gcc.c (execute): Emit friendlier message if inferior is killed
12000 by an external cause.
12002 2017-08-15 Richard Biener <rguenther@suse.de>
12004 PR tree-optimization/81790
12005 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
12006 CONSTRUCTORs from simplifying and VN.
12008 2017-08-14 Martin Sebor <msebor@redhat.com>
12010 * builtin-attrs.def: Add comments.
12012 2017-08-14 Martin Sebor <msebor@redhat.com>
12015 * doc/extend.texi (attribute nonstring): Document new attribute.
12017 2017-08-14 Martin Sebor <msebor@redhat.com>
12020 * tree-diagnostic.c (default_tree_printer): Handle %G.
12021 * gimple-pretty-print.h (percent_G_format): Declare new function.
12022 * gimple-pretty-print.c (percent_G_format): Define.
12023 * tree-pretty-print.c (percent_K_format): Add argument.
12025 2017-08-14 Martin Sebor <msebor@redhat.com>
12027 PR translation/79998
12028 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
12029 Remove a stray space.
12031 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
12034 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
12035 (*iordi_1_bts): Ditto.
12036 (*xordi_1_btc): Ditto.
12038 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12041 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
12042 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12044 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
12045 quoted strings, and make more translator-friendly.
12046 (darwin_rs6000_override_options): Likewise.
12047 (rs6000_option_override_internal): Likewise.
12048 (rs6000_return_in_memory): Fix overlong line.
12049 (init_cmulative_args): Use quoted strings, and make more
12050 translator-friendly.
12051 (rs6000_pass_by_reference): Fix overlong line.
12052 (def_builtin): Use quoted strings.
12053 (altivec_expand_predicate_builtin): Use quoted strings, and make
12054 more translator-friendly.
12055 (htm_expand_builtin): Use quoted strings.
12056 (cpu_expand_builtin): Use quoted strings, and make more
12057 translator-friendly.
12058 (altivec_expand_builtin): Likewise.
12059 (paired_expand_predicate_builtin): Likewise.
12060 (rs6000_invalid_builtin): Likewise.
12061 (builtin_function_type): Use quoted strings.
12062 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
12063 more translator-friendly.
12064 (rs6000_trampoline_init): Likewise.
12065 (rs6000_handle_altivec_attribute): Likewise.
12066 (rs6000_inner_target_options): Use quoted strings.
12067 (rs6000_disable_incompatible_switches): Likewise.
12068 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
12069 strings, and make more translator-friendly.
12070 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
12072 2017-08-14 Bin Cheng <bin.cheng@arm.com>
12074 PR tree-optimization/81799
12075 * tree-loop-distribution.c (version_loop_by_alias_check): Force
12076 cond_expr to simple gimple operand.
12078 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
12080 PR middle-end/46932
12081 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
12083 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
12087 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
12088 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
12089 TARGET_GASISR_PROLOGUES.
12090 * config/avr/avr.c (avr_option_override): Same.
12091 (avr_pass_pre_proep::execute): Same.
12093 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
12096 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
12097 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
12099 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
12101 * config/i386/i386.md (*load_tp_<mode>): Redefine as
12102 define_insn_and_split. Split to a memory load from 0 in
12103 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
12104 using PTR mode iterator.
12105 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
12106 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
12107 (*add_tp_<mode>): Redefine as define_insn_and_split.
12108 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
12109 address space. Merge with *add_tp_x32 using PTR mode iterator.
12110 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
12111 Split to an add with a memory load from 0 in
12112 DEFAULT_TLS_SEG_REG address space.
12114 2017-08-12 Andrew Pinski <apinski@cavium.com>
12116 * config/aarch64/aarch64-option-extensions.def (rdma):
12117 Fix feature string to what Linux prints out in /proc/cpuinfo.
12119 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
12122 * dwarf2out.c (modified_type_die): For C typedef types that have
12123 an ultimate origin, process the ultimate origin instead of the
12125 (gen_typedef_die): Assert that input DECLs have no ultimate
12127 (gen_type_die_with_usage): For typedef variants that have an
12128 ultimate origin, just call gen_decl_die on the original DECL.
12129 (process_scope_var): Avoid creating DIEs for local typedefs and
12130 concrete static variables.
12132 2017-08-12 Alan Modra <amodra@gmail.com>
12136 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
12137 match gnu-user.h startfile.
12138 (ENDFILE_LINUX_SPEC): Similarly.
12140 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
12143 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
12145 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
12147 2017-08-11 Tamar Christina <tamar.christina@arm.com>
12148 * config/aarch64/aarch64.md (mov<mode>): Change.
12149 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12150 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
12151 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
12153 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
12155 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
12156 for storage order barriers.
12158 2017-08-11 Martin Liska <mliska@suse.cz>
12161 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
12162 variables of void type.
12164 2017-08-11 Martin Liska <mliska@suse.cz>
12166 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
12167 TARGET_SUPPORTS_ALIASES.
12168 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
12169 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
12170 (optimize_weakref): Likewise.
12171 * symtab.c (symtab_node::noninterposable_alias): Likewise.
12172 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
12173 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
12175 2017-08-11 Martin Liska <mliska@suse.cz>
12178 * config/i386/i386.c (make_resolver_func): Do complete
12179 refactoring of the function.
12181 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12184 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
12185 * config/i386/i386.c (ix86_stack_protect_guard): Use
12186 ix86_stack_protect_guard_symbol_str to generate varible declaration.
12187 * doc/invoke.texi (x86 Options): Document
12188 -mstack-protector-guard-symbol= option.
12190 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12192 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
12193 * config/i386/i386.c (ix86_split_stack_guard): New function.
12194 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
12195 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
12196 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
12197 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
12198 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
12199 (split_stack_space_check): Call ix86_split_stack_guard.
12201 2017-08-10 Martin Sebor <msebor@redhat.com>
12203 * print-tree.c (print_node): Print location using the established
12205 Replace spaces with colons.
12206 (debug_raw, debug): Ditto.
12208 2017-08-10 Martin Sebor <msebor@redhat.com>
12211 * pretty-print.c (pp_format): Correct the handling of %s precision.
12213 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
12216 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12218 (ix86_finalize_stack_frame_flags): This. Also clear
12219 frame_pointer_needed if -fno-omit-frame-pointer is used without
12221 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12222 with ix86_finalize_stack_frame_flags.
12223 (ix86_expand_epilogue): Likewise.
12224 (ix86_expand_split_stack_prologue): Likewise.
12225 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
12227 2017-08-10 Martin Liska <mliska@suse.cz>
12230 * c-attribs.c (handle_target_attribute):
12231 Report warning for an empty string argument of target attribute.
12233 2017-08-09 Jakub Jelinek <jakub@redhat.com>
12236 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
12238 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
12239 or DECL_NONLOCAL labels.
12240 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
12241 or DECL_NONLOCAL labels here.
12243 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
12245 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
12246 to indicate when early gimple folding has been disabled.
12247 (rs6000_gimple_fold_builtin): Add debug content.
12248 (rs6000_invalid_builtin): Fix whitespace.
12249 (rs6000_expand_builtin): Fix whitespace.
12250 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
12252 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
12255 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
12256 SAVE_MULTIPLE if not all the registers that saves, should be saved.
12258 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
12260 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
12261 (qdf24xx): Likewise.
12262 * config/aarch64/aarch64.md: Include falkor.md.
12263 * config/aarch64/falkor.md: New.
12265 2017-08-09 Marek Polacek <polacek@redhat.com>
12268 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
12269 * diagnostic.c (emit_diagnostic): Add a comment.
12270 (emit_diagnostic_valist): New function.
12272 2017-08-09 Marek Polacek <polacek@redhat.com>
12275 * input.c (make_location): New overload.
12276 * input.h (make_location): Declare.
12278 2017-08-08 Alan Modra <amodra@gmail.com>
12279 H.J. Lu <hongjiu.lu@intel.com>
12282 * gcc.c (NO_PIE_SPEC): Delete.
12283 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
12285 (LINK_PIE_SPEC): ..to here.
12286 (LINK_COMMAND_SPEC): Support -no-pie.
12287 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
12288 chain of crtbegin*.o selection, update for PIE_SPEC changes and
12290 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
12291 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
12292 (ENDFILE_CRTEND_SPEC): Similarly.
12294 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
12297 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
12298 (mstack-protector-guard-offset=): Ditto.
12299 * config/i386/i386.c (ix86_option_override): Handle
12300 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
12302 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
12303 ix86_stack_protect_guard_offset variables.
12304 (TARGET_STACK_PROTECT_GUARD): Always define.
12305 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
12306 and -mstack-protector-guard-offset= options.
12308 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12310 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
12311 boundary case for the last candidate.
12313 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12315 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
12316 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
12318 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12320 PR middle-end/19706
12321 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
12322 * config/aarch64/aarch64-builtins.c
12323 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
12324 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
12325 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
12327 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12328 Andrew Pinski <pinskia@gmail.com>
12330 PR middle-end/19706
12331 * internal-fn.def (XORSIGN): New.
12332 * optabs.def (xorsign_optab): New.
12333 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
12334 (convert_expand_mult_copysign): New.
12335 (pass_optimize_widening_mul::execute): Call
12336 convert_expand_mult_copysign.
12338 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12340 PR tree-optimization/81354
12341 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
12342 Insert on edges rather than explicitly creating landing pads.
12343 (analyze_candidates_and_replace): Commit edge inserts.
12345 2017-08-08 Richard Biener <rguenther@suse.de>
12347 PR middle-end/81719
12348 * tree-ssa-loop-niter.c: Include tree-dfa.h.
12349 (expand_simple_operations): Also look through ADDR_EXPRs with
12350 MEM_REF bases treating them as POINTER_PLUS_EXPR.
12352 2017-08-08 Richard Biener <rguenther@suse.de>
12354 PR tree-optimization/81723
12355 * tree-vect-slp.c (struct bst_traits): New hash traits.
12356 (bst_fail): New global.
12357 (vect_build_slp_tree_2): New worker, split out from ...
12358 (vect_build_slp_tree): ... this now wrapping it with using
12359 bst_fail set to cache SLP tree build fails. Properly handle
12361 (vect_analyze_slp_instance): Allocate and free bst_fail.
12363 2017-08-08 Martin Liska <mliska@suse.cz>
12366 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
12367 LABEL_DECLs that can be from a different function.
12369 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12371 PR tree-optimization/81744
12372 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
12373 loop's number of iterations.
12375 2017-08-08 Martin Liska <mliska@suse.cz>
12377 * asan.c: Include header files.
12378 * attribs.c (build_decl_attribute_variant): New function moved
12380 (build_type_attribute_qual_variant): Likewise.
12381 (cmp_attrib_identifiers): Likewise.
12382 (simple_cst_list_equal): Likewise.
12383 (omp_declare_simd_clauses_equal): Likewise.
12384 (attribute_value_equal): Likewise.
12385 (comp_type_attributes): Likewise.
12386 (build_type_attribute_variant): Likewise.
12387 (lookup_ident_attribute): Likewise.
12388 (remove_attribute): Likewise.
12389 (merge_attributes): Likewise.
12390 (merge_type_attributes): Likewise.
12391 (merge_decl_attributes): Likewise.
12392 (merge_dllimport_decl_attributes): Likewise.
12393 (handle_dll_attribute): Likewise.
12394 (attribute_list_equal): Likewise.
12395 (attribute_list_contained): Likewise.
12396 * attribs.h (lookup_attribute): New function moved from tree.[ch].
12397 (lookup_attribute_by_prefix): Likewise.
12398 * bb-reorder.c: Include header files.
12399 * builtins.c: Likewise.
12400 * calls.c: Likewise.
12401 * cfgexpand.c: Likewise.
12402 * cgraph.c: Likewise.
12403 * cgraphunit.c: Likewise.
12404 * convert.c: Likewise.
12405 * dwarf2out.c: Likewise.
12406 * final.c: Likewise.
12407 * fold-const.c: Likewise.
12408 * function.c: Likewise.
12409 * gimple-expr.c: Likewise.
12410 * gimple-fold.c: Likewise.
12411 * gimple-pretty-print.c: Likewise.
12412 * gimple.c: Likewise.
12413 * gimplify.c: Likewise.
12414 * hsa-common.c: Likewise.
12415 * hsa-gen.c: Likewise.
12416 * internal-fn.c: Likewise.
12417 * ipa-chkp.c: Likewise.
12418 * ipa-cp.c: Likewise.
12419 * ipa-devirt.c: Likewise.
12420 * ipa-fnsummary.c: Likewise.
12421 * ipa-inline.c: Likewise.
12422 * ipa-visibility.c: Likewise.
12424 * lto-cgraph.c: Likewise.
12425 * omp-expand.c: Likewise.
12426 * omp-general.c: Likewise.
12427 * omp-low.c: Likewise.
12428 * omp-offload.c: Likewise.
12429 * omp-simd-clone.c: Likewise.
12430 * opts-global.c: Likewise.
12431 * passes.c: Likewise.
12432 * predict.c: Likewise.
12433 * sancov.c: Likewise.
12434 * sanopt.c: Likewise.
12435 * symtab.c: Likewise.
12436 * toplev.c: Likewise.
12437 * trans-mem.c: Likewise.
12438 * tree-chkp.c: Likewise.
12439 * tree-eh.c: Likewise.
12440 * tree-into-ssa.c: Likewise.
12441 * tree-object-size.c: Likewise.
12442 * tree-parloops.c: Likewise.
12443 * tree-profile.c: Likewise.
12444 * tree-ssa-ccp.c: Likewise.
12445 * tree-ssa-live.c: Likewise.
12446 * tree-ssa-loop.c: Likewise.
12447 * tree-ssa-sccvn.c: Likewise.
12448 * tree-ssa-structalias.c: Likewise.
12449 * tree-ssa.c: Likewise.
12450 * tree-streamer-in.c: Likewise.
12451 * tree-vectorizer.c: Likewise.
12452 * tree-vrp.c: Likewise.
12453 * tsan.c: Likewise.
12454 * ubsan.c: Likewise.
12455 * varasm.c: Likewise.
12456 * varpool.c: Likewise.
12457 * tree.c: Remove functions moved to attribs.[ch].
12458 * tree.h: Likewise.
12459 * config/aarch64/aarch64.c: Add attrs.h header file.
12460 * config/alpha/alpha.c: Likewise.
12461 * config/arc/arc.c: Likewise.
12462 * config/arm/arm.c: Likewise.
12463 * config/avr/avr.c: Likewise.
12464 * config/bfin/bfin.c: Likewise.
12465 * config/c6x/c6x.c: Likewise.
12466 * config/cr16/cr16.c: Likewise.
12467 * config/cris/cris.c: Likewise.
12468 * config/darwin.c: Likewise.
12469 * config/epiphany/epiphany.c: Likewise.
12470 * config/fr30/fr30.c: Likewise.
12471 * config/frv/frv.c: Likewise.
12472 * config/ft32/ft32.c: Likewise.
12473 * config/h8300/h8300.c: Likewise.
12474 * config/i386/winnt.c: Likewise.
12475 * config/ia64/ia64.c: Likewise.
12476 * config/iq2000/iq2000.c: Likewise.
12477 * config/lm32/lm32.c: Likewise.
12478 * config/m32c/m32c.c: Likewise.
12479 * config/m32r/m32r.c: Likewise.
12480 * config/m68k/m68k.c: Likewise.
12481 * config/mcore/mcore.c: Likewise.
12482 * config/microblaze/microblaze.c: Likewise.
12483 * config/mips/mips.c: Likewise.
12484 * config/mmix/mmix.c: Likewise.
12485 * config/mn10300/mn10300.c: Likewise.
12486 * config/moxie/moxie.c: Likewise.
12487 * config/msp430/msp430.c: Likewise.
12488 * config/nds32/nds32-isr.c: Likewise.
12489 * config/nds32/nds32.c: Likewise.
12490 * config/nios2/nios2.c: Likewise.
12491 * config/nvptx/nvptx.c: Likewise.
12492 * config/pa/pa.c: Likewise.
12493 * config/pdp11/pdp11.c: Likewise.
12494 * config/powerpcspe/powerpcspe.c: Likewise.
12495 * config/riscv/riscv.c: Likewise.
12496 * config/rl78/rl78.c: Likewise.
12497 * config/rx/rx.c: Likewise.
12498 * config/s390/s390.c: Likewise.
12499 * config/sh/sh.c: Likewise.
12500 * config/sol2.c: Likewise.
12501 * config/sparc/sparc.c: Likewise.
12502 * config/spu/spu.c: Likewise.
12503 * config/stormy16/stormy16.c: Likewise.
12504 * config/tilegx/tilegx.c: Likewise.
12505 * config/tilepro/tilepro.c: Likewise.
12506 * config/v850/v850.c: Likewise.
12507 * config/vax/vax.c: Likewise.
12508 * config/visium/visium.c: Likewise.
12509 * config/xtensa/xtensa.c: Likewise.
12511 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12514 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
12515 constraints since the -mupper-regs-* switches have been
12517 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
12518 into a vector from a double word element that was extracted from
12519 another vector, and eliminate extra XXPERMDI instructions.
12520 (vsx_concat_<mode>_2): Likewise.
12521 (vsx_concat_<mode>_3): Likewise.
12522 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
12523 concat to allow optimizing inserts from previous extracts.
12525 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
12527 * config/i386/i386.c (ix86_stack_protect_guard): Generate
12528 memory reference to a SSP offset in TLS address space.
12529 (ix86_print_operand) <case '@'>: Remove.
12530 (ix86_print_operand_punct_valid_p): Remove '@' code.
12531 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
12532 UNSPEC_SP_TLS_TEST.
12533 (stack_tls_protect_set_<mode>): Remove.
12534 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
12535 (stack_tls_protect_test_<mode>): Remove.
12536 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
12538 2017-08-07 Olivier Hainque <hainque@adacore.com>
12541 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
12543 2017-08-07 Douglas Rupp <rupp@adacore.com>
12545 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
12546 variable was referenced as multidir in command.
12548 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12551 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
12554 2017-08-07 Martin Liska <mliska@suse.cz>
12556 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
12557 * config/rl78/rl78.c: Add include of attribs.h.
12558 * config/sh/sh.c: Likewise.
12559 * config/v850/v850.c: Likewise.
12561 2017-08-07 Tom de Vries <tom@codesourcery.com>
12563 PR middle-end/78266
12564 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
12566 2017-08-07 Martin Liska <mliska@suse.cz>
12568 * config/mips/mips.c: Include attribs.h.
12570 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
12573 * doc/invoke.texi: Document change in behvaior for -Ofast for
12576 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
12578 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
12580 (aarch64_pop_regs): Likewise.
12581 (aarch64_gen_load_pair): Likewise.
12582 (aarch64_save_callee_saves): Likewise.
12583 (aarch64_restore_callee_saves): Likewise.
12585 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12587 * config/i386/i386.c: Revert the last change.
12589 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12592 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12594 (ix86_finalize_stack_frame_flags): This. Also clear
12595 frame_pointer_needed if -fno-omit-frame-pointer is used without
12597 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12598 with ix86_finalize_stack_frame_flags.
12599 (ix86_expand_epilogue): Likewise.
12600 (ix86_expand_split_stack_prologue): Likewise.
12602 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12605 * config/i386/i386.c (get_builtin_code_for_version): Set priority
12606 to P_AES for Westmere.
12608 2017-08-07 Jonathan Yong <10walls@gmail.com>
12610 * config/i386/mingw.opt (fset-stack-executable): Removed.
12611 * config/i386/cygming.opt (fset-stack-executable): Moved
12613 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
12615 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
12617 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
12619 2017-08-07 Marek Polacek <polacek@redhat.com>
12621 PR middle-end/81737
12622 * fold-const.c (fold_indirect_ref_1): Check type_domain.
12624 2017-08-07 Martin Liska <mliska@suse.cz>
12626 * attribs.h (canonicalize_attr_name): New function.
12627 (cmp_attribs): Move from c-format.c and adjusted.
12628 (is_attribute_p): Moved from tree.h.
12629 * tree-inline.c: Add new includes.
12630 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
12631 (private_is_attribute_p): Remove.
12632 (private_lookup_attribute): Likewise.
12633 (private_lookup_attribute_by_prefix): Simplify.
12634 (remove_attribute): Use is_attribute_p.
12635 * tree.h: Remove removed declarations.
12637 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12639 PR middle-end/81698
12640 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
12641 instead of computing it in the function. Formatting fix.
12642 (expand_case): Don't rely on default_edge being the first edge,
12643 clear it if removing it, pass default_edge to
12644 emit_case_dispatch_table.
12645 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
12648 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
12650 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
12651 insn in the function, emit NOP after the insn.
12653 2017-08-06 Tom de Vries <tom@codesourcery.com>
12655 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
12658 2017-08-06 Tom de Vries <tom@codesourcery.com>
12660 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
12663 2017-08-04 Yury Gribov <tetra2005@gmail.com>
12665 PR tree-optimization/57371
12666 * match.pd: New pattern.
12668 2017-08-04 Marek Polacek <polacek@redhat.com>
12670 PR middle-end/81695
12671 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
12672 perform the computation in offset_int.
12674 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12676 PR tree-optimization/81136
12677 * tree-vectorizer.h: Include tree-hash-traits.h.
12678 (vec_base_alignments): New typedef.
12679 (vec_info): Add a base_alignments field.
12680 (vect_record_base_alignments): Declare.
12681 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
12683 (DR_IS_CONDITIONAL_IN_STMT): New macro.
12684 (create_data_ref): Add an is_conditional_in_stmt argument.
12685 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
12686 the is_conditional_in_stmt field.
12687 (data_ref_loc): Add an is_conditional_in_stmt field.
12688 (get_references_in_stmt): Set the is_conditional_in_stmt field.
12689 (find_data_references_in_stmt): Update call to create_data_ref.
12690 (graphite_find_data_references_in_stmt): Likewise.
12691 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
12692 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
12693 (vect_record_base_alignment): New function.
12694 (vect_record_base_alignments): Likewise.
12695 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
12696 for nested statements even if we fail to compute a misalignment.
12697 Use pooled base alignments for unconditional references.
12698 (vect_find_same_alignment_drs): Compare base addresses instead
12700 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
12701 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
12703 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12705 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
12706 Add an explicit name for the enum. Use auto_vec for slp_instances
12707 and grouped_stores.
12708 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
12710 (_bb_vec_info): Add a constructor and destructor.
12711 (vinfo_for_stmt): Return NULL for uids of -1 as well.
12712 (destroy_loop_vec_info): Delete.
12713 (vect_destroy_datarefs): Likewise.
12714 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
12715 (vec_info::vec_info): New function.
12716 (vec_info::~vec_info): Likewise.
12717 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
12718 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
12719 destroy_loop_vec_info.
12720 * tree-vect-loop.c (new_loop_vec_info): Replace with...
12721 (_loop_vec_info::_loop_vec_info): ...this.
12722 (destroy_loop_vec_info): Replace with...
12723 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
12724 the stmt_vec_infos. Leave handling of vec_info information to its
12725 destructor. Remove explicit vector releases.
12726 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
12727 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
12728 * tree-vect-slp.c (new_bb_vec_info): Replace with...
12729 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
12730 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
12731 (destroy_bb_vec_info): Replace with...
12732 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
12733 information to its destructor.
12734 (vect_slp_analyze_bb_1): Use new and delete instead of
12735 new_bb_vec_info and destroy_bb_vec_info.
12736 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
12739 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12741 * tree-data-ref.h (subscript): Add access_fn field.
12742 (data_dependence_relation): Add could_be_independent_p.
12743 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
12744 (same_access_functions): Move to tree-data-ref.c.
12745 * tree-data-ref.c (ref_contains_union_access_p): New function.
12746 (access_fn_component_p): Likewise.
12747 (access_fn_components_comparable_p): Likewise.
12748 (dr_analyze_indices): Add a reference to access_fn_component_p.
12749 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
12751 (constant_access_functions): Likewise.
12752 (add_other_self_distances): Likewise.
12753 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
12754 (initialize_data_dependence_relation): Use XCNEW and remove
12755 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
12756 of access functions that have the same type. Allow the
12757 subsequence to end with different bases in some circumstances.
12758 Record the chosen access functions in SUB_ACCESS_FN.
12759 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
12760 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
12761 (subscript_dependence_tester_1): Likewise dra and drb.
12762 (build_classic_dist_vector): Update calls accordingly.
12763 (subscript_dependence_tester): Likewise.
12764 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
12765 DDR_COULD_BE_INDEPENDENT_P.
12766 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
12767 comp_alias_ddrs instead of may_alias_ddrs.
12768 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
12770 (vect_analyze_data_ref_dependence): Use it if
12771 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
12772 distance vectors if that fails.
12773 (dependence_distance_ge_vf): New function.
12774 (vect_prune_runtime_alias_test_list): Use it. Don't clear
12775 LOOP_VINFO_MAY_ALIAS_DDRS.
12777 2017-08-04 Richard Biener <rguenther@suse.de>
12779 PR middle-end/81705
12780 * fold-const.c (fold_binary_loc): Properly restrict
12781 minus_var0 && minus_var1 case when associating undefined overflow
12784 2017-08-04 Tom de Vries <tom@codesourcery.com>
12786 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
12788 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12790 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12791 Don't start diagnostic messages with a capital letter.
12792 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12794 (rs6000_invalid_builtin): Likewise.
12795 (rs6000_trampoline_init): Likewise.
12797 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12800 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
12801 after setting changeable df flags.
12803 2017-08-03 Richard Biener <rguenther@suse.de>
12805 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
12806 up if the use is in USE - X.
12808 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
12810 * toplev.c (dumpfile.h): New include.
12811 (internal_error_reentered): New static function. Use it...
12812 (internal_error_function): ...here to handle reentered internal_error.
12814 2017-08-03 Richard Biener <rguenther@suse.de>
12816 PR middle-end/81148
12817 * fold-const.c (split_tree): Add minus_var and minus_con
12818 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
12819 here but always use minus_*.
12820 (associate_trees): Assert we never associate with MINUS_EXPR
12821 and NULL first operand. Do not recurse for PLUS_EXPR operands
12822 when associating as MINUS_EXPR either.
12823 (fold_binary_loc): Track minus_var and minus_con.
12825 2017-08-03 Tom de Vries <tom@codesourcery.com>
12828 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
12829 ACCEL_COMPILER, apply finish_options on
12830 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
12832 2017-08-03 Tom de Vries <tom@codesourcery.com>
12835 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
12836 function_entry_patch_area_size > 0.
12838 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12841 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
12842 instead of 10??LU, perform unit multiplication in wide_int,
12843 don't change alloc_object_size_limit if the limit is larger
12846 PR tree-optimization/81655
12847 PR tree-optimization/81588
12848 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
12849 the case when ranges[i].low and high are 1 for unsigned type with
12852 PR middle-end/81052
12853 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
12854 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
12856 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12858 * tree-vrp.h: Add include guard.
12860 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
12863 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
12864 (ud2): New insn pattern.
12865 * config/i386/i386.c (ix86_expand_epilogue):
12866 For naked functions, generate ud2 instead of trap insn.
12868 2017-08-02 Marek Polacek <polacek@redhat.com>
12871 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
12873 2017-08-02 Tom de Vries <tom@codesourcery.com>
12874 Cesar Philippidis <cesar@codesourcery.com>
12876 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
12877 Add missing edge probabilities.
12879 2017-08-02 Tamar Christina <tamar.christina@arm.com>
12881 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
12882 Correct endianness.
12884 2017-08-02 Jakub Jelinek <jakub@redhat.com>
12886 PR middle-end/79499
12887 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
12888 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
12889 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
12890 prologue_seq sequences - if any.
12892 2017-08-02 Richard Biener <rguenther@suse.de>
12894 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
12895 via vectors if supported, integer extracts via punning if supported
12896 or otherwise vector extracts.
12898 2017-08-02 Richard Biener <rguenther@suse.de>
12900 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
12902 (bitmap_insert_into_set): ... this.
12904 2017-08-02 Richard Biener <rguenther@suse.de>
12906 PR tree-optimization/81633
12908 2015-08-17 Alan Hayward <alan.hayward@arm.com>
12910 PR tree-optimization/71752
12911 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
12913 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
12915 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
12916 (machine_function::call_ms2sysv_pad_out): Remove field.
12917 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
12918 (ix86_compute_frame_layout): Likewise.
12920 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
12923 * config/i386/i386.c (ix86_set_func_type): Disallow naked
12924 attribute with interrupt attribute.
12926 2017-08-01 Andrew Pinski <apinski@cavium.com>
12928 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
12929 BIT_INSERT_EXPR's operand 1
12930 to see if the types precision matches.
12932 2017-08-01 Martin Liska <mliska@suse.cz>
12934 PR middle-end/70140
12935 * builtins.c (expand_builtin_memcpy_args): Remove.
12936 (expand_builtin_memcpy): Call newly added function
12937 expand_builtin_memory_copy_args.
12938 (expand_builtin_memcpy_with_bounds): Likewise.
12939 (expand_builtin_mempcpy): Remove last argument.
12940 (expand_builtin_mempcpy_with_bounds): Likewise.
12941 (expand_builtin_memory_copy_args): New function created from
12942 expand_builtin_mempcpy_args with small modifications.
12943 (expand_builtin_mempcpy_args): Remove.
12944 (expand_builtin_stpcpy): Remove unused argument.
12945 (expand_builtin): Likewise.
12946 (expand_builtin_with_bounds): Likewise.
12948 2017-08-01 Martin Liska <mliska@suse.cz>
12951 Make mempcpy more optimal (PR middle-end/70140).
12953 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12956 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12957 __builtin_vec_cmpne verify both arguments are compatible vectors
12958 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
12959 verify arg1_type is a pointer or array type. For __builtin_vec_st,
12960 move computation of aligned to after checking the argument types.
12964 * config/rs6000/vsx.md (vextract_fp_from_shorth,
12965 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
12968 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
12969 Doug Rupp <rupp@adacore.com>
12970 Olivier Hainque <hainque@adacore.com>
12972 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
12973 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
12975 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
12976 for TARGET_OS_CPP_BUILTIN.
12977 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
12978 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
12980 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
12981 passing required abi options to the assembler for EABI configurations.
12982 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
12983 of .text.hot and .text.unlikely sections for kernel modules when
12984 using ARM style exceptions.
12985 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
12986 options. Add EXTRA_CC1_SPEC.
12987 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
12989 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
12991 (ARM_TARGET2_DWARF_FORMAT): Define.
12992 * config/arm/t-vxworks: Adjust multilib control to removal of the
12993 Diab command line options.
12995 2017-08-01 Martin Liska <mliska@suse.cz>
12997 PR gcov-profile/81561
12998 * gcov.c (unblock): Make unblocking safe as we need to preserve
12999 index correspondence of blocks and block_lists.
13001 2017-08-01 Richard Biener <rguenther@suse.de>
13003 PR tree-optimization/81181
13004 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
13005 (compute_antic): ... end of iteration here.
13007 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
13009 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
13010 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
13011 (ftree-slp-vectorize): Likewise.
13012 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
13013 can no longer be set independent of flag_tree_loop_vectorize.
13014 * omp-general.c (emp_max_vf): Likewise.
13015 * opts.c (enable_fdo_optimizations): Remove references to
13016 flag_tree_vectorize, these are now implicit.
13017 (common_handle_option): Remove handling for OPT_ftree_vectorize,
13018 and leave it for the options machinery.
13020 2017-08-01 Martin Liska <mliska@suse.cz>
13022 PR middle-end/70140
13023 * builtins.c (expand_builtin_memcpy_args): Remove.
13024 (expand_builtin_memcpy): Call newly added function
13025 expand_builtin_memory_copy_args.
13026 (expand_builtin_memcpy_with_bounds): Likewise.
13027 (expand_builtin_mempcpy): Remove last argument.
13028 (expand_builtin_mempcpy_with_bounds): Likewise.
13029 (expand_builtin_memory_copy_args): New function created from
13030 expand_builtin_mempcpy_args with small modifications.
13031 (expand_builtin_mempcpy_args): Remove.
13032 (expand_builtin_stpcpy): Remove unused argument.
13033 (expand_builtin): Likewise.
13034 (expand_builtin_with_bounds): Likewise.
13036 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13039 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
13040 print "ds:" only for immediates in generic address space.
13042 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
13045 * config/i386/i386.c (ix86_funciton_naked): New prototype.
13046 (ix86_function_ok_for_sibcall): Return false for naked functions.
13048 2017-08-01 Richard Biener <rguenther@suse.de>
13050 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
13051 (compute_antic): Seed worklist with exit block predecessors.
13052 * cfganal.c (dfs_find_deadend): For a cycle return the source
13053 of the edge closing it.
13055 2017-08-01 Tamar Christina <tamar.christina@arm.com>
13057 * config/aarch64/aarch64.c
13058 (aarch64_can_const_movi_rtx_p): Move 0 check.
13060 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13062 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
13063 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
13065 * match.pd: Ditto in address comparison pattern.
13067 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13069 PR tree-optimization/81627
13070 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
13071 closed ssa form for store-store chain.
13073 2017-08-01 Bin Cheng <bin.cheng@arm.com>
13075 PR tree-optimization/81620
13076 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
13077 for store-store chain.
13079 2017-08-01 Jakub Jelinek <jakub@redhat.com>
13081 PR tree-optimization/81588
13082 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
13083 ranges[i].in_p, invert comparison code ccode. For >/>=,
13084 swap rhs1 and rhs2 and comparison code unconditionally,
13085 for </<= don't do that. Don't swap rhs1/rhs2 again if
13086 ranges[i].in_p, instead invert comparison code ccode if
13087 opcode or oe->rank is BIT_IOR_EXPR.
13090 * optabs.def (vec_extract_optab, vec_init_optab): Change from
13091 a direct optab to conversion optab.
13092 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
13093 with GET_MODE_INNER as last argument instead of optab_handler.
13094 * expmed.c (extract_bit_field_1): Likewise. Use vector from
13095 vector extraction if possible and optab is available.
13096 * expr.c (store_constructor): Use convert_optab_handler instead
13097 of optab_handler. Use vector initialization from smaller
13098 vectors if possible and optab is available.
13099 * tree-vect-stmts.c (vectorizable_load): Likewise.
13100 * doc/md.texi (vec_extract, vec_init): Document that the optabs
13101 now have two modes.
13102 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
13103 of vec_init from half-sized vectors with the same element mode.
13104 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
13105 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
13106 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
13107 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
13108 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
13109 after mode in gen_vec_extract* calls.
13110 (vec_extract<mode>): Renamed to ...
13111 (vec_extract<mode><ssescalarmodelower>): ... this.
13112 (vec_extract<mode><ssehalfvecmodelower>): New expander.
13113 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
13114 element mode after mode in gen_vec_init* calls.
13115 (VEC_INIT_HALF_MODE): New mode iterator.
13116 (vec_init<mode>): Renamed to ...
13117 (vec_init<mode><ssescalarmodelower>): ... this.
13118 (vec_init<mode><ssehalfvecmodelower>): New expander.
13119 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
13120 (vec_extractv2sfsf): ... this.
13121 (vec_initv2sf): Renamed to ...
13122 (vec_initv2sfsf): ... this.
13123 (vec_extractv2si): Renamed to ...
13124 (vec_extractv2sisi): ... this.
13125 (vec_initv2si): Renamed to ...
13126 (vec_initv2sisi): ... this.
13127 (vec_extractv4hi): Renamed to ...
13128 (vec_extractv4hihi): ... this.
13129 (vec_initv4hi): Renamed to ...
13130 (vec_initv4hihi): ... this.
13131 (vec_extractv8qi): Renamed to ...
13132 (vec_extractv8qiqi): ... this.
13133 (vec_initv8qi): Renamed to ...
13134 (vec_initv8qiqi): ... this.
13135 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
13136 (vec_init<mode>): Renamed to ...
13137 (vec_init<mode><VEC_base_l>): ... this.
13138 (vec_extract<mode>): Renamed to ...
13139 (vec_extract<mode><VEC_base_l>): ... this.
13140 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
13141 (vec_initv2sfsf): ... this.
13142 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
13143 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13144 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
13145 element mode after mode in gen_vec_init* calls.
13146 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
13147 (vec_init<mode><Vel>): ... this.
13148 (vec_extract<mode>): Renamed to ...
13149 (vec_extract<mode><Vel>): ... this.
13150 * config/aarch64/iterators.md (Vel): New mode attribute.
13151 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
13152 Add element mode after mode in gen_vec_extract* calls.
13153 * config/s390/vector.md (non_vec_l): New mode attribute.
13154 (vec_extract<mode>): Renamed to ...
13155 (vec_extract<mode><non_vec_l>): ... this.
13156 (vec_init<mode>): Renamed to ...
13157 (vec_init<mode><non_vec_l>): ... this.
13158 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
13159 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
13161 * config/arm/iterators.md (V_elem_l): New mode attribute.
13162 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
13163 (vec_extract<mode><V_elem_l>): ... this.
13164 (vec_extractv2di): Renamed to ...
13165 (vec_extractv2didi): ... this.
13166 (vec_init<mode>): Renamed to ...
13167 (vec_init<mode><V_elem_l>): ... this.
13168 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
13169 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
13170 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
13171 Add element mode after gen_vec_extract* calls.
13172 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
13173 (vec_init<mode><unitmode>): ... this.
13174 (vec_extract<mode>): Renamed to ...
13175 (vec_extract<mode><unitmode>): ... this.
13176 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
13177 (vec_init<mode><unitmode>): ... this.
13178 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
13179 (vec_initv2sfsf): ... this.
13180 (vec_extractv2sf): Renamed to ...
13181 (vec_extractv2sfsf): ... this.
13182 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
13183 Add element mode after gen_vec_extract* calls.
13184 * config/mips/mips.md (unitmode): New mode iterator.
13185 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
13186 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
13187 * config/spu/spu.md (inner_l): New mode attribute.
13188 (vec_init<mode>): Renamed to ...
13189 (vec_init<mode><inner_l>): ... this.
13190 (vec_extract<mode>): Renamed to ...
13191 (vec_extract<mode><inner_l>): ... this.
13192 * config/sparc/sparc.md (veltmode): New mode iterator.
13193 (vec_init<VMALL:mode>): Renamed to ...
13194 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
13195 * config/ia64/vect.md (vec_initv2si): Renamed to ...
13196 (vec_initv2sisi): ... this.
13197 (vec_initv2sf): Renamed to ...
13198 (vec_initv2sfsf): ... this.
13199 (vec_extractv2sf): Renamed to ...
13200 (vec_extractv2sfsf): ... this.
13201 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
13202 (vec_init<mode>): Renamed to ...
13203 (vec_init<mode><VEC_base_l>): ... this.
13204 (vec_extract<mode>): Renamed to ...
13205 (vec_extract<mode><VEC_base_l>): ... this.
13206 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
13207 (vec_initv2sfsf): ... this.
13208 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
13209 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13210 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
13211 gen_vec_init* calls.
13213 2017-08-01 Richard Biener <rguenther@suse.de>
13215 PR tree-optimization/81297
13216 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
13217 TREE_OVERFLOW from INTEGER_CSTs.
13219 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
13221 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
13223 2017-07-31 Carl Love <cel@us.ibm.com>
13225 * config/rs6000/rs6000-c: Add support for built-in functions
13226 vector signed char vec_xl_be (signed long long, signed char *);
13227 vector unsigned char vec_xl_be (signed long long, unsigned char *);
13228 vector signed int vec_xl_be (signed long long, signed int *);
13229 vector unsigned int vec_xl_be (signed long long, unsigned int *);
13230 vector signed long long vec_xl_be (signed long long, signed long long *);
13231 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
13232 vector signed short vec_xl_be (signed long long, signed short *);
13233 vector unsigned short vec_xl_be (signed long long, unsigned short *);
13234 vector double vec_xl_be (signed long long, double *);
13235 vector float vec_xl_be (signed long long, float *);
13236 * config/rs6000/altivec.h (vec_xl_be): Add #define.
13237 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
13238 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
13240 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
13241 (altivec_expand_builtin): Add switch statement to call
13242 altivec_expand_xl_be for each builtin.
13243 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
13244 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
13245 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
13246 __builtin_vsx_le_be_v16qi.
13247 * doc/extend.texi: Update the built-in documentation file for the
13248 new built-in functions.
13250 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13253 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
13255 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13257 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13259 * config.gcc: Add z14.
13260 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
13261 CPU model numbers for z13s and z14.
13262 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
13264 * config/s390/s390-opts.h (enum processor_type): Rename
13265 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13266 * config/s390/s390.c (processor_table): Add field for CPU name to
13267 be passed to Binutils.
13268 (s390_asm_output_machine_for_arch): Use the new field in
13269 processor_table for Binutils.
13270 (s390_expand_builtin): Replace arch12 with z14.
13271 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13272 (s390_get_sched_attrmask): Likewise.
13273 (s390_get_unit_mask): Likewise.
13274 * config/s390/s390.opt: Add z14 to processor_type enum.
13276 2017-07-31 Martin Jambor <mjambor@suse.cz>
13279 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
13280 regardless of optimization level.
13282 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
13283 Martin Liska <mliska@suse.cz>
13285 * predict.def: Remove old comment and adjust probability.
13286 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
13287 PREDICT statements.
13289 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13292 * config/i386/i386.c (ix86_function_naked): New function.
13293 (ix86_can_use_return_insn_p): Return false for naked functions.
13294 (ix86_expand_prologue): Skip prologue for naked functions.
13295 (ix86_expand_epilogue): Skip epilogue for naked functions
13296 and emit trap instruction.
13297 (ix86_warn_func_return): New function.
13298 (ix86_attribute_table): Add "naked" attribute specification.
13299 (TARGET_WARN_FUNC_RETURN): Define.
13300 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
13302 2017-07-31 Martin Liska <mliska@suse.cz>
13304 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
13305 (dump_gimple_bb_header): Always dump BB info.
13306 (pp_cfg_jump): Do not append info about BB when dumping a jump.
13308 2017-07-31 Martin Liska <mliska@suse.cz>
13311 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
13312 also with current_function_decl non-null equality.
13314 2017-07-31 Jakub Jelinek <jakub@redhat.com>
13317 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
13318 change type to the element type, instead add eltype variable and
13319 use it where we are interested in the element type.
13321 PR tree-optimization/81603
13322 * ipa-polymorphic-call.c
13323 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
13324 offset arithmetic in offset_int, bail out if the resulting bit offset
13325 doesn't fit into shwi.
13327 2017-07-31 Martin Liska <mliska@suse.cz>
13329 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
13330 (gimplify_save_expr): Fix comment.
13332 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
13335 * config/i386/i386.c (ix86_function_arg): Update arguments for
13337 (ix86_compute_frame_layout): Set the initial stack offset to
13338 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
13339 INCOMING_FRAME_SP_OFFSET.
13340 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
13341 stack before exception handler returns.
13342 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
13343 the 'ERROR_CODE' for exception handler.
13345 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
13347 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
13348 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
13349 (ASM_OUTPUT_REG_POP): Ditto.
13350 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
13351 instead of asm_fprintf to output pure string.
13353 2017-07-29 Jakub Jelinek <jakub@redhat.com>
13355 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
13356 to imported_module_or_decl hook.
13357 (debug_nothing_tree_tree_tree_bool): Remove.
13358 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
13359 * debug.c (do_nothing_debug_hooks): Use
13360 debug_nothing_tree_tree_tree_bool_bool instead of
13361 debug_nothing_tree_tree_tree_bool.
13362 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13363 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
13364 * sdbout.c (sdb_debug_hooks): Likewise.
13365 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
13366 (gen_namespace_die): Add DW_AT_export_symbols attribute if
13368 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
13369 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
13370 attribute, don't add anything.
13372 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13374 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
13375 (fold_build2_stat_loc): Likewise.
13376 (fold_build3_stat_loc): Likewise.
13377 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
13378 (fold_build1_loc): Remove macro.
13379 (fold_build2_loc): Likewise.
13380 (fold_build3_loc): Likewise.
13382 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13384 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
13385 (gimple_build_debug_bind_source_stat): Likewise.
13386 * gimple.h (gimple_build_debug_bind): Remove macro.
13387 (gimple_build_debug_bind_source): Likewise.
13389 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13391 * bitmap.c (bitmap_alloc): Adjust.
13392 (bitmap_gc_alloc): Likewise.
13393 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
13395 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13397 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
13398 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
13399 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
13400 (bitmap_gc_alloc_stat): Likewise.
13401 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
13403 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13405 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
13406 * rtl.h (shallow_copy_rtx): Remove macro.
13408 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13410 * emit-rtl.c (gen_raw_REG): Adjust.
13411 * gengenrtl.c (gendef): Likewise.
13412 * rtl.c (rtx_alloc_stat): Remove _stat from name.
13413 * rtl.h (rtx_alloc): Remove macro.
13415 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13417 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
13418 (build_tree_list_stat): Likewise.
13419 * tree.h (build_tree_list): Remove macro.
13420 (build_tree_list_vec): Likewise.
13422 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13424 * tree.c (make_vector_stat): Remove _stat from name.
13425 (build_vector_stat): Likewise.
13426 * tree.h (make_vector_stat): Remove macro.
13427 (build_vector_stat): Likewise.
13429 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13431 * tree.h (build_var_debug_value): Remove prototype.
13433 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13435 * tree.c (tree_cons_stat): Remove _stat from name.
13436 * tree.h (tree_cons): Remove macro.
13438 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13440 * tree.c (build_vl_exp_stat): Remove _stat from name.
13441 * tree.h (build_vl_exp): Remove macro.
13443 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13445 * tree.c (build_decl_stat): Remove _stat from name.
13446 * tree.h (build_decl): Remove macro.
13448 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13450 * gimple.c (gimple_build_with_ops_stat): Adjust.
13451 (gimple_alloc_stat): Remove _stat from name.
13452 * gimple.h (gimple_alloc): Remove macro.
13454 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13456 * tree.c (make_tree_vec_stat): Remove _stat from name.
13457 (grow_tree_vec_stat): Likewise.
13458 * tree.h (make_tree_vec_stat): Adjust prototype.
13459 (grow_tree_vec_stat): Likewise.
13460 (make_tree_vec): Remove macro.
13461 (grow_tree_vec): Likewise.
13463 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13465 * fold-const.c (fold_build1_stat_loc): Adjust.
13466 (fold_build2_stat_loc): Likewise.
13467 (fold_build3_stat_loc): Likewise.
13468 * tree.c (build0_stat): Remove _stat from name.
13469 (build1_stat): Likewise.
13470 (build2_stat): Likewise.
13471 (build3_stat): Likewise.
13472 (build4_stat): Likewise.
13473 (build5_stat): Likewise.
13474 * tree.h (build1_loc): Remove macro, and rename _stat function
13476 (build2_loc): Likewise.
13477 (build3_loc): Likewise.
13478 (build4_loc): Likewise.
13479 (build5_loc): Likewise.
13481 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13483 * tree.c (make_int_cst_stat): Remove _stat from name.
13484 * tree.h (make_int_cst_stat): Adjust prototype.
13485 (make_int_cst): Remove macro.
13487 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13489 * tree.c (make_tre_binfo_stat): Remove _stat from name.
13490 * tree.h (make_tree_binfo_stat): Adjust prototype.
13491 (make_tree_binfo): Remove.
13493 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13495 * tree.c (copy_node_stat): Rename to copy_node.
13496 (build_distinct_type_copy): Adjust.
13497 * tree.h (copy_node_stat): Adjust prototype.
13498 (copy_node): Remove macro.
13500 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13502 * tree.c (make_node_stat): rename to make_node.
13503 (build_tree_list_stat): Adjust.
13504 (build0_stat): Likewise.
13505 (build2_stat): Likewise.
13506 (build3_stat): Likewise.
13507 (build4_stat): Likewise.
13508 (build5_stat): Likewise.
13509 (build_decl_stat): Likewise.
13510 * tree.h (make_node_stat): Adjust prototype.
13511 (make_node): remove macro.
13513 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
13515 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
13516 (PPC_FEATURE2_SCV): Likewise.
13517 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
13519 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13521 * config/aarch64/aarch64.c
13522 (aarch64_internal_mov_immediate): Add new special pattern.
13523 * config/aarch64/aarch64.md (*movdi_aarch64):
13524 Add reg/32bit const mov case.
13526 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13527 Richard Sandiford <richard.sandiford@linaro.org>
13529 * config/aarch64/aarch64.md (mov<mode>): Generalize.
13530 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13531 Add integer and movi cases.
13532 (movi-split-hf-df-sf split, fp16): New.
13533 (enabled): Added TARGET_FP_F16INST.
13534 * config/aarch64/iterators.md (GPF_HF): New.
13535 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
13537 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13539 * config/aarch64/aarch64.c
13540 (aarch64_simd_container_mode): Add prototype.
13541 (aarch64_expand_mov_immediate): Add HI support.
13542 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
13543 (aarch64_can_const_movi_rtx_p): New.
13544 (aarch64_preferred_reload_class):
13545 Remove restrictions of using FP registers for certain SIMD operations.
13546 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
13547 (aarch64_valid_floating_const): Add integer move validation.
13548 (aarch64_simd_imm_scalar_p): Remove.
13549 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
13550 (aarch64_legitimate_constant_p): Expand list of supported cases.
13551 * config/aarch64/aarch64-protos.h
13552 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
13553 (aarch64_reinterpret_float_as_int): New.
13554 (aarch64_simd_imm_scalar_p): Remove.
13555 * config/aarch64/constraints.md (Uvi): New.
13556 (Dd): Split into Ds and new Dd.
13557 * config/aarch64/aarch64.md (*movsi_aarch64):
13559 (*movdi_aarch64): Add SIMD mov case.
13561 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13563 * tree-predcom.c: (struct chain): Handle store-store chain in which
13564 stores for elimination only store loop invariant values.
13565 (execute_pred_commoning_chain): Ditto.
13566 (prepare_initializers_chain_store_elim): Ditto.
13567 (prepare_finalizers): Ditto.
13568 (is_inv_store_elimination_chain): New function.
13569 (initialize_root_vars_store_elim_1): New function.
13571 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13573 * tree-predcom.c: Revise general description of the pass.
13574 (enum chain_type): New enum type for store elimination.
13575 (struct chain): New field supporting store elimination.
13576 (struct component): Ditto.
13577 (dump_chain): Dump store-stores chain.
13578 (release_chain): Release resources.
13579 (split_data_refs_to_components): Compute and create component
13580 contains only stores for elimination.
13581 (get_chain_last_ref_at): New function.
13582 (make_invariant_chain): Initialization.
13583 (make_rooted_chain): Specify chain type in parameter and record it.
13584 (add_looparound_copies): Skip for store-stores chain.
13585 (determine_roots_comp): Compute type of chain and pass it to
13587 (initialize_root_vars_store_elim_2): New function.
13588 (finalize_eliminated_stores): New function.
13589 (remove_stmt): Handle store for elimination.
13590 (execute_pred_commoning_chain): Execute predictive commoning on
13591 store-store chains.
13592 (determine_unroll_factor): Skip unroll for store-stores chain.
13593 (prepare_initializers_chain_store_elim): New function.
13594 (prepare_initializers_chain): Hanlde store-store chain.
13595 (prepare_finalizers_chain, prepare_finalizers): New function.
13596 (tree_predictive_commoning_loop): Return integer value indicating
13597 if loop is unrolled or lcssa form is corrupted.
13598 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
13600 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13602 * tree-predcom.c (initialize_root): Delete.
13603 (execute_pred_commoning_chain): Initialize root vars and replace
13604 reference of non-combined chain directly, rather than call above
13607 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13609 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
13610 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
13612 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13614 * tree-predcom.c (struct chain): New field init_seq.
13615 (release_chain): Release init_seq.
13616 (prepare_initializers_chain): Record intialization stmts in above
13618 (insert_init_seqs): New function.
13619 (tree_predictive_commoning_loop): Call insert_init_seqs.
13621 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13623 * tree-predcom.c (determine_roots_comp): Skip trivial components.
13625 2017-07-28 Richard Biener <rguenther@suse.de>
13627 * match.pd: Remove superfluous :c.
13628 * genmatch.c (simplify::id): Add member.
13629 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
13631 (current_id): New global.
13632 (dt_node::parent): Move from ...
13633 (dt_operand::parent): ... here. Add for_id member.
13634 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
13635 (decision_tree::find_node): Relax order requirement when
13636 merging DT_TRUE nodes to ones inbetween the current simplify
13637 and the one we try to merge with. Add diagnostic whenever
13638 we need to enforce pattern order by not merging.
13639 (decision_tree::insert): Set current_id.
13640 (decision_tree::print_node): Dump parent node and for_id.
13641 (parser::last_id): Add member.
13642 (parser::push_simplify): Assign unique id.
13643 (parser::parser): Initialize last_id.
13645 2017-07-28 Martin Liska <mliska@suse.cz>
13648 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
13649 gimple_build_debug_bind.
13651 2017-07-28 Richard Biener <rguenther@suse.de>
13653 PR tree-optimization/81502
13654 * match.pd: Add pattern combining BIT_INSERT_EXPR with
13656 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
13658 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
13659 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
13660 for BIT_FIELD_REF args.
13661 * fold-const.c (make_bit_field_ref): Likewise.
13662 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
13664 2017-07-28 Jakub Jelinek <jakub@redhat.com>
13667 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
13668 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
13669 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
13670 Or it into SANITIZER_UNDEFINED.
13671 * ubsan.c: Include gimple-fold.h and varasm.h.
13672 (ubsan_expand_ptr_ifn): New function.
13673 (instrument_pointer_overflow): New function.
13674 (maybe_instrument_pointer_overflow): New function.
13675 (instrument_object_size): Formatting fix.
13676 (pass_ubsan::execute): Call instrument_pointer_overflow
13677 and maybe_instrument_pointer_overflow.
13678 * internal-fn.c (expand_UBSAN_PTR): New function.
13679 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
13680 * sanitizer.def (__ubsan_handle_pointer_overflow,
13681 __ubsan_handle_pointer_overflow_abort): New builtins.
13682 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
13683 * internal-fn.def (UBSAN_PTR): New internal function.
13684 * opts.c (sanitizer_opts): Add pointer-overflow.
13685 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
13686 * fold-const.c (build_range_check): Compute pointer range check in
13687 integral type if pointer arithmetics would be needed. Formatting
13690 2017-07-28 Martin Liska <mliska@suse.cz>
13693 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
13694 parameters that are of a variable-length.
13696 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
13698 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
13700 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
13701 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
13702 (CRT_CALL_STATIC_FUNCTION): Likewise.
13703 (ASM_DEFAULT_SPEC): New define.
13704 (ASM_SPEC32): Likewise.
13705 (ASM_SPEC64): Likewise.
13706 (ASM_SPEC_COMMON): Likewise.
13707 (ASM_SPEC): Likewise.
13708 (INVALID_64BIT): Likewise.
13709 (LINK_OS_DEFAULT_SPEC): Likewise.
13710 (LINK_OS_SPEC32): Likewise.
13711 (LINK_OS_SPEC64): Likewise.
13712 (POWERPC_LINUX): Likewise.
13713 (PTRDIFF_TYPE): Likewise.
13714 (RESTORE_FP_PREFIX): Likewise.
13715 (RESTORE_FP_SUFFIX): Likewise.
13716 (SAVE_FP_PREFIX): Likewise.
13717 (SAVE_FP_SUFFIX): Likewise.
13718 (SIZE_TYPE): Likewise.
13719 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
13720 (TARGET_64BIT): Likewise.
13721 (TARGET_64BIT): Likewise.
13722 (TARGET_AIX): Likewise.
13723 (WCHAR_TYPE_SIZE): Likewise.
13724 (WCHAR_TYPE): Undefine.
13725 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
13726 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
13727 (CPP_OS_RTEMS_SPEC): Delete.
13728 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
13729 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
13731 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
13733 2017-07-28 Jakub Jelinek <jakub@redhat.com>
13735 PR tree-optimization/81578
13736 * tree-parloops.c (build_new_reduction): Bail out if
13737 reduction_code isn't one of the standard OpenMP reductions.
13738 Move the details printing after that decision.
13740 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13742 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
13743 related to reload_in_progress.
13744 (splat_input_operand): Likewise.
13745 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
13747 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
13749 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
13750 (TARGET_INSTANTIATE_DECLS): Likewise.
13751 (legitimate_indexed_address_p): Delete reload_in_progress code.
13752 (rs6000_debug_legitimate_address_p): Likewise.
13753 (rs6000_eliminate_indexed_memrefs): Likewise.
13754 (rs6000_emit_le_vsx_store): Likewise.
13755 (rs6000_emit_move_si_sf_subreg): Likewise.
13756 (rs6000_emit_move): Likewise.
13757 (register_to_reg_type): Likewise.
13758 (rs6000_pre_atomic_barrier): Likewise.
13759 (rs6000_machopic_legitimize_pic_address): Likewise.
13760 (rs6000_allocate_stack_temp): Likewise.
13761 (rs6000_address_for_fpconvert): Likewise.
13762 (rs6000_address_for_altivec): Likewise.
13763 (rs6000_secondary_memory_needed_rtx): Delete function.
13764 (rs6000_check_sdmode): Likewise.
13765 (rs6000_alloc_sdmode_stack_slot): Likewise.
13766 (rs6000_instantiate_decls): Likewise.
13767 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
13768 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
13769 Delete reload_in_progress.
13770 (*vec_reload_and_plus_<mptrsize>): Likewise.
13771 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
13772 (vsx_div_v2di): Likewise.
13773 (vsx_udiv_v2di): Likewise.
13775 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13777 * config/rs6000/rs6000.opt (mlra): Replace with stub.
13778 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
13779 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
13780 (rs6000_debug_reg_global): Delete print of LRA status.
13781 (rs6000_option_override_internal): Delete dead LRA related code.
13782 (rs6000_lra_p): Delete function.
13783 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
13785 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
13787 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
13788 * config/riscv/rtems.h: New file.
13790 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13791 Sudakshina Das <sudi.das@arm.com>
13793 * config/aarch64/aarch64.md
13794 (define_split for and<mode>3nr_compare): Move
13795 non aarch64_logical_operand to a register.
13796 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
13797 register immediate operand to a register.
13798 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
13800 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13802 PR middle-end/81564
13803 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
13805 2017-07-27 Richard Biener <rguenther@suse.de>
13807 PR tree-optimization/81573
13808 PR tree-optimization/81494
13809 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
13810 multi defuse cycle case.
13812 2017-07-27 Richard Biener <rguenther@suse.de>
13814 PR tree-optimization/81571
13815 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
13818 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
13820 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
13821 earlier and only if MASK_FPU is set. Adjust formatting.
13823 2017-07-27 Martin Liska <mliska@suse.cz>
13825 * opt-functions.awk: Add validation of value of Init.
13826 * optc-gen.awk: Pass new argument.
13828 2017-07-27 Martin Liska <mliska@suse.cz>
13830 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
13831 Fix wrong condition.
13833 2017-07-27 Martin Liska <mliska@suse.cz>
13835 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
13836 BBs and edges seen by autoFDO.
13838 2017-07-27 Richard Biener <rguenther@suse.de>
13840 PR tree-optimization/81502
13841 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
13842 with incompatible but same sized type.
13843 (execute_update_addresses_taken): Likewise.
13845 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
13847 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
13848 flag_tree_loop_vectorize rather than flag_tree_vectorize.
13850 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13853 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
13854 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
13855 Change s_operand to memory_operand.
13857 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
13859 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
13860 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
13861 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
13862 Emit instructions rather than returning an expression. Handle TFmode
13863 and KFmode by casting to TImode.
13864 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
13865 (rs6000_emit_le_vsx_store): Likewise.
13866 * config/rs6000/vsx.md (VSX_TI): New iterator.
13867 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
13868 (*vsx_le_undo_permute_<mode>): Likewise.
13869 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
13870 emit the split sequence.
13871 (*vsx_le_perm_store_<mode>): Likewise.
13873 2017-07-27 Jakub Jelinek <jakub@redhat.com>
13875 PR tree-optimization/81555
13876 PR tree-optimization/81556
13877 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
13878 if true, force CHANGED for the recursive invocation.
13879 (reassociate_bb): Remember original length of ops array, pass
13880 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
13882 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
13883 attributes for noipa attribute. For naked attribute use
13884 lookup_attribute first before lookup_attribute_spec.
13885 * final.c (rest_of_handle_final): Disable IPA RA for functions with
13887 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
13888 for functions with noipa attribute.
13889 (cgraph_externally_visible_p): Return true for functions with noipa
13891 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
13892 for functions with noipa attribute.
13893 * doc/extend.texi: Document noipa function attribute.
13894 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
13895 also for functions with noipa attribute.
13896 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
13898 2017-07-26 Andrew Pinski <apinski@cavium.com>
13900 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
13901 vec_unalign_load_cost and vec_unalign_store_cost.
13903 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13905 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
13906 -mvsx-small-integer option.
13907 (ISA_3_0_MASKS_IEEE): Likewise.
13908 (OTHER_VSX_VECTOR_MASKS): Likewise.
13909 (POWERPC_MASKS): Likewise.
13910 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
13911 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
13912 code, only testing for DImode being allowed in non-VSX floating
13914 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
13915 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
13917 (rs6000_option_override_internal): Delete -mvsx-small-integer.
13918 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
13919 TARGET_P8_VECTOR test.
13920 (rs6000_secondary_reload_simple_move): Likewise.
13921 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
13922 since TARGET_P9_VECTOR was already tested.
13923 (rs6000_opt_masks): Remove -mvsx-small-integer.
13924 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
13925 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13927 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
13928 test for TARGET_VEXTRACTUB was used, and that uses
13930 (p9 extract splitter): Likewise.
13931 (vsx_extract_<mode>_di_p9): Likewise.
13932 (vsx_extract_<mode>_store_p9): Likewise.
13933 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
13934 for TARGET_P9_VECTOR was used. Delete code that is now dead with
13935 the elimination of TARGET_VSX_SMALL_INTEGER.
13936 (vsx_extract_<mode>_p8): Likewise.
13937 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
13938 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
13939 (vsx_set_<mode>_p9): Likewise.
13940 (vsx_set_v4sf_p9): Likewise.
13941 (vsx_set_v4sf_p9_zero): Likewise.
13942 (vsx_insert_extract_v4sf_p9): Likewise.
13943 (vsx_insert_extract_v4sf_p9_2): Likewise.
13944 * config/rs6000/rs6000.md (sign extend splitter): Change
13945 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
13946 (floatsi<mode>2_lfiwax_mem): Likewise.
13947 (floatunssi<mode>2_lfiwzx_mem): Likewise.
13948 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
13949 since a test for TARGET_P9_VECTOR was used.
13950 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13951 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
13952 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13953 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13954 TARGET_P8_VECTOR test.
13955 (fix_trunc<mode>si2_stfiwx): Likewise.
13956 (fix_trunc<mode>si2_internal): Likewise.
13957 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
13958 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13960 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13961 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13962 TARGET_P8_VECTOR test.
13963 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13964 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
13965 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13967 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13968 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
13969 since a test for TARGET_P9_VECTOR was used.
13970 (splitter for loading small constants): Likewise.
13972 2017-07-26 Andrew Pinski <apinski@cavium.com>
13974 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
13977 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
13980 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
13981 (fp_valid_at): Likewise.
13983 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13985 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
13986 (qdf24xx_addrcost_table): Likewise.
13987 (cortexa57_tunings): Update to use generic_branch_cost.
13988 (cortexa72_tunings): Likewise.
13989 (cortexa73_tunings): Likewise.
13990 (qdf24xx_tunings): Likewise.
13992 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13994 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
13995 (thunderx2t99_branch_cost): Likewise.
13996 (cortexa35_tunings): Update to use generic_branch_cost.
13997 (cortexa53_tunings): Likewise.
13998 (cortexa57_tunings): Likewise.
13999 (cortexa72_tunings): Likewise.
14000 (cortexa73_tunings): Likewise.
14001 (thunderx2t99_tunings): Likewise.
14003 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14005 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
14006 (sparc_option_override): Honour MASK_FSMULD.
14007 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
14008 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
14009 * config/sparc/sparc.opt (mfsmuld): New option.
14010 * doc/invoke.texi (mfsmuld): Document option.
14012 2017-07-26 Marek Polacek <polacek@redhat.com>
14014 PR middle-end/70992
14015 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
14017 2017-07-26 Richard Biener <rguenther@suse.de>
14019 * gimple-match-head.c (do_valueize): Return OP if valueize
14021 (get_def): New helper to get at the def stmt of a SSA name
14022 if valueize allows.
14023 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
14024 do_valueize to get at the def stmt.
14025 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
14027 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
14029 PR middle-end/46932
14030 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
14032 2017-07-26 Martin Liska <mliska@suse.cz>
14035 * function.c (expand_function_start): Make expansion of
14036 nonlocal_goto_save_area after parm_birth_insn.
14038 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
14040 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
14041 from all CPU target flags enable members.
14043 2017-07-26 Richard Biener <rguenther@suse.de>
14045 * genmatch.c (dt_simplify::gen): Make iterator vars const.
14046 (decision_tree::gen): Make 'type' const.
14047 (write_predicate): Likewise.
14049 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14051 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
14052 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
14053 (rs6000_option_override_internal): Likewise.
14054 (rs6000_expand_vector_set): Likewise.
14055 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
14056 (TARGET_UPPER_REGS_SF): Likewise.
14057 (TARGET_UPPER_REGS_DI): Likewise.
14058 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
14059 (TARGET_DIRECT_MOVE_64BIT): Likewise.
14060 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
14061 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
14062 (Splitters for DI constants in Altivec registers): Likewise.
14063 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
14064 (vsx_set_v4sf_p9): Likewise.
14065 (vsx_set_v4sf_p9_zero): Likewise.
14066 (vsx_insert_extract_v4sf_p9): Likewise.
14067 (vsx_insert_extract_v4sf_p9_2): Likewise.
14069 2017-07-25 Carl Love <cel@us.ibm.com>
14071 * doc/extend.texi: Update the built-in documentation file for the
14072 existing built-in functions
14073 vector signed char vec_cnttz (vector signed char);
14074 vector unsigned char vec_cnttz (vector unsigned char);
14075 vector signed short vec_cnttz (vector signed short);
14076 vector unsigned short vec_cnttz (vector unsigned short);
14077 vector signed int vec_cnttz (vector signed int);
14078 vector unsigned int vec_cnttz (vector unsigned int);
14079 vector signed long long vec_cnttz (vector signed long long);
14080 vector unsigned long long vec_cnttz (vector unsigned long long);
14082 2017-07-25 Andrew Pinski <apinski@cavium.com>
14084 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
14085 accesses where the use is for the first operand of a BIT_INSERT.
14087 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14090 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14091 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14093 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14095 * config/i386/gstabs.h: Delete.
14096 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
14098 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
14100 * config/i386/i386.c (ix86_decompose_address): Do not check for
14101 register RTX when looking at index_reg or base_reg.
14102 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
14104 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
14106 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
14107 to update EH info here.
14109 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14111 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
14113 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14115 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
14117 2017-07-25 Torsten Duwe <duwe@suse.de>
14119 * common.opt: Introduce -fpatchable-function-entry
14120 command line option, and its variables function_entry_patch_area_size
14121 and function_entry_patch_area_start.
14122 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
14123 including a two-value parser.
14124 * target.def (print_patchable_function_entry): New target hook.
14125 * targhooks.h (default_print_patchable_function_entry): New function.
14126 * targhooks.c (default_print_patchable_function_entry): Likewise.
14127 * toplev.c (process_options): Switch off IPA-RA if
14128 patchable function entries are being generated.
14129 * varasm.c (assemble_start_function): Look at the
14130 patchable-function-entry command line switch and current
14131 function attributes and maybe generate NOP instructions by
14132 calling the print_patchable_function_entry hook.
14133 * doc/extend.texi: Document patchable_function_entry attribute.
14134 * doc/invoke.texi: Document -fpatchable_function_entry
14135 command line option.
14136 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
14138 * doc/tm.texi: Re-generate.
14140 2017-07-25 Jakub Jelinek <jakub@redhat.com>
14143 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
14144 TARGET_AVX512DQ rather than TARGET_AVX512BW.
14146 2017-07-25 Tamar Christina <tamar.christina@arm.com>
14148 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
14150 2017-07-25 Richard Biener <rguenther@suse.de>
14152 PR tree-optimization/81455
14153 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
14154 not walk in cycles when looking for guards.
14156 2017-07-25 Richard Biener <rguenther@suse.de>
14158 PR tree-optimization/81529
14159 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
14160 when optimizing backedge uses.
14162 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
14164 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
14166 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
14167 to dl_section_ref. On AIX, append an expression to subtract
14168 the size of the section length to dl_section_ref.
14170 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
14172 * configure.ac: If any of the config.* scripts fail, exit 1.
14173 * configure: Regenerate.
14175 2017-07-25 Richard Biener <rguenther@suse.de>
14177 PR middle-end/81546
14178 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
14179 of immediate uses, be more verbose on errors.
14181 2017-07-25 Richard Biener <rguenther@suse.de>
14183 PR tree-optimization/81510
14184 * tree-vect-loop.c (vect_is_simple_reduction): When the
14185 reduction stmt is not inside the loop bail out.
14187 2017-07-25 Richard Biener <rguenther@suse.de>
14189 PR tree-optimization/81303
14190 * tree-vect-loop-manip.c (vect_loop_versioning): Build
14191 profitability check against LOOP_VINFO_NITERSM1.
14193 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14195 * domwalk.c (cmp_bb_postorder): Simplify.
14196 (sort_bbs_postorder): New function. Use it...
14197 (dom_walker::walk): ...here to optimize common cases.
14199 2017-07-25 Martin Liska <mliska@suse.cz>
14202 * ipa-visibility.c (function_and_variable_visibility): Make the
14203 redirection just on target that supports aliasing.
14204 Fix GNU coding style.
14206 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
14209 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
14211 * config/arm/rtems.h: Likewise.
14212 * config/bfin/rtems.h: Likewise.
14213 * config/i386/rtemself.h: Likewise.
14214 * config/lm32/rtems.h: Likewise.
14215 * config/m32c/rtems.h: Likewise.
14216 * config/m68k/rtemself.h: Likewise.
14217 * config/microblaze/rtems.h: Likewise.
14218 * config/mips/rtems.h: Likewise.
14219 * config/moxie/rtems.h: Likewise.
14220 * config/nios2/rtems.h: Likewise.
14221 * config/powerpcspe/rtems.h: Likewise.
14222 * config/rs6000/rtems.h: Likewise.
14223 * config/rtems.h: Likewise.
14224 * config/sh/rtems.h: Likewise.
14225 * config/sh/rtemself.h: Likewise.
14226 * config/sparc/rtemself.h: Likewise.
14228 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14231 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
14232 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
14233 * tree-ssa-structalias.c (alias_get_name): Same.
14235 2017-07-25 Bin Cheng <bin.cheng@arm.com>
14238 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
14239 instructions if no du chain is found.
14241 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14243 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
14245 2017-07-25 Richard Biener <rguenther@suse.de>
14247 PR middle-end/81505
14248 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
14251 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14253 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
14254 upper-regs options.
14255 (ISA_2_7_MASKS_SERVER): Likewise.
14256 (ISA_3_0_MASKS_IEEE): Likewise.
14257 (OTHER_P8_VECTOR_MASKS): Likewise.
14258 (OTHER_VSX_VECTOR_MASKS): Likewise.
14259 (POWERPC_MASKS): Likewise.
14260 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
14261 duplicate list of options.
14262 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
14263 explicit -mupper-regs options.
14264 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
14265 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
14266 alias for -mupper-regs-df.
14267 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14268 (rs6000_init_hard_regno_mode_ok): Likewise.
14269 (rs6000_option_override_internal): Likewise.
14270 (rs6000_opt_masks): Likewise.
14271 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
14272 options in terms of whether -mvsx or -mpower8-vector was used.
14273 (TARGET_UPPER_REGS_DI): Likewise.
14274 (TARGET_UPPER_REGS_SF): Likewise.
14275 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
14276 -mupper-regs-* options.
14278 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14280 * passes.c (emergency_dump_function): Print some empty lines and a
14281 header before the RTL dump.
14283 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14285 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
14287 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
14290 * config/aarch64/aarch64.c (aarch64_classify_symbol):
14291 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
14293 2017-07-24 Carl Love <cel@us.ibm.com>
14295 * config/rs6000/rs6000-c.c: Add support for built-in functions
14296 vector float vec_extract_fp32_from_shorth (vector unsigned short);
14297 vector float vec_extract_fp32_from_shortl (vector unsigned short);
14298 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
14299 vec_extract_fp_from_shortl): Add defines for the two builtins.
14300 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
14301 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
14303 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
14304 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
14305 * doc/extend.texi: Update the built-in documentation file for the
14306 new built-in function.
14308 2017-07-24 Jakub Jelinek <jakub@redhat.com>
14311 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
14313 * doc/generic.texi: Likewise.
14314 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14315 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14317 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
14319 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
14320 (aarch64_mls_elt_merge<mode>): Likewise.
14322 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
14324 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14325 having __cxa_atexit.
14327 2017-07-23 Michael Collison <michael.collison@arm.com>
14329 * config/arm/arm.c (arm_option_override): Deprecate
14330 use of -mstructure-size-boundary.
14331 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
14332 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
14334 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14337 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14338 Reduce cost estimate for direct moves.
14340 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
14343 * config/i386/i386.c (ix86_option_override_internal): Disable
14344 BMI, BMI2 and TBM instructions for -m16.
14346 2017-07-21 Carl Love <cel@us.ibm.com>
14348 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14349 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14350 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14351 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14352 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14353 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14354 VMULOSW): New enum "unspec" values.
14355 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14356 altivec_vmulosw): New patterns.
14357 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14358 VMULOSW): Add definitions.
14360 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
14362 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
14363 (qdf24xx): Likewise.
14364 * config/aarch64/aarch64-options-extensions.def (rdma); New.
14365 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
14366 (AARCH64_FL_V8_1): Renumber.
14367 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
14368 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
14369 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
14370 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
14371 rdma to feature modifiers list.
14373 2017-07-21 Yury Gribov <tetra2005@gmail.com>
14375 PR middle-end/56727
14376 * ipa-visibility (function_and_variable_visibility): Convert
14377 recursive PLT call to direct call if appropriate.
14379 2017-07-21 Andrew Pinski <apinski@cavium.com>
14381 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
14382 operand 1 to see if the types precision matches.
14383 * fold-const.c (operand_equal_p): Likewise.
14385 2017-07-21 Richard Biener <rguenther@suse.de>
14387 PR tree-optimization/81303
14388 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
14389 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
14390 (vect_peeling_hash_get_lowest_cost): Adjust.
14391 (vect_enhance_data_refs_alignment): Likewise. Use
14392 vect_get_peeling_costs_all_drs to compute the penalty for no
14393 peeling to match up costs.
14395 2017-07-21 Richard Biener <rguenther@suse.de>
14397 PR tree-optimization/81500
14398 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
14399 we didn't identify a reduction path.
14401 2017-07-21 Tom de Vries <tom@codesourcery.com>
14402 Cesar Philippidis <cesar@codesourcery.com>
14404 PR gcov-profile/81442
14405 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
14408 2017-07-21 Tom de Vries <tom@codesourcery.com>
14411 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
14413 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
14414 nvptx_override_options_after_change.
14416 2017-07-21 Ulrich Drepper <drepper@redhat.com>
14418 * dwarf2out.c (output_file_names): Avoid double testing for
14419 dwarf_version >= 5.
14421 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
14423 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
14425 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14427 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
14429 (try_crossjump_to_edge): Do not punt on partitioned functions.
14431 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14433 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
14434 Put all BBs reachable only via paths crossing cold region to cold
14436 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
14438 2016-07-21 Richard Biener <rguenther@suse.de>
14440 PR tree-optimization/81303
14441 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
14442 into account prologue and epilogue iterations when raising
14443 min_profitable_iters to sth at least covering one vector iteration.
14445 2017-07-21 Tamar Christina <tamar.christina@arm.com>
14447 * config/arm/arm.c (arm_test_cpu_arch_dat):
14450 2017-07-20 Nathan Sidwell <nathan@acm.org>
14452 Remove TYPE_METHODS.
14453 * tree.h (TYPE_METHODS): Delete.
14454 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
14455 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
14456 (dbxout_type_methods): Scan TYPE_FIELDS.
14457 (dbxout_type): Don't check TYPE_METHODS here.
14458 * function.c (use_register_for_decl): Always ignore register for
14459 class types when not optimizing.
14460 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
14461 * tree.c (free_lang_data_in_type): Stitch out member functions and
14462 templates from TYPE_FIELDS.
14463 (build_distinct_type_copy, verify_type_variant,
14464 verify_type): Member fns are on TYPE_FIELDS.
14465 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
14466 * tree-pretty-print.c (dump_generic_node): Likewise.
14468 2017-07-20 Jakub Jelinek <jakub@redhat.com>
14471 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
14472 V2TImode and V4TImode.
14473 (ix86_expand_vector_extract): Likewise.
14474 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
14475 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
14476 (ssescalarmode): Handle V4TImode and V2TImode.
14477 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
14478 (*vec_extractv2ti, *vec_extractv4ti): New insns.
14479 (VEXTRACTI128_MODE): New mode iterator.
14480 (splitter for *vec_extractv?ti first element): New.
14481 (VEC_INIT_MODE): New mode iterator.
14482 (vec_init<mode>): Consolidate 3 expanders into one using
14483 VEC_INIT_MODE mode iterator.
14485 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14487 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
14488 non_spilled_static_chain_regno_p.
14490 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14492 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
14494 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
14496 * bb-reorder.c (connect_traces): Allow copying of blocks within
14499 2017-07-20 Richard Biener <rguenther@suse.de>
14501 * gimple.h (gimple_phi_result): Add gphi * overload.
14502 (gimple_phi_result_ptr): Likewise.
14503 (gimple_phi_arg): Likewise. Adjust index assert to only
14504 allow actual argument accesses rather than all slots available
14506 (gimple_phi_arg_def): Add gphi * overload.
14507 * tree-phinodes.c (make_phi_node): Initialize only actual
14509 (resize_phi_node): Clear memory not covered by old node,
14510 do not initialize excess argument slots.
14511 (reserve_phi_args_for_new_edge): Initialize new argument slot
14514 2017-07-20 Bin Cheng <bin.cheng@arm.com>
14516 PR tree-optimization/81388
14518 2016-07-21 Bin Cheng <bin.cheng@arm.com>
14520 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
14521 by removing computation of may_be_zero.
14523 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14524 Tom de Vries <tom@codesourcery.com>
14526 PR middle-end/81030
14527 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
14528 when gimple level profile disagrees with what RTL expander did.
14530 2017-07-20 Richard Biener <rguenther@suse.de>
14532 PR tree-optimization/61171
14533 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
14534 (vect_analyze_stmt): Add slp instance parameter.
14535 (vectorizable_reduction): Likewise.
14536 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
14537 (vect_is_simple_reduction): Deal with chains not detected
14538 as SLP reduction chain, specifically not properly associated
14539 chains containing a mix of plus/minus.
14540 (get_reduction_op): Remove.
14541 (get_initial_defs_for_reduction): Simplify, pass in whether
14542 this is a reduction chain, pass in the SLP node for the PHIs.
14543 (vect_create_epilog_for_reduction): Get the SLP instance as
14545 (vectorizable_reduction): Get the SLP instance as arg.
14546 During analysis remember the SLP node with the PHIs in the
14547 instance. Simplify getting at the vectorized reduction PHIs.
14548 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
14549 through SLP instance.
14550 (vect_slp_analyze_operations): Likewise.
14551 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
14552 (vect_transform_stmt): Likewise.
14554 2017-07-20 Tom de Vries <tom@codesourcery.com>
14556 PR tree-optimization/81489
14557 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
14558 read of phi arg location to before loop that modifies phi.
14560 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14562 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
14565 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14567 PR middle-end/81331
14568 * except.c (execute): Fix ordering issue.
14570 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14572 PR rtl-optimization/81423
14573 * combine.c (make_compound_operation_int): Don't try to optimize
14574 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
14576 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14578 PR rtl-optimization/81423
14579 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
14580 with a constant that is -1 in the truncated to mode.
14582 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14584 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
14585 (determine_unlikely_bbs): ... here.
14586 * predict.h (propagate_unlikely_bbs_forward): Declare.
14587 * cfgexpand.c (pass_expand::execute): Use it.
14588 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
14590 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
14591 propagate_unlikely_bbs_forward.
14593 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14595 PR middle-end/81331
14596 * except.c (maybe_add_nop_after_section_switch): New function.
14599 2017-07-19 Tom de Vries <tom@codesourcery.com>
14601 * gimple.h (gimple_phi_set_arg): Make assert more strict.
14603 2017-07-19 Tom de Vries <tom@codesourcery.com>
14605 * gimple.h (gimple_phi_arg): Make assert more strict.
14607 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
14609 * config.gcc (powerpc*-*-*): Add mmintrin.h.
14610 * config/rs6000/mmintrin.h: New file.
14611 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
14613 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14615 PR tree-optimization/81346
14616 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
14618 2017-07-19 Tom de Vries <tom@codesourcery.com>
14620 * config/nvptx/nvptx.md (VECIM): Add V2DI.
14622 2017-07-19 Tom de Vries <tom@codesourcery.com>
14624 * config/nvptx/nvptx-modes.def: Add V2DImode.
14625 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
14626 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
14627 (nvptx_output_mov_insn): Handle lack of mov.b128.
14628 (nvptx_print_operand): Handle 'H' and 'L' codes.
14629 (nvptx_vector_mode_supported): Allow V2DImode.
14630 (nvptx_preferred_simd_mode): New function.
14631 (nvptx_data_alignment): New function.
14632 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
14633 nvptx_preferred_simd_mode.
14634 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
14636 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
14638 2017-07-19 Tom de Vries <tom@codesourcery.com>
14640 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
14641 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
14642 (nvptx_vector_mode_supported): New function. Allow V2SImode.
14643 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
14644 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
14645 (mov<VECIM>_insn): New define_insn.
14646 (define_expand "mov<VECIM>): New define_expand.
14648 2017-07-19 Tom de Vries <tom@codesourcery.com>
14650 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
14652 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14654 PR tree-optimization/81346
14655 * fold-const.h (fold_div_compare, range_check_type): Declare.
14656 * fold-const.c (range_check_type): New function.
14657 (build_range_check): Use range_check_type.
14658 (fold_div_compare): No longer static, rewritten into
14659 a match.pd helper function.
14660 (fold_comparison): Don't call fold_div_compare here.
14661 * match.pd (X / C1 op C2): New optimization using fold_div_compare
14662 as helper function.
14664 2017-07-19 Nathan Sidwell <nathan@acm.org>
14666 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
14667 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
14668 * tree.c (find_decls_types_r, verify_type): Use
14669 TYPE_{MIN,MAX}_VALUE_RAW.
14670 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
14671 (hash_tree): Likewise.
14672 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14674 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14677 2017-07-18 Tom de Vries <tom@codesourcery.com>
14679 PR middle-end/81464
14680 * omp-expand.c (expand_omp_for_static_chunk): Handle
14681 equal-argument loop exit phi.
14683 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
14686 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
14687 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
14688 operand 2 predicate.
14689 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
14690 operand 2 predicate.
14691 (ror,rol -> rorx splitters): Use const_int_operand as
14692 operand 2 predicate.
14694 2017-06-18 Richard Biener <rguenther@suse.de>
14696 PR tree-optimization/81410
14697 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
14698 the gap in the ! slp_perm SLP case after each group.
14700 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14702 PR middle-end/81463
14703 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
14706 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14708 PR middle-end/81462
14709 * predict.c (set_even_probabilities): Cleanup; do not affect
14710 probabilities that are already known.
14711 (combine_predictions_for_bb): Call even when count is set.
14713 2017-07-18 Nathan Sidwell <nathan@acm.org>
14715 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
14718 2017-07-18 Bin Cheng <bin.cheng@arm.com>
14721 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
14722 optimization for loop niter analysis.
14724 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
14727 * config/avr/avr.c (avr_optimize_casesi): Don't use
14728 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
14730 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
14732 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
14733 body_cost_vec from _vect_peel_extended_info.
14734 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
14735 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
14738 2017-07-18 Bin Cheng <bin.cheng@arm.com>
14740 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
14742 2017-07-18 Richard Biener <rguenther@suse.de>
14744 PR tree-optimization/80620
14745 PR tree-optimization/81403
14746 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
14747 info when re-using a VN table entry.
14749 2017-07-18 Richard Biener <rguenther@suse.de>
14751 PR tree-optimization/81418
14752 * tree-vect-loop.c (vectorizable_reduction): Properly compute
14753 vectype_in. Verify that with lane-reducing reduction operations
14754 we have a single def-use cycle.
14756 2017-07-17 Carl Love <cel@us.ibm.com>
14758 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
14760 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14761 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14762 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14763 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14764 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14765 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14766 VMULOSW): New enum "unspec" values.
14767 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14768 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14769 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14770 altivec_vmulosw): New patterns.
14771 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14772 VMULOSW): Add definitions.
14773 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
14775 * config/alpha/alpha.c: Include predict.h.
14777 2017-07-17 Yury Gribov <tetra2005@gmail.com>
14779 * tree-vrp.c (compare_assert_loc): Fix comparison function
14780 to return predictable results.
14782 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14784 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
14786 (subdi3): Likewise.
14787 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
14788 * doc/invoke.texi (mexpand-adddi): Update text.
14790 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14792 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
14793 that also clobbers the CC register. The old expand code is moved
14795 (*arc_clzsi2): ... here.
14796 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
14797 the CC register. The old expand code is moved to ...
14798 (arc_ctzsi2): ... here.
14800 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14802 * config/arc/arc.opt (mindexed-loads): Use initial value
14803 TARGET_INDEXED_LOADS_DEFAULT.
14804 (mauto-modify-reg): Use initial value
14805 TARGET_AUTO_MODIFY_REG_DEFAULT.
14806 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14807 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14808 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14809 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14811 2017-07-17 Martin Liska <mliska@suse.cz>
14814 * opts.c (finish_options): Do not allow -fgnu-tm
14815 w/ -fsanitize={kernel-,}address. Say sorry.
14817 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14820 * tree-loop-distribution.c (classify_partition): Only assert on
14821 numer of iterations.
14822 (merge_dep_scc_partitions): Delete prameter. Update function call.
14823 (distribute_loop): Remove code handling loop with unknown niters.
14824 (pass_loop_distribution::execute): Skip loop with unknown niters.
14826 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14829 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
14830 function sort_partitions_by_post_order.
14832 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14834 PR tree-optimization/81374
14835 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
14836 the max index of basic blocks, rather than number of basic blocks.
14838 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14840 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
14842 (arc_legitimate_pic_operand_p): Likewise.
14843 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
14845 (arc_needs_pcl_p): Likewise.
14846 (arc_legitimate_pc_offset_p): Likewise.
14847 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
14848 function is also used in constrains.md.
14849 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
14850 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
14851 PLUS. Only return true/false in known cases, otherwise assert.
14852 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
14853 is already called in arc_legitimate_constant_p.
14854 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
14856 (LEGITIMATE_PIC_OPERAND_P): Use
14857 arc_raw_symbolic_reference_mentioned_p function.
14858 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
14863 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14864 Andrew Burgess <andrew.burgess@embecosm.com>
14866 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
14867 (arc_return_address_register): New function.
14868 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
14869 (arc_handle_fndecl_attribute): Add naked attribute.
14870 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14871 (TARGET_WARN_FUNC_RETURN): Likewise.
14872 (arc_allocate_stack_slots_for_args): New function.
14873 (arc_warn_func_return): Likewise.
14874 (machine_function): Change type fn_type.
14875 (arc_compute_function_type): Consider new naked function type,
14876 change function return type.
14877 (arc_must_save_register): Adapt to handle new
14878 arc_compute_function_type's return type.
14879 (arc_expand_prologue): Likewise.
14880 (arc_expand_epilogue): Likewise.
14881 (arc_return_address_regs): Delete.
14882 (arc_return_address_register): New function.
14883 (arc_epilogue_uses): Use above function.
14884 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
14885 (arc_function_type): Change encoding, add naked type.
14886 (ARC_INTERRUPT_P): Change to handle the new encoding.
14887 (ARC_FAST_INTERRUPT_P): Likewise.
14888 (ARC_NORMAL_P): Define.
14889 (ARC_NAKED_P): Likewise.
14890 (arc_compute_function_type): Delete prototype.
14891 * config/arc/arc.md (in_ret_delay_slot): Use
14892 arc_return_address_register function.
14893 (simple_return): Likewise.
14894 (p_return_i): Likewise.
14896 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14898 PR tree-optimization/81428
14899 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
14900 can't be built for those types.
14902 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14904 Remove stuff dead since r239246.
14906 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
14907 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
14908 (avr_inform_devices): Remove dead stuff.
14910 2017-07-17 Tamar Christina <tamar.christina@arm.com>
14912 * config/arm/arm_neon.h: Fix softp typo.
14914 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14916 PR tree-optimization/81365
14917 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
14918 aggregate moves onto bb predecessor edges, make sure there are no
14919 loads that could alias the lhs in between the start of bb and the
14922 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14925 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
14926 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
14927 [LSHIFTRT, outer_code = TRUNCATE]: Same.
14929 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14931 PR tree-optimization/81396
14932 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
14933 (init_symbolic_number): Initialize it to 1.
14934 (perform_symbolic_merge): Add n_ops from both operands into the new
14936 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
14937 without base_addr as useless if they need more than one operation.
14938 (bswap_replace): Handle !bswap case for NULL base_addr.
14940 2017-07-17 Tom de Vries <tom@codesourcery.com>
14943 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
14946 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
14948 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
14949 conditional builtin define __FIX_LEON3FT_B2BST.
14951 2017-07-17 Daniel Cederman <cederman@gaisler.com>
14953 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
14954 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
14957 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14959 PR rtl-optimization/81424
14960 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
14961 to remove potential trapping from operands if -fnon-call-exceptions.
14963 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14965 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
14966 profile_proability for scalling.
14967 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
14969 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14971 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
14973 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14975 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
14976 fixpoint arithmetics.
14978 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14980 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
14981 fixpoint arithmetics.
14983 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14985 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
14986 fixpoint arithmetics.
14988 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14990 * profile-count.h (profile_probability::from_reg_br_prob_note,
14991 profile_probability::to_reg_br_prob_note): New functions.
14992 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
14993 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
14994 * predict.c (probability_reliable_p): Update.
14995 (edge_probability_reliable_p): Update.
14996 (br_prob_note_reliable_p): Update.
14997 (invert_br_probabilities): Update.
14998 (add_reg_br_prob_note): New function.
14999 (combine_predictions_for_insn): Update.
15000 * asan.c (asan_clear_shadow): Update.
15001 * cfgbuild.c (compute_outgoing_frequencies): Update.
15002 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
15003 (update_br_prob_note): Update.
15004 (rtl_verify_edges): Update.
15005 (purge_dead_edges): Update.
15006 (fixup_reorder_chain): Update.
15007 * emit-rtl.c (try_split): Update.
15008 * ifcvt.c (cond_exec_process_insns): Update.
15009 (cond_exec_process_if_block): Update.
15010 (dead_or_predicable): Update.
15011 * internal-fn.c (expand_addsub_overflow): Update.
15012 (expand_neg_overflow): Update.
15013 (expand_mul_overflow): Update.
15014 * loop-doloop.c (doloop_modify): Update.
15015 * loop-unroll.c (compare_and_jump_seq): Update.
15016 * optabs.c (emit_cmp_and_jump_insn_1): Update.
15017 * predict.h: Update.
15018 * reorg.c (mostly_true_jump): Update.
15020 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
15021 * config/alpha/alpha.c (emit_unlikely_jump): Update.
15022 * config/arc/arc.c: (emit_unlikely_jump): Update.
15023 * config/arm/arm.c: (emit_unlikely_jump): Update.
15024 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
15025 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
15026 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
15027 (ix86_print_operand): Update.
15028 (ix86_split_fp_branch): Update.
15029 (predict_jump): Update.
15030 * config/ia64/ia64.c (ia64_print_operand): Update.
15031 * config/mmix/mmix.c (mmix_print_operand): Update.
15032 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
15033 (rs6000_expand_split_stack_prologue): Update.
15034 * config/rs6000/rs6000.c: Update.
15035 * config/s390/s390.c (s390_expand_vec_strlen): Update.
15036 (s390_expand_vec_movstr): Update.
15037 (s390_expand_cs_tdsi): Update.
15038 (s390_expand_split_stack_prologue): Update.
15039 * config/sh/sh.c (sh_print_operand): Update.
15040 (expand_cbranchsi4): Update.
15041 (expand_cbranchdi4): Update.
15042 * config/sparc/sparc.c (output_v9branch): Update.
15043 * config/spu/spu.c (get_branch_target): Update.
15044 (ea_load_store_inline): Update.
15045 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
15046 * config/tilepro/tilepro.c: Update.
15048 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
15050 * gimplify.c (mostly_copy_tree_r): Revert latest change.
15051 (gimplify_save_expr): Likewise.
15053 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15055 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
15057 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15059 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
15061 * timevar.def (TV_IPA_FNSUMMARY): Define.
15063 2017-07-16 Daniel Cederman <cederman@gaisler.com>
15065 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
15066 to back store errata sensitive sequence from being generated.
15067 (sqrtdf2_fix): Likewise.
15069 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15071 * tree-ssa-threadupdate.c (compute_path_counts,
15072 update_joiner_offpath_counts): Use profile_probability.
15074 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
15077 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15079 * config/arm/arm-c.c (arm_cpu_builtins): Define
15080 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15082 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
15084 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15085 array entries to represent __ieee128 versions of the
15086 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
15087 scalar_extract_sig, and scalar_insert_exp built-in functions.
15088 (altivec_resolve_overloaded_builtin): Add special case handling
15089 for the __builtin_scalar_insert_exp function, as represented by
15090 the P9V_BUILTIN_VEC_VSIEDP constant.
15091 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
15092 exponent support for __ieee128 argument.
15093 (VSESQP): Add scalar extract signature support for __ieee128
15095 (VSTDCNQP): Add scalar test negative support for __ieee128
15097 (VSIEQP): Add scalar insert exponent support for __int128 argument
15098 with __ieee128 result.
15099 (VSIEQPF): Add scalar insert exponent support for __ieee128
15100 argument with __ieee128 result.
15101 (VSTDCQP): Add scalar test data class support for __ieee128
15103 (VSTDCNQP): Add overload support for scalar test negative with
15104 __ieee128 argument.
15105 (VSTDCQP): Add overload support for scalar test data class
15106 __ieee128 argument.
15107 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
15108 UNSPEC_VSX_SXSIGDP.
15109 (UNSPEC_VSX_SIEXPQP): New constant.
15110 (xsxexpqp): New insn for VSX scalar extract exponent quad
15112 (xsxsigqp): New insn for VSX scalar extract significand quad
15114 (xsiexpqpf): New insn for VSX scalar insert exponent quad
15115 precision with floating point argument.
15116 (xststdcqp): New expand for VSX scalar test data class quad
15118 (xststdcnegqp): New expand for VSX scalar test negative quad
15120 (xststdcqp): New insn to match expansions for VSX scalar test data
15121 class quad precision and VSX scalar test negative quad precision.
15122 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
15123 special case operand checking to enforce that second operand of
15124 VSX scalar test data class with quad precision argument is a 7-bit
15126 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15127 prototypes and descriptions of __ieee128 versions of
15128 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
15129 scalar_test_data_class, and scalar_test_neg built-in functions.
15131 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15133 PR tree-optimization/81162
15134 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
15135 replace a negate with an add.
15137 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
15139 * doc/invoke.texi (arm/-mcpu): Document +crypto.
15141 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15143 * config/arm/arm-c.c (arm_cpu_builtins): Define
15144 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15146 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15148 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
15149 (armv8-r): Set ARM Cortex-R52 as default CPU.
15150 * config/arm/arm-tables.opt: Regenerate.
15151 * config/arm/arm-tune.md: Regenerate.
15152 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15154 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
15155 extension for -mcpu=cortex-r52.
15157 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15159 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
15160 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
15161 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
15162 (fp-armv8): Define it as FP_ARMv8 only.
15163 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
15164 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
15166 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
15167 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
15168 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
15169 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
15170 than TARGET_FPU_ARMV8.
15171 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
15172 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
15173 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
15175 * config/arm/neon.md (neon_vrint): Likewise.
15176 (neon_vcvt): Likewise.
15177 (neon_<fmaxmin_op><mode>): Likewise.
15178 (<fmaxmin><mode>3): Likewise.
15179 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
15180 * config/arm/predicates.md (arm_cond_move_operator): Check against
15181 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
15183 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
15185 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
15186 to top of function.
15188 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15190 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
15191 loop in comment with memset.
15193 2017-07-14 Martin Liska <mliska@suse.cz>
15195 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
15196 * dwarf2out.c (is_java): Remove the function.
15197 (output_pubname): Remove usage of the function.
15198 (lower_bound_default): Remove usage of DW_LANG_Java.
15199 (gen_compile_unit_die): Likewise.
15200 * gcc.c: Remove compiler defaults for .java and .zip files.
15201 * gimple-expr.c (remove_suffix): Change as there's no longer
15202 extension than 4-letter one.
15203 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
15204 (gimplify_save_expr): Likewise.
15205 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
15206 as it's possible even for other languages than Java.
15207 * langhooks.h (struct lang_hooks): Remove Java from a comment.
15208 * lto-opts.c (lto_write_options): Remove reference to Java.
15209 * opts.c (strip_off_ending): Update file extension handling.
15210 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
15211 * tree-eh.c (lower_resx): Likewise.
15212 * tree.c (free_lang_data_in_type): Remove dead code.
15213 (find_decls_types_r): Likewise.
15214 (build_common_builtin_nodes): Remove Java from a comment.
15215 (verify_type): Remove dead code.
15216 * varasm.c (assemble_external): Remove Java from a comment.
15218 2017-07-14 Martin Liska <mliska@suse.cz>
15220 * opts.c (finish_options): Add quotes.
15221 (common_handle_option): Likewise.
15223 2017-07-14 Martin Liska <mliska@suse.cz>
15225 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
15226 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
15227 Remove N_SO_PASCAL.
15228 * dwarf2out.c (lower_bound_default): Do not handle
15230 (gen_compile_unit_die): Likewise.
15231 * gcc.c: Remove default extension binding for GNU Pascal.
15232 * stmt.c: Remove Pascal language from a comment.
15233 * xcoffout.c: Likewise.
15235 2017-07-13 David Malcolm <dmalcolm@redhat.com>
15238 * diagnostic-show-locus.c (fixit_cmp): New function.
15239 (layout::layout): Sort m_fixit_hints.
15240 (column_range::column_range): Assert that the values are valid.
15241 (struct char_span): New struct.
15242 (correction::overwrite): New method.
15243 (struct source_line): New struct.
15244 (line_corrections::add_hint): Add assertions. Reimplement memcpy
15245 calls in terms of classes source_line and char_span, and
15246 correction::overwrite.
15247 (selftest::test_overlapped_fixit_printing_2): New function.
15248 (selftest::diagnostic_show_locus_c_tests): Call it.
15250 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
15252 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
15253 early if there is no lhs.
15255 2017-07-13 Martin Liska <mliska@suse.cz>
15257 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
15258 (gen_reference_type_die): Likewise.
15259 * stor-layout.c: Remove Pascal-related comment.
15261 2017-07-13 Martin Liska <mliska@suse.cz>
15263 * opts.c (finish_options): Add quotes to error messages.
15264 (parse_sanitizer_options): Likewise.
15266 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
15268 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
15270 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
15272 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
15274 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
15276 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
15278 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
15280 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
15283 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
15284 provides the hardware capability bits, define the macro
15285 __BUILTIN_CPU_SUPPORTS__.
15286 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
15287 if GLIBC does not provide the hardware capability bits. Add a
15288 gcc_unreachable call if the built-in cpu function is neither
15289 __builtin_cpu_is nor __builtin_cpu_supports.
15290 (rs6000_get_function_versions_dispatcher): Change the warning
15291 that an old GLIBC is used which does not export the capability
15292 bits to be an error.
15293 * doc/extend.texi (target_clones attribute): Document the
15294 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
15295 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
15296 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
15297 the macros defined by GCC if the newer GLIBC is available.
15299 2017-07-12 Jeff Law <law@redhat.com>
15301 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
15302 remaining includes slightly.
15303 * config/riscv/riscv-builtins.c: Include profile-count.h.
15305 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15308 * config/avr/avr.c (avr_set_current_function): In diagnostic
15309 messages: Quote keywords and (parts of) identifiers.
15310 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
15313 2017-07-12 Carl Love <cel@us.ibm.com>
15315 * config/rs6000/rs6000-c.c: Add support for built-in functions
15316 vector bool char vec_revb (vector bool char);
15317 vector bool short vec_revb (vector short char);
15318 vector bool int vec_revb (vector bool int);
15319 vector bool long long vec_revb (vector bool long long);
15320 * doc/extend.texi: Update the built-in documentation file for the
15321 new built-in functions.
15323 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15325 * config/s390/s390.md: Remove movcc splitter.
15327 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15329 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
15330 load/store on condition.
15332 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15335 * config/avr/avr.c (avr_encode_section_info)
15336 [progmem && !TREE_READONLY]: Error if progmem object needs
15339 2017-07-11 Michael Collison <michael.collison@arm.com>
15341 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
15344 2017-07-11 Carl Love <cel@us.ibm.com>
15346 * config/rs6000/rs6000-c.c: Add support for builtins
15347 vector unsigned int vec_parity_lsbb (vector signed int);
15348 vector unsigned int vec_parity_lsbb (vector unsigned int);
15349 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
15350 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
15351 vector unsigned long long vec_parity_lsbb (vector signed long long);
15352 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
15353 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
15354 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
15355 * doc/extend.texi: Update the built-in documentation file for the
15356 new built-in functions.
15358 2017-07-11 David Malcolm <dmalcolm@redhat.com>
15360 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
15361 (layout::m_primary_loc): New field.
15362 (layout::layout): Initialize new field. Move location filtering
15363 logic from here to...
15364 (layout::maybe_add_location_range): ...this new method. Add
15365 support for filtering to just the lines already specified by other
15367 (layout::will_show_line_p): New method.
15368 (gcc_rich_location::add_location_if_nearby): New method.
15369 (selftest::test_add_location_if_nearby): New test function.
15370 (selftest::diagnostic_show_locus_c_tests): Call it.
15371 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
15374 2017-07-11 Tom de Vries <tom@codesourcery.com>
15376 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
15377 (bb_first_real_insn): New function.
15378 (nvptx_single): Add extra initialization of broadcasted condition
15381 2017-07-11 Nathan Sidwell <nathan@acm.org>
15383 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
15385 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
15387 * doc/extend.texi (AVR Function Attributes): Remove weblink to
15388 Binutils doc as TEXI will mess them up.
15389 * doc/invoke.texi (AVR Options): Same here.
15391 2017-07-11 Daniel Cederman <cederman@gaisler.com>
15393 * config/sparc/sparc.opt (mfix-ut700): New option.
15394 (mfix-gr712rc): Likewise.
15395 (sparc_fix_b2bst): New variable.
15396 * doc/invoke.texi (SPARC options): Document them.
15397 (ARM options): Fix warnings.
15398 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
15399 instructions to prevent sequences that can trigger the store-store
15400 errata for certain LEON3FT processors.
15401 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
15402 (sparc_option_override): Set sparc_fix_b2bst appropriately.
15403 * config/sparc/sparc.md (fix_b2bst): New attribute.
15404 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
15406 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
15409 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
15411 (*rsqrtsf2_sse): Ditto.
15413 (div<mode>3): Macroize insn from divdf3 and divsf3
15414 using MODEF mode iterator.
15416 2017-07-10 Martin Sebor <msebor@redhat.com>
15418 PR tree-optimization/80397
15419 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
15420 instead of testing for equality to INTEGER_TYPE.
15422 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
15424 * config.gcc: Remove uclibc from arc target spec.
15426 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
15428 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
15430 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15433 * lto-wrapper.c (remove_option): New function.
15434 (merge_and_complain): Merge PIC/PIE options more realistically.
15436 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15438 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
15442 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
15443 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
15444 * config.in: Regenerate.
15445 * configure: Regenerate.
15446 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
15447 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
15448 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
15449 (TARGET_GASISR_PROLOGUES): ...target mask.
15450 * common/config/avr/avr-common.c
15451 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
15452 Set -mgas-isr-prologues.
15453 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
15454 INSERT_PASS_BEFORE for it.
15455 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
15456 * config/avr/avr.c (avr_option_override)
15457 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
15458 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
15459 (avr_attribute_table) <no_gccisr>: Add new function attribute.
15460 (avr_set_current_function) <is_no_gccisr>: Init machine field.
15461 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
15463 (make_avr_pass_pre_proep): New function.
15464 (emit_push_sfr) <treg>: Add argument to function and use it
15465 instead of TMP_REG.
15466 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
15467 and set machine->gasisr.yes.
15468 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
15469 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
15470 __gcc_isr.n_pushed to .L__stack_usage.
15471 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
15472 (avr_asm_final_postscan_insn): ...this new static function.
15473 * config/avr/avr.h (machine_function)
15474 <is_no_gccisr, use_L__stack_usage>: New fields.
15475 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
15476 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
15477 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
15478 (gasisr, *gasisr): New expander and insn.
15479 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15480 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
15481 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
15483 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
15485 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
15488 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15490 Move jump-tables out of .text again.
15493 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
15494 (ASM_OUTPUT_ADDR_VEC): New function.
15495 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
15496 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
15497 INSN_ADDRESSes as asm comment.
15498 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
15499 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
15500 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
15501 * config/avr/avr.md (*tablejump): Adjust comment.
15502 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
15503 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
15505 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
15506 (avr_output_addr_vec): New proto.
15507 (avr_log_t) <insn_addresses>: New field.
15509 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
15512 * config/i386/i386.c (ix86_function_arg_advance): Set
15513 outgoing_args_on_stack to true if there are outgoing arguments
15515 (ix86_function_arg): Likewise.
15516 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
15517 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
15518 * config/i386/i386.h (machine_function): Add
15519 outgoing_args_on_stack.
15521 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
15523 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15524 supporting pthreds.
15525 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
15527 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
15529 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
15530 (REAL_H): Remove $(MACHMODE_H).
15531 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
15533 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
15534 $(MACHMODE_H) and double-int.h.
15535 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
15537 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
15540 2017-07-07 Andrew Pinski <apinski@cavium.com>
15542 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
15543 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
15545 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15547 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
15548 Add warning if GCC was not configured to link against a GLIBC that
15549 exports the hardware capability bits.
15550 (make_resolver_func): Make resolver function private and not a
15551 COMDAT function. Create the name with clone_function_name instead
15552 of make_unique_name.
15555 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
15556 correct operand in doing the split.
15558 2017-07-07 Carl Love <cel@us.ibm.com>
15560 * config/rs6000/rs6000-c: Add support for built-in function
15561 vector unsigned short vec_pack_to_short_fp32 (vector float,
15563 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
15564 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
15565 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
15566 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
15567 (convert_4f32_8i16): Add define_expand.
15568 * doc/extend.texi: Update the built-in documentation file for the
15569 new built-in function.
15571 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15573 * config/sparc/m8.md: New file.
15574 * config/sparc/sparc.md: Include m8.md.
15576 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15578 * config/sparc/sparc.opt: New option -mvis4b.
15579 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
15580 (sparc_option_override): Handle VIS4B.
15581 (enum sparc_builtins): Define
15582 SPARC_BUILTIN_DICTUNPACK{8,16,32},
15583 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
15584 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
15585 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
15586 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
15587 (check_constant_argument): New function.
15588 (sparc_vis_init_builtins): Define builtins
15589 __builtin_vis_dictunpack{8,16,32},
15590 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
15591 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
15592 __builtin_vis_fpcmpde{8,16,32}shl and
15593 __builtin_vis_fpcmpur{8,16,32}shl.
15594 (sparc_expand_builtin): Check that the constant operands to
15595 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
15596 constant and in range.
15597 * config/sparc/sparc-c.c (sparc_target_macros): Handle
15599 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
15600 (SPARC_IMM5_P): Likewise.
15601 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
15602 (enabled): Handle vis4b.
15603 (UNSPEC_DICTUNPACK): New unspec.
15604 (UNSPEC_FPCMPSHL): Likewise.
15605 (UNSPEC_FPUCMPSHL): Likewise.
15606 (UNSPEC_FPCMPDESHL): Likewise.
15607 (UNSPEC_FPCMPURSHL): Likewise.
15608 (cpu_feature): New CPU feature `vis4b'.
15609 (dictunpack{8,16,32}): New insns.
15610 (FPCSMODE): New mode iterator.
15611 (fpcscond): New code iterator.
15612 (fpcsucond): Likewise.
15613 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
15614 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
15615 (fpcmpde{8,16,32}{si,di}shl): Likewise.
15616 (fpcmpur{8,16,32}{si,di}shl): Likewise.
15617 * config/sparc/constraints.md: Define constraints `q' for unsigned
15618 2-bit integer constants and `t' for unsigned 5-bit integer
15620 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
15622 (imm5_operand_dictunpack16): Likewise.
15623 (imm5_operand_dictunpack32): Likewise.
15624 (imm2_operand): Likewise.
15625 * doc/invoke.texi (SPARC Options): Document -mvis4b.
15626 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
15627 ditunpack* and fpcmp*shl builtins.
15629 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15631 * config.gcc: Handle m8 in --with-{cpu,tune} options.
15632 * config.in: Add HAVE_AS_SPARC6 define.
15633 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
15635 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
15637 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
15638 (CPP_CPU_SPEC): Handle m8.
15639 (ASM_CPU_SPEC): Likewise.
15640 * config/sparc/sparc-opts.h (enum processor_type): Add
15642 * config/sparc/sparc.c (m8_costs): New struct.
15643 (sparc_option_override): Handle TARGET_CPU_m8.
15644 (sparc32_initialize_trampoline): Likewise.
15645 (sparc64_initialize_trampoline): Likewise.
15646 (sparc_issue_rate): Likewise.
15647 (sparc_register_move_cost): Likewise.
15648 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
15649 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
15650 (ASM_CPU64_DEFAULT_SPEC): Likewise.
15651 (CPP_CPU_SPEC): Handle M8.
15652 (ASM_CPU_SPEC): Likewise.
15653 (AS_M8_FLAG): Define.
15654 * config/sparc/sparc.md: Add m8 to the cpu attribute.
15655 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
15656 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
15658 * configure: Regenerate.
15659 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
15662 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15664 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
15666 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
15667 ("*movdi_insn_sp32"): Do not set v3pipe.
15668 ("*movsi_insn"): Likewise.
15669 ("*movdi_insn_sp64"): Likewise.
15670 ("*movsf_insn"): Likewise.
15671 ("*movdf_insn_sp32"): Likewise.
15672 ("*movdf_insn_sp64"): Likewise.
15673 ("*zero_extendsidi2_insn_sp64"): Likewise.
15674 ("*sign_extendsidi2_insn"): Likewise.
15675 ("*mov<VM32:mode>_insn"): Likewise.
15676 ("*mov<VM64:mode>_insn_sp64"): Likewise.
15677 ("*mov<VM64:mode>_insn_sp32"): Likewise.
15678 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
15679 ("<vlop:code><VL:mode>3"): Likewise.
15680 ("*not_<vlop:code><VL:mode>3"): Likewise.
15681 ("*nand<VL:mode>_vis"): Likewise.
15682 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
15683 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
15684 ("one_cmpl<VL:mode>2"): Likewise.
15685 ("faligndata<VM64:mode>_vis"): Likewise.
15686 ("alignaddrsi_vis"): Likewise.
15687 ("alignaddrdi_vis"): Likweise.
15688 ("alignaddrlsi_vis"): Likewise.
15689 ("alignaddrldi_vis"): Likewise.
15690 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
15691 ("bmaskdi_vis"): Likewise.
15692 ("bmasksi_vis"): Likewise.
15693 ("bshuffle<VM64:mode>_vis"): Likewise.
15694 ("cmask8<P:mode>_vis"): Likewise.
15695 ("cmask16<P:mode>_vis"): Likewise.
15696 ("cmask32<P:mode>_vis"): Likewise.
15697 ("pdistn<P:mode>_vis"): Likewise.
15698 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
15700 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15702 * config/sparc/sparc.md ("subtype"): New insn attribute.
15703 ("*wrgsr_sp64"): Set insn subtype.
15704 ("*rdgsr_sp64"): Likewise.
15705 ("alignaddrsi_vis"): Likewise.
15706 ("alignaddrdi_vis"): Likewise.
15707 ("alignaddrlsi_vis"): Likewise.
15708 ("alignaddrldi_vis"): Likewise.
15709 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
15710 ("fexpand_vis"): Likewise.
15711 ("fpmerge_vis"): Likewise.
15712 ("faligndata<VM64:mode>_vis"): Likewise.
15713 ("bshuffle<VM64:mode>_vis"): Likewise.
15714 ("cmask8<P:mode>_vis"): Likewise.
15715 ("cmask16<P:mode>_vis"): Likewise.
15716 ("cmask32<P:mode>_vis"): Likewise.
15717 ("fchksm16_vis"): Likewise.
15718 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
15719 ("fmean16_vis"): Likewise.
15720 ("fp<plusminus_insn>64_vis"): Likewise.
15721 ("<plusminus_insn>v8qi3"): Likewise.
15722 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
15723 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
15724 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
15725 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
15726 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
15727 ("*movqi_insn"): Likewise.
15728 ("*movhi_insn"): Likewise.
15729 ("*movsi_insn"): Likewise.
15730 ("movsi_pic_gotdata_op"): Likewise.
15731 ("*movdi_insn_sp32"): Likewise.
15732 ("*movdi_insn_sp64"): Likewise.
15733 ("movdi_pic_gotdata_op"): Likewise.
15734 ("*movsf_insn"): Likewise.
15735 ("*movdf_insn_sp32"): Likewise.
15736 ("*movdf_insn_sp64"): Likewise.
15737 ("*zero_extendhisi2_insn"): Likewise.
15738 ("*zero_extendqihi2_insn"): Likewise.
15739 ("*zero_extendqisi2_insn"): Likewise.
15740 ("*zero_extendqidi2_insn"): Likewise.
15741 ("*zero_extendhidi2_insn"): Likewise.
15742 ("*zero_extendsidi2_insn_sp64"): Likewise.
15743 ("ldfsr"): Likewise.
15744 ("prefetch_64"): Likewise.
15745 ("prefetch_32"): Likewise.
15746 ("tie_ld32"): Likewise.
15747 ("tie_ld64"): Likewise.
15748 ("*tldo_ldub_sp32"): Likewise.
15749 ("*tldo_ldub1_sp32"): Likewise.
15750 ("*tldo_ldub2_sp32"): Likewise.
15751 ("*tldo_ldub_sp64"): Likewise.
15752 ("*tldo_ldub1_sp64"): Likewise.
15753 ("*tldo_ldub2_sp64"): Likewise.
15754 ("*tldo_ldub3_sp64"): Likewise.
15755 ("*tldo_lduh_sp32"): Likewise.
15756 ("*tldo_lduh1_sp32"): Likewise.
15757 ("*tldo_lduh_sp64"): Likewise.
15758 ("*tldo_lduh1_sp64"): Likewise.
15759 ("*tldo_lduh2_sp64"): Likewise.
15760 ("*tldo_lduw_sp32"): Likewise.
15761 ("*tldo_lduw_sp64"): Likewise.
15762 ("*tldo_lduw1_sp64"): Likewise.
15763 ("*tldo_ldx_sp64"): Likewise.
15764 ("*mov<VM32:mode>_insn"): Likewise.
15765 ("*mov<VM64:mode>_insn_sp64"): Likewise.
15766 ("*mov<VM64:mode>_insn_sp32"): Likewise.
15768 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15770 * config/sparc/sparc.md ("type"): New insn type viscmp.
15771 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
15773 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
15774 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
15775 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
15776 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
15778 ("n7_vis_logical_11cycle"): Likewise.
15779 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
15780 * config/sparc/niagara2.md ("niag3_vis": Likewise.
15781 * config/sparc/niagara.md ("niag_vis"): Likewise.
15782 * config/sparc/ultra3.md ("us3_fga"): Likewise.
15783 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
15785 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15787 * config/sparc/sparc.md: New instruction type `bmask'.
15788 (bmaskdi_vis): Use the `bmask' type.
15789 (bmasksi_vis): Likewise.
15790 * config/sparc/ultra3.md (us3_array): Likewise.
15791 * config/sparc/niagara7.md (n7_array): Likewise.
15792 * config/sparc/niagara4.md (n4_array): Likewise.
15793 * config/sparc/niagara2.md (niag2_vis): Likewise.
15794 (niag3_vis): Likewise.
15795 * config/sparc/niagara.md (niag_vis): Likewise.
15797 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15799 * ipa-comdats.c: Remove optimize check from gate.
15800 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
15801 for functions not optimized.
15802 (ipa_fn_summary_read): Skip optimize check.
15803 (ipa_fn_summary_write): Likewise.
15804 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
15806 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
15808 (can_inline_edge_p): Check flag_pcc_struct_return for match.
15809 (check_callers): Give up on caller which is not optimized.
15810 (inline_small_functions): Likewise.
15811 (ipa_inline): Do not give up when not optimizing.
15812 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
15813 away unoptimizes cdtors.
15814 (whole_program_function_and_variable_visibility): Do
15815 ipa_discover_readonly_nonaddressable_vars in LTO mode.
15816 * ipa.c (process_references): Do not check optimize.
15817 (symbol_table::remove_unreachable_nodes): Update optimize check.
15818 (set_writeonly_bit): Update optimize check.
15819 (pass_ipa_cdtor_merge::gate): Do not check optimize.
15820 (pass_ipa_single_use::gate): Remove.
15822 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15824 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
15825 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
15826 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
15827 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
15828 permute_load, permute_store, adjust_extract, adjust_splat,
15829 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
15830 replace_swap_with_copy, dump_swap_insn_table,
15831 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
15832 recombine_lvx_pattern, recombine_stvx_pattern,
15833 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
15834 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
15835 to file rs6000-p8swap.c.
15836 * config/rs6000/rs6000-p8swap.c: New file.
15837 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
15838 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
15839 and rs6000*-*-* targets.
15841 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15843 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
15845 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15847 * lto-wrapper.c (merge_and_complain): Do not merge
15848 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
15849 fsigned_zeros, ftrapping_math, fwrapv.
15850 (append_compiler_options): Do not track these options.
15851 (append_linker_options): Likewie
15853 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15855 * cgraphunit.c (cgraph_node::finalize_function): When
15856 !flag_toplevel_reorde set no_reorder flag.
15857 (varpool_node::finalize_decl): Likewise.
15858 (symbol_table::compile): Drop no toplevel reorder path.
15860 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15862 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
15863 edges; zero probability is not better than uninitialized.
15865 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15867 * asan.h (asan_sanitize_allocas_p): Declare.
15868 * asan.c (asan_sanitize_allocas_p): New function.
15869 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
15870 (handle_builtin_alloca): Likewise.
15871 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
15872 if !asan_sanitize_allocas_p.
15873 * params.def (asan-instrument-allocas): Add new option.
15874 * params.h (ASAN_PROTECT_ALLOCAS): Define.
15875 * opts.c (common_handle_option): Disable allocas sanitization for
15878 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15880 * asan.c: Include gimple-fold.h.
15881 (get_last_alloca_addr): New function.
15882 (handle_builtin_stackrestore): Likewise.
15883 (handle_builtin_alloca): Likewise.
15884 (asan_emit_allocas_unpoison): Likewise.
15885 (get_mem_refs_of_builtin_call): Add new parameter, remove const
15886 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
15887 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
15888 (instrument_builtin_call): Pass gimple iterator to
15889 get_mem_refs_of_builtin_call.
15890 (last_alloca_addr): New global.
15891 * asan.h (asan_emit_allocas_unpoison): Declare.
15892 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
15893 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
15894 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
15895 if function calls alloca.
15896 * gimple-fold.c (replace_call_with_value): Remove static keyword.
15897 * gimple-fold.h (replace_call_with_value): Declare.
15898 * internal-fn.c: Include asan.h.
15899 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
15900 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
15902 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15904 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
15905 (C_SELFTEST_FLAGS): New.
15906 (CPP_SELFTEST_FLAGS): New.
15907 (SELFTEST_DEPS): New, from deps of s-selftest.
15908 (C_SELFTEST_DEPS): New, from deps of s-selftest.
15909 (CPP_SELFTEST_DEPS): New.
15910 (selftest): Add dependency on s-selftest-c++.
15911 (s-selftest): Rename to...
15912 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
15913 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
15914 than SELFTEST_FLAGS.
15915 (selftest-gdb): Rename to...
15916 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
15918 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
15919 (selftest-valgrind): Rename to...
15920 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
15922 (selftest-valgrind): Reintroduce as an alias for
15923 selftest-c-valgrind.
15924 (s-selftest-c++): New.
15925 (selftest-c++-gdb): New.
15926 (selftest-c++-valgrind): New.
15928 2017-07-06 Olivier Hainque <hainque@adacore.com>
15930 * gcc.c (process_command): When deciding if undefined variables
15931 should be ignored when processing specs, accept "gcc -v" as well.
15933 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15935 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
15936 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
15938 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15940 * config/arm/arm-cpus.in (armv8-r): Add new entry.
15941 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
15942 * config/arm/arm-tables.opt: Regenerate.
15943 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
15945 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
15947 2017-07-06 Carl Love <cel@us.ibm.com>
15949 * ChangeLog: Clean up from mid air collision
15951 2017-07-06 Carl Love <cel@us.ibm.com>
15953 * config/rs6000/rs6000-c.c: Add support for built-in functions
15954 vector signed int vec_subc (vector signed int, vector signed int);
15955 vector signed __int128 vec_subc (vector signed __int128,
15956 vector signed __int128);
15957 vector unsigned __int128 vec_subc (vector unsigned __int128,
15958 vector unsigned __int128);
15959 vector signed int vec_sube (vector signed int, vector signed int,
15960 vector signed int);
15961 vector unsigned int vec_sube (vector unsigned int,
15962 vector unsigned int,
15963 vector unsigned int);
15964 vector signed __int128 vec_sube (vector signed __int128,
15965 vector signed __int128,
15966 vector signed__int128);
15967 vector unsigned __int128 vec_sube (vector unsigned __int128,
15968 vector unsigned __int128,
15969 vector unsigned __int128);
15970 vector signed int vec_subec (vector signed int, vector signed int,
15971 vector signed int);
15972 vector unsigned int vec_subec (vector unsigned int,
15973 vector unsigned int,
15974 vector unsigned int);
15975 vector signed __int128 vec_subec (vector signed __int128,
15976 vector signed __int128,
15977 vector signed__int128);
15978 vector unsigned __int128 vec_subec (vector unsigned __int128,
15979 vector unsigned __int128,
15980 vector unsigned __int128);
15981 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
15982 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
15983 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
15984 BU_ALTIVEC_OVERLOAD_X definitions.
15985 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
15986 * doc/extend.texi: Update the built-in documentation file for the new
15987 built-in functions.
15989 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15992 * diagnostic-show-locus.c (layout::layout): Use start and finish
15993 spelling location for the start and finish of each range.
15994 * genmatch.c (linemap_client_expand_location_to_spelling_point):
15995 Add unused aspect param.
15996 * input.c (expand_location_1): Add "aspect" param, and use it
15997 to access the correct part of the location.
15998 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
16000 (expand_location_to_spelling_point): Likewise.
16001 (linemap_client_expand_location_to_spelling_point): Add "aspect"
16002 param, and pass it to expand_location_1.
16004 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
16006 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
16007 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
16008 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
16009 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
16010 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
16011 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
16012 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
16013 _mm_maskz_getmant_ss): New intrinsics.
16014 (__builtin_ia32_getexpss128_mask): Changed to ...
16015 __builtin_ia32_getexpss128_round ... this.
16016 (__builtin_ia32_getexpsd128_mask): Changed to ...
16017 __builtin_ia32_getexpsd128_round ... this.
16018 * config/i386/i386-builtin-types.def
16019 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
16020 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
16021 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
16022 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
16023 __builtin_ia32_getmantss_mask_round): New builtins.
16024 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
16025 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
16026 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
16027 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
16028 * config/i386/sse.md
16029 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
16030 avx512f_sgetexp<mode><mask_scalar_name>
16031 <round_saeonly_scalar_name> ... this.
16032 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
16033 %0, %1, %2<round_saeonly_op3>}): Changed to ...
16034 vgetexp<ssescalarmodesuffix>
16035 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16036 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
16037 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
16038 avx512f_vgetmant<mode><mask_scalar_name>
16039 <round_saeonly_scalar_name> ... this.
16040 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
16041 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
16042 vgetmant<ssescalarmodesuffix>
16043 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
16044 %0<mask_scalar_operand4>, %1, %2
16045 <round_saeonly_scalar_mask_op4>, %3} ... this.
16046 * config/i386/subst.md (mask_scalar_operand4,
16047 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
16048 round_saeonly_scalar_nimm_predicate): New subst attributes.
16050 2017-07-06 Julia Koval <julia.koval@intel.com>
16052 * config/i386/i386.c (ix86_erase_embedded_rounding):
16053 Remove code for old rounding pattern.
16055 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
16057 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
16059 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
16061 * doc/sourcebuild.texi (Test Directives, Variants of
16062 dg-require-support): Add documentation for dg-require-stack-check.
16064 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
16066 * config/i386/subst.md (mask_scalar, round_scalar,
16067 round_saeonly_scalar): New meta-templates.
16068 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
16069 round_scalar_mask_operand3, round_scalar_mask_op3,
16070 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
16071 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
16072 round_saeonly_scalar_constraint,
16073 round_saeonly_scalar_prefix): New subst attribute.
16074 * config/i386/sse.md
16075 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
16076 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
16077 <round_scalar_name> ... this.
16078 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
16079 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
16080 <round_scalar_name> ... this.
16081 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
16082 <sse>_vm<code><mode>3<mask_scalar_name>
16083 <round_saeonly_scalar_name> ... this.
16084 (v<plusminus_mnemonic><ssescalarmodesuffix>
16085 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16086 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16087 v<plusminus_mnemonic><ssescalarmodesuffix>
16088 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16089 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16090 (v<multdiv_mnemonic><ssescalarmodesuffix>
16091 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
16092 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
16093 v<multdiv_mnemonic><ssescalarmodesuffix>
16094 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16095 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16096 (v<maxmin_float><ssescalarmodesuffix>
16097 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
16098 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
16099 v<maxmin_float><ssescalarmodesuffix>
16100 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16101 %0<mask_scalar_operand3>, %1, %<iptr>2
16102 <round_saeonly_scalar_mask_op3>} ... this.
16104 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
16106 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
16107 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
16109 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
16110 Alan Hayward <alan.hayward@arm.com>
16111 David Sherwood <david.sherwood@arm.com>
16113 * combine.c (simplify_if_then_else): Remove "enum" before
16115 * compare-elim.c (can_eliminate_compare): Likewise.
16116 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16118 (aarch64_lookup_simd_builtin_type): Likewise.
16119 (aarch64_simd_builtin_type): Likewise.
16120 (aarch64_init_simd_builtin_types): Likewise.
16121 (aarch64_simd_expand_args): Likewise.
16122 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
16124 (aarch64_reverse_mask): Likewise.
16125 (aarch64_simd_emit_reg_reg_move): Likewise.
16126 (aarch64_gen_adjusted_ldpstp): Likewise.
16127 (aarch64_ccmp_mode_to_code): Likewise.
16128 (aarch64_operands_ok_for_ldpstp): Likewise.
16129 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16130 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
16132 (aarch64_min_divisions_for_recip_mul): Likewise.
16133 (aarch64_reassociation_width): Likewise.
16134 (aarch64_get_condition_code_1): Likewise.
16135 (aarch64_simd_emit_reg_reg_move): Likewise.
16136 (aarch64_simd_attr_length_rglist): Likewise.
16137 (aarch64_reverse_mask): Likewise.
16138 (aarch64_operands_ok_for_ldpstp): Likewise.
16139 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16140 (aarch64_gen_adjusted_ldpstp): Likewise.
16141 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
16143 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
16144 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16145 (arm_lookup_simd_builtin_type): Likewise.
16146 (arm_simd_builtin_type): Likewise.
16147 (arm_init_simd_builtin_types): Likewise.
16148 (arm_expand_builtin_args): Likewise.
16149 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
16150 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
16151 (ft32_setup_incoming_varargs): Likewise.
16152 (ft32_function_arg): Likewise.
16153 (ft32_function_arg_advance): Likewise.
16154 (ft32_pass_by_reference): Likewise.
16155 (ft32_arg_partial_bytes): Likewise.
16156 (ft32_valid_pointer_mode): Likewise.
16157 (ft32_addr_space_pointer_mode): Likewise.
16158 (ft32_addr_space_legitimate_address_p): Likewise.
16159 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
16161 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
16162 (ix86_emit_outlined_ms2sysv_restore): Likewise.
16163 (iamcu_alignment): Likewise.
16164 (canonicalize_vector_int_perm): Likewise.
16165 (ix86_noce_conversion_profitable_p): Likewise.
16166 (ix86_mpx_bound_mode): Likewise.
16167 (ix86_operands_ok_for_move_multiple): Likewise.
16168 * config/microblaze/microblaze-protos.h
16169 (microblaze_expand_conditional_branch_reg): Likewise.
16170 * config/microblaze/microblaze.c
16171 (microblaze_expand_conditional_branch_reg): Likewise.
16172 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16174 (rs6000_reassociation_width): Likewise.
16175 (rs6000_invalid_binary_op): Likewise.
16176 (fusion_p9_p): Likewise.
16177 (emit_fusion_p9_load): Likewise.
16178 (emit_fusion_p9_store): Likewise.
16179 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
16181 (riscv_hard_regno_mode_ok_p): Likewise.
16182 (riscv_address_insns): Likewise.
16183 (riscv_split_symbol): Likewise.
16184 (riscv_legitimize_move): Likewise.
16185 (riscv_function_value): Likewise.
16186 (riscv_hard_regno_nregs): Likewise.
16187 (riscv_expand_builtin): Likewise.
16188 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
16189 (riscv_build_integer): Likewise.
16190 (riscv_split_integer): Likewise.
16191 (riscv_legitimate_constant_p): Likewise.
16192 (riscv_cannot_force_const_mem): Likewise.
16193 (riscv_regno_mode_ok_for_base_p): Likewise.
16194 (riscv_valid_base_register_p): Likewise.
16195 (riscv_valid_offset_p): Likewise.
16196 (riscv_valid_lo_sum_p): Likewise.
16197 (riscv_classify_address): Likewise.
16198 (riscv_legitimate_address_p): Likewise.
16199 (riscv_address_insns): Likewise.
16200 (riscv_load_store_insns): Likewise.
16201 (riscv_force_binary): Likewise.
16202 (riscv_split_symbol): Likewise.
16203 (riscv_force_address): Likewise.
16204 (riscv_legitimize_address): Likewise.
16205 (riscv_move_integer): Likewise.
16206 (riscv_legitimize_const_move): Likewise.
16207 (riscv_legitimize_move): Likewise.
16208 (riscv_address_cost): Likewise.
16209 (riscv_subword): Likewise.
16210 (riscv_output_move): Likewise.
16211 (riscv_canonicalize_int_order_test): Likewise.
16212 (riscv_emit_int_order_test): Likewise.
16213 (riscv_function_arg_boundary): Likewise.
16214 (riscv_pass_mode_in_fpr_p): Likewise.
16215 (riscv_pass_fpr_single): Likewise.
16216 (riscv_pass_fpr_pair): Likewise.
16217 (riscv_get_arg_info): Likewise.
16218 (riscv_function_arg): Likewise.
16219 (riscv_function_arg_advance): Likewise.
16220 (riscv_arg_partial_bytes): Likewise.
16221 (riscv_function_value): Likewise.
16222 (riscv_pass_by_reference): Likewise.
16223 (riscv_setup_incoming_varargs): Likewise.
16224 (riscv_print_operand): Likewise.
16225 (riscv_elf_select_rtx_section): Likewise.
16226 (riscv_save_restore_reg): Likewise.
16227 (riscv_for_each_saved_reg): Likewise.
16228 (riscv_register_move_cost): Likewise.
16229 (riscv_hard_regno_mode_ok_p): Likewise.
16230 (riscv_hard_regno_nregs): Likewise.
16231 (riscv_class_max_nregs): Likewise.
16232 (riscv_memory_move_cost): Likewise.
16233 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
16234 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
16235 (rl78_addr_space_address_mode): Likewise.
16236 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16238 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16239 (rs6000_reassociation_width): Likewise.
16240 (rs6000_invalid_binary_op): Likewise.
16241 (fusion_p9_p): Likewise.
16242 (emit_fusion_p9_load): Likewise.
16243 (emit_fusion_p9_store): Likewise.
16244 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
16245 (ok_for_simple_move_operands): Likewise.
16246 (ok_for_simple_move_strict_operands): Likewise.
16247 (ok_for_simple_arith_logic_operands): Likewise.
16248 (visium_legitimize_reload_address): Likewise.
16249 (visium_select_cc_mode): Likewise.
16250 (output_cbranch): Likewise.
16251 (visium_split_double_move): Likewise.
16252 (visium_expand_copysign): Likewise.
16253 (visium_expand_int_cstore): Likewise.
16254 (visium_expand_fp_cstore): Likewise.
16255 * config/visium/visium.c (visium_pass_by_reference): Likewise.
16256 (visium_function_arg): Likewise.
16257 (visium_function_arg_advance): Likewise.
16258 (visium_libcall_value): Likewise.
16259 (visium_setup_incoming_varargs): Likewise.
16260 (visium_legitimate_constant_p): Likewise.
16261 (visium_legitimate_address_p): Likewise.
16262 (visium_legitimize_address): Likewise.
16263 (visium_secondary_reload): Likewise.
16264 (visium_register_move_cost): Likewise.
16265 (visium_memory_move_cost): Likewise.
16266 (prepare_move_operands): Likewise.
16267 (ok_for_simple_move_operands): Likewise.
16268 (ok_for_simple_move_strict_operands): Likewise.
16269 (ok_for_simple_arith_logic_operands): Likewise.
16270 (visium_function_value_1): Likewise.
16271 (rtx_ok_for_offset_p): Likewise.
16272 (visium_legitimize_reload_address): Likewise.
16273 (visium_split_double_move): Likewise.
16274 (visium_expand_copysign): Likewise.
16275 (visium_expand_int_cstore): Likewise.
16276 (visium_expand_fp_cstore): Likewise.
16277 (visium_split_cstore): Likewise.
16278 (visium_select_cc_mode): Likewise.
16279 (visium_split_cbranch): Likewise.
16280 (output_cbranch): Likewise.
16281 (visium_print_operand_address): Likewise.
16282 * expmed.c (flip_storage_order): Likewise.
16283 * expmed.h (emit_cstore): Likewise.
16284 (flip_storage_order): Likewise.
16285 * genrecog.c (validate_pattern): Likewise.
16286 * hsa-gen.c (gen_hsa_addr): Likewise.
16287 * internal-fn.c (expand_arith_overflow): Likewise.
16288 * ira-color.c (allocno_copy_cost_saving): Likewise.
16289 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16290 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
16291 (process_invariant_for_inheritance): Likewise.
16292 * lra-eliminations.c (move_plus_up): Likewise.
16293 * omp-low.c (lower_oacc_reductions): Likewise.
16294 * simplify-rtx.c (simplify_subreg): Likewise.
16295 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
16296 (TARGET_CHKP_BOUND_MODE): Likewise..
16297 * targhooks.c (default_chkp_bound_mode): Likewise.
16298 (default_setup_incoming_vararg_bounds): Likewise.
16299 * targhooks.h (default_chkp_bound_mode): Likewise.
16300 (default_setup_incoming_vararg_bounds): Likewise.
16301 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
16302 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
16303 (have_whole_vector_shift): Likewise.
16304 * tree-vect-stmts.c (vectorizable_load): Likewise.
16305 * doc/tm.texi: Regenerate.
16307 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16309 Graceful degrade if Binutils PR21472 is not available.
16312 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
16313 .rodata in flash test fails.
16314 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
16315 * confgure: Regenerate.
16316 * config.in: Regenerate.
16317 * config/avr/avr.c (avr_asm_named_section)
16318 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
16319 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
16320 (avr_asm_init_sections): Same.
16322 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16324 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
16325 (fma<VH:mode>4_intrinsic): Likewise.
16326 (*fmsub<VCVTF:mode>4): Likewise.
16327 (*fmsub<VH:mode>4_intrinsic): Likewise.
16329 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16332 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
16333 Don't depend on "optimize > 0".
16334 (out_movhi_r_mr, out_movqi_mr_r): Same.
16335 (out_movhi_mr_r, out_movqi_r_mr): Same.
16336 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
16337 io_address_operand on "optimize > 0".
16339 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16341 * tree-loop-distribution.c: Add general explanantion on the pass.
16342 (generate_loops_for_partition): Mark distributed loop.
16343 (pg_add_dependence_edges): New parameter. Handle alias data
16344 dependence specially and record it in the parameter if asked.
16345 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
16346 (init_partition_graph_vertices, add_partition_graph_edge): New.
16347 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
16348 (free_partition_graph_vdata, build_partition_graph): New.
16349 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
16350 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
16351 (data_ref_segment_size, latch_dominated_by_data_ref): New.
16352 (compute_alias_check_pairs, version_loop_by_alias_check): New.
16353 (version_for_distribution_p, finalize_partitions): New.
16354 (distribute_loop): Handle alias data dependence specially. Factor
16355 out loop fusion code as functions and call these functions.
16357 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16359 * tree-loop-distribution.c (classify_partition): New parameter and
16360 better handle reduction statement.
16361 (rdg_build_partitions): Revise comment.
16362 (distribute_loop): Compute statements in all partitions and pass it
16363 to classify_partition.
16365 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16367 * tree-loop-distribution.c (enum partition_type): New.
16368 (struct partition): New field type.
16369 (partition_merge_into): Add parameter. Update partition type.
16370 (data_dep_in_cycle_p, update_type_for_merge): New functions.
16371 (build_rdg_partition_for_vertex): Compute partition type.
16372 (rdg_build_partitions): Dump partition type.
16373 (distribute_loop): Update calls to partition_merge_into.
16375 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16377 * tree-loop-distribution.c (struct ddr_hasher): New.
16378 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
16380 (classify_partition): Call get_data_dependence.
16381 (pg_add_dependence_edges): Ditto.
16382 (distribute_loop): Release data dependence hash table.
16384 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16386 * tree-loop-distribution.c (ref_base_address): Delete.
16387 (similar_memory_accesses): Rename ...
16388 (share_memory_accesses): ... to this. Check if partitions access
16389 the same memory reference.
16390 (distribute_loop): Call share_memory_accesses.
16392 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16394 * tree-loop-distribution.c (struct partition): New field recording
16395 its data reference.
16396 (partition_alloc, partition_free): Init and release data refs.
16397 (partition_merge_into): Merge data refs.
16398 (build_rdg_partition_for_vertex): Collect data refs for partition.
16399 (pg_add_dependence_edges): Change parameters from vector to bitmap.
16401 (distribute_loop): Remve data refs from vertice data of partition
16404 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16406 * tree-loop-distribution.c (params.h): Include header file.
16407 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
16408 (datarefs_vec): New global var.
16409 (create_rdg_vertices): Use datarefs_vec directly.
16410 (free_rdg): Don't free data references.
16411 (build_rdg): Update use. Don't free data references.
16412 (distribute_loop): Compute global variable for data references.
16413 Bail out if there are too many data references.
16415 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16417 * tree-loop-distribution.c (loop_nest): New global var.
16418 (build_rdg): Use loop directly, rather than loop nest.
16419 (pg_add_dependence_edges): Remove loop nest parameter. Use global
16421 (distribute_loop): Compute global variable loop nest. Update use.
16423 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16425 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
16426 (partition_merge_into): New parameter. Dump reason for fusion.
16427 (distribute_loop): Update use of partition_merge_into.
16429 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16431 * tree-loop-distribution.c (bb_top_order_index): New.
16432 (bb_top_order_index_size, bb_top_order_cmp): New.
16433 (stmts_from_loop): Use topological order.
16434 (pass_loop_distribution::execute): Compute and release topological
16435 order for basic blocks.
16437 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16439 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
16442 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16444 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
16445 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
16446 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
16447 * internal-fn.def (LOOP_DIST_ALIAS): New.
16448 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
16449 (fold_loop_internal_call): ... this.
16450 (vect_loop_dist_alias_call): New function.
16451 (set_uid_loop_bbs): Call fold_loop_internal_call.
16452 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
16455 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16458 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
16459 Require dead FLAGS_REG at the beginning of a peephole.
16461 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16464 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
16465 arguments in the call to __builtin_ia32_sbb_u32.
16466 (_subborrow_u64): Swap _X and _Y arguments in the call to
16467 __builtin_ia32_sbb_u64.
16469 2017-07-04 Jakub Jelinek <jakub@redhat.com>
16472 * tree-vrp.c (compare_assert_loc): Turn into a function template
16473 with stable template parameter. Only test if a->e is NULL,
16474 !a->e == !b->e has been verified already. Use e == NULL or
16475 e != NULL instead of e or ! e tests. If stable is true, don't use
16476 iterative_hash_expr, on the other side allow a or b or both NULL
16477 and sort the NULLs last.
16478 (process_assert_insertions): Sort using compare_assert_loc<false>
16479 instead of compare_assert_loc, later sort using
16480 compare_assert_loc<true> before calling process_assert_insertions_for
16481 in a loop. Use break instead of continue once seen NULL pointer.
16483 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16485 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16486 Cortex-R7 and Cortex-R8 processors.
16488 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16490 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
16491 uninitialized while src is not.
16493 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
16495 * common/config/arm/arm-common.c: Adjust include path for
16497 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
16498 (arm-cpu.h): Create in build directory. Adjust dependency rules.
16499 (arm-cpu-data.h): Likewise.
16500 (arm-cpu-cdata.h): Likewise.
16501 * config/arm/arm-cpu.h: Delete.
16502 * config/arm/arm-cpu-cdata.h: Delete.
16503 * config/arm/arm-cpu-data.h: Delete.
16505 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
16507 * config/arm/arm-cpus.in (cortex-a55): New.
16508 (cortex-a75): Likewise.
16509 (cortex-a75.cortex-a55): Likewise.
16510 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
16512 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
16513 * config/arm/arm-cpu-cdata.h: Regenerate.
16514 * config/arm/arm-cpu-data.h: Regenerate.
16515 * config/arm/arm-cpu.h: Regenerate.
16516 * config/arm/arm-tables.opt: Regenerate.
16517 * config/arm/arm-tune.md: Regenerate.
16519 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16521 * haifa-sched.c (sched_create_recovery_edges): Update profile.
16523 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16525 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
16528 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
16530 PR tree-optimization/81292
16531 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
16532 full_string_p, also call adjust_related_strinfos if the adjustment
16533 is simple, otherwise invalidate related strinfos.
16535 2017-07-04 Martin Liska <mliska@suse.cz>
16538 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
16539 newly created variable as DECL_IGNORED_P.
16541 2017-07-04 Martin Liska <mliska@suse.cz>
16544 * ipa-inline.c (inline_small_functions):
16545 Use xstrdup_for_dump.
16547 2017-07-04 Tom de Vries <tom@codesourcery.com>
16549 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
16551 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16554 * config/darwin.c (darwin_function_switched_text_sections):
16557 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
16559 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
16561 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16563 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
16565 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16567 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
16568 min_profitable_iters, and th as inclusive lower bounds.
16569 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
16570 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
16571 for min_profitable_iters and min_profitable_estimate.
16572 (vect_transform_loop): Treat th as an inclusive lower bound.
16573 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16575 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16578 * config/darwin.c (darwin_function_switched_text_sections):
16579 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
16580 in two pieces, and suppress the use of buf.
16582 2017-07-03 Nathan Sidwell <nathan@acm.org>
16584 * hash-table.h (hash_table_mod1): Fix indentation.
16586 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16588 PR middle-end/81290
16589 * predict.c (force_edge_cold): Be more careful about propagation
16591 * profile-count.h (profile_probability::guessed,
16592 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
16594 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
16596 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
16598 * doc/invoke.texi (rcpc architecture extension): Document it.
16600 2017-07-03 Richard Biener <rguenther@suse.de>
16602 PR tree-optimization/60510
16603 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
16604 the scalar reduction PHI and use it.
16605 (vectorizable_reduction): Properly guard the single_defuse_cycle
16606 path for non-SLP reduction chains where we cannot use it.
16607 Rework reduc_def/index and vector type deduction. Rework
16608 vector operand gathering during reduction op code-gen.
16609 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
16610 chains dissolve the chain and leave it to non-SLP reduction
16613 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16615 * tree-data-ref.h (dr_alignment): Declare.
16616 * tree-data-ref.c (dr_alignment): New function.
16617 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
16618 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
16620 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
16622 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16624 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
16625 and base_misalignment fields.
16626 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
16627 * tree-data-ref.c: Include builtins.h.
16628 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
16629 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
16630 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
16631 * tree-vect-data-refs.c: Include tree-cfg.h.
16632 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
16633 fields instead of calculating an alignment here.
16634 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
16635 innermost_loop_behavior fields.
16637 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16639 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
16641 (DR_STEP_ALIGNMENT): New macro.
16642 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
16643 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
16644 (create_data_ref): Print it.
16645 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
16646 to tell whether the step preserves vector (mis)alignment.
16647 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16648 Move the check for an integer step and generalise to all INTEGER_CST.
16649 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
16650 Print the outer step alignment.
16652 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16654 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
16655 with offset_alignment.
16656 (DR_ALIGNED_TO): Delete.
16657 (DR_OFFSET_ALIGNMENT): New macro.
16658 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
16659 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
16660 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
16661 (create_data_ref): Likewise.
16662 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16663 (vect_analyze_data_refs): Likewise.
16664 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
16665 creating dummy innermost behavior.
16667 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16669 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
16670 with a "innermost_loop_behavior *" and refeence tree.
16671 * tree-data-ref.c (dr_analyze_innermost): Likewise.
16672 (create_data_ref): Update call accordingly.
16673 * tree-predcom.c (find_looparound_phi): Likewise.
16675 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16677 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
16678 fields with dr_wrt_vec_loop.
16679 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
16680 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
16681 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
16682 (vect_dr_behavior): New function.
16683 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
16684 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
16685 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
16686 track whether the step preserves the misalignment.
16687 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
16688 Use vect_dr_behavior.
16689 (vect_setup_realignment): Update call accordingly.
16690 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
16691 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
16692 call to vect_create_addr_base_for_vector_ref.
16693 (vect_create_cond_for_align_checks): Likewise.
16694 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
16695 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
16696 (vect_recog_mask_conversion_pattern): Likewise.
16697 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
16698 (new_stmt_vec_info): Remove redundant zeroing.
16700 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16702 * common/config/arm/arm-common.c (arm_be8_option): New function.
16703 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
16704 (ISA_ARMv6): Add isa_bit_be8.
16705 * config/arm/arm.h (arm_be8_option): Add prototype.
16706 (BE8_SPEC_FUNCTION): New define.
16707 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
16708 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
16709 (mlittle-endian): Similarly.
16710 (mbe8, mbe32): New options.
16711 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
16712 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
16714 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16716 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
16718 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16720 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
16721 (cleanup_tree_cfg_bb): Use it.
16722 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
16724 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
16726 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16729 * loop-doloop.c (add_test): Update profile.
16731 2017-07-03 Martin Liska <mliska@suse.cz>
16734 * sanopt.c (rewrite_usage_of_param): New function.
16735 (sanitize_rewrite_addressable_params): Likewise.
16736 (pass_sanopt::execute): Call rewrite_usage_of_param.
16738 2017-07-03 Richard Biener <rguenther@suse.de>
16740 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
16741 back to using VIEW_CONVERT_EXPR.
16743 2017-07-03 Martin Liska <mliska@suse.cz>
16746 * doc/extend.texi: Document when a resolver function is
16747 generated for target_clones.
16749 2017-07-03 Martin Liska <mliska@suse.cz>
16751 * asan.c (asan_emit_stack_protection): Unpoison just red zones
16752 and shadow memory of auto variables which are subject of
16753 use-after-scope sanitization.
16754 (asan_expand_mark_ifn): Add do set only when is_poison.
16756 2016-07-03 Richard Biener <rguenther@suse.de>
16758 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
16760 (vect_force_simple_reduction): Record reduction def -> phi mapping.
16761 (vectorizable_reduction): Perform reduction PHI creation when
16762 visiting a reduction PHI and adjust and simplify code generation
16763 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
16764 (vect_transform_loop): Visit reduction PHIs.
16765 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
16766 defs into the SLP tree.
16767 (vect_build_slp_tree): Reduction defs terminate the recursion.
16768 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
16770 (vect_get_vec_defs_for_stmt_copy): Export.
16771 (vect_get_vec_defs): Likewise.
16772 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
16774 (vect_get_vec_defs_for_stmt_copy): Declare.
16775 (vect_get_vec_defs): Likewise.
16777 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16779 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
16780 parameter with a "loop" parameter and use it instead of the
16781 loop containing DR_STMT. Don't check simple_iv when doing
16782 BB analysis. Describe the two analysis modes in the comment.
16784 2017-07-03 Tom de Vries <tom@codesourcery.com>
16786 PR tree-optimization/69468
16787 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
16788 (find_same_succ_bb): Handle ignore_edge_flags.
16790 2017-07-03 Tom de Vries <tom@codesourcery.com>
16792 PR tree-optimization/81192
16793 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
16795 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
16797 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
16799 2017-07-03 Tom de Vries <tom@codesourcery.com>
16801 PR tree-optimization/81192
16802 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
16803 BB_SAME_SUCC (bb) == NULL.
16805 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16807 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
16810 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16812 * dumpfile.c: Include profile-count.h
16813 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
16815 (insert_cond_bb): Update profile.
16816 * tree-cfg.h (insert_cond_bb): Update prototype.
16817 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
16818 * tree-dump.c: Do not include tree-cfg.
16820 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16822 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
16824 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16826 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
16829 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16831 * tree-complex.c (expand_complex_div_wide): update profile.
16833 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16834 Alan Hayward <alan.hayward@arm.com>
16835 David Sherwood <david.sherwood@arm.com>
16837 * Makefile.in (MACHMODE_H): Remove insn-modes.h
16838 (CORETYPES_H): New define.
16839 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
16840 (insn-modes-inline.h, s-modes-inline-h): New rules.
16841 (generated_files): Add insn-modes-inline.h.
16842 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
16843 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
16844 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
16845 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
16846 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
16847 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
16848 (build/gencodes.o, build/genconditions.o): Likewise.
16849 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
16850 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
16851 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
16852 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
16853 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
16854 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
16855 * coretypes.h: Include everything up to real.h for generators.
16856 Include insn-modes.h first. Include wide-int-print.h after
16857 wide-int.h. Include insn-modes-inline.h and then machmode.h.
16858 * machmode.h: Don't include insn-modes.h here.
16859 * function-tests.c: Remove includes of signop.h, machmode.h,
16860 double-int.h and wide-int.h.
16862 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
16864 * optc-save-gen.awk: Likewise.
16865 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
16866 * godump.c: Remove include of wide-int-print.h.
16867 * pretty-print.h: Likewise.
16868 * wide-int-print.cc: Likewise.
16869 * wide-int.cc: Likewise.
16870 * hash-map-tests.c: Remove include of signop.h.
16871 * hash-set-tests.c: Likewise.
16872 * rtl-tests.c: Likewise.
16873 * mkconfig.sh: Remove include of machmode.h.
16874 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
16876 (emit_insn_modes_inline_h): ...this new function. Emit the code
16877 into an insn-modes-inline.h header file, adding appropriate
16878 include guards and end comments.
16879 (emit_insn_modes_c_header): Remove include of machmode.h.
16880 (emit_min_insn_modes_c_header): Include coretypes.h rather than
16882 (main): Handle -i flag and call emit_insn_modes_inline_h when
16885 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16887 * tree-ssa-strlen.c (strinfo): Rename the length field to
16888 nonzero_chars. Add a full_string_p field.
16889 (compare_nonzero_chars, zero_length_string_p): New functions.
16890 (get_addr_stridx): Add an offset_out parameter.
16891 Use compare_nonzero_chars.
16892 (get_stridx): Update accordingly. Use compare_nonzero_chars.
16893 (new_strinfo): Update after above changes to strinfo.
16894 (set_endptr_and_length): Set full_string_p.
16895 (get_string_length): Update after above changes to strinfo.
16896 (unshare_strinfo): Update call to new_strinfo.
16897 (maybe_invalidate): Likewise.
16898 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
16899 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
16900 as a uhwi instead of an shwi. Update after above changes to
16901 strinfo and new_strinfo.
16902 (zero_length_string): Assert that chainsi contains full strings.
16903 Use zero_length_string_p. Update call to new_strinfo.
16904 (adjust_related_strinfos): Update after above changes to strinfo.
16905 Copy full_string_p from origsi.
16906 (adjust_last_stmt): Use zero_length_string_p.
16907 (handle_builtin_strlen): Update after above changes to strinfo and
16908 new_strinfo. Install the lhs as the string length if the previous
16909 entry didn't describe a full string.
16910 (handle_builtin_strchr): Update after above changes to strinfo
16912 (handle_builtin_strcpy): Likewise.
16913 (handle_builtin_strcat): Likewise.
16914 (handle_builtin_malloc): Likewise.
16915 (handle_pointer_plus): Likewise.
16916 (handle_builtin_memcpy): Likewise. Track nonzero characters
16917 that aren't necessarily followed by a nul terminator.
16918 (handle_char_store): Likewise.
16920 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16922 PR tree-optimization/80769
16923 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
16924 for malloc and calloc. Document the new invariant that all related
16925 strinfos have delayed lengths or none do.
16926 (verify_related_strinfos): Move earlier in file.
16927 (set_endptr_and_length): New function, split out from...
16928 (get_string_length): ...here. Also set the lengths of related
16930 (zero_length_string): Assert that chainsi has known (rather than
16932 (adjust_related_strinfos): Likewise.
16934 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16936 PR tree-optimization/81136
16937 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
16938 assert that two references with the same misalignment have the same
16939 compile-time misalignment if those compile-time misalignments
16942 2017-07-01 Andi Kleen <ak@linux.intel.com>
16944 * print-tree.c (print_node): Print all attributes.
16946 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16948 * cfg.c (scale_bbs_frequencies): New function.
16949 * cfg.h (scale_bbs_frequencies): Declare it.
16950 * cfgloopanal.c (single_likely_exit): Cleanup.
16951 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
16953 (scale_loop_profile): Likewise.
16954 (loop_version): Likewise.
16955 (create_empty_loop_on_edge): Update.
16956 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
16957 scale_loop_frequencies, scale_loop_profile, loopify,
16958 loop_version): Update prototypes.
16959 * modulo-sched.c (sms_schedule): Update.
16960 * predict.c (unlikely_executed_edge_p): Also check probability.
16961 (probably_never_executed_edge_p): Fix typo.
16962 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16963 * tree-parloops.c (gen_parallel_loop): Update.
16964 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
16965 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16966 * tree-ssa-loop-split.c (split_loop): Update.
16967 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16968 * tree-vect-loop-manip.c (vect_do_peeling): Update.
16969 (vect_loop_versioning): Update.
16970 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16972 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16974 * trans-mem.c (split_bb_make_tm_edge): Update profile.
16976 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16978 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
16979 to keep profile consistent.
16981 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16983 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
16984 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
16985 * profile-count.h (max_safe_multiplier): Make unsigned.
16986 (profile_count::guessed_zero): New.
16988 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16990 * bb-reorder.c (fix_up_crossing_landing_pad,
16991 fix_crossing_conditional_branches): Use make_single_succ_edge
16992 to keep profile consistent.
16994 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16996 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
16999 2017-07-01 Jakub Jelinek <jakub@redhat.com>
17002 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
17003 the right scopes, make sure cond_jump isn't preserved between multiple
17004 iterations. Search for fallthru edge whenever there are 3+ edges and
17005 use find_fallthru_edge for it.
17007 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17009 Patch by Alexander Monakov <amonakov@ispras.ru>
17010 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
17011 probabilities consistently.
17013 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17015 * pa.c (pa_expand_compare_and_swap_loop): Update call of
17016 emit_cmp_and_jump_insns.
17018 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17021 * tree-inline.c (expand_call_inline): Combine profile statuses.
17023 2017-06-30 Andrew Pinski <apinski@cavium.com>
17025 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
17026 fold_stmt returned true.
17028 2017-06-30 Nathan Sidwell <nathan@acm.org>
17030 * ggc.h (empty_string): Delete.
17031 * cfgexpand.c (expand_asm_stmt): Use plain "".
17032 * optabs.c (expand_asm_memory_barrier): Likewise.
17033 * stringpool.c (empty_string): Delete.
17034 (digit_vector, digit_string): Delete.
17035 (ggc_alloc_string): Use plain "", don't optimize single digit
17036 strings. Use ggc_alloc_atomic.
17038 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
17040 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
17041 comparison set and one other set, use the cost of the non-comparison
17044 2017-06-30 Nathan Sidwell <nathan@acm.org>
17046 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
17049 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
17051 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
17052 loops. Remove now unneeded calls to gimple_switch_set_label() that
17053 just set removed labels to NULL_TREE.
17055 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
17057 * tree-ssanames.c (set_range_info_raw): Abstract from ...
17058 (set_range_info): ...here. Only call set_range_info_raw if domain
17060 (set_nonzero_bits): Call set_range_info_raw.
17061 * tree-ssanames.h (set_range_info_raw): New.
17063 2017-06-30 Jakub Jelinek <jakub@redhat.com>
17066 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
17067 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
17068 of nonimmediate_operand and <store_mask_constraint> instead of m
17069 for the input operand. For V8FI iterator, always split if input
17070 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
17071 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
17072 <store_mask_predicate> instead of register_operand and
17073 <store_mask_constraint> instead of v for the input operand. Make
17074 sure both operands aren't MEMs for if not <mask_applied>.
17076 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
17078 * lto-wrapper.c (copy_file) Close both file descriptors before
17081 2017-06-30 Martin Liska <mliska@suse.cz>
17084 * multiple_target.c (create_dispatcher_calls): Make ifunc
17085 also for function that don't have calls or are not referenced.
17087 2017-06-30 Richard Biener <rguenther@suse.de>
17089 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
17090 analyze the first scalar stmt. Move vector type computation
17091 for the BB case here from ...
17092 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
17093 live operation processing in the SLP case properly.
17095 2017-06-30 Richard Biener <rguenther@suse.de>
17097 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
17099 2017-06-30 Martin Liska <mliska@suse.cz>
17102 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
17103 before BUILT_IN_UNWIND_RESUME when ASAN is used.
17105 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
17107 * doc/invoke.texi (AArch64): Add missing options and remove redundant
17110 2017-06-30 Richard Biener <rguenther@suse.de>
17112 PR tree-optimization/81249
17113 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
17114 condition reduction result to original scalar type.
17116 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17118 * profile-count.h (enum profile_quality): Fix typos and whitespace
17121 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17123 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
17124 type for branch probabilities.
17126 2017-06-29 Julian Brown <julian@codesourcery.com>
17127 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17129 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
17130 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
17131 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
17132 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
17134 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17136 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
17137 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
17138 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
17139 CC usage from generic code to here.
17140 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
17141 CC usage into the target macros.
17143 2017-06-29 Maya Rashish <coypu@sdf.org>
17145 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
17148 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17150 * arm/arm-builtins.c: Include profile-count.h
17151 * except.c (sjlj_emit_function_enter): Use
17152 profile_probability::unlikely.
17154 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17156 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
17157 and tocrel_offset be pointer args rather than implicitly using
17159 (legitimate_constant_pool_address_p, rs6000_emit_move,
17160 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
17161 tocrel_offset and use in toc_relative_expr_p call.
17162 (print_operand, print_operand_address): Use static tocrel_base_oac
17163 and tocrel_offset_oac.
17164 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
17167 2017-06-29 Maya Rashish <coypu@sdf.org>
17169 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
17171 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
17173 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
17174 objects, take into account only the alignment of 'op0' and 'mode1' if
17177 2017-06-29 Steve Ellcey <sellcey@cavium.com>
17179 * ccmp.c (ccmp_tree_comparison_p): New function.
17180 (ccmp_candidate_p): Update to use above function.
17181 (get_compare_parts): New function.
17182 (expand_ccmp_next): Update to use new functions.
17183 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
17185 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
17186 take mode as argument.
17187 * ccmp.h (expand_ccmp_expr): Add mode as argument.
17188 * expr.c (expand_expr_real_1): Pass mode as argument.
17190 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
17192 * combine.c (combine_instructions): Print insns to dump_file, together
17195 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17197 * asan.c (asan_emit_stack_protection): Update.
17198 (create_cond_insert_point): Update.
17199 * auto-profile.c (afdo_propagate_circuit): Update.
17200 * basic-block.h (struct edge_def): Turn probability to
17201 profile_probability.
17202 (EDGE_FREQUENCY): Update.
17203 * bb-reorder.c (find_traces_1_round): Update.
17204 (better_edge_p): Update.
17205 (sanitize_hot_paths): Update.
17206 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
17207 (make_single_succ_edge): Update.
17208 (check_bb_profile): Update.
17209 (dump_edge_info): Update.
17210 (update_bb_profile_for_threading): Update.
17211 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
17212 probabilitycount to 0.
17213 * cfgbuild.c (compute_outgoing_frequencies): Update.
17214 * cfgcleanup.c (try_forward_edges): Update.
17215 (outgoing_edges_match): Update.
17216 (try_crossjump_to_edge): Update.
17217 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
17218 (expand_gimple_tailcall): Update.
17219 (construct_init_block): Use make_single_succ_edge.
17220 (construct_exit_block): Use make_single_succ_edge.
17221 * cfghooks.c (verify_flow_info): Update.
17222 (redirect_edge_succ_nodup): Update.
17223 (split_edge): Update.
17224 (account_profile_record): Update.
17225 * cfgloopanal.c (single_likely_exit): Update.
17226 * cfgloopmanip.c (scale_loop_profile): Update.
17227 (set_zero_probability): Remove.
17228 (duplicate_loop_to_header_edge): Update.
17229 * cfgloopmanip.h (loop_version): Update prototype.
17230 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
17231 (force_nonfallthru_and_redirect): Update.
17232 (update_br_prob_note): Update.
17233 (rtl_verify_edges): Update.
17234 (purge_dead_edges): Update.
17235 (rtl_lv_add_condition_to_bb): Update.
17236 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
17237 * cgraphunit.c (init_lowered_empty_function): Update.
17238 (cgraph_node::expand_thunk): Update.
17239 * cilk-common.c: Include profile-count.h
17240 * dojump.c (inv): Remove.
17241 (jumpifnot): Update.
17242 (jumpifnot_1): Update.
17243 (do_jump_1): Update.
17245 (do_jump_by_parts_greater_rtx): Update.
17246 (do_compare_rtx_and_jump): Update.
17247 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
17248 do_jump_1. do_compare_rtx_and_jump): Update prototype.
17249 * dwarf2cfi.c: Include profile-count.h
17250 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
17251 (sjlj_emit_dispatch_table): Likewise.
17252 * explow.c: Include profile-count.h
17253 * expmed.c (emit_store_flag_force): Update.
17254 (do_cmp_and_jump): Update.
17255 * expr.c (compare_by_pieces_d::generate): Update.
17256 (compare_by_pieces_d::finish_mode): Update.
17257 (emit_block_move_via_loop): Update.
17258 (store_expr_with_bounds): Update.
17259 (store_constructor): Update.
17260 (expand_expr_real_2): Update.
17261 (expand_expr_real_1): Update.
17262 * expr.h (try_casesi, try_tablejump): Update prototypes.
17263 * gimple-pretty-print.c (dump_probability): Update.
17264 (dump_profile): New.
17265 (dump_gimple_label): Update.
17266 (dump_gimple_bb_header): Update.
17267 * graph.c (draw_cfg_node_succ_edges): Update.
17268 * hsa-gen.c (convert_switch_statements): Update.
17269 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
17270 (find_if_case_1): Update.
17271 (find_if_case_2): Update.
17272 * internal-fn.c (expand_arith_overflow_result_store): Update.
17273 (expand_addsub_overflow): Update.
17274 (expand_neg_overflow): Update.
17275 (expand_mul_overflow): Update.
17276 (expand_vector_ubsan_overflow): Update.
17277 * ipa-cp.c (good_cloning_opportunity_p): Update.
17278 * ipa-split.c (split_function): Use make_single_succ_edge.
17279 * ipa-utils.c (ipa_merge_profiles): Update.
17280 * loop-doloop.c (add_test): Update.
17281 (doloop_modify): Update.
17282 * loop-unroll.c (compare_and_jump_seq): Update.
17283 (unroll_loop_runtime_iterations): Update.
17284 * lra-constraints.c (lra_inheritance): Update.
17285 * lto-streamer-in.c (input_cfg): Update.
17286 * lto-streamer-out.c (output_cfg): Update.
17287 * mcf.c (adjust_cfg_counts): Update.
17288 * modulo-sched.c (sms_schedule): Update.
17289 * omp-expand.c (expand_omp_for_init_counts): Update.
17290 (extract_omp_for_update_vars): Update.
17291 (expand_omp_ordered_sink): Update.
17292 (expand_omp_for_ordered_loops): Update.
17293 (expand_omp_for_generic): Update.
17294 (expand_omp_for_static_nochunk): Update.
17295 (expand_omp_for_static_chunk): Update.
17296 (expand_cilk_for): Update.
17297 (expand_omp_simd): Update.
17298 (expand_omp_taskloop_for_outer): Update.
17299 (expand_omp_taskloop_for_inner): Update.
17300 * omp-simd-clone.c (simd_clone_adjust): Update.
17301 * optabs.c (expand_doubleword_shift): Update.
17302 (expand_abs): Update.
17303 (emit_cmp_and_jump_insn_1): Update.
17304 (expand_compare_and_swap_loop): Update.
17305 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
17306 * predict.c (predictable_edge_p): Update.
17307 (edge_probability_reliable_p): Update.
17308 (set_even_probabilities): Update.
17309 (combine_predictions_for_insn): Update.
17310 (combine_predictions_for_bb): Update.
17311 (propagate_freq): Update.
17312 (estimate_bb_frequencies): Update.
17313 (force_edge_cold): Update.
17314 * profile-count.c (profile_count::dump): Add missing space into dump.
17315 (profile_count::debug): Add newline.
17316 (profile_count::differs_from_p): Explicitly convert to unsigned.
17317 (profile_count::stream_in): Update.
17318 (profile_probability::dump): New member function.
17319 (profile_probability::debug): New member function.
17320 (profile_probability::differs_from_p): New member function.
17321 (profile_probability::differs_lot_from_p): New member function.
17322 (profile_probability::stream_in): New member function.
17323 (profile_probability::stream_out): New member function.
17324 * profile-count.h (profile_count_quality): Rename to ...
17325 (profile_quality): ... this one.
17326 (profile_probability): New.
17327 (profile_count): Update.
17328 * profile.c (compute_branch_probabilities): Update.
17329 * recog.c (peep2_attempt): Update.
17330 * sched-ebb.c (schedule_ebbs): Update.
17331 * sched-rgn.c (find_single_block_region): Update.
17332 (compute_dom_prob_ps): Update.
17333 (schedule_region): Update.
17334 * sel-sched-ir.c (compute_succs_info): Update.
17335 * stmt.c (struct case_node): Update.
17336 (do_jump_if_equal): Update.
17337 (get_outgoing_edge_probs): Update.
17338 (conditional_probability): Update.
17339 (emit_case_dispatch_table): Update.
17340 (expand_case): Update.
17341 (expand_sjlj_dispatch_table): Update.
17342 (emit_case_nodes): Update.
17343 * targhooks.c: Update.
17344 * tracer.c (better_p): Update.
17345 (find_best_successor): Update.
17346 * trans-mem.c (expand_transaction): Update.
17347 * tree-call-cdce.c: Update.
17348 * tree-cfg.c (gimple_split_edge): Upate.
17349 (move_sese_region_to_fn): Upate.
17350 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
17351 * tree-eh.c (lower_resx): Upate.
17352 (cleanup_empty_eh_move_lp): Upate.
17353 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17354 * tree-inline.c (copy_edges_for_bb): Update.
17355 (copy_cfg_body): Update.
17356 * tree-parloops.c (gen_parallel_loop): Update.
17357 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
17358 (gimple_gen_time_profiler): Update.
17359 * tree-ssa-dce.c (remove_dead_stmt): Update.
17360 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17361 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
17362 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
17363 (unloop_loops): Update.
17364 (try_peel_loop): Update.
17365 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17366 * tree-ssa-loop-split.c (connect_loops): Update.
17367 (split_loop): Update.
17368 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17369 (hoist_guard): Update.
17370 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
17371 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
17372 (value_replacement): Update.
17373 * tree-ssa-reassoc.c (branch_fixup): Update.
17374 * tree-ssa-tail-merge.c (replace_block_by): Update.
17375 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
17376 (create_edge_and_update_destination_phis): Update.
17377 (compute_path_counts): Update.
17378 (recompute_probabilities): Update.
17379 (update_joiner_offpath_counts): Update.
17380 (freqs_to_counts_path): Update.
17381 (duplicate_thread_path): Update.
17382 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
17383 (struct switch_conv_info): Update.
17384 (gen_inbound_check): Update.
17385 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17386 (vect_do_peeling): Update.
17387 (vect_loop_versioning): Update.
17388 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17389 (optimize_mask_stores): Update.
17390 * ubsan.c (ubsan_expand_null_ifn): Update.
17391 * value-prof.c (gimple_divmod_fixed_value): Update.
17392 (gimple_divmod_fixed_value_transform): Update.
17393 (gimple_mod_pow2): Update.
17394 (gimple_mod_pow2_value_transform): Update.
17395 (gimple_mod_subtract): Update.
17396 (gimple_mod_subtract_transform): Update.
17397 (gimple_ic): Update.
17398 (gimple_stringop_fixed_value): Update.
17399 (gimple_stringops_transform): Update.
17400 * value-prof.h: Update.
17402 2017-06-29 Carl Love <cel@us.ibm.com>
17404 * config/rs6000/rs6000-c.c: Add support for built-in functions
17405 vector signed int vec_signed (vector float);
17406 vector signed long long vec_signed (vector double);
17407 vector signed int vec_signed2 (vector double, vector double);
17408 vector signed int vec_signede (vector double);
17409 vector signed int vec_signedo (vector double);
17410 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
17411 instruction generator.
17412 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17413 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
17414 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
17416 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
17417 vunsignede_v2df): Add define_expands.
17418 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
17419 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
17420 VEC_UNSIGNEDO): Add definitions.
17421 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17422 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
17423 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
17424 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
17425 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
17426 * config/rs6000/altivec.h (vec_signed, vec_signed2,
17427 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
17428 vec_unsignede, vec_unsignedo): Add builtin defines.
17429 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
17431 * doc/extend.texi: Update the built-in documentation file for the
17432 new built-in functions.
17434 2017-06-29 Richard Biener <rguenther@suse.de>
17436 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
17437 reduction chains to LOOP_VINFO_REDUCTIONS.
17438 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
17439 SLP reductions after processing reduction chains.
17441 2017-06-29 Nathan Sidwell <nathan@acm.org>
17443 * builtins.c (fold_builtin_FUNCTION): Use
17444 lang_hooks.decl_printable_name.
17446 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
17448 PR middle-end/81194
17449 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
17450 with only one label.
17451 * stmt.c (expand_case): Assert NCASES is greater than one.
17453 2017-06-29 Richard Biener <rguenther@suse.de>
17455 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
17457 (group_case_labels): Likewise.
17458 (find_taken_edge): Push sanity checking on val to workers...
17459 (find_taken_edge_cond_expr): ... here
17460 (find_taken_edge_switch_expr): ... and here, handle cases
17461 with just a default label.
17462 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
17463 (group_case_labels): Likewise.
17464 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
17465 group_case_labels does anything cleanup the CFG again.
17467 2017-06-29 Bin Cheng <bin.cheng@arm.com>
17469 PR tree-optimization/81196
17470 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
17471 exit condition comparing two IVs.
17473 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
17475 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
17476 profile to the dummy entry at the end of the list of architectures.
17477 * config/arm/arm-cpu-cdata.h: Regenerated.
17479 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17480 Michael Collison <michael.collison@arm.com>
17483 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
17485 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
17486 (*aarch64_reg_<mode>3_minus_mask): New pattern.
17487 (*aarch64_<optab>_reg_di3_mask2): New pattern.
17488 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
17489 of shift when the shift amount is masked with constant equal to
17490 the size of the mode.
17491 * config/aarch64/predicates.md (subreg_lowpart_operator): New
17494 2017-06-29 Martin Liska <mliska@suse.cz>
17496 * config/i386/i386.opt: Change range from [1,5] to [0,5].
17498 2017-06-29 Yury Gribov <tetra2005@gmail.com>
17501 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
17504 (edge_growth_cache_entry::edge_growth_cache_entry): New
17506 (reset_edge_growth_cache): Update to use constructor.
17508 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17510 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
17511 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17512 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17514 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
17516 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
17517 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
17519 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
17521 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
17522 (*-linux-uclibc*): Add t-uclibc tmake_file.
17523 * config/t-musl: New.
17524 * config/t-uclibc: New.
17526 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
17528 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
17530 (gen_comm_data): Emit architectural setting of arch_prof.
17531 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
17533 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
17534 (armv8-m.base, armv8-m.main): Likewise.
17535 * arm-protos.h (arm_build_target): Add profile field.
17536 (arch_option): Likewise.
17537 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
17539 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
17540 arm_active_target.profile.
17542 2017-06-28 Richard Biener <rguenther@suse.de>
17544 PR middle-end/81227
17545 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
17546 TYPE_OVERFLOW_WRAPS.
17547 * match.pd (negate_expr_p): Likewise.
17548 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
17549 fold_build2, not fold_binary.
17551 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17553 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
17554 Convert memory address to Pmode.
17555 (aarch64_print_operand): Assert MEM operands are always Pmode.
17557 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17560 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
17561 Remove redundant if.
17562 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
17563 * config/arm/aarch-common-protos.h
17564 (aarch_forward_to_shift_is_not_shifted_re): Remove.
17565 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
17567 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
17570 * multiple_target.c (create_dispatcher_calls): Set the default
17571 clone to be static, not public.
17573 2017-06-28 Richard Biener <rguenther@suse.de>
17575 * tree-vect-loop.c (vectorizable_reduction): Move special
17576 cond reduction IV var creation ...
17577 (vect_create_epilog_for_reduction): ... here. Remove induction_index
17578 parameter. Use STMT_VINFO_VECTYPE.
17579 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
17582 2017-06-28 Martin Liska <mliska@suse.cz>
17585 * ipa-visibility.c (non_local_p): Handle visibility.
17587 2017-06-28 Martin Liska <mliska@suse.cz>
17590 * common.opt: Add IntegerRange to various options.
17591 * opt-functions.awk (integer_range_info): New function.
17592 * optc-gen.awk: Add integer_range_info to cl_options struct.
17593 * opts-common.c (decode_cmdline_option): Handle
17594 CL_ERR_INT_RANGE_ARG.
17595 (cmdline_handle_error): Likewise.
17596 * opts.c (print_filtered_help): Show valid interval in
17597 when --help is provided.
17598 * opts.h (struct cl_option): Add range_min and range_max fields.
17599 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
17601 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
17603 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
17604 (x * C EQ/NE y * C): New transformation.
17606 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
17608 * genmultilib (combination_space): Accept '+' in option names.
17610 2017-06-28 Martin Liska <mliska@suse.cz>
17613 * asan.c (instrument_derefs): Bail out inner references
17614 that are hard register variables.
17616 2017-06-28 Jakub Jelinek <jakub@redhat.com>
17619 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
17620 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
17622 2017-06-28 Richard Biener <rguenther@suse.de>
17624 * tree-vectorizer.h (vect_get_vec_defs): Remove.
17625 (vect_get_slp_defs): Adjust.
17626 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
17628 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
17630 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
17631 get_initial_defs_for_reduction instead of vect_get_vec_defs.
17632 (vectorizable_reduction): Adjust.
17633 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
17635 (vect_get_slp_defs): Likewise.
17636 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
17637 (vectorizable_bswap): Adjust.
17638 (vectorizable_call): Likewise.
17639 (vectorizable_conversion): Likewise.
17640 (vectorizable_assignment): Likewise.
17641 (vectorizable_shift): Likewise.
17642 (vectorizable_operation): Likewise.
17643 (vectorizable_store): Likewise.
17644 (vectorizable_condition): Likewise.
17645 (vectorizable_comparison): Likewise.
17647 2017-06-28 Michael Collison <michael.collison@arm.com>
17650 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
17652 (arm_gen_movmemqi): Removed unused variable 'i'.
17653 Convert 'for' loop into 'while' loop.
17654 (arm_expand_prologue): Remove last unnecessary set of insn.
17655 (thumb_pop): Remove unused variable 'pushed_words'.
17656 (thumb_exit): Remove last unnecessary set of regs_to_pop.
17658 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17660 * config/s390/predicates.md: Use s390_rel_address_ok_p.
17661 * config/s390/s390-protos.h: Add prototype of
17662 s390_rel_address_ok_p.
17663 * config/s390/s390.c (s390_got_symbol): New function.
17664 (s390_rel_address_ok_p): New function.
17665 (legitimize_pic_address): Use s390_rel_address_ok_p.
17666 (s390_load_got): Use s390_got_symbol.
17667 (s390_option_override): Issue error if
17668 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
17669 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
17671 * config/s390/s390.opt: New option mpic-data-is-text-relative.
17673 2017-06-27 Andrew Pinski <apinski@cavium.com>
17675 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
17676 (X * copysign (1.0, X)): New pattern.
17677 (X * copysign (1.0, -X)): New pattern.
17678 (copysign (-1.0, CST)): New pattern.
17680 2017-06-27 Joseph Myers <joseph@codesourcery.com>
17682 * genmultilib (combination_space): Remove variable.
17683 Validate reuse rules against regular expression for any sequence
17684 of multilib options in any order.
17686 2017-06-27 Michael Collison <michael.collison@arm.com>
17688 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
17689 call aarch64_split_simd_combine.
17690 * (aarch64_combine_internal<mode>): Delete pattern.
17691 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
17692 Allow register and subreg operands.
17694 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17696 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
17697 specific need, just fallback on defaults.
17698 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
17700 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17701 Olivier Hainque <hainque@adacore.com>
17703 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
17705 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
17706 targets. Pick a default if no particular attempt applied.
17707 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
17710 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17712 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
17713 (x86_64-wrs-vxworks7): Likewise.
17715 2017-06-27 Marek Polacek <polacek@redhat.com>
17718 * ubsan.c (instrument_null): Check get_base_address's result for null.
17720 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
17722 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
17724 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
17726 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
17727 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
17728 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
17729 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
17730 New function types.
17731 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
17732 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
17733 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
17734 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
17735 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
17736 BUILT_IN_FEUPDATEENV): New builtins.
17737 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
17738 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
17739 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
17740 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
17742 (builtin_structptr_types): Adjust size.
17743 * tree.c (builtin_structptr_types): Add four entries.
17745 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17746 Olivier Hainque <hainque@adacore.com>
17748 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
17749 (TLS_SYM): New local macro, forcing reference to __tls__ on
17750 link command lines for VxWorks 7 RTPs, triggering initialization
17752 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
17753 OS features TLS support, true for RTPs on VxWorks 7.
17754 * config/vxworks.c (vxworks_override_options): Setup emutls
17757 2017-06-27 Jakub Jelinek <jakub@redhat.com>
17759 * predict.c (test_prediction_value_range): Use -1U instead of -1
17760 to avoid narrowing conversion warning.
17761 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
17762 to avoid narrowing conversion warning.
17763 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
17765 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
17767 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17769 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
17770 64bit configurations.
17771 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
17772 (SIZE_TYPE): Likewise.
17773 * config/vxworks.c (vxworks_emutls_var_fields): Use
17774 long_unsigned_type_node instead of unsigned_type_node as the offset
17775 field type, which is "pointer" mode in emutls.c.
17777 2017-06-27 Jakub Jelinek <jakub@redhat.com>
17780 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
17782 PR middle-end/81207
17783 * gimple-fold.c (replace_call_with_call_and_fold): Handle
17784 gimple_vuse copying separately from gimple_vdef copying.
17786 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17788 * value-prof.c (free_hist): Remove call to memset and the enclosing if
17791 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
17792 Olivier Hainque <hainque@adacore.com>
17794 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
17795 for all vxworks7 targets.
17796 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
17797 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
17798 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
17799 variations for VX6/VX7 and 32/64bits later on in ...
17800 (VXWORKS_LIB_SPEC): Leverage new macros.
17801 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
17802 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
17804 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
17806 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
17807 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
17809 2017-06-26 Carl Love <cel@us.ibm.com>
17811 * config/rs6000/rs6000-c.c: Add support for built-in functions
17812 vector bool char vec_reve (vector bool char);
17813 vector signed char vec_reve (vector signed char);
17814 vector unsigned char vec_reve (vector unsigned char);
17815 vector bool int vec_reve (vector bool int);
17816 vector signed int vec_reve (vector signed int);
17817 vector unsigned int vec_reve (vector unsigned int);
17818 vector bool long long vec_reve (vector bool long long);
17819 vector signed long long vec_reve (vector signed long long);
17820 vector unsigned long long vec_reve (vector unsigned long long);
17821 vector bool short vec_reve (vector bool short);
17822 vector signed short vec_reve (vector signed short);
17823 vector double vec_reve (vector double);
17824 vector float vec_reve (vector float);
17825 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
17826 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
17827 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
17828 (altivec_vreve): New pattern.
17829 * config/rs6000/altivec.h (vec_reve): New define.
17830 * doc/extend.texi (vec_rev): Update the built-in documentation file
17831 for the new built-in functions.
17833 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17835 PR tree-optimization/71815
17836 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
17838 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
17840 (slsr_process_phi): Likewise.
17841 (replace_uncond_cands_and_profitable_phis): Don't replace a
17842 multiply candidate with a stride of 1 (copy or cast).
17843 (phi_incr_cost): Call uses_consumed_by_stmt rather than
17845 (lowest_cost_path): Likewise.
17846 (total_savings): Likewise.
17848 2017-06-26 Richard Biener <rguenther@suse.de>
17851 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
17852 Use def_builtin_pure for all gather builtins.
17854 2017-06-26 Richard Biener <rguenther@suse.de>
17856 PR tree-optimization/81203
17857 * tree-tailcall.c (find_tail_calls): Do not move stmts into
17858 non-dominating BBs.
17860 2017-06-26 Marek Polacek <polacek@redhat.com>
17863 * doc/invoke.texi: Document -Wmultistatement-macros.
17865 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
17867 * doc/sourcebuild.texi (ARM-specific attributes): Document new
17868 arm_neon_ok_no_float_abi effective target.
17870 2017-06-26 Richard Biener <rguenther@suse.de>
17872 PR tree-optimization/80928
17873 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
17874 (copy_bbs): Set BB_DUPLICATED flag early.
17875 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
17877 (execute_on_shrinking_pred): Likewise.
17878 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
17879 BB_DUPLICATED blocks.
17880 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
17881 iterate over all PHIs considering removal of *gsi.
17883 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
17885 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
17888 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17890 * config/rs6000/rs6000-string.c: (expand_block_clear,
17891 do_load_for_compare, select_block_compare_mode,
17892 compute_current_alignment, expand_block_compare,
17893 expand_strncmp_align_check, expand_strn_compare,
17894 expand_block_move, rs6000_output_load_multiple)
17895 Move functions related to string/block move/compare
17896 to a separate file.
17897 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
17898 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
17899 for this function which is now used in two files.
17900 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
17901 * config.gcc: Add rs6000-string.o to extra_objs for
17902 targets powerpc*-*-* and rs6000*-*-*.
17904 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
17907 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
17908 32-bit, since indexed is not valid for DImode.
17909 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
17910 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
17911 (define_peephole2 for Altivec d-form load): Add 32-bit support.
17912 (define_peephole2 for Altivec d-form store): Likewise.
17915 * multiple_target.c (create_dispatcher_calls): Only create the
17916 dispatcher call if the function is the default clone of a
17917 versioned function.
17919 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
17921 PR middle-end/80902
17922 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
17923 a call, force the call to not be a tail call.
17925 2017-06-23 Jeff Law <law@redhat.com>
17927 * doc/contrib.texi: Add entry for Steven Pemberton's work on
17930 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
17932 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
17933 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
17934 handling for early expansion of vector shifts (sl,sr,sra,rl).
17935 (builtin_function_type): Add vector shift right instructions
17936 to the unsigned argument list.
17938 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
17940 rtl-optimizatoin/79286
17941 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
17942 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
17943 trap. PIC register plus a const unspec without offset can never trap.
17945 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
17947 * tree.h (builtin_structptr_type): New type.
17948 (builtin_structptr_types): Declare new array.
17949 * tree.c (builtin_structptr_types): New array.
17950 (free_lang_data, build_common_tree_nodes): Use it.
17952 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
17955 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
17958 2017-06-22 Matt Turner <mattst88@gmail.com>
17960 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
17961 Lake models to skylake case. Assume skylake for unknown
17962 models with clflushopt.
17964 2017-06-22 Jeff Law <law@redhat.com>
17966 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
17967 frame sizes that do not satisfy aarch64_uimm12_shift.
17969 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
17971 * profile-count.h (apply_probability,
17972 apply_scale, probability_in): Fix checks for zero.
17974 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17976 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
17977 * doc/cppdiropts.texi (-I @var{dir}): Document it.
17979 2016-06-22 Richard Biener <rguenther@suse.de>
17981 * tree-vect-loop.c (vect_model_reduction_cost): Handle
17982 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
17983 REDUC_MAX_EXPR support.
17984 (vectorizable_reduction): Likewise.
17985 (vect_create_epilog_for_reduction): Likewise.
17987 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
17989 * match.pd (A / (1 << B) -> A >> B): New.
17990 * generic-match-head.c: Include optabs-tree.h.
17991 * gimple-match-head.c: Likewise.
17992 * optabs-tree.h (target_supports_op_p): New.
17993 * optabs-tree.c (target_supports_op_p): New.
17995 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17997 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
17998 $gcc_cv_ld --help output.
17999 (gcc_cv_ld_demangle): Likewise.
18000 (gcc_cv_ld_eh_frame_hdr): Likewise.
18001 (gcc_cv_ld_pie): Likewise.
18002 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
18003 (gcc_cv_ld_buildid): Likewise.
18004 (gcc_cv_ld_sysroot): Likewise.
18005 (ld_bndplt_support): Likewise.
18006 (ld_pushpopstate_support): Likewise.
18007 * configure: Regenerate.
18008 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
18010 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18013 * config/i386/sse.md (round<mode>2): Renumber match_dup and
18014 operands indexes to avoid gap between operands and match_dups.
18016 2017-06-21 Andrew Pinski <apinski@cavium.com>
18018 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
18019 Increment Arith_shift and Arith_shift_reg by 1.
18020 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
18022 * config/aarch64/aarch64.c (thunderx_tunings): Enable
18023 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
18024 (aarch64_strip_extend): Add new argument and test for it.
18025 (aarch64_cheap_mult_shift_p): New function.
18026 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
18027 add a cost if it is true.
18028 Update calls to aarch64_strip_extend.
18029 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
18031 2017-06-21 Andrew Pinski <apinski@cavium.com>
18033 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
18035 (thunderxt88): Likewise.
18036 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
18037 (thunderx_prefetch_tune): New variable.
18038 (thunderx2t99_prefetch_tune): Update for the correct values.
18039 (thunderxt88_tunings): New variable.
18040 (thunderx_tunings): Use thunderx_prefetch_tune instead of
18041 generic_prefetch_tune.
18042 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
18044 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18046 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
18047 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
18048 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
18049 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
18050 (aarch64_atomic_cas<mode>, GPI): Likewise.
18052 2017-06-21 Martin Liska <mliska@suse.cz>
18054 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
18055 statements on cold and hot labels.
18056 * predict.c (tree_estimate_probability_bb): Remove the
18057 prediction from this place.
18059 2017-06-21 Martin Liska <mliska@suse.cz>
18061 PR tree-optimization/79489
18062 * gimplify.c (maybe_add_early_return_predict_stmt): New
18064 (gimplify_return_expr): Call the function.
18065 * predict.c (tree_estimate_probability_bb): Remove handling
18067 * predict.def: Update comment about early return predictor.
18068 * gimple-predict.h (is_gimple_predict): New function.
18069 * predict.def: Change default value of early return to 66.
18070 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
18072 * passes.def: Put pass_strip_predict_hints to the beginning of
18075 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
18077 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
18078 FUNCTION_DECL declarations.
18079 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
18081 (dwaf2out_decl): Likewise.
18082 * godump.c (go_early_global_decl): Skip call to the real debug hook
18083 for FUNCTION_DECL declarations.
18084 * passes.c (rest_of_decl_compilation): Skip call to the
18085 early_global_decl debug hook for FUNCTION_DECL declarations, unless
18086 -fdump-go-spec is passed.
18088 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18090 * config/i386/i386.c (struct builtin_isa): New field pure_p.
18091 Reorder for compactness.
18092 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
18093 (def_builtin_pure, def_builtin_pure2): New functions.
18094 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
18096 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18098 * match.pd (nop_convert): New predicate.
18099 ((A +- CST1) +- CST2): Allow some NOP conversions.
18101 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18104 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
18105 with ctors/dtors if GOVD_SHARED is set.
18107 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18109 * config/aarch64/aarch64.md (movti_aarch64):
18110 Emit mov rather than orr.
18111 (movtf_aarch64): Likewise.
18112 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
18113 Emit mov rather than orr.
18115 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18117 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
18118 Swap alternatives, make integer dup more expensive.
18120 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18122 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
18123 Return true for non-tls symbols.
18125 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
18127 * config/aarch64/aarch64-cores.def (cortex-a55): New.
18128 (cortex-a75): Likewise.
18129 (cortex-a75.cortex-a55): Likewise.
18130 * config/aarch64/aarch64-tune.md: Regenerate.
18131 * doc/invoke.texi (-mtune): Document new values for -mtune.
18133 2017-06-21 Tom de Vries <tom@codesourcery.com>
18135 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
18136 stack_size feature.
18137 (Effective-Target Keywords, Other attributes): Suggest using
18138 dg-add-options stack_size feature to get stack limit in stack_size
18139 effective target documentation.
18141 2017-06-21 Julian Brown <julian@codesourcery.com>
18142 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18144 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
18145 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
18146 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
18148 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
18149 attribute type list for neon_multiply.
18150 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
18151 attribute type list for neon_multiply.
18152 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
18153 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
18154 attribute type list for neon_multiply.
18155 * config/arm/types.md (crypto_pmull): Add.
18156 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
18157 attribute type list.
18159 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
18161 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
18164 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18166 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
18167 to make sure not to dereference a NULL cost_classes_ptr pointer.
18169 2017-06-20 Carl Love <cel@us.ibm.com>
18171 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18172 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18173 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18174 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18175 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18176 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18177 VMULOSW): New enum "unspec" values.
18178 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18179 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18180 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18181 altivec_vmulosw): New patterns.
18182 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18183 VMULOSW): Add definitions.
18185 2017-06-20 Julia Koval <julia.koval@intel.com>
18187 * config/i386/i386.c: Fix rounding expand for new pattern.
18188 * config/i386/subst.md: Fix pattern (parallel -> unspec).
18190 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18192 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
18193 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
18195 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18197 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
18200 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18202 * config/aarch64/aarch64-cores.def: Rearrange to sort by
18203 architecture, then by implementer ID.
18204 * config/aarch64/aarch64-tune.md: Regenerate.
18206 2017-06-20 Richard Biener <rguenther@suse.de>
18208 PR middle-end/81097
18209 * fold-const.c (split_tree): Fold to type before negating.
18211 2017-06-20 David Malcolm <dmalcolm@redhat.com>
18213 * diagnostic-show-locus.c
18214 (selftest::test_fixit_deletion_affecting_newline): New function.
18215 (selftest::diagnostic_show_locus_c_tests): Call it.
18217 2017-06-20 Andreas Schwab <schwab@suse.de>
18220 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
18223 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18225 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
18226 __ARM_FEATURE_COPROC according to support.
18228 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18230 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18231 Rewritten to avoid overflow for > 32-bit pointers.
18234 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
18235 by removing enum keyword.
18236 (ubsan_type_descriptor): Likewise. Formatting fix.
18239 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
18240 splitter): Require TARGET_SSE2 in the condition.
18242 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
18245 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
18246 for doing vector set of SFmode on ISA 3.0.
18247 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
18248 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
18250 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
18251 SFmode value into a V4SF variable that was extracted from another
18252 V4SF variable without converting the element to double precision
18253 and back to single precision vector format.
18254 (vsx_insert_extract_v4sf_p9_2): Likewise.
18256 2017-06-19 Jakub Jelinek <jakub@redhat.com>
18258 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
18259 in UWHI to avoid undefined overflow.
18262 * ubsan.h (enum ubsan_encode_value_phase): New.
18263 (ubsan_encode_value): Change second argument to
18264 enum ubsan_encode_value_phase with default value of
18265 UBSAN_ENCODE_VALUE_GENERIC.
18266 * ubsan.c (ubsan_encode_value): Change second argument to
18267 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
18268 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
18269 create_tmp_var_raw instead of create_tmp_var and use a
18271 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
18272 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
18273 ubsan_encode_value callers.
18276 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
18277 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
18278 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
18280 2017-06-19 Richard Biener <rguenther@suse.de>
18282 PR middle-end/81118
18283 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
18284 estimates if we changed anything.
18286 2017-06-19 Richard Biener <rguenther@suse.de>
18288 PR tree-optimization/80887
18289 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
18290 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
18291 simplified lookups, then reset mprts_hook.
18292 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
18294 (try_to_simplify): Likewise.
18296 2017-06-19 Martin Liska <mliska@suse.cz>
18299 * gimplify.c (gimplify_switch_expr):
18300 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
18302 2017-06-19 Martin Liska <mliska@suse.cz>
18304 * doc/install.texi: Document that PGO runs in 4 stages.
18306 2017-06-19 Martin Liska <mliska@suse.cz>
18309 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
18310 to dispatcher function name.
18311 * multiple_target.c (replace_function_decl): New function.
18312 (create_dispatcher_calls): Redirect both edges and references.
18314 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
18316 * profile-count.c (profile_count::dump): Dump quality.
18317 (profile_count::differs_from_p): Update for unsigned val.
18318 * profile-count.h (profile_count_quality): New enum.
18319 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
18321 2017-06-19 Richard Biener <rguenther@suse.de>
18323 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
18324 struct function as arg.
18325 (estimate_numbers_of_iterations): Export overload with loop arg.
18326 (free_numbers_of_iterations_estimates_loop): Use an overload of
18327 free_numbers_of_iterations_estimates instead.
18328 * tree-cfg.c (remove_bb): Adjust.
18329 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
18330 * tree-parloops.c (gen_parallel_loop): Likewise.
18331 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
18333 (tree_unroll_loops_completely): Likewise.
18334 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
18335 Use an overload instead and export.
18336 (estimated_loop_iterations): Adjust.
18337 (max_loop_iterations): Likewise.
18338 (likely_max_loop_iterations): Likewise.
18339 (estimate_numbers_of_iterations): Take struct function as arg
18341 (loop_exits_before_overflow): Adjust.
18342 (free_numbers_of_iterations_estimates_loop): Use an overload.
18343 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
18344 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
18346 2017-06-19 Richard Biener <rguenther@suse.de>
18349 * ipa-prop.c (find_constructor_constant_at_offset): Handle
18350 RANGE_EXPR conservatively.
18352 2017-06-16 Carl Love <cel@us.ibm.com>
18354 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18355 definitions for vec_float, vec_float2, vec_floato,
18356 vec_floate built-ins.
18357 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
18358 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
18360 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
18361 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
18362 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
18363 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
18364 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
18365 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
18366 vec_floato): Add builtin defines.
18367 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
18368 Update the built-in documentation file for the new built-in
18371 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18373 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
18374 (mthumb): Mark as the negative of -marm.
18376 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18378 * doc/invoke.texi (ARM Options, -mcpu): Document supported
18380 (ARM Options, -mtune): Document that this accepts the same
18381 extension options as -mcpu.
18382 (ARM Options, -mfpu): Document addition of -mfpu=auto.
18384 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18386 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
18387 permitted extensions.
18389 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18391 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
18392 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
18393 (armv8-m.main): Add option +nodsp.
18394 * config/arm/arm-cpu-cdata.h: Regenerated.
18396 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18398 * config/arm/t-fuchsia: New file.
18399 * config.gcc (arm*-*-fuchsia*): Use it.
18401 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18403 * config/arm/t-symbian: Rewrite for new option infrastructure.
18405 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18407 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
18408 (MULTILIB_REQUIRED): Likewise.
18410 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18412 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
18413 (MULTILIB_RESUE): Likewise.
18414 (MULTILIB_MATCHES): Likewise.
18415 (MULTLIB_REQUIRED): Likewise.
18417 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18419 * config/arm/t-rtems: Rewrite for new option framework.
18421 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18423 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
18424 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
18425 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
18426 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
18427 * config/arm/t-multilib: ... here.
18428 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
18429 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
18430 armv7-a and armv8*-a when A-profile libraries have not been built.
18431 * config/arm/t-rmprofile: Rewrite.
18433 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18435 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
18436 with a backslash. Remove the backslash after substituting unescaped
18438 * doc/fragments.texi (MULTILIB_REUSE): Document it.
18440 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18442 * config.gcc: (arm*-*-*): When building a-profile libraries, force
18443 the driver to pass through the default setting of -mfloat-abi.
18444 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
18446 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
18447 (all_feat_combs): New rule.
18448 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
18450 * config/arm/t-aprofile: Rewrite.
18452 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18454 * config/arm/arm.h (FPUTYPE_AUTO): Define.
18455 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
18456 fpu is not specified by the user/command-line.
18457 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
18458 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
18459 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
18460 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
18461 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
18462 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
18464 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18466 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
18467 * config/arm/t-arm-elf: Rewritten.
18469 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18471 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
18472 have some floating-point instructions.
18473 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
18474 (TARGET_MAYBE_HARD_FLOAT): New macro.
18475 * config/arm/arm-builtins.c (arm_init_builtins): Use
18476 TARGET_MAYBE_HARD_FLOAT.
18477 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
18479 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18481 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
18482 (configargs.h): Include it.
18483 (arm_print_hint_for_fpu_option): New function.
18484 (arm_parse_fpu_option): New function.
18485 (candidate_extension): New class.
18486 (arm_canon_for_multilib): New function.
18487 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
18488 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
18489 (ARCH_CANONICAL_SPECS): New macro.
18490 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
18492 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18494 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
18495 are set after handling multilib fragments. Set target_cpu_default2
18498 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18500 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
18502 (arm*-*-*): Set target_cpu_default2 to a quoted string.
18503 * config/arm/parsecpu.awk (check_cpu): Validate any extension
18505 (check_arch): Likewise.
18506 * config/arm/arm.c (arm_configure_build_target): Handle
18507 TARGET_CPU_DEFAULT being a string constant. Scan any feature
18508 options in the default.
18510 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18512 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
18513 when an option is an alias of another.
18514 * config/arm/parsecpu.awk (optalias): New parser token.
18515 (gen_comm_data): Mark non-alias options as such. Emit entries
18516 for extension aliases.
18517 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
18518 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
18519 (armv6kz, armv6zk, armv6t2): Likewise.
18520 (armv7): Make vfpv3-d16 an alias.
18521 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
18523 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
18524 Sort in canonical order.
18525 (armv8-a): Sort in canonical order.
18526 (armv8.1-a, armv8.2-a): Likewise.
18527 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
18529 (cortex-a9): Sort in canonical order.
18530 * config/arm/arm.c (selftests.h): Include it.
18531 (arm_test_cpu_arch_data): New function.
18532 (arm_run_self_tests): New function.
18533 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
18534 (targetm): Move declaration to the end of the file.
18535 * arm-cpu-cdata.h: Regenerated.
18537 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18539 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
18540 call to target_mode_check describing the type of option passed.
18541 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
18542 (arm_target_thumb_only): Use arm_parse_arch_option_name or
18543 arm_parse_cpu_option_name to match parameters against list of
18545 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
18546 arm_arch_core_flags data structure.
18547 * config/arm/arm-cpu_cdata.h: Regenerated.
18549 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18551 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
18553 (arm_print_hint_for_cpu_option): Likewise.
18554 (arm_print_hint_for_arch_option): Likewise.
18555 (arm_parse_cpu_option_name): Likewise.
18556 (arm_parse_arch_option_name): Likewise.
18557 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
18558 of entries in the all_fpus list.
18559 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
18560 (arm_parse_cpu_option_name): Declare.
18561 (arm_parse_arch_option_name): Declare.
18562 (arm_parse_option_features): Declare.
18563 (arm_intialize_isa): Declare.
18564 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
18566 (gen_comm_data): ... here. Make definitions non-static.
18567 * config/arm/arm-cpu-data.h: Regenerated.
18568 * config/arm/arm-cpu-cdata.h: Regenerated.
18570 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18572 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
18573 (cpu_arch_extension): New structure.
18574 (cpu_arch_option, arch_option, cpu_option): New structures.
18575 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
18576 architecture types.
18577 (gen_data): Generate new format data tables.
18578 * config/arm/arm.c (cpu_tune): New structure.
18579 (cpu_option, processors): Delete.
18580 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
18581 (arm_print_hint_for_cpu_option): ... this and ...
18582 (arm_print_hint_for_arch_option): ... this.
18583 (arm_parse_arch_cpu_name): Delete. Replace with ...
18584 (arm_parse_cpu_option_name): ... this and ...
18585 (arm_parse_arch_option_name): ... this.
18586 (arm_unrecognized_feature): Change type of target parameter to
18588 (arm_parse_arch_cpu_features): Delete. Replace with ...
18589 (arm_parse_option_features): ... this.
18590 (arm_configure_build_target): Rework to use new configuration data
18592 (arm_print_tune_info): Rework for new configuration data tables.
18593 * config/arm/arm-cpu-data.h: Regenerated.
18594 * config/arm/arm-cpu.h: Regenerated.
18596 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18598 * Makefile.in (OBJS): Move sbitmap.o from here ...
18599 (OBJS-libcommon): ... to here.
18601 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18603 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
18604 (ISA_ALL_CRYPTO): New macro.
18605 (ISA_ALL_SIMD): New macro
18606 (ISA_ALL_FP): New macro.
18607 * config/arm/arm.c (fpu_bitlist): Update initializer.
18608 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
18610 (arm9e): Add fpu. Add option for nofp
18611 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
18612 (arm926ej-s, arm1026ej-s): Likewise.
18613 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
18614 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
18615 neon-fp16, neon-vfpv4, nofp and nosimd.
18616 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
18617 (cortex-a8): Add fpu. Add option for nofp.
18618 (cortex-a9): Add fpu. Add options for nosimd and nofp.
18619 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
18620 (cortex-r4f): Add fpu.
18621 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
18622 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
18624 (cortex-r8): Likewise.
18625 (cortex-m4): Add fpu. Add option for nofp.
18626 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
18627 (cortex-a17.cortex-a7): Likewise.
18628 (cortex-a32): Add fpu. Add options for crypto and nofp.
18629 (cortex-a35, cortex-a53): Likewise.
18630 (cortex-a57): Add fpu. Add option for crypto.
18631 (cortex-a72, cortex-a73): Likewise.
18632 (exynos-m1): Likewise.
18633 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
18634 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
18635 (cortex-m33): Add fpu. Add option for nofp.
18636 * config/arm/arm-cpu-cdata.h: Regenerated
18637 * config/arm/arm-cpu-data.h: Regenerated.
18639 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18641 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
18642 (armv5te, armv5tej): Likewise.
18643 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
18644 (armv7): Add options fp and vfpv3-d16.
18645 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
18646 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
18648 (armv7ve): Likewise.
18649 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
18650 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
18651 (armv8-a): Add nocrypto option.
18652 (armv8.1-a, armv8.2-a): Likewise.
18653 (armv8-m.main): add options fp, fp.dp and nofp.
18655 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18657 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
18659 (armv8-a+crc): Delete.
18660 (armv8.1-a): Add options simd, crypto and nofp.
18661 (armv8.2-a): Add options fp16, simd, crypto and nofp.
18662 (armv8.2-a+fp16): Delete.
18663 (armv8-m.main): Add option dsp.
18664 (armv8-m.main+dsp): Delete.
18665 (cortex-a8): Add fpu. Add nofp option.
18666 (cortex-a9): Add fpu. Add nofp and nosimd options.
18667 * config/arm/parsecpu.awk (gen_data): Generate option tables and
18668 link to main cpu and architecture data structures.
18669 (gen_comm_data): Only put isa attributes from the main architecture
18671 (option): New statement for architecture and CPU entries.
18672 * arm.c (struct cpu_option): New structure.
18673 (struct processors): Add entry for options.
18674 (arm_unrecognized_feature): New function.
18675 (arm_parse_arch_cpu_name): Ignore any characters after the first
18677 (arm_parse_arch_cpu_feature): New function.
18678 (arm_configure_build_target): Separate out any CPU and architecture
18679 features and parse separately. Don't error out if -mfpu=auto is
18680 used with only an architecture string.
18681 (arm_print_asm_arch_directives): New function.
18682 (arm_file_start): Call it.
18683 * config/arm/arm-cpu-cdata.h: Regenerated.
18684 * config/arm/arm-cpu-data.h: Likewise.
18685 * config/arm/arm-tables.opt: Likewise.
18687 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18689 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
18690 assembler when it is not -mfpu=auto.
18692 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18694 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
18695 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
18696 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
18697 (ASM_CPU_SPEC): Rewrite.
18698 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
18699 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
18700 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
18701 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
18702 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
18703 copied string is NUL-terminated. Also strip any characters prefixed
18705 (arm_rewrite_selected_arch): New function.
18706 (arm_rewrite_march): New function.
18708 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18710 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
18711 (x_arm_cpu_string, x_arm_tune_string): Likewise.
18712 (march, mcpu, mtune): Convert to string-based options.
18713 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
18714 (arm_parse_arch_cpu_name): New function.
18715 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
18716 identify selected architecture or CPU.
18717 (arm_option_save): New function.
18718 (TARGET_OPTION_SAVE): Redefine.
18719 (arm_option_restore): Restore string options.
18720 (arm_option_print): Print string options.
18722 2017-06-16 Martin Sebor <msebor@redhat.com>
18724 PR tree-optimization/80933
18725 PR tree-optimization/80934
18726 * builtins.c (fold_builtin_3): Do not handle bcmp here.
18727 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
18728 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
18729 (gimple_fold_builtin): Call them.
18731 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18733 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
18734 as unlikely; update profile.
18736 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18738 * predict.c (force_edge_cold): Handle declaring edges impossible
18741 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18743 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
18745 (try_unroll_loop_completely): Fix reporting.
18747 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18749 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
18751 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
18754 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
18755 if given a non-constant argument for an intrinsic which requires a
18758 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18760 * profile.c (compare_freqs): New function.
18761 (branch_prob): Sort edge list.
18762 (find_spanning_tree): Assume that the list is priority sorted.
18764 2017-06-16 Richard Biener <rguenther@suse.de>
18766 PR tree-optimization/81090
18767 * passes.def (pass_record_bounds): Remove.
18768 * tree-pass.h (make_pass_record_bounds): Likewise.
18769 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
18770 make_pass_record_bounds): Likewise.
18771 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
18772 not free niter estimates at the beginning but at the end.
18773 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
18775 2017-06-16 Richard Biener <rguenther@suse.de>
18777 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
18778 initializer to workaround ICE in host GCC 4.8.
18780 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18782 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
18784 (clone_inlined_nodes): Update.
18786 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18788 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
18789 prefetch settings, and enable prefetching by default at -O3.
18791 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18793 * config/aarch64/aarch64.c (aarch64_override_options_internal):
18794 Set flag_prefetch_loop_arrays according to tuning data.
18796 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18798 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
18799 New tune structure.
18800 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
18801 [Unrelated to main purpose of the patch] Place the pointer field last
18802 to enable type checking errors when tune structure are wrongly merged.
18803 * config/aarch64/aarch64.c (generic_prefetch_tune,)
18804 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
18805 (thunderx2t99_prefetch_tune): New tune constants.
18806 (tune_params *_tunings): Update all tunings (no functional change).
18807 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
18808 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
18809 from tunings structures.
18811 2017-06-16 Jakub Jelinek <jakub@redhat.com>
18814 * ubsan.c (instrument_null): Add T argument, use it instead
18815 of computing it based on IS_LHS.
18816 (instrument_object_size): Likewise.
18817 (pass_ubsan::execute): Adjust instrument_null and
18818 instrument_object_size callers to pass gimple_get_lhs or
18819 gimple_assign_rhs1 result to it. Use instrument_null instead of
18820 calling get_base_address and instrument_mem_ref. Handle
18821 aggregate call arguments for object-size sanitization.
18823 2017-06-16 Yury Gribov <tetra2005@gmail.com>
18825 PR tree-optimization/81089
18826 * tree-vrp.c (is_masked_range_test): Validate operands of
18829 2017-06-15 Martin Sebor <msebor@redhat.com>
18832 * dumpfile.c (dump_register): Avoid calling memset to initialize
18833 a class with a default ctor.
18834 * gcc.c (struct compiler): Remove const qualification.
18835 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
18836 * hash-table.h: Ditto.
18837 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
18839 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
18840 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
18842 * params.h (struct param_info): Make struct members non-const.
18843 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
18844 with default initialization.
18845 * vec.h (vec_copy_construct, vec_default_construct): New helper
18847 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
18848 with vec_copy_construct.
18849 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
18850 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
18851 * doc/invoke.texi (-Wclass-memaccess): Document.
18853 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18855 * emit-rtl.h (is_leaf): Update comment about local
18856 register allocator.
18858 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18861 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
18862 for a variable to have a section before checking if the section has a
18864 Set section to.persistent if persistent attribute is set.
18865 Warn if .persistent attribute is used on an automatic variable.
18867 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
18869 PR rtl-optimization/80474
18870 * reorg.c (update_block): Do not ignore instructions in a delay slot.
18872 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
18874 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
18877 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
18879 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
18880 (casesi): Emit bounds checking as RTL.
18881 (casesi_internal_mips16_<mode>): Remove bounds checking.
18883 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
18885 * config/xtensa/xtensa.c (xtensa_option_override): Append
18886 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
18887 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
18888 xtensa_doloop_hooks): Define unconditionally.
18889 (xtensa_reorg_loops): Only call reorg_loops in the presence of
18891 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
18892 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
18893 for it in xtensa_option_override.
18894 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
18895 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
18897 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
18899 * doc/cppopts.texi: Document '-' special value to -MF.
18901 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
18903 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
18904 (cortex_a53_fconst): Likewise.
18905 (cortex_a53_fpmul): Likewise.
18906 (cortex_a53_f_load_64): Likewise.
18907 (cortex_a53_f_load_many): Likewise.
18908 (cortex_a53_advsimd_alu): Likewise.
18909 (cortex_a53_advsimd_alu_q): Likewise.
18910 (cortex_a53_advsimd_mul): Likewise.
18911 (cortex_a53_advsimd_mul_q): Likewise.
18912 (fpmac bypass): Add new bypass for fpmac-fpmac case.
18913 Add missing fmul, r2f_cvt and fconst cases.
18915 2017-06-14 Richard Biener <rguenther@suse.de>
18917 PR middle-end/81088
18918 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
18920 (fold_binary_loc): When associating do not treat pre-existing
18921 TREE_OVERFLOW on literal constants as a reason to allow
18922 TREE_OVERFLOW on associated literal constants.
18924 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18926 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
18927 (MASK_FEATURES): New macro.
18928 * config/sparc/sparc.c (sparc_option_override): Remove the special
18929 handling of -mfpu and generalize it to all MASK_FEATURES switches.
18931 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18933 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
18934 a division of 0 if non-call exceptions are enabled.
18936 2017-06-14 Andrew Pinski <apinski@cavium.com>
18937 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18940 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
18941 Improve vector initialization code gen for only variable case.
18943 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18945 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
18947 2017-06-14 Richard Biener <rguenther@suse.de>
18949 PR tree-optimization/81083
18950 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
18953 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18955 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
18956 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
18957 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
18958 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
18959 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
18960 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
18962 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18964 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
18965 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
18967 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18969 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
18971 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18973 * config/rs6000/t-rtems: Don't handle SPE.
18975 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18977 * config/rs6000/t-linux: Don't handle SPE.
18979 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18981 * config/rs6000/eabispe.h: Delete file.
18983 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18985 * config/rs6000/t-spe: Delete file.
18987 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18989 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
18990 (rs6000_legitimate_offset_address_p): Return false for anything in
18991 V2SImode or V2SFmode.
18993 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18995 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
18996 except V2SF and V2SI. Rearrange the vector modes, and add comments.
18997 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
18999 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
19000 (rs6000_legitimate_offset_address_p): Ditto.
19001 (rs6000_emit_move): Ditto.
19002 (rs6000_init_builtins): Remove V4HI_type_node.
19004 2017-06-13 Martin Liska <mliska@suse.cz>
19007 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
19008 (gate_asan): Likewise.
19009 * asan.h (asan_no_sanitize_address_p): Remove the function.
19010 (sanitize_flags_p): New function.
19011 * builtins.def: Fix coding style.
19012 * common.opt: Use renamed enum value.
19013 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
19014 * doc/extend.texi: Document no_sanitize attribute.
19015 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
19016 to SANITIZE_UNDEFINED_NONDEFAULT.
19017 * gcc.c (sanitize_spec_function): Use the renamed enum value.
19018 * gimple-fold.c (optimize_atomic_compare_exchange_p):
19019 Use sanitize_flags_p.
19020 * gimplify.c (gimplify_function_tree): Likewise.
19021 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
19022 * opts.c (parse_no_sanitize_attribute): New function.
19023 (common_handle_option): Use renamed enum value.
19024 * opts.h (parse_no_sanitize_attribute): Declare.
19025 * tree.c (sanitize_flags_p): New function.
19026 * tree.h: Declared here.
19027 * tsan.c: Use sanitize_flags_p.
19028 * ubsan.c (ubsan_expand_null_ifn): Likewise.
19029 (instrument_mem_ref): Likewise.
19030 (instrument_bool_enum_load): Likewise.
19031 (do_ubsan_in_current_function): Remove the function.
19032 (pass_ubsan::execute): Use sanitize_flags_p.
19033 * ubsan.h: Remove do_ubsan_in_current_function
19034 * tree-cfg.c (print_no_sanitize_attr_value): New function.
19035 (dump_function_to_file): Use it here.
19037 2017-06-13 Martin Jambor <mjambor@suse.cz>
19039 PR tree-optimization/80803
19040 PR tree-optimization/81063
19041 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
19042 (propagate_subaccesses_across_link): Enqueue subtree whenever
19043 necessary instead of relying on the caller.
19045 2017-06-13 Martin Jambor <mjambor@suse.cz>
19047 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
19048 that have a first_link.
19049 (sort_and_splice_var_accesses): Do not check first_link before
19051 (subtree_mark_written_and_enqueue): Likewise.
19052 (propagate_all_subaccesses): Likewise and do not stop at first
19053 parent with a first_link.
19055 2017-06-13 Martin Jambor <mjambor@suse.cz>
19057 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
19060 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19062 * match.pd: New pattern.
19064 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19066 * tree-vrp.c (is_masked_range_test): New function.
19067 (register_edge_assert_for): Determine ranges for
19070 2017-06-13 Yury Gribov <tetra2005@gmail.com>
19072 PR tree-optimization/67328
19073 * fold-const.c (maskable_range_p): New function.
19074 (build_range_check): Generate bittests if possible.
19076 2017-06-13 Martin Liska <mliska@suse.cz>
19078 * gimple-pretty-print.c (dump_probability): Add new argument.
19079 (dump_edge_probability): Dump both probability and count.
19080 (dump_gimple_label): Likewise.
19081 (dump_gimple_bb_header): Likewise.
19083 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
19086 * config/avr/avr-devices.c: Fix indentation.
19087 * config/avr/gen-avr-mmcu-specs.c: Dito.
19089 2017-06-13 Richard Biener <rguenther@suse.de>
19091 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
19092 instead get vector type from stmt_info.
19093 (vectorizable_reduction): Adjust. Remove dead code.
19095 2017-06-13 Richard Biener <rguenther@suse.de>
19097 PR middle-end/81065
19098 * fold-const.c (extract_muldiv_1): Remove bogus distribution
19099 case of C * (x * C2 + C3).
19100 (fold_addr_of_array_ref_difference): Properly fold index difference.
19102 2017-06-12 David S. Miller <davem@davemloft.net>
19105 * config/sparc/sparc.md (return expander): Emit frame blockage if
19106 function uses alloca.
19108 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
19110 * combine.c (make_field_assignment): Check len rather than the mode
19111 precision when calling force_to_mode.
19113 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
19115 Support multilibs and devices that see flash in RAM address range.
19118 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
19119 (avr_mcu_t) <flash_pm_offset>: New field.
19120 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
19121 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
19122 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
19123 (AVR_TINY_PM_OFFSET): Remove macro.
19124 * config/avr/avr.opt (-mshort-calls): New option.
19125 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19126 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
19127 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
19128 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
19129 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
19130 instead of avr_arch->have_jmp_call.
19131 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
19132 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
19133 avr_arch->flash_pm_offset to define.
19134 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
19135 new field flash_pm_offset. Add entry for avrxmega3.
19136 (avr_texinfo): Add entry for avrxmega3.
19137 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
19138 attiny212, attiny214,
19139 attiny412, attiny414, attiny416, attiny417,
19140 attiny814, attiny816, attiny817,
19141 attiny1614, attiny1616, attiny1617,
19142 attiny3214, attiny3216, attiny3217.
19143 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
19144 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
19145 (avr_print_operand_address) [AVR_TINY]: Same.
19146 (avr_asm_init_sections) <readonly_data_section>: Only patch
19147 callback if avr_arch->flash_pm_offset = 0.
19148 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
19149 for rodata if avr_arch->flash_pm_offset != 0.
19150 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
19151 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
19152 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
19153 (m_options): Append opt_rcall.
19154 (m_dirnames): Append dir_rcall.
19155 * config/avr/t-multilib: Regenerate.
19157 * configure.ac [target=avr]: Check whether avrxmega3 default
19158 linker description file works as needed.
19159 * configure: Regenerate.
19160 * doc/avr-mmcu.texi: Regenerate.
19161 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
19162 <__AVR_ARCH__>: Document avrxmega3 and 103.
19163 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
19164 <__AVR_SHORT_CALLS__>: Document it.
19165 <__AVR_PM_BASE_ADDRESS__>: Document it.
19166 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
19167 (AVR Variable Attributes) <progmem>: Document this is
19168 not needed for avrxmega3.
19169 (AVR Named Address Spaces) <__flash>: Dito.
19171 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
19173 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
19175 2017-06-12 Doug Rupp <rupp@adacore.com>
19177 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
19178 Append vxworks-stdint.h to the tm_file list.
19179 * config/vxworks-stdint.h: New file.
19181 2017-06-12 Martin Liska <mliska@suse.cz>
19183 PR tree-optimization/81041
19184 * tree-profile.c (gimple_gen_ic_func_profiler):
19185 Create an extra BB in profile-generate
19186 (gimple_gen_time_profiler): Likewise.
19188 2017-06-12 Jakub Jelinek <jakub@redhat.com>
19190 PR tree-optimization/81003
19191 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
19192 (update_range_test): Use it instead of force_gimple_operand_gsi.
19194 2017-06-12 Richard Biener <rguenther@suse.de>
19196 PR tree-optimization/81053
19197 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
19198 with backedge value not defined in loop. Simplify def stmt
19201 2017-06-11 Tom de Vries <tom@codesourcery.com>
19204 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
19206 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
19207 nvptx_cannot_force_const_mem.
19209 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19211 * opts.c (finish_options): Move test for flag_split_stack after
19212 it has been initialized.
19214 2017-06-11 Jason Merrill <jason@redhat.com>
19216 * tree.h (id_equal): New.
19217 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
19218 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
19219 instead of strcmp of IDENTIFIER_POINTER.
19221 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19223 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
19224 (mark_all_inlined_calls_cdtor): Fix formating.
19225 (inline_transform): Rescale profile before inlining.
19227 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19229 * cgraph.h (cgraph_edge::clone): Update prototype.
19230 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
19231 (cgraph_node::create_clone): Update.
19232 (cgraph_node::create_version_clone): Update.
19233 * tree-inline.c (copy_bb): Update.
19234 (expand_call_inline): Update.
19236 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19238 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
19239 factored out from ...
19240 (rs6000_emit_prologue): ... here.
19242 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19244 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
19245 factored out from ...
19246 (rs6000_emit_prologue): ... here.
19248 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19250 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
19252 (handle_missing_profiles): Fix computation of tp_first_run.
19253 (counts_to_freqs): Do not touch freqs when count is 0.
19255 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19257 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
19260 2017-06-10 Tom de Vries <tom@codesourcery.com>
19262 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19263 attributes): Document signal effective target.
19265 2017-06-10 Tom de Vries <tom@codesourcery.com>
19267 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19268 Document effective target stack_size.
19270 2017-06-09 David Malcolm <dmalcolm@redhat.com>
19272 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
19273 to the edit_context if they can be auto-applied.
19275 2017-06-9 Ian Lance Taylor <iant@golang.org>
19277 * opts.c (finish_options): If -fsplit-stack, disable implicit
19278 -forder-blocks-and-partition.
19279 * doc/invoke.texi (Optimize Options): Document that when using
19280 -fsplit-stack -forder-blocks-and-partition is not implicitly
19283 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19285 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
19286 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
19287 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
19288 * builtins.def (abort, trap, unreachable): Declare cold.
19289 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
19290 * tree-core.h (ECF_COLD): New.
19291 * tree.c (set_call_expr_flags): Handle ECF_COLD.
19292 (build_common_builtin_nodes): Mark unreachable and abort as cold.
19294 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19296 * predict.c (unlikely_executed_stmt_p): Cleanup.
19298 2017-06-09 Richard Biener <rguenther@suse.de>
19300 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
19301 model if the ref is always written to.
19303 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19305 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
19307 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19309 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
19312 2017-06-09 Tom de Vries <tom@codesourcery.com>
19315 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
19316 "target cannot support label values" when encountering LABEL_REF.
19318 2017-06-09 Martin Liska <mliska@suse.cz>
19320 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
19321 (gimple_gen_ic_func_profiler): Emit direct comparison
19322 of __gcov_indirect_call_callee with NULL.
19323 (gimple_gen_time_profiler): Change probability from
19324 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
19326 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19328 * profile.c (edge_gcov_counts): Turn to pointer.
19329 (compute_branch_probabilities, compute_branch_probabilities): Update.
19330 (branch_prob): Do not clear edge_gcov_count.
19331 * profile.h (edge_gcov_counts): Turn to pointer.
19332 (edge_gcov_count): Update.
19334 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19336 * gimple.h (gimple_check_failed): Mark cold.
19338 2017-06-09 Richard Biener <rguenther@suse.de>
19340 PR tree-optimization/66623
19341 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
19342 refactor check_reduction into two parts, properly computing
19343 whether we have to check reduction validity for outer loop
19346 2017-06-09 Richard Biener <rguenther@suse.de>
19348 PR tree-optimization/79483
19349 * graphite-scop-detection.c (order): New global.
19350 (get_order): Compute bb to order mapping that satisfies code
19351 generation constraints.
19352 (cmp_pbbs): New helper.
19353 (build_scops): Start domwalk at entry block, sort generated
19356 2017-06-09 Richard Biener <rguenther@suse.de>
19358 PR middle-end/81007
19359 * ipa-polymorphic-call.c
19360 (ipa_polymorphic_call_context::restrict_to_inner_class):
19361 Skip FIELD_DECLs with error_mark_node type.
19362 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
19365 2017-06-09 Martin Liska <mliska@suse.cz>
19367 * predict.c (struct branch_predictor): New struct.
19368 (test_prediction_value_range): New test.
19369 (predict_c_tests): New function.
19370 * selftest-run-tests.c (selftest::run_tests): Run the function.
19371 * selftest.h: Declare new tests.
19373 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
19376 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
19377 gen_add3_insn did not fail.
19378 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
19379 r0, construct that number in a temporary reg and add that reg to r0.
19380 If asked to put the result in r0 as well, fail.
19382 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
19384 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19385 for early expansion of vec_eqv.
19387 2017-06-08 Jakub Jelinek <jakub@redhat.com>
19389 PR middle-end/81005
19390 * ubsan.c (instrument_null): Avoid pointless code temporary.
19391 (pass_ubsan::execute): Instrument aggregate arguments of calls.
19393 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
19397 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
19400 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
19403 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19405 * predict.c (unlikely_executed_edge_p): Move ahead.
19406 (probably_never_executed_edge_p): Use it.
19408 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19410 PR middle-end/79988
19411 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
19412 gimple_call_builtin_p call.
19414 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19416 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
19417 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
19418 rtl_check_failed_type2, rtl_check_failed_code1,
19419 rtl_check_failed_code2, rtl_check_failed_code_mode,
19420 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
19421 rtvec_check_failed_bounds, rtl_check_failed_flag,
19422 _fatal_insn_not_found, _fatal_insn): Likewise.
19423 * tree.h (tree_contains_struct_check_failed,
19424 tree_check_failed, tree_not_check_failed,
19425 tree_class_check_failed, tree_range_check_failed,
19426 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
19427 tree_vec_elt_check_failed, phi_node_elt_check_failed,
19428 tree_operand_check_failed, omp_clause_check_failed,
19429 omp_clause_operand_check_failed, omp_clause_range_check_failed):
19432 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19434 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
19435 flag_branch_probabilities.
19436 * ipa-inline.c (edge_badness): Likewise.
19437 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
19438 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
19439 * predict.c (maybe_hot_frequency_p): Likewise.
19440 (probably_never_executed): Likewise.
19441 * sched-ebb.c (schedule_ebbs): Likewise.
19442 * sched-rgn.c (find_single_block_region): Likewise.
19443 * tracer.c (tail_duplicate): Likewise.
19445 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19447 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
19448 longer requires x_flag_profile_use.
19450 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19452 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
19453 instead of flag_reorder_blocks_and_partition.
19454 * dbxout.c (dbxout_function_end): Likewise.
19455 * dwarf2out.c (gen_subprogram_die): Likewise.
19456 * haifa-sched.c (sched_create_recovery_edges): Likewise.
19457 * hw-doloop.c (reorg_loops): Likewise.
19458 * varasm.c (assemble_start_function,
19459 assemble_end_function): Likewise.
19460 (decide_function_section): Do not check for
19461 flag_reorder_blocks_and_partition.
19463 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19465 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19467 (chkp_get_hard_register_fake_addr_expr): Ditto.
19468 (chkp_build_addr_expr): Add check for hard reg case.
19469 (chkp_parse_array_and_component_ref): Ditto.
19470 (chkp_find_bounds_1): Ditto.
19471 (chkp_process_stmt): Don't generate bounds store for
19474 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19476 * predict.c (maybe_hot_bb_p): Do not check profile status.
19477 (maybe_hot_edge_p): Likewise.
19478 (probably_never_executed): Check for zero counts even if profile
19480 (unlikely_executed_edge_p): New function.
19481 (unlikely_executed_stmt_p): New function.
19482 (unlikely_executed_bb_p): New function.
19483 (set_even_probabilities): Use unlikely predicates.
19484 (combine_predictions_for_bb): Likewise.
19485 (predict_paths_for_bb): Likewise.
19486 (predict_paths_leading_to_edge): Likewise.
19487 (determine_unlikely_bbs): New function.
19488 (estimate_bb_frequencies): Use it.
19489 (compute_function_frequency): Use zero counts even if profile is
19491 * profile-count.h: Fix typo.
19493 2017-08-08 Julia Koval <julia.koval@intel.com>
19495 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
19496 _mm512_mask_cvtsepi16_storeu_epi8,
19497 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
19498 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
19499 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
19500 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
19501 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
19502 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
19503 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
19504 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
19505 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
19506 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
19507 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
19508 __builtin_ia32_pmovuswb256mem_mask,
19509 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
19510 __builtin_ia32_pmovwb512mem_mask): New builtins.
19512 2017-08-08 Julia Koval <julia.koval@intel.com>
19514 PR target/73350,80862
19515 * config/i386/subst.md (round): Fix round pattern.
19516 * config/i386/i386.c (ix86_erase_embedded_rounding):
19517 Fix erasing rounding for the fixed pattern.
19519 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19521 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
19523 2017-06-08 Martin Liska <mliska@suse.cz>
19525 PR gcov-profile/80911
19526 * gcov.c (block_info::block_info): New constructor.
19528 2017-06-07 Carl Love <cel@us.ibm.com>
19530 * config/rs6000/rs6000-c: The return type of the following
19531 built-in functions was implemented as int not long long. Fix sign
19532 of return value for the unsigned version of vec_mulo and vec_mule.
19533 vector unsigned long long vec_bperm (vector unsigned long long,
19534 vector unsigned char)
19535 vector signed long long vec_mule (vector signed int,
19537 vector unsigned long long vec_mule (vector unsigned int,
19538 vector unsigned int)
19539 vector signed long long vec_mulo (vector signed int,
19541 vector unsigned long long vec_mulo (vector unsigned int,
19542 vector unsigned int)
19543 * doc/extend.texi: Fix the documentation for the built-in
19546 2017-06-07 Carl Love <cel@us.ibm.com>
19549 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
19552 2017-06-07 Carl Love <cel@us.ibm.com>
19554 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
19555 support, Generate doublehv for signed int/float for BE case only.
19557 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
19559 * doc/invoke.texi (mcx16): Rewrite.
19561 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19563 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
19564 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
19565 *mov<mode>_softfloat, and an anonymous splitter): Use
19566 nonimmediate_operand instead of rs6000_nonimmediate_operand.
19568 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19570 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
19572 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19573 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
19574 (rs6000_debug_reg_global): Adjust.
19575 (rs6000_init_hard_regno_mode_ok): Adjust.
19576 (rs6000_dbx_register_number): Adjust.
19577 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
19578 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19579 Remove SPE_ACC and SPEFSCR.
19580 (REG_ALLOC_ORDER): Ditto.
19581 (FRAME_POINTER_REGNUM): Change to 111.
19582 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
19583 (REG_CLASS_NAMES): Ditto.
19584 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
19585 (REGISTER_NAMES): Ditto.
19586 (ADDITIONAL_REG_NAMES): Ditto.
19587 (rs6000_reg_names): Ditto.
19588 * config/rs6000/rs6000.md: Renumber some register number
19591 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19593 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
19595 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19596 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
19598 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
19599 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19600 Delete the SPE high registers.
19601 (REG_ALLOC_ORDER): Ditto.
19602 (enum reg_class): Remove SPE_HIGH_REGS.
19603 (REG_CLASS_NAMES): Ditto.
19604 (REG_CLASS_CONTENTS): Delete the SPE high registers.
19605 (REGISTER_NAMES): Ditto.
19606 (rs6000_reg_names): Ditto.
19607 * doc/tm.texi.in: Remove SPE as example.
19608 * doc/tm.texi: Regenerate.
19610 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19612 * config/rs6000/8540.md (ppc8540_brinc): Delete.
19613 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
19614 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
19615 * config/rs6000/rs6000.md (type): Remove "brinc".
19617 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19619 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
19620 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
19621 * config/rs6000/linuxspe.h: Delete file.
19622 * config/rs6000/rs6000.md: Don't include spe.md.
19623 * config/rs6000/spe.h: Delete file.
19624 * config/rs6000/spe.md: Delete file.
19625 * config/rs6000/t-rs6000: Remove spe.md.
19627 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19629 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
19630 (reg_or_none500mem_operand): Delete.
19631 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
19632 instead of reg_or_none500mem_operand.
19634 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19636 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19637 handling of SPE flags.
19638 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
19640 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19642 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
19644 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
19645 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
19646 paired_divv2sf3): Similar.
19647 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
19648 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
19649 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
19651 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
19652 Rename the paired_* instruction patterns.
19653 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
19655 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
19656 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
19657 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
19658 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
19659 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
19660 PAIRED_VECTOR_MODE.
19661 (struct machine_function): Delete field spe_insn_chain_scanned_p.
19662 (spe_func_has_64bit_regs_p): Delete.
19663 (spe_expand_predicate_builtin): Delete.
19664 (spe_expand_evsel_builtin): Delete.
19665 (TARGET_DWARF_REGISTER_SPAN): Do not define.
19666 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
19667 (invalid_e500_subreg): Delete.
19668 (rs6000_legitimize_address): Always force_reg op2 as well, for
19669 paired single memory accesses.
19670 (rs6000_member_type_forces_blk): Delete.
19671 (rs6000_spe_function_arg): Delete.
19672 (rs6000_expand_unop_builtin): Delete SPE handling.
19673 (rs6000_expand_binop_builtin): Ditto.
19674 (spe_expand_stv_builtin): Delete.
19675 (bdesc_2arg_spe): Delete.
19676 (spe_expand_builtin): Delete.
19677 (spe_expand_predicate_builtin): Delete.
19678 (spe_expand_evsel_builtin): Delete.
19679 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
19680 (spe_init_builtins): Delete.
19681 (spe_func_has_64bit_regs_p): Delete.
19682 (savres_routine_name): Delete "info" parameter. Adjust callers.
19683 (rs6000_emit_stack_reset): Ditto.
19684 (rs6000_dwarf_register_span): Delete.
19685 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
19686 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
19687 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
19689 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
19691 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
19692 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
19693 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
19694 mulv2sf3, divv2sf3): Delete expanders.
19696 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19698 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
19700 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19702 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
19703 * config/rs6000/rs6000.c: Ditto.
19705 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19707 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
19708 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
19709 comparison_operator.
19711 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19713 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
19714 * config/rs6000/rs6000.opt: Ditto.
19715 * config/rs6000/t-rtems: Ditto.
19717 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19719 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
19720 TARGET_E500_SINGLE by 0, simplify.
19721 * config/rs6000/rs6000.c: Ditto.
19722 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
19723 (spe_build_register_parallel): Delete.
19724 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
19725 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
19726 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
19727 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
19728 (E500_CONVERT): Delete.
19729 * config/rs6000/spe.md: Remove many patterns and all define_constants.
19731 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19733 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
19734 * config/rs6000/dfp.md: Ditto.
19735 (negdd2, *negdd2_fpr): Merge.
19736 (absdd2, *absdd2_fpr): Merge.
19737 (negtd2, *negtd2_fpr): Merge.
19738 (abstd2, *abstd2_fpr): Merge.
19739 * config/rs6000/e500.h: Delete file.
19740 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
19741 TARGET_FPRS by 1 and simplify.
19742 * config/rs6000/rs6000-c.c: Ditto.
19743 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
19744 TARGET_DF_SPE by 0.
19745 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
19747 * config/rs6000/rs6000.md: Ditto.
19748 (floatdidf2, *floatdidf2_fpr): Merge.
19749 (move_from_CR_gt_bit): Delete.
19750 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
19751 (E500_CR_IOR_COMPARE): Delete.
19752 (All patterns that require !TARGET_FPRS): Delete.
19753 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
19755 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19757 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
19759 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19761 * graphds.c (add_edge): Intitialize edge's attached data.
19762 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
19763 pointer parameter. Call pointed function on each edge during
19764 graph traversing. Skip traversing the edge when the function
19766 (graphds_dfs, graphds_scc): Ditto.
19767 (for_each_edge): New parameter. Pass the new parameter to callback
19769 * graphds.h (skip_edge_callback): New function pointer type.
19770 (graphds_dfs, graphds_scc): New function pointer parameter.
19771 (graphds_edge_callback, for_each_edge): New parameter.
19773 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19775 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
19776 out code checking if runtime alias check is possible to below ...
19777 Call the new function.
19778 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
19779 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
19781 2017-06-07 Marek Polacek <polacek@redhat.com>
19784 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
19785 TYPE_OVERFLOW_WRAPS check.
19787 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19789 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
19790 if versioning is required.
19791 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
19792 peeling with the check for versioning.
19794 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19796 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
19797 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
19798 Set true to new parameter if new ssa variable is defined.
19799 (vect_gen_vector_loop_niters): Refactor. Set range information
19800 for the new vector loop bound variable.
19801 (vect_do_peeling): Ditto.
19803 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19805 * tree-affine.c (ssa.h): Include header file.
19806 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
19807 has wrapping overflow behavior.
19809 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19811 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
19813 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19815 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
19816 (tree_to_aff_combination): ... here.
19818 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19820 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
19821 reg_pressure model function.
19822 (ivopts_global_cost_for_size): Delete.
19823 (determine_set_costs, iv_ca_recount_cost): Call new model function
19824 ivopts_estimate_reg_pressure.
19826 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19828 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
19829 expensive than udiv. Remove floating point cases from mod.
19831 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19833 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
19834 Increase idiv cost.
19836 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19838 * config/aarch64/aarch64.md
19839 (copysignsf3): Fix mask generation.
19841 2017-06-07 Jakub Jelinek <jakub@redhat.com>
19843 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
19845 (class dump_manager): Add register_dumps method.
19846 * dumpfile.c: Include langhooks.h.
19847 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
19848 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
19849 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
19850 (dump_manager::dump_register): Start with 512 entries instead of 32.
19851 (dump_manager::register_dumps): New method.
19852 * toplev.c (general_init): Instead of invoking register_dumps
19853 langhook, invoke register_dumps method on the dump manager.
19854 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
19857 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
19859 * doc/md.texi: Clarify the restrictions on a define_insn condition.
19860 Say that # requires an associated define_split to exist, and that
19861 the define_split must be suitable for use after register allocation.
19863 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19865 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
19866 (compute_outgoing_frequencies): Also initialize zero counts.
19867 (find_many_sub_basic_blocks): Do not produce uninitialized profile
19868 around loops; preserve more of profile when nothing changes.
19870 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
19872 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
19874 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
19875 * config/arm/arm-cpu-cdata.h: Regenerate.
19876 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
19877 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
19878 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
19879 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
19880 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
19882 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
19884 * config/arm/t-rmprofile: Likewise.
19885 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
19887 2017-06-06 David S. Miller <davem@davemloft.net>
19890 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
19891 blockage if function uses alloca.
19893 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19895 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
19896 New "uid" fields to hold pretty-print IDs of group and ref.
19897 Memory references are now identified as <group_id>:<ref_id>
19898 instead of using [random] addresses.
19899 (dump_mem_details): Simplify, no functional change.
19900 (dump_mem_ref): Simplify and make output more concise.
19901 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
19902 (find_or_create_group): Initialize group uid.
19903 (record_ref): Initialize ref uid. Improve debug output.
19904 (prune_group_by_reuse, should_issue_prefetch_p,)
19905 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
19906 (mark_nontemporal_store, determine_loop_nest_reuse):
19907 Improve debug output.
19909 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19911 * dbgcnt.def (prefetch): New debug counter.
19912 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
19913 (schedule_prefetches): Stop issueing prefetches if debug counter
19916 2017-06-06 Tom de Vries <tom@codesourcery.com>
19918 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
19919 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
19921 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19923 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
19924 Use aarch64_reg_or_zero predicate for operand 4.
19925 (aarch64_compare_and_swap<mode> define_insn_and_split):
19926 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
19927 (aarch64_store_exclusive<mode>): Likewise for operand 2.
19929 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19931 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
19932 (arm_compute_save_core_reg_mask): This.
19933 (thumb1_compute_save_reg_mask): Rename into ...
19934 (thumb1_compute_save_core_reg_mask): This.
19935 (arm_compute_save_reg0_reg12_mask): Adapt comment.
19936 (arm_compute_frame_layout): Likewise.
19938 2017-06-06 Richard Biener <rguenther@suse.de>
19940 PR tree-optimization/80974
19941 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
19942 keep or clear leaders SSA info.
19944 2017-06-06 Tom de Vries <tom@codesourcery.com>
19946 * config/nvptx/nvptx.c (split_mode_p): New function.
19947 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
19949 2017-06-06 Tom de Vries <tom@codesourcery.com>
19951 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
19953 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19956 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
19959 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19961 * shrink-wrap.c (handle_simple_exit): Update profile.
19962 (try_shrink_wrapping): Upate profile.
19964 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19966 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
19967 (tree_guess_outgoing_edge_probabilities): New.
19968 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
19969 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
19971 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19973 * ipa-split.c (split_function): Initialize return bb profile.
19975 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19977 * profile.c (compute_branch_probabilities): Also initialize
19978 EXIT_BLOCK profile.
19980 2017-06-06 Richard Biener <rguenther@suse.de>
19982 PR tree-optimization/80928
19983 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
19984 (vect_analyze_loop_operations): Properly guard analysis for
19986 (vect_transform_loop): Likewise.
19987 (vect_analyze_loop_2): Also reset SLP type on PHIs.
19988 (vect_model_induction_cost): Do not cost for pure SLP.
19989 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
19990 of induction in inner loop vectorization.
19991 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
19992 (vect_get_and_check_slp_defs): Handle vect_induction_def.
19993 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
19995 (vect_analyze_slp_cost_1): Cost induction.
19996 (vect_detect_hybrid_slp_stmts): Handle PHIs.
19997 (vect_get_slp_vect_defs): Likewise.
19998 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
19999 (vect_transform_stmt): Handle SLP reductions.
20000 * tree-vectorizer.h (vectorizable_induction): Adjust.
20002 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20004 * config/rs6000/rs6000.c (make_resolver_func): Update
20005 init_lowered_empty_function call.
20007 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
20009 * doc/invoke.texi: Document the -fprofile-abs-path option.
20010 * common.opt (fprofile-abs-path): New option.
20011 * gcov-io.h (gcov_write_filename): Declare.
20012 * gcov-io.c (gcov_write_filename): New function.
20013 * coverage.c (coverage_begin_function): Use gcov_write_filename.
20014 * profile.c (output_location): Likewise.
20016 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20018 * shring-wrap.c: Revert accidental commit.
20020 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
20022 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
20024 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20026 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
20028 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
20029 profile in callgraph edge.
20030 * profile-count.h (apply_probability): If THIS is 0, then result is 0
20031 (apply_scale): Likewise.
20032 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
20033 Also scale profile when inlining function with zero profile.
20034 (initialize_cfun): Update exit block profile even when it is zero.
20035 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
20036 when profile is read.
20038 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
20040 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
20041 (CLONE_*): New constants to define the processors we can generate
20042 code for with the target_clone attribute.
20043 (rs6000_clone_map): New array to identify which clone processors
20044 the current program is running on.
20045 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
20046 target_clone attribute.
20047 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
20048 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
20049 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
20050 (cpu_expand_builtin): Add support for target_clone attribute.
20051 (rs6000_valid_attribute_p): Allow "default" attribute.
20052 (get_decl_name): New debug function to simplify printing the
20053 current function name in debugging statements.
20054 (rs6000_clone_priority): New functions to support the target_clone
20055 attribute, and be able to generate code to switch between ISA 2.05
20056 through ISA 3.0 (power6 through power9).
20057 (rs6000_compare_version_priority): Likewise.
20058 (rs6000_get_function_versions_dispatcher): Likewise.
20059 (make_resolver_func): Likewise.
20060 (add_condition_to_bb): Likewise.
20061 (dispatch_function_versions): Likewise.
20062 (rs6000_generate_version_dispatcher_body): Likewise.
20063 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
20064 (fusion_gpr_load_p): Fix a spacing issue.
20065 * doc/extend.texi (Common Function Attributes): Document that the
20066 PowerPC supports the target_clone attribute.
20068 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
20070 * config/arm/arm.h: explain F symbol found in description of ARM
20071 register allocation in its legend.
20073 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20075 * config/mips/frame-header-opt.c: Include profile-count.h.
20076 * config/riscv/riscv.c: Include profile-count.h
20078 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
20080 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
20082 (sm_set_flag_if_changed): Add bbs field.
20083 (execute_sm_if_changed_flag_set): Pass BBS.
20084 (execute_sm): Update.
20086 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20088 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
20091 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20093 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
20094 (peephole2): New peephole2 to emit the above.
20095 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
20097 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20099 * config/aarch64/aarch64.c (define_peephole2 above
20100 *sub_<shift>_<mode>): New peephole.
20102 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20104 * config/i386/i386.c (make_resolver_func): Update.
20105 * Makefile.in: Add profile-count.h and profile-count.o
20106 * auto-profile.c (afdo_indirect_call): Update to new API.
20107 (afdo_set_bb_count): Update.
20108 (afdo_propagate_edge): Update.
20109 (afdo_propagate_circuit): Update.
20110 (afdo_calculate_branch_prob): Update.
20111 (afdo_annotate_cfg): Update.
20112 * basic-block.h: Include profile-count.h
20113 (struct edge_def): Turn count to profile_count.
20114 (struct basic_block_def): Likewie.
20115 (REG_BR_PROB_BASE): Move to profile-count.h
20116 (RDIV): Move to profile-count.h
20117 * bb-reorder.c (max_entry_count): Turn to profile_count.
20118 (find_traces): Update.
20119 (rotate_loop):Update.
20120 (connect_traces):Update.
20121 (sanitize_hot_paths):Update.
20122 * bt-load.c (migrate_btr_defs): Update.
20123 * cfg.c (RDIV): Remove.
20124 (init_flow): Use alloc_block.
20125 (alloc_block): Uninitialize count.
20126 (unchecked_make_edge): Uninitialize count.
20127 (check_bb_profile): Update.
20128 (dump_edge_info): Update.
20129 (dump_bb_info): Update.
20130 (update_bb_profile_for_threading): Update.
20131 (scale_bbs_frequencies_int): Update.
20132 (scale_bbs_frequencies_gcov_type): Update.
20133 (scale_bbs_frequencies_profile_count): New.
20134 * cfg.h (update_bb_profile_for_threading): Update.
20135 (scale_bbs_frequencies_profile_count): Declare.
20136 * cfgbuild.c (compute_outgoing_frequencies): Update.
20137 (find_many_sub_basic_blocks): Update.
20138 * cfgcleanup.c (try_forward_edges): Update.
20139 (try_crossjump_to_edge): Update.
20140 * cfgexpand.c (expand_gimple_tailcall): Update.
20141 (construct_exit_block): Update.
20142 * cfghooks.c (verify_flow_info): Update.
20143 (dump_bb_for_graph): Update.
20144 (split_edge): Update.
20145 (make_forwarder_block): Update.
20146 (duplicate_block): Update.
20147 (account_profile_record): Update.
20148 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
20149 (get_estimated_loop_iterations): Update.
20150 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
20151 (single_likely_exit): Update.
20152 * cfgloopmanip.c (scale_loop_profile): Update.
20154 (set_zero_probability): Update.
20155 (lv_adjust_loop_entry_edge): Update.
20156 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20157 (purge_dead_edges): Update.
20158 (rtl_account_profile_record): Update.
20159 * cgraph.c (cgraph_node::create): Uninitialize count.
20160 (symbol_table::create_edge): Uninitialize count.
20161 (cgraph_update_edges_for_call_stmt_node): Update.
20162 (cgraph_edge::dump_edge_flags): Update.
20163 (cgraph_node::dump): Update.
20164 (cgraph_edge::maybe_hot_p): Update.
20165 * cgraph.h: Include profile-count.h
20166 (create_clone), create_edge, create_indirect_edge): Update.
20167 (cgraph_node): Turn count to profile_count.
20168 (cgraph_edge0: Likewise.
20169 (make_speculative, clone): Update.
20170 (create_edge): Update.
20171 (init_lowered_empty_function): Update.
20172 * cgraphclones.c (cgraph_edge::clone): Update.
20173 (duplicate_thunk_for_node): Update.
20174 (cgraph_node::create_clone): Update.
20175 * cgraphunit.c (cgraph_node::analyze): Update.
20176 (cgraph_node::expand_thunk): Update.
20177 * final.c (dump_basic_block_info): Update.
20178 * gimple-streamer-in.c (input_bb): Update.
20179 * gimple-streamer-out.c (output_bb): Update.
20180 * graphite.c (print_global_statistics): Update.
20181 (print_graphite_scop_statistics): Update.
20182 * hsa-brig.c: Include basic-block.h.
20183 * hsa-dump.c: Include basic-block.h.
20184 * hsa-gen.c (T sum_slice): Update.
20185 (convert_switch_statements):Update.
20186 * hsa-regalloc.c: Include basic-block.h.
20187 * ipa-chkp.c (chkp_produce_thunks): Update.
20188 * ipa-cp.c (struct caller_statistics): Update.
20189 (init_caller_stats): Update.
20190 (gather_caller_stats): Update.
20191 (ipcp_cloning_candidate_p): Update.
20192 (good_cloning_opportunity_p): Update.
20193 (get_info_about_necessary_edges): Update.
20194 (dump_profile_updates): Update.
20195 (update_profiling_info): Update.
20196 (update_specialized_profile): Update.
20197 (perhaps_add_new_callers): Update.
20198 (decide_about_value): Update.
20199 (ipa_cp_c_finalize): Update.
20200 * ipa-devirt.c (struct odr_type_warn_count): Update.
20201 (struct decl_warn_count): Update.
20202 (struct final_warning_record): Update.
20203 (possible_polymorphic_call_targets): Update.
20204 (ipa_devirt): Update.
20205 * ipa-fnsummary.c (redirect_to_unreachable): Update.
20206 * ipa-icf.c (sem_function::merge): Update.
20207 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
20208 * ipa-inline.c (compute_uninlined_call_time): Update.
20209 (compute_inlined_call_time): Update.
20210 (want_inline_small_function_p): Update.
20211 (want_inline_self_recursive_call_p): Update.
20212 (edge_badness): Update.
20213 (lookup_recursive_calls): Update.
20214 (recursive_inlining): Update.
20215 (inline_small_functions): Update.
20216 (dump_overall_stats): Update.
20217 (dump_inline_stats): Update.
20218 * ipa-profile.c (ipa_profile_generate_summary): Update.
20219 (ipa_propagate_frequency): Update.
20220 (ipa_profile): Update.
20221 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20222 * ipa-utils.c (ipa_merge_profiles): Update.
20223 * loop-doloop.c (doloop_modify): Update.
20224 * loop-unroll.c (report_unroll): Update.
20225 (unroll_loop_runtime_iterations): Update.
20226 * lto-cgraph.c (lto_output_edge): Update.
20227 (lto_output_node): Update.
20228 (input_node): Update.
20229 (input_edge): Update.
20230 (merge_profile_summaries): Update.
20231 * lto-streamer-in.c (input_cfg): Update.
20232 * lto-streamer-out.c (output_cfg): Update.
20233 * mcf.c (create_fixup_graph): Update.
20234 (adjust_cfg_counts): Update.
20235 (sum_edge_counts): Update.
20236 * modulo-sched.c (sms_schedule): Update.
20237 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
20238 * predict.c (maybe_hot_count_p): Update.
20239 (probably_never_executed): Update.
20240 (dump_prediction): Update.
20241 (combine_predictions_for_bb): Update.
20242 (propagate_freq): Update.
20243 (handle_missing_profiles): Update.
20244 (counts_to_freqs): Update.
20245 (rebuild_frequencies): Update.
20246 (force_edge_cold): Update.
20247 * predict.h: Include profile-count.h
20248 (maybe_hot_count_p, counts_to_freqs): UPdate.
20249 * print-rtl-function.c: Do not include cfg.h
20250 * print-rtl.c: Include basic-block.h
20251 * profile-count.c: New file.
20252 * profile-count.h: New file.
20253 * profile.c (is_edge_inconsistent): Update.
20254 (correct_negative_edge_counts): Update.
20255 (is_inconsistent): Update.
20256 (set_bb_counts): Update.
20257 (read_profile_edge_counts): Update.
20258 (compute_frequency_overlap): Update.
20259 (compute_branch_probabilities): Update; Initialize and deinitialize
20261 (branch_prob): Update.
20262 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
20263 (edge_gcov_count): New.
20264 (bb_gcov_count): New.
20265 * shrink-wrap.c (try_shrink_wrapping): Update.
20266 * tracer.c (better_p): Update.
20267 * trans-mem.c (expand_transaction): Update.
20268 (ipa_tm_insert_irr_call): Update.
20269 (ipa_tm_insert_gettmclone_call): Update.
20270 * tree-call-cdce.c: Update.
20271 * tree-cfg.c (gimple_duplicate_sese_region): Update.
20272 (gimple_duplicate_sese_tail): Update.
20273 (gimple_account_profile_record): Update.
20274 (execute_fixup_cfg): Update.
20275 * tree-inline.c (copy_bb): Update.
20276 (copy_edges_for_bb): Update.
20277 (initialize_cfun): Update.
20278 (freqs_to_counts): Update.
20279 (copy_cfg_body): Update.
20280 (expand_call_inline): Update.
20281 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20282 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
20283 (try_unroll_loop_completely): Update.
20284 (try_peel_loop): Update.
20285 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20286 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
20287 * tree-ssa-loop-split.c (connect_loops): Update.
20288 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
20289 * tree-ssa-reassoc.c (branch_fixup): Update.
20290 * tree-ssa-tail-merge.c (replace_block_by): Update.
20291 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
20292 (compute_path_counts): Update.
20293 (update_profile): Update.
20294 (recompute_probabilities): Update.
20295 (update_joiner_offpath_counts): Update.
20296 (estimated_freqs_path): Update.
20297 (freqs_to_counts_path): Update.
20298 (clear_counts_path): Update.
20299 (ssa_fix_duplicate_block_edges): Update.
20300 (duplicate_thread_path): Update.
20301 * tree-switch-conversion.c (case_bit_test_cmp): Update.
20302 (struct switch_conv_info): Update.
20303 * tree-tailcall.c (decrease_profile): Update.
20304 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20305 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20306 * value-prof.c (check_counter): Update.
20307 (gimple_divmod_fixed_value): Update.
20308 (gimple_mod_pow2): Update.
20309 (gimple_mod_subtract): Update.
20310 (gimple_ic_transform): Update.
20311 (gimple_stringop_fixed_value): Update.
20312 * value-prof.h (gimple_ic): Update.
20314 2017-06-02 Carl Love <cel@us.ibm.com>
20316 * config/rs6000/rs6000-c: Add support for built-in functions
20317 vector double vec_doublee (vector signed int);
20318 vector double vec_doublee (vector unsigned int);
20319 vector double vec_doublee (vector float);
20320 vector double vec_doubleh (vector signed int);
20321 vector double vec_doubleh (vector unsigned int);
20322 vector double vec_doubleh (vector float);
20323 vector double vec_doublel (vector signed int);
20324 vector double vec_doublel (vector unsigned int);
20325 vector double vec_doublel (vector float);
20326 vector double vec_doubleo (vector signed int);
20327 vector double vec_doubleo (vector unsigned int);
20328 vector double vec_doubleo (vector float);.
20329 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
20330 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
20332 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
20333 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
20334 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
20336 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
20337 vec_doublel, vec_doubleh.
20338 * doc/extend.texi: Update the built-in documentation file for the
20339 new built-in functions.
20341 2017-06-02 David Malcolm <dmalcolm@redhat.com>
20344 * ipa-inline-analysis.c (free_growth_caches): Set
20345 edge_removal_hook_holder to NULL after removing it.
20347 2017-06-02 Sudakshina Das <sudi.das@arm.com>
20349 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
20350 comparision with zero.
20352 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20353 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20354 for early expansion of vec_min and vec_max builtins.
20355 (builtin_function_type): Add min/max unsigned variants to those
20356 identified as having unsigned arguments.
20358 2017-06-02 Olivier Hainque <hainque@adacore.com>
20360 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
20362 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20364 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
20365 Use VALL_F16 iterator rather than VALL.
20367 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20369 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
20370 Emit CBNZ inside loop when doing a strong exchange and comparing
20371 against zero. Generate the CC flags after the loop.
20373 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
20375 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
20376 (dl_section_ref): New.
20377 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
20378 On AIX, append an expression to subtract the size of the
20379 section length to dl_section_ref.
20381 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20383 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20384 for early expansion of vector absolute builtins.
20386 2017-06-02 Richard Biener <rguenther@suse.de>
20388 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
20389 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
20391 2017-06-02 Richard Biener <rguenther@suse.de>
20393 PR tree-optimization/80948
20394 * tree-tailcall.c (find_tail_calls): Track stmts to move in
20395 stmt order as well.
20397 2017-06-02 Richard Biener <rguenther@suse.de>
20399 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
20401 * tree-vect-stmts.c (process_use): Do not mark backedge defs
20402 for inductions as relevant.
20404 2017-06-02 Richard Biener <rguenther@suse.de>
20406 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
20407 (vectorizable_induction): ... this. Remove dead code.
20409 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
20411 * builtins. (expand_builtin_alloca): Remove second parameter and
20412 infer its value from the first parameter instead.
20413 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
20415 2017-06-02 Jakub Jelinek <jakub@redhat.com>
20417 PR rtl-optimization/80903
20418 * loop-doloop.c (add_test): Unshare sequence.
20420 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
20422 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
20424 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
20426 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
20428 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
20429 xlogue_layout::get_instance, logue_layout::xlogue_layout,
20430 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
20431 (xlogue_layout::get_stub_rtx): Make static.
20432 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
20433 (xlogue_layout::compute_stub_managed_regs): Rename to...
20434 (xlogue_layout::count_stub_managed_regs): ...this.
20435 (xlogue_layout::is_stub_managed_reg): New function.
20436 (xlogue_layout::m_stub_names): Rename to...
20437 (xlogue_layout::s_stub_names): ...this, make static.
20438 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
20439 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
20440 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
20441 xlogue_layout::s_stub_names): Instantiate statics.
20442 (stub_managed_regs): Remove.
20443 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
20444 (disable_call_ms2sysv_xlogues): Rename to...
20445 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
20446 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
20448 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
20449 change after reload_completed.
20450 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
20452 (ix86_expand_prologue): Likewise.
20453 (ix86_expand_epilogue): Likewise.
20454 (ix86_expand_split_stack_prologue): Likewise.
20455 (ix86_compute_frame_layout): Remove frame parameter ...
20456 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
20457 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
20459 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
20460 (ix86_frame): Move from here ...
20461 * config/i386/i386.h (ix86_frame): ... to here.
20462 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
20463 complete ix86_frame data structure instead. Remove some_ld_name.
20465 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20467 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
20468 symbols that hold a DECL_VALUE_EXPR.
20470 2017-06-01 Martin Jambor <mjambor@suse.cz>
20472 PR tree-optimization/80898
20473 * tree-sra.c (process_subtree_disqualification): Removed.
20474 (disqualify_candidate): Do not acll
20475 process_subtree_disqualification.
20476 (subtree_mark_written_and_enqueue): New function.
20477 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
20478 RHS has been disqualified and re-queue LHS if necessary. Apart
20479 from that, ignore disqualified RHS.
20481 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20483 * config/s390/s390.c (s390_emit_epilogue): Disable early return
20484 address fetch for z10 or later.
20486 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20488 * config/arc/arc.md (tst_movb): Add guard when splitting.
20490 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20492 * config/arc/arc.c (arc_can_eliminate): Test against
20493 arc_frame_pointer_needed.
20495 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20497 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
20498 to prevent store reordering.
20499 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
20500 (type): Add block type.
20501 (stack_tie): Define special instruction to be used in
20504 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20506 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
20507 constraint. It is not valid for the pattern.
20508 (noncommutative_binary_comparison): Likewise.
20510 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20512 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
20515 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20517 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
20518 be used by the reg-alloc.
20520 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20522 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
20523 reg-alloc when having mul64 or mul32x16 instructions.
20524 (mulsidi3): Likewise.
20525 (umulsidi3): Likewise.
20526 (mulsi32x16): New pattern.
20527 (mulsi64): Likewise.
20528 (mulsidi64): Likewise.
20529 (umulsidi64): Likewise.
20530 (MUL32x16_REG): Define.
20531 (mul64_600): Use MUL32x16_REG.
20532 (mac64_600): Likewise.
20533 (umul64_600): Likewise.
20534 (umac64_600): Likewise.
20536 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20538 * config/arc/arc.md (mulsi3_700): Make it commutative.
20540 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
20542 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
20544 (*sign_extendsidi2_insn): Likewise for movstosw.
20546 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20548 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
20549 the type of the input discriminant value. Convert the
20550 discriminant value of signedness vary.
20552 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
20554 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
20555 Add to -Wall section.
20557 2017-06-01 Richard Biener <rguenther@suse.de>
20559 PR middle-end/66313
20560 * fold-const.c (fold_plusminus_mult_expr): If the factored
20561 factor may be zero use a wrapping type for the inner operation.
20562 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
20563 and handle moved defs.
20564 (process_assignment): Properly guard the unary op case. Return a
20565 tri-state indicating that moving the stmt before the call may allow
20566 to continue. Pass through to_move.
20567 (find_tail_calls): Handle moving unrelated defs before
20570 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
20573 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
20574 splitter result in the canonical way.
20576 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
20578 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
20579 also for 32bit target. Update insn attributes.
20580 (zero-extendsidi2 splitter): Allow all registers for operand 1.
20582 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
20584 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
20585 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
20586 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
20587 (_mm_maskz_min_ss): New intrinsics.
20589 2017-05-31 Martin Liska <mliska@suse.cz>
20591 * tree-vect-loop.c (vect_create_epilog_for_reduction):
20592 Change comment style to one we normally use.
20593 (vectorizable_reduction): Likewise.
20594 (vectorizable_induction): Likewise.
20595 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20596 (vectorizable_call): Likewise.
20597 (vectorizable_simd_clone_call): Likewise.
20598 (vectorizable_conversion): Likewise.
20599 (vectorizable_assignment): Likewise.
20600 (vectorizable_shift): Likewise.
20601 (vectorizable_operation): Likewise.
20602 (vectorizable_store): Likewise.
20603 (vectorizable_load): Likewise.
20604 * tree-vectorizer.h: Likewise.
20606 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
20608 * passes.c (emergency_dump_function): New.
20609 * tree-pass.h (emergency_dump_function): Declare.
20610 * plugin.c (plugins_internal_error_function): Remove.
20611 * plugin.h (plugins_internal_error_function): Remove declaration.
20612 * toplev.c (internal_error_function): New static function. Use it...
20613 (general_init): ...here.
20615 2017-05-31 Graham Markall <graham.markall@embecosm.com>
20617 * config/arc/arc.c (arc_print_operand): Handle constant operands.
20618 (arc_rtx_costs): Add costs for new patterns.
20619 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
20620 * config/arc/predicates.md: Add _1_2_3_operand predicate.
20622 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20624 * tree-ssa-strlen.c (get_next_strinfo): New function.
20625 (get_stridx_plus_constant): Use it.
20626 (zero_length_string): Likewise.
20627 (adjust_related_strinfos): Likewise.
20628 (adjust_last_stmt): Likewise.
20630 2017-05-31 Richard Biener <rguenther@suse.de>
20633 * config/i386/i386.c (ix86_expand_builtin): Remove assert
20634 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
20636 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20638 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
20639 loop_vinfo argument and use of dependence distance vectors.
20640 Check instead whether the two references differ only in their
20641 initial value and assume that they have the same alignment if the
20642 difference is a multiple of the vector alignment.
20643 (vect_analyze_data_refs_alignment): Update call accordingly.
20645 2017-05-31 Martin Liska <mliska@suse.cz>
20648 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
20650 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20652 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
20653 (create_intersect_range_checks): Move from ...
20654 * tree-data-ref.c (create_intersect_range_checks_index)
20655 (create_intersect_range_checks): ... to here.
20656 (create_runtime_alias_checks): New function factored from ...
20657 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
20658 here. Call above function.
20659 * tree-data-ref.h (create_runtime_alias_checks): New function.
20661 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20663 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
20664 segment length for dr_b and compute it in wide_int.
20666 2017-05-31 Richard Biener <rguenther@suse.de>
20668 PR tree-optimization/80906
20669 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
20670 and pass through iv_map.
20671 (copy_bb_and_scalar_dependences): Adjust.
20672 (translate_pending_phi_nodes): Likewise.
20673 (copy_loop_close_phi_args): Handle code-generating IVs instead
20676 2017-05-30 David Malcolm <dmalcolm@redhat.com>
20678 * diagnostic-color.c (color_dict): Add "type-diff".
20679 (parse_gcc_colors): Update comment.
20680 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
20681 -fdiagnostics-show-template-tree and -fno-elide-type.
20682 (GCC_COLORS): Add type-diff to example.
20684 (-fdiagnostics-show-template-tree): New.
20685 (-fno-elide-type): New.
20686 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
20687 the pp_format_decoder callback. Call any m_format_postprocessor's
20689 (pretty_printer::pretty_printer): Initialize
20690 m_format_postprocessor.
20691 (pretty_printer::~pretty_printer): Delete any
20692 m_format_postprocessor.
20693 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
20694 (class format_postprocessor): New class.
20695 (struct pretty_printer::format_decoder): Document the new parameters.
20696 (struct pretty_printer::m_format_postprocessor): New field.
20697 * tree-diagnostic.c (default_tree_printer): Update for new
20698 bool and const char ** params.
20699 * tree-diagnostic.h (default_tree_printer): Likewise.
20701 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
20703 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
20704 (lwa_operand): Delete rs6000_gen_cell_microcode test.
20705 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20706 rs6000_gen_cell_microcode code.
20707 (rs6000_final_prescan_insn): Delete.
20708 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
20709 "warn-cell-microcode" entries.
20710 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
20711 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
20712 throughout. Change cc_reg_not_micro_cr0_operand to
20713 cc_reg_not_cr0_operand throughout.
20714 (*extendhi<mode>2_noload): Delete.
20715 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
20716 (mwarn-cell-microcode): Delete.
20717 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
20718 -mgen-cell-microcode and -mwarn-cell-microcode.
20720 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
20723 * config/i386/constraints.md (Yd): New constraint.
20725 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
20726 and (?Yd, r) alternatives. Update insn attributes.
20727 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
20728 and (?*Yd, r) alternatives. Update insn attributes.
20729 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
20731 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
20733 * gimplify.c (gimplify_modify_expr): Don't create a
20734 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
20737 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
20739 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
20741 2017-05-30 Richard Biener <rguenther@suse.de>
20743 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
20744 and reduc_def fields.
20745 (STMT_VINFO_REDUC_TYPE): New define.
20746 (STMT_VINFO_REDUC_DEF): Likewise.
20747 (vect_force_simple_reduction): Adjust prototype.
20748 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
20749 (vect_is_simple_reduction): Remove check_reduction argument.
20750 (vect_force_simple_reduction): Adjust and set
20751 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
20752 (vectorizable_reduction): Do not re-do reduction analysis
20753 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
20754 * tree-parloops.c (gather_scalar_reductions): Adjust.
20756 2017-05-30 Richard Biener <rguenther@suse.de>
20758 PR middle-end/80901
20759 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
20762 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
20764 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
20765 Introduce unknown_misalignment parameter and remove vf.
20766 (vect_peeling_hash_get_lowest_cost):
20767 Pass unknown_misalignment parameter.
20768 (vect_enhance_data_refs_alignment):
20769 Fix unsupportable data ref treatment.
20771 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20773 * tree-vect-data-refs.c (vect_get_data_access_cost):
20774 Workaround for SLP handling.
20775 (vect_enhance_data_refs_alignment):
20776 Compute costs for doing no peeling at all, compare to the best
20777 peeling costs so far and avoid peeling if cheaper.
20779 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20781 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
20782 Return peeling info and set costs to zero for unlimited cost
20784 (vect_enhance_data_refs_alignment): Also inspect all datarefs
20785 with unknown misalignment. Compute and costs for unknown
20786 misalignment, compare them to the costs for known misalignment
20787 and choose the cheapest for peeling.
20789 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20791 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
20792 (vect_get_peeling_costs_all_drs): Create function.
20793 (vect_peeling_hash_get_lowest_cost):
20794 Use vect_get_peeling_costs_all_drs.
20795 (vect_peeling_supportable): Create function.
20796 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
20798 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20800 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
20801 DR_HAS_NEGATIVE_STEP.
20802 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
20803 (vect_enhance_data_refs_alignment): Use.
20804 (vect_duplicate_ssa_name_ptr_info): Use.
20805 * tree-vectorizer.h (dr_misalignment): Use.
20806 (known_alignment_for_access_p): Use.
20808 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20811 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
20813 (has_section_name): New function.
20815 2017-05-30 Martin Liska <mliska@suse.cz>
20818 * auto-profile.c (get_function_decl_from_block): Fix
20821 2017-05-30 Richard Biener <rguenther@suse.de>
20823 PR middle-end/80876
20824 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
20826 2017-05-30 Martin Liska <mliska@suse.cz>
20828 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
20829 * dumpfile.h (struct dump_file_info): Remove ctors.
20831 2017-05-30 Martin Liska <mliska@suse.cz>
20833 * predict.def: Fix GNU coding style.
20835 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20837 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20838 Mark 'to' argument with ATTRIBUTE_UNUSED.
20840 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20842 * config/xtensa/xtensa.c (xtensa_emit_call): Use
20843 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
20844 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
20847 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
20849 * doc/install.texi (Options specification): Restore entry of
20850 --enable-sjlj-exceptions.
20852 2017-05-27 Michael Eager <eager@eagercon.com>
20855 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
20857 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
20859 * config/microblaze/microblaze.h
20860 (FIXED_REGISTERS): Update in macro.
20861 (CALL_USED_REGISTERS): Update in macro.
20863 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
20865 * doc/install.texi: Add links to macOS binary distributions.
20867 2017-05-27 Jakub Jelinek <jakub@redhat.com>
20871 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20873 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20875 2017-05-26 Martin Liska <mliska@suse.cz>
20877 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
20879 2017-05-26 Martin Liska <mliska@suse.cz>
20881 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
20882 always leading ';; '.
20883 (dump_bb_info): Likewise.
20884 (brief_dump_cfg): Likewise.
20885 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
20886 * dumpfile.c: Remove usage of TDF_VERBOSE.
20887 * dumpfile.h (enum dump_kind): Likewise.
20888 (dump_gimple_bb_header): Do not use TDF_COMMENT.
20889 * print-tree.c (debug_verbose): Remove.
20890 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
20891 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
20892 * tree-diagnostic.c (default_tree_printer): Replace
20893 TDF_DIAGNOSTIC with TDF_SLIM.
20895 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20897 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
20898 in parameter loop, rather than loop_vinfo.
20899 (create_intersect_range_checks): Ditto.
20900 (vect_create_cond_for_alias_checks): Update call to above functions.
20902 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20904 PR tree-optimization/80815
20905 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
20906 for merging runtime alias checks. Handle negative DR_STEPs.
20908 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20910 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
20912 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
20913 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
20914 out code pruning runtime alias checks.
20915 * tree-data-ref.c (prune_runtime_alias_test_list): New function
20916 factored out from above.
20917 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20919 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20921 (prune_runtime_alias_test_list): New decalaration.
20923 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20925 * tree-vect-data-refs.c (compare_tree): Rename and move ...
20926 * tree-data-ref.c (data_ref_compare_tree): ... to here.
20927 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
20928 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
20929 (operator==, comp_dr_with_seg_len_pair): Ditto.
20930 (vect_prune_runtime_alias_test_list): Ditto.
20932 2017-05-26 Martin Liska <mliska@suse.cz>
20935 * params.def: Bound partial-inlining-entry-probability param.
20937 2017-05-26 Marek Polacek <polacek@redhat.com>
20940 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
20943 2017-05-26 Richard Biener <rguenther@suse.de>
20945 PR tree-optimization/80842
20946 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
20949 2017-05-26 Richard Biener <rguenther@suse.de>
20951 PR tree-optimization/80844
20952 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
20954 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
20956 * doc/md.texi (Machine Constraints): Update x86 family
20957 machine constraints section to match 'config/i386/constraints.md'.
20959 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
20961 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
20963 2017-05-25 Nathan Sidwell <nathan@acm.org>
20965 * doc/invoke.texi (--enable-languages): Update documentation.
20967 2017-05-25 Martin Liska <mliska@suse.cz>
20969 * dumpfile.c: Add TDF_FOLDING.
20970 * dumpfile.h (enum dump_kind): Likewise.
20971 * genmatch.c (dt_simplify::gen_1): Use it.
20973 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20975 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
20977 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20979 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20980 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
20982 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20984 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
20985 * match.pd (X == C): Rewrite it here.
20986 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
20987 with_certain_nonzero_bits2): New predicates.
20988 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
20990 2017-05-24 Nathan Sidwell <nathan@acm.org>
20992 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
20995 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
20998 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21000 * config/powerpcspe: New port. Files are copied from the rs6000
21001 port, with "rs6000" in filenames replaced by "powerpcspe".
21003 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
21005 PR rtl-optimization/80754
21006 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
21008 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
21010 * config/sparc/sparc.md (length): Return the correct value for -mflat
21011 sibcalls to match output_sibcall.
21013 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
21017 * config/rs6000/rs6000.c (struct machine_function): Add new field
21019 (rs6000_get_separate_components): Init that field, use it.
21020 (rs6000_components_for_bb): Use the field.
21022 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
21024 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
21026 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
21028 PR middle-end/80823
21029 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
21031 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
21034 * config/s390/s390.c (s390_check_qrst_address): Check incoming
21035 address against address_operand predicate.
21036 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
21038 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
21040 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
21041 parameters passed indirectly.
21043 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
21045 * config/i386/i386.md (*movdi_internal): Remove SSE4
21046 alternative 18 (?r, *v). Update insn attributes.
21047 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
21048 Update insn attributes.
21049 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
21050 Update insn attributes.
21051 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
21052 alternative 1 (r, v). Remove isa attribute.
21053 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
21054 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
21055 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
21057 2017-05-23 Tom de Vries <tom@codesourcery.com>
21059 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
21062 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21064 * cgraphunit.c (symbol_table::process_new_functions): Update.
21065 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
21066 (inline_generate_summary): Rename to ...
21067 (ipa_fn_summary_generate): ... this one.
21068 (inline_read_summary): Rename to ...
21069 (ipa_fn_summary_read): ... this one.
21070 (inline_write_summary): Rename to ...
21071 (ipa_fn_summary_write): ... this one.
21072 (inline_free_summary): Rename to ...
21073 (ipa_free_fn_summary): ... this one.
21074 (pass_data_local_fn_summary, pass_local_fn_summary,
21075 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
21076 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
21077 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
21078 make_pass_ipa_fn_summary): New.
21079 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
21080 inline_write_summary, inline_free_summary): Remove.
21081 (ipa_free_fn_summary) : New.
21082 * ipa-inline.c (ipa_inline): Update.
21083 (pass_ipa_inline): Do not generate summaries.
21084 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
21086 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
21087 and add pass_ipa_fn_summary.
21088 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
21090 (make_pass_inline_parameters): Remove.
21092 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
21094 * omp-low.c (struct omp_context): Remove "default_kind" member.
21097 * omp-offload.c (execute_oacc_device_lower): Remove the
21098 parallelism dimensions function attributes for unparallelized
21099 OpenACC kernels constructs.
21101 2017-05-23 Martin Liska <mliska@suse.cz>
21103 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
21105 (cgraph_edge::make_speculative): Likewise.
21106 (cgraph_edge::resolve_speculation): Likewise.
21107 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21108 (cgraph_node::dump): Likewise.
21109 * cgraph.h: Likewise.
21110 * cgraphunit.c (analyze_functions): Likewise.
21111 (symbol_table::compile): Likewise.
21112 * ipa-cp.c (print_all_lattices): Likewise.
21113 (determine_versionability): Likewise.
21114 (initialize_node_lattices): Likewise.
21115 (ipcp_verify_propagated_values): Likewise.
21116 (estimate_local_effects): Likewise.
21117 (update_profiling_info): Likewise.
21118 (create_specialized_node): Likewise.
21119 (perhaps_add_new_callers): Likewise.
21120 (decide_about_value): Likewise.
21121 (decide_whether_version_node): Likewise.
21122 (identify_dead_nodes): Likewise.
21123 (ipcp_store_bits_results): Likewise.
21124 * ipa-devirt.c (dump_targets): Likewise.
21125 (ipa_devirt): Likewise.
21126 * ipa-icf.c (sem_item::dump): Likewise.
21127 (sem_function::equals): Likewise.
21128 (sem_variable::equals): Likewise.
21129 (sem_item_optimizer::read_section): Likewise.
21130 (sem_item_optimizer::execute): Likewise.
21131 (congruence_class::dump): Likewise.
21132 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
21133 (dump_inline_summary): Likewise.
21134 (estimate_node_size_and_time): Likewise.
21135 (inline_analyze_function): Likewise.
21136 * ipa-inline-transform.c (inline_call): Likewise.
21137 * ipa-inline.c (report_inline_failed_reason): Likewise.
21138 (want_early_inline_function_p): Likewise.
21139 (edge_badness): Likewise.
21140 (update_edge_key): Likewise.
21141 (inline_small_functions): Likewise.
21142 * ipa-profile.c (ipa_profile): Likewise.
21143 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
21144 (ipa_make_edge_direct_to_target): Likewise.
21145 (remove_described_reference): Likewise.
21146 (ipa_impossible_devirt_target): Likewise.
21147 (propagate_controlled_uses): Likewise.
21148 (ipa_print_node_params): Likewise.
21149 (ipcp_transform_function): Likewise.
21150 * ipa-pure-const.c (pure_const_read_summary): Likewise.
21151 (propagate_pure_const): Likewise.
21152 * ipa-reference.c (generate_summary): Likewise.
21153 (read_write_all_from_decl): Likewise.
21154 (propagate): Likewise.
21155 (ipa_reference_read_optimization_summary): Likewise.
21156 * ipa-utils.c (ipa_merge_profiles): Likewise.
21157 * ipa.c (walk_polymorphic_call_targets): Likewise.
21158 (symbol_table::remove_unreachable_nodes): Likewise.
21159 (ipa_single_use): Likewise.
21160 * passes.c (execute_todo): Likewise.
21161 * predict.c (drop_profile): Likewise.
21162 * symtab.c (symtab_node::get_dump_name): New function.
21163 (symtab_node::dump_name): Likewise.
21164 (symtab_node::dump_asm_name): Likewise.
21165 (symtab_node::dump_references): Likewise.
21166 (symtab_node::dump_referring): Likewise.
21167 (symtab_node::dump_base): Likewise.
21168 (symtab_node::debug_symtab): Likewise.
21169 * tree-sra.c (convert_callers_for_node): Likewise.
21170 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
21171 * value-prof.c (init_node_map): Likewise.
21173 2017-05-23 Martin Liska <mliska@suse.cz>
21175 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
21176 and symtab_node::debug_symtab to symbol_table::debug.
21177 * cgraphunit.c (analyze_functions): Use the renamed function.
21178 (symbol_table::compile): Likewise.
21179 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
21180 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
21181 * passes.c (execute_todo): Likewise.
21182 * symtab.c (symbol_table::dump): New function.
21183 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
21185 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21187 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
21188 that nonconst implies exec.
21190 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21192 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21193 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21194 (inline_edge_summary_vec): Turn into ...
21195 (ipa_call_summaries): ... this one.
21196 (redirect_to_unreachable, edge_set_predicate,
21197 evaluate_properties_for_edge, inline_summary_alloc,
21198 reset_ipa_call_summary, reset_inline_summary,
21199 inline_summary_t::duplicate): Update.
21200 (inline_edge_duplication_hook): Turn to ...
21201 (ipa_call_summary_t::duplicate): ... this one.
21202 (inline_edge_removal_hook): Turn to ...
21203 (ipa_call_summary_t::remove): ... this one.
21204 (dump_inline_edge_summary): Turn to ...
21205 (dump_ipa_call_summary): ... this one.
21206 (estimate_function_body_sizes): Update.
21207 (inline_update_callee_summaries): Update.
21208 (remap_edge_change_prob): Update.
21209 (remap_edge_summaries): Update.
21210 (inline_merge_summary): Update.
21211 (do_estimate_edge_time): Update.
21212 (inline_generate_summary): Update.
21213 (inline_read_section): Update.
21214 (inline_read_summary): Update.
21215 (inline_free_summary): Update.
21216 * ipa-inline.c (can_inline_edge_p): Update.
21217 (compute_inlined_call_time): Update.
21218 (want_inline_small_function_p): Update.
21219 (edge_badness): Update.
21220 (early_inliner): Update.
21221 * ipa-inline.h (inline_edge_summary): Turn to ...
21222 (ipa_call_summary): ... this one.
21223 (ipa_call_summary_t): New class.
21224 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21225 (ipa_call_summaries): New.
21226 (inline_edge_summary): Remove.
21227 (estimate_edge_growth): Update.
21228 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21229 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21230 * ipa-split.c (execute_split_functions): Update.
21231 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21233 2017-05-23 Tom de Vries <tom@codesourcery.com>
21235 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21236 attributes): Document rdrand effective target.
21238 2017-05-23 Tom de Vries <tom@codesourcery.com>
21240 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21241 attributes): Sort alphabetically.
21243 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
21245 * config/avr/genmultilib.awk: Use gsub instead of gensub.
21247 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
21250 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
21251 V2DF/V2DI splat into two separate patterns, one that handles
21252 registers, and the other that only handles memory. Drop support
21253 for splatting from a GPR on ISA 2.07 and then splitting the
21254 splat into direct move and splat.
21255 (vsx_splat_<mode>_reg): Likewise.
21256 (vsx_splat_<mode>_mem): Likewise.
21258 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
21260 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
21262 2017-05-22 Jakub Jelinek <jakub@redhat.com>
21264 PR middle-end/80809
21265 * omp-low.c (finish_taskreg_remap): New function.
21266 (finish_taskreg_scan): If unit size of ctx->record_type
21267 is non-constant, unshare the size expression and replace
21268 decls in it with possible outer var refs.
21270 PR middle-end/80809
21271 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
21272 GOVD_SHARED rather than GOVD_PRIVATE with it.
21273 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
21274 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
21276 PR middle-end/80853
21277 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
21278 as last argument to build_outer_var_ref for pointer bases of array
21279 section reductions.
21281 2017-05-19 Martin Sebor <msebor@redhat.com>
21283 * print-tree.c (print_node): Print DECL_READ_P flag.
21285 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21287 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
21288 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
21289 * cgraph.c: Likewise.
21290 * cgraphunit.c: Likewise.
21291 * gengtype.c: Likewise.
21292 * ipa-cp.c: Likewise.
21293 * ipa-devirt.c: Likewise.
21294 * ipa-icf.c: Likewise.
21295 * ipa-predicate.c: Likewise.
21296 * ipa-profile.c: Likewise.
21297 * ipa-prop.c: Likewise.
21298 * ipa-split.c: Likewise.
21300 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
21301 edge_predicate_pool, dump_inline_hints,
21302 inline_summary::account_size_time, redirect_to_unreachable,
21303 edge_set_predicate, set_hint_predicate,
21304 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
21305 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
21306 inline_summary_t::remove, remap_hint_predicate_after_duplication,
21307 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
21308 ipa_call_summary_t::remove, initialize_growth_caches,
21309 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
21310 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
21311 mark_modified, unmodified_parm_1, unmodified_parm,
21312 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
21313 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
21314 compute_bb_predicates, will_be_nonconstant_expr_predicate,
21315 will_be_nonconstant_predicate, record_modified_bb_info,
21316 get_minimal_bb, record_modified, param_change_prob,
21317 phi_result_unknown_predicate, predicate_for_phi_result,
21318 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
21319 estimate_function_body_sizes, compute_inline_parameters,
21320 compute_inline_parameters_for_curren, pass_data_inline_parameters,
21321 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
21322 inline_update_callee_summaries, remap_edge_change_prob,
21323 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
21324 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
21325 inline_analyze_function, inline_summary_t::insert,
21326 inline_generate_summary, read_ipa_call_summary, inline_read_section,
21327 inline_read_summary, write_ipa_call_summary, inline_write_summary,
21328 inline_free_summary): Move to ipa-fnsummary.h
21329 (predicate_t): Remove.
21330 * ipa-fnsummary.c: New file.
21331 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
21332 (enum inline_hints_vals, inline_hints, agg_position_info,
21333 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
21334 inline_summaries, ipa_call_summary, ipa_call_summary_t,
21335 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
21336 dump_inline_summary, dump_inline_hints, inline_generate_summary,
21337 inline_read_summary, inline_write_summary, inline_free_summary,
21338 inline_analyze_function, initialize_inline_failed,
21339 inline_merge_summary, inline_update_overall_summary,
21340 compute_inline_parameters): Move to ipa-fnsummary.h
21341 * ipa-fnsummary.h: New file.
21342 * ipa-inline-transform.h: Include ipa-inline.h.
21343 * ipa-inline.c: LIkewise.
21345 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21347 * ipa-inline.c (edge_badness): Use inlined_time instead of
21348 inline_summaries->get.
21350 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21352 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
21354 2017-05-22 Nathan Sidwell <nathan@acm.org>
21356 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
21357 (fdump-lang): Document 'raw' option.
21358 * dumpfile.h (TDI_tu): Delete.
21359 * dumpfile.c (dump_files): Remove translation-unit.
21360 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
21362 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
21364 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
21365 command option from $(AWK) call.
21366 * config/avr/genmultilib.awk: Simplify and rewrite so that it
21367 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
21368 [FORMAT]: Remove handling of variable.
21369 * config/avr/t-multilib: Regenerate.
21371 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21373 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
21375 (dump_inline_summary): Do not print self_time.
21376 (estimate_function_body_sizes): Do not set self_time.
21377 (compute_inline_parameters): Likewise.
21378 (inline_read_section, inline_write_summary): Do not stream self_time.
21379 * ipa-inline.h (inline_summary): Drop self_time.
21381 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21383 * ipa-inline-analysis.c (account_size_time): Rename to ...
21384 (inline_summary::account_size_time): ... this one.
21385 (reset_ipa_call_summary): Turn to ...
21386 (ipa_call_summary::reset): ... this one.
21387 (reset_inline_summary): Turn to ...
21388 (inline_summary::reset): ... this one.
21389 (inline_summary_t::remove): Update.
21390 (inline_summary_t::duplicate): Update.
21391 (ipa_call_summary_t::remove): Update.
21392 (dump_inline_summary): Update.
21393 (estimate_function_body_sizes): Update.
21394 (compute_inline_parameters): Update.
21395 (estimate_node_size_and_time): Update.
21396 (inline_merge_summary): Update.
21397 (inline_update_overall_summary): Update.
21398 (inline_read_section): Update.
21399 (inline_write_summary): Update.
21400 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
21401 add account_size_time and reset member functions.
21402 (ipa_call_summary): Add reset function.
21403 * ipa-predicate.h (predicate::operator &): Constify.
21405 2017-05-22 Richard Biener <rguenther@suse.de>
21407 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
21409 2017-05-19 Jason Merrill <jason@redhat.com>
21411 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
21413 2017-05-19 Marek Polacek <polacek@redhat.com>
21416 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
21417 TYPE_OVERFLOW_WRAPS checks.
21419 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
21421 * tree-core.h (enum omp_clause_default_kind): Add
21422 "OMP_CLAUSE_DEFAULT_PRESENT".
21423 * tree-pretty-print.c (dump_omp_clause): Handle it.
21424 * gimplify.c (enum gimplify_omp_var_data): Add
21425 "GOVD_MAP_FORCE_PRESENT".
21426 (gimplify_adjust_omp_clauses_1): Map it to
21427 "GOMP_MAP_FORCE_PRESENT".
21428 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
21430 * gimplify.c (oacc_default_clause): Clarify.
21432 2017-05-19 Nathan Sidwell <nathan@acm.org>
21434 LANG_HOOK_REGISTER_DUMPS
21435 * toplev.c (general_init): Call register dump lang hook.
21436 * doc/invoke.texi: Document -fdump-lang option family.
21437 * dumpfile.c (dump_files): Remove class dump here.
21438 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
21439 * dumpfile.h (tree_dump_index): Remove TDI_class.
21440 * langhooks-def.h (lhd_register_dumps): Declare.
21441 (LANG_HOOKS_REGISTER_DUMPS): Define.
21442 (LANG_HOOKS_INITIALIZER): Add it.
21443 * langhooks.c (lhd_register_dumps): Define.
21444 * langhooks.h (struct lang_hooks): Add register_dumps.
21446 2017-05-19 Nathan Sidwell <nathan@acm.org>
21448 * context.h (context::set_passes): New.
21449 * context.c (context::context): Do not create pass manager.
21450 * toplev.c (general_init): Create pass manager here.
21452 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
21454 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
21455 use this splitter if two add or or instructions would also work for
21456 the constant we want to generate.
21458 2017-05-19 Richard Biener <rguenther@suse.de>
21461 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
21462 predicate evaluation.
21464 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21466 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
21468 * ipa-inline.c (want_inline_small_function_p): Do not cast to
21471 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21473 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21474 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21475 (inline_edge_summary_vec): Turn into ...
21476 (ipa_call_summaries): ... this one.
21477 (redirect_to_unreachable, edge_set_predicate,
21478 evaluate_properties_for_edge, inline_summary_alloc,
21479 reset_ipa_call_summary, reset_inline_summary,
21480 inline_summary_t::duplicate): Update.
21481 (inline_edge_duplication_hook): Turn to ...
21482 (ipa_call_summary_t::duplicate): ... this one.
21483 (inline_edge_removal_hook): Turn to ...
21484 (ipa_call_summary_t::remove): ... this one.
21485 (dump_inline_edge_summary): Turn to ...
21486 (dump_ipa_call_summary): ... this one.
21487 (estimate_function_body_sizes): Update.
21488 (inline_update_callee_summaries): Update.
21489 (remap_edge_change_prob): Update.
21490 (remap_edge_summaries): Update.
21491 (inline_merge_summary): Update.
21492 (do_estimate_edge_time): Update.
21493 (inline_generate_summary): Update.
21494 (inline_read_section): Update.
21495 (inline_read_summary): Update.
21496 (inline_free_summary): Update.
21497 * ipa-inline.c (can_inline_edge_p): Update.
21498 (compute_inlined_call_time): Update.
21499 (want_inline_small_function_p): Update.
21500 (edge_badness): Update.
21501 (early_inliner): Update.
21502 * ipa-inline.h (inline_edge_summary): Turn to ...
21503 (ipa_call_summary): ... this one.
21504 (ipa_call_summary_t): New class.
21505 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21506 (ipa_call_summaries): New.
21507 (inline_edge_summary): Remove.
21508 (estimate_edge_growth): Update.
21509 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21510 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21511 * ipa-split.c (execute_split_functions): Update.
21512 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21514 2017-05-19 Richard Biener <rguenther@suse.de>
21516 PR middle-end/80764
21517 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
21519 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
21521 * config/rs6000/rs6000.c (struct machine_function): Add field
21522 fpr_is_wrapped_separately.
21523 (rs6000_get_separate_components): Use 64 components. Handle the
21524 new FPR components.
21525 (rs6000_components_for_bb): Handle the FPR components.
21526 (rs6000_emit_prologue_components): Handle the FPR components.
21527 (rs6000_emit_epilogue_components): Handle the FPR components.
21528 (rs6000_set_handled_components): Handle the FPR components.
21529 (rs6000_emit_prologue): Don't output prologue code for those FPRs
21530 that are already separately shrink-wrapped.
21531 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
21532 that are already separately shrink-wrapped.
21534 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21537 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
21540 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
21541 (define_peephole2 for Altivec d-form load): Add peepholes to catch
21542 cases where the register allocator uses a move and an offsettable
21543 memory operation to/from a FPR register on ISA 2.06/2.07.
21544 (define_peephole2 for Altivec d-form store): Likewise.
21546 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
21549 * config/i386/mmx.md (*mov<mode>_internal): Enable
21550 alternatives 11, 12, 13 and 14 also for 32bit targets.
21551 Remove alternatives 15, 16, 17 and 18.
21552 * config/i386/sse.md (vec_concatv2di): Change
21553 alternative (!x, *y) to (x, ?!*Yn).
21555 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
21557 * dumpfile.h (enum dump_kind): Remove stray comma.
21559 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21561 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
21562 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
21563 predicate::num_conditions
21564 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
21565 (CHANGED): turn into predicate::changed.
21566 (agg_position_info): Move to ipa-predicate.h
21567 (add_condition, predicate::add_clause, predicate::operator &=,
21568 predicate::or_with, predicate::evaluate, predicate::probability,
21569 dump_condition, dump_clause, predicate::dump,
21570 predicate::remap_after_duplication, predicate::remap_after_inlining,
21571 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
21572 (evaluate_conditions_for_known_args): Update.
21573 (set_cond_stmt_execution_predicate): Update.
21574 * ipa-inline.h: Include ipa-predicate.h
21575 (condition, inline_param_summary, conditions, agg_position_info,
21576 predicate): Move to ipa-predicate.h
21577 * ipa-predicate.c: New file.
21578 * ipa-predicate.h: New file.
21580 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
21582 * final.c (leaf_function_p): Check we are not in a sequence.
21584 2017-05-18 Martin Liska <mliska@suse.cz>
21586 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
21587 * dumpfile.c (dump_register): Use new enum dump_kind.
21588 (get_dump_file_name): Likewise.
21589 (dump_enable_all): Likewise.
21590 (dump_switch_p_1): Likewise.
21591 (enable_rtl_dump_file): Remove usage of TDF_RTL.
21592 * dumpfile.h (enum dump_kind): New enum type.
21593 (struct dump_file_info): Create constructor and
21594 format fields and comments.
21595 * passes.c (pass_manager::register_one_dump_file):
21597 * statistics.c (statistics_early_init): Likewise.
21598 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
21599 TDF_TREE with TDF_SLIM.
21600 (gather_memory_references_ref): Likewise.
21602 2017-05-18 Martin Liska <mliska@suse.cz>
21604 * vec.h (struct vnull): Use it.
21606 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21608 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
21609 (true_predicate, false_predicate, true_predicate_p,
21610 false_predicate_p): Remove.
21611 (single_cond_predicate, not_inlined_predicate): Turn to member function
21613 (add_condition): Update.
21614 (add_clause): Turn to...
21615 (predicate::add_clause): ... this one; update; allow passing NULL
21617 (and_predicates): Turn to ...
21618 (predicate::operator &=): ... this one.
21619 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
21620 (or_predicates): Turn to ...
21621 (predicate::or_with): ... this one.
21622 (evaluate_predicate): Turn to ...
21623 (predicate::evaluate): ... this one.
21624 (predicate_probability): Turn to ...
21625 (predicate::probability): ... this one.
21626 (dump_condition): Update.
21627 (dump_predicate): Turn to ...
21628 (predicate::dump): ... this one.
21629 (account_size_time): Update.
21630 (edge_set_predicate): Update.
21631 (set_hint_predicate): UPdate.
21632 (evaluate_conditions_for_known_args): Update.
21633 (evaluate_properties_for_edge): Update.
21634 (remap_predicate_after_duplication): Turn to...
21635 (predicate::remap_after_duplication): ... this one.
21636 (remap_hint_predicate_after_duplication): Update.
21637 (inline_summary_t::duplicate): UPdate.
21638 (dump_inline_edge_summary): Update.
21639 (dump_inline_summary): Update.
21640 (set_cond_stmt_execution_predicate): Update.
21641 (set_switch_stmt_execution_predicate): Update.
21642 (compute_bb_predicates): Update.
21643 (will_be_nonconstant_expr_predicate): Update.
21644 (will_be_nonconstant_predicate): Update.
21645 (phi_result_unknown_predicate): Update.
21646 (predicate_for_phi_result): Update.
21647 (array_index_predicate): Update.
21648 (estimate_function_body_sizes): Update.
21649 (estimate_node_size_and_time): Update.
21650 (estimate_ipcp_clone_size_and_time): Update.
21651 (remap_predicate): Rename to ...
21652 (predicate::remap_after_inlining): ... this one.
21653 (remap_hint_predicate): Update.
21654 (inline_merge_summary): Update.
21655 (inline_update_overall_summary): Update.
21656 (estimate_size_after_inlining): Update.
21657 (read_predicate): Rename to ...
21658 (predicate::stream_in): ... this one.
21659 (read_inline_edge_summary): Update.
21660 (write_predicate): Rename to ...
21661 (predicate::stream_out): ... this one.
21662 (write_inline_edge_summary): Update.
21663 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
21664 (clause_t): Turn to uint32_t
21665 (predicate): Turn to class; implement constructor and operators
21667 (size_time_entry): Update.
21668 (inline_summary): Update.
21669 (inline_edge_summary): Update.
21671 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
21673 * fold-const.c (fold_binary_loc): Move transformation...
21674 * match.pd (C - X CMP X): ... here.
21676 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
21678 * config/sparc/sparc.c (sparc_option_override): Set function
21679 alignment for -mcpu=niagara7 to 64 to match the I$ line.
21680 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
21682 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
21684 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
21686 2017-05-18 Marek Polacek <polacek@redhat.com>
21689 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
21690 (pass_ubsan::execute): Call gimple_assign_single_p instead of
21691 gimple_assign_load_p.
21693 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
21695 PR middle-end/80692
21696 * real.c (do_compare): Give decimal_do_compare preference over
21697 comparing just the signs.
21699 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
21701 * doc/md.texi (Canonicalization of Instructions): Describe the
21702 canonical form of instructions that inherently set a condition
21705 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
21707 PR middle-end/80775
21708 * tree-cfg.c: Move deletion of unreachable case statements to after
21709 the merging of consecutive case labels.
21711 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
21713 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
21714 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
21715 restoring of callee-saved registers.
21717 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
21719 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
21720 * config/visium/visium.c (single_set_and_flags): Likewise.
21721 * config/visium/visium.md (Substitutions): Likewise.
21723 2017-05-17 Martin Liska <mliska@suse.cz>
21725 * cfg.c: Introduce dump_flags_t type and
21726 use it instead of int type.
21728 * cfghooks.c: Likewise.
21729 * cfghooks.h (struct cfg_hooks): Likewise.
21730 * cfgrtl.c: Likewise.
21731 * cfgrtl.h: Likewise.
21732 * cgraph.c (cgraph_node::get_body): Likewise.
21733 * coretypes.h: Likewise.
21734 * domwalk.c: Likewise.
21735 * domwalk.h: Likewise.
21736 * dumpfile.c (struct dump_option_value_info): Likewise.
21737 (dump_enable_all): Likewise.
21738 (dump_switch_p_1): Likewise.
21739 (opt_info_switch_p): Likewise.
21740 * dumpfile.h (enum tree_dump_index): Likewise.
21741 (struct dump_file_info): Likewise.
21742 * genemit.c: Likewise.
21743 * generic-match-head.c: Likewise.
21744 * gengtype.c (open_base_files): Likewise.
21745 * gimple-pretty-print.c: Likewise.
21746 * gimple-pretty-print.h: Likewise.
21747 * graph.c (print_graph_cfg): Likewise.
21748 * graphite-scop-detection.c (dot_all_sese): Likewise.
21749 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
21750 * loop-unroll.c (report_unroll): Likewise.
21751 * passes.c (pass_manager::register_one_dump_file): Likewise.
21752 * print-tree.c: Likewise.
21753 * statistics.c: Likewise.
21754 * tree-cfg.c: Likewise.
21755 * tree-cfg.h: Likewise.
21756 * tree-dfa.c: Likewise.
21757 * tree-dfa.h: Likewise.
21758 * tree-dump.c (dump_function): Likewise.
21759 * tree-dump.h (struct dump_info): Likewise.
21760 * tree-pretty-print.c: Likewise.
21761 * tree-pretty-print.h: Likewise.
21762 * tree-ssa-live.c: Likewise.
21763 * tree-ssa-live.h: Likewise.
21764 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
21765 * tree-vect-loop.c: Likewise.
21766 * tree-vect-slp.c: Likewise.
21768 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
21769 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21771 PR tree-optimization/80457
21772 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
21773 of all arguments to a statement as scalar_to_vec operations.
21774 (vectorizable_call): Adjust call to vect_model_simple_cost for
21776 (vectorizable_conversion): Likewise.
21777 (vectorizable_assignment): Likewise.
21778 (vectorizable_shift): Likewise.
21779 (vectorizable_operation): Likewise.
21780 (vectorizable_comparison): Likewise.
21781 (vect_is_simple_cond): Record the def types for operands.
21782 (vectorizable_condition): Likewise, call vect_model_simple_cost.
21783 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
21784 for statement argument count.
21786 2017-05-16 Carl Love <cel@us.ibm.com>
21788 * config/rs6000/rs6000-c: Add support for built-in functions
21789 vector unsigned long long vec_bperm (vector unsigned long long,
21790 vector unsigned char)
21791 vector signed long long vec_mule (vector signed int,
21793 vector unsigned long long vec_mule (vector unsigned int,
21794 vector unsigned int)
21795 vector signed long long vec_mulo (vector signed int,
21797 vector unsigned long long vec_mulo (vector unsigned int,
21798 vector unsigned int)
21799 vector signed char vec_sldw (vector signed char,
21800 vector signed char,
21802 vector unsigned char vec_sldw (vector unsigned char,
21803 vector unsigned char,
21805 vector signed short vec_sldw (vector signed short,
21806 vector signed short,
21808 vector unsigned short vec_sldw (vector unsigned short,
21809 vector unsigned short,
21811 vector signed int vec_sldw (vector signed int,
21814 vector unsigned int vec_sldw (vector unsigned int,
21815 vector unsigned int,
21817 vector signed long long vec_sldw (vector signed long long,
21818 vector signed long long,
21820 vector unsigned long long vec_sldw (vector unsigned long long,
21821 vector unsigned long long,
21823 * config/rs6000/rs6000-c: Add support for built-in functions
21824 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
21825 * config/rs6000/altivec.h: Add defintion for vec_sldw.
21826 * doc/extend.texi: Update the built-in documentation for the
21827 new built-in functions.
21829 2017-05-16 Marek Polacek <polacek@redhat.com>
21833 * tree.c (save_expr): Don't fold the expression.
21835 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
21837 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
21838 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
21839 and (?*y,m). Update insn attributes.
21841 2017-05-16 Martin Liska <mliska@suse.cz>
21843 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
21844 flags argument of print_gimple_stmt, print_gimple_expr,
21845 print_generic_stmt and print_generic_expr.
21846 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
21847 * coretypes.h: Likewise.
21848 * except.c (dump_eh_tree): Likewise.
21849 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
21850 * gimple-pretty-print.h: Likewise.
21851 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
21852 (backprop::push_to_worklist): Likewise.
21853 (backprop::pop_from_worklist): Likewise.
21854 (backprop::process_use): Likewise.
21855 (backprop::intersect_uses): Likewise.
21856 (note_replacement): Likewise.
21857 * gimple-ssa-store-merging.c
21858 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
21859 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
21860 (pass_store_merging::execute): Likewise.
21861 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
21862 (ssa_base_cand_dump_callback): Likewise.
21863 (dump_incr_vec): Likewise.
21864 (replace_refs): Likewise.
21865 (replace_mult_candidate): Likewise.
21866 (create_add_on_incoming_edge): Likewise.
21867 (create_phi_basis): Likewise.
21868 (insert_initializers): Likewise.
21869 (all_phi_incrs_profitable): Likewise.
21870 (introduce_cast_before_cand): Likewise.
21871 (replace_one_candidate): Likewise.
21872 * gimplify.c (gimplify_expr): Likewise.
21873 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
21874 (set_rename): Likewise.
21875 (rename_uses): Likewise.
21876 (copy_loop_phi_nodes): Likewise.
21877 (add_close_phis_to_merge_points): Likewise.
21878 (copy_loop_close_phi_args): Likewise.
21879 (copy_cond_phi_args): Likewise.
21880 (graphite_copy_stmts_from_block): Likewise.
21881 (translate_pending_phi_nodes): Likewise.
21882 * graphite-poly.c (print_pdr): Likewise.
21883 (dump_gbb_cases): Likewise.
21884 (dump_gbb_conditions): Likewise.
21885 (print_scop_params): Likewise.
21886 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
21887 (build_cross_bb_scalars_use): Likewise.
21888 (gather_bbs::before_dom_children): Likewise.
21889 * hsa-dump.c (dump_hsa_immed): Likewise.
21890 * ipa-cp.c (print_ipcp_constant_value): Likewise.
21891 (get_replacement_map): Likewise.
21892 * ipa-inline-analysis.c (dump_condition): Likewise.
21893 (estimate_function_body_sizes): Likewise.
21894 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
21895 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
21896 * ipa-prop.c (ipa_dump_param): Likewise.
21897 (ipa_print_node_jump_functions_for_edge): Likewise.
21898 (ipa_modify_call_arguments): Likewise.
21899 (ipa_modify_expr): Likewise.
21900 (ipa_dump_param_adjustments): Likewise.
21901 (ipa_dump_agg_replacement_values): Likewise.
21902 (ipcp_modif_dom_walker::before_dom_children): Likewise.
21903 * ipa-pure-const.c (check_stmt): Likewise.
21904 (pass_nothrow::execute): Likewise.
21905 * ipa-split.c (execute_split_functions): Likewise.
21906 * omp-offload.c (dump_oacc_loop_part): Likewise.
21907 (dump_oacc_loop): Likewise.
21908 * trans-mem.c (tm_log_emit): Likewise.
21909 (tm_memopt_accumulate_memops): Likewise.
21910 (dump_tm_memopt_set): Likewise.
21911 (dump_tm_memopt_transform): Likewise.
21912 * tree-cfg.c (gimple_verify_flow_info): Likewise.
21913 (print_loop): Likewise.
21914 * tree-chkp-opt.c (chkp_print_addr): Likewise.
21915 (chkp_gather_checks_info): Likewise.
21916 (chkp_get_check_result): Likewise.
21917 (chkp_remove_check_if_pass): Likewise.
21918 (chkp_use_outer_bounds_if_possible): Likewise.
21919 (chkp_reduce_bounds_lifetime): Likewise.
21920 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
21921 (chkp_mark_completed_bounds): Likewise.
21922 (chkp_register_incomplete_bounds): Likewise.
21923 (chkp_mark_invalid_bounds): Likewise.
21924 (chkp_maybe_copy_and_register_bounds): Likewise.
21925 (chkp_build_returned_bound): Likewise.
21926 (chkp_get_bound_for_parm): Likewise.
21927 (chkp_build_bndldx): Likewise.
21928 (chkp_get_bounds_by_definition): Likewise.
21929 (chkp_generate_extern_var_bounds): Likewise.
21930 (chkp_get_bounds_for_decl_addr): Likewise.
21931 * tree-chrec.c (chrec_apply): Likewise.
21932 * tree-data-ref.c (dump_data_reference): Likewise.
21933 (dump_subscript): Likewise.
21934 (dump_data_dependence_relation): Likewise.
21935 (analyze_overlapping_iterations): Likewise.
21936 * tree-inline.c (expand_call_inline): Likewise.
21937 (tree_function_versioning): Likewise.
21938 * tree-into-ssa.c (dump_defs_stack): Likewise.
21939 (dump_currdefs): Likewise.
21940 (dump_names_replaced_by): Likewise.
21941 (dump_update_ssa): Likewise.
21942 (update_ssa): Likewise.
21943 * tree-object-size.c (pass_object_sizes::execute): Likewise.
21944 * tree-parloops.c (build_new_reduction): Likewise.
21945 (try_create_reduction_list): Likewise.
21946 (ref_conflicts_with_region): Likewise.
21947 (oacc_entry_exit_ok_1): Likewise.
21948 (oacc_entry_exit_single_gang): Likewise.
21949 * tree-pretty-print.h: Likewise.
21950 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
21951 (get_scalar_evolution): Likewise.
21952 (add_to_evolution): Likewise.
21953 (get_loop_exit_condition): Likewise.
21954 (analyze_evolution_in_loop): Likewise.
21955 (analyze_initial_condition): Likewise.
21956 (analyze_scalar_evolution): Likewise.
21957 (instantiate_scev): Likewise.
21958 (number_of_latch_executions): Likewise.
21959 (gather_chrec_stats): Likewise.
21960 (final_value_replacement_loop): Likewise.
21961 (scev_const_prop): Likewise.
21962 * tree-sra.c (dump_access): Likewise.
21963 (disqualify_candidate): Likewise.
21964 (create_access): Likewise.
21965 (reject): Likewise.
21966 (maybe_add_sra_candidate): Likewise.
21967 (create_access_replacement): Likewise.
21968 (analyze_access_subtree): Likewise.
21969 (analyze_all_variable_accesses): Likewise.
21970 (sra_modify_assign): Likewise.
21971 (initialize_constant_pool_replacements): Likewise.
21972 (find_param_candidates): Likewise.
21973 (decide_one_param_reduction): Likewise.
21974 (replace_removed_params_ssa_names): Likewise.
21975 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
21976 * tree-ssa-copy.c (dump_copy_of): Likewise.
21977 (copy_prop_visit_cond_stmt): Likewise.
21978 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
21979 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
21980 (record_equivalences_from_stmt): Likewise.
21981 * tree-ssa-dse.c (compute_trims): Likewise.
21982 (delete_dead_call): Likewise.
21983 (delete_dead_assignment): Likewise.
21984 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
21985 (forward_propagate_into_cond): Likewise.
21986 (pass_forwprop::execute): Likewise.
21987 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
21988 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
21990 (move_computations_worker): Likewise.
21991 (execute_sm): Likewise.
21992 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
21993 (remove_exits_and_undefined_stmts): Likewise.
21994 (remove_redundant_iv_tests): Likewise.
21995 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
21996 (adjust_iv_update_pos): Likewise.
21997 * tree-ssa-math-opts.c (bswap_replace): Likewise.
21998 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
21999 (value_replacement): Likewise.
22000 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
22001 * tree-ssa-pre.c (print_pre_expr): Likewise.
22002 (get_representative_for): Likewise.
22003 (create_expression_by_pieces): Likewise.
22004 (insert_into_preds_of_block): Likewise.
22005 (eliminate_insert): Likewise.
22006 (eliminate_dom_walker::before_dom_children): Likewise.
22007 (eliminate): Likewise.
22008 (remove_dead_inserted_code): Likewise.
22009 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
22010 * tree-ssa-reassoc.c (get_rank): Likewise.
22011 (eliminate_duplicate_pair): Likewise.
22012 (eliminate_plus_minus_pair): Likewise.
22013 (eliminate_not_pairs): Likewise.
22014 (undistribute_ops_list): Likewise.
22015 (eliminate_redundant_comparison): Likewise.
22016 (update_range_test): Likewise.
22017 (optimize_range_tests_var_bound): Likewise.
22018 (optimize_vec_cond_expr): Likewise.
22019 (rewrite_expr_tree): Likewise.
22020 (rewrite_expr_tree_parallel): Likewise.
22021 (linearize_expr): Likewise.
22022 (break_up_subtract): Likewise.
22023 (linearize_expr_tree): Likewise.
22024 (attempt_builtin_powi): Likewise.
22025 (attempt_builtin_copysign): Likewise.
22026 (transform_stmt_to_copy): Likewise.
22027 (transform_stmt_to_multiply): Likewise.
22028 (dump_ops_vector): Likewise.
22029 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
22030 (print_scc): Likewise.
22031 (set_ssa_val_to): Likewise.
22032 (visit_reference_op_store): Likewise.
22033 (visit_use): Likewise.
22034 (sccvn_dom_walker::before_dom_children): Likewise.
22035 (run_scc_vn): Likewise.
22036 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
22038 (expr_hash_elt::print): Likewise.
22039 (const_and_copies::pop_to_marker): Likewise.
22040 (const_and_copies::record_const_or_copy_raw): Likewise.
22041 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
22042 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
22043 (dump_predicates): Likewise.
22044 (find_uninit_use): Likewise.
22045 (warn_uninitialized_phi): Likewise.
22046 (pass_late_warn_uninitialized::execute): Likewise.
22047 * tree-ssa.c (verify_vssa): Likewise.
22048 (verify_ssa): Likewise.
22049 (maybe_optimize_var): Likewise.
22050 * tree-vrp.c (dump_value_range): Likewise.
22051 (dump_all_value_ranges): Likewise.
22052 (dump_asserts_for): Likewise.
22053 (register_edge_assert_for_2): Likewise.
22054 (vrp_visit_cond_stmt): Likewise.
22055 (vrp_visit_switch_stmt): Likewise.
22056 (vrp_visit_stmt): Likewise.
22057 (vrp_visit_phi_node): Likewise.
22058 (simplify_cond_using_ranges_1): Likewise.
22059 (fold_predicate_in): Likewise.
22060 (evrp_dom_walker::before_dom_children): Likewise.
22061 (evrp_dom_walker::push_value_range): Likewise.
22062 (evrp_dom_walker::pop_value_range): Likewise.
22063 (execute_early_vrp): Likewise.
22065 2017-05-16 Richard Biener <rguenther@suse.de>
22067 * dwarf2out.c (loc_list_from_tree_1): Do not create
22068 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
22070 2017-05-16 Richard Biener <rguenther@suse.de>
22072 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
22074 (note_variable_value_in_expr): If we resolved the decl ref
22075 do not push to the stack.
22077 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
22079 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
22080 operations in fast-math mode.
22081 (vaddq_f16): Likewise.
22082 (vmul_f16): Likewise.
22083 (vmulq_f16): Likewise.
22084 (vsub_f16): Likewise.
22085 (vsubq_f16): Likewise.
22086 * config/arm/neon.md (add<mode>3): New.
22088 (fma:<VH:mode>3): New. Also remove outdated comment.
22091 2017-05-16 Martin Liska <mliska@suse.cz>
22095 * ipa-devirt.c (warn_types_mismatch): Fix typo.
22096 (odr_types_equivalent_p): Likewise.
22098 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
22100 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
22102 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
22105 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
22106 non-interunit SSE move alternatives with '?'.
22107 (zero-extendsidi peephole2): New peephole to skip intermediate
22108 general register in SSE zero-extend sequence.
22110 2017-05-15 Jeff Law <law@redhat.com>
22112 * reorg.c (relax_delay_slots): Create a new variable to hold
22113 the temporary target rather than clobbering TARGET_LABEL.
22115 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
22116 missing argument to extract_bit_field call.
22117 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
22119 2017-05-15 Martin Liska <mliska@suse.cz>
22122 * gcc.c (process_command): Do not allow empty argument of -o option.
22124 2017-05-15 Renlin Li <renlin.li@arm.com>
22126 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
22127 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
22128 * config/aarch64/constraints.md (Usf): Add long call check.
22129 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
22130 (call_value): Likewise.
22131 (sibcall): Likewise.
22132 (sibcall_value): Likewise.
22134 (call_value_insn): New.
22135 (sibcall_insn): Update rtx pattern.
22136 (sibcall_value_insn): Likewise.
22137 (call_internal): Remove.
22138 (call_value_internal): Likewise.
22139 (sibcall_internal): Likewise.
22140 (sibcall_value_internal): Likewise.
22141 (call_reg): Likewise.
22142 (call_symbol): Likewise.
22143 (call_value_reg): Likewise.
22144 (call_value_symbol): Likewise.
22146 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
22149 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
22151 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
22153 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
22154 compatible with CCGOCmode and with CCZmode.
22156 2017-05-14 Martin Sebor <msebor@redhat.com>
22158 PR middle-end/77671
22159 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
22160 (gimple_fold_builtin_snprintf): Same.
22161 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
22162 (gimple_fold_builtin_snprintf): Same.
22163 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
22164 of character types.
22165 (is_call_safe): New function.
22166 (try_substitute_return_value): Call it.
22167 (try_simplify_call): New function.
22168 (pass_sprintf_length::handle_gimple_call): Call it.
22170 2017-05-14 Martin Sebor <msebor@redhat.com>
22172 PR middle-end/80669
22173 * builtins.c (expand_builtin_stpncpy): Simplify.
22175 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
22177 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
22178 * config/i386/i386.h
22179 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
22180 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
22181 (struct machine_function): Add new members call_ms2sysv,
22182 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
22183 (struct machine_frame_state): New fields sp_realigned and
22184 sp_realigned_offset.
22185 * config/i386/i386.c
22186 (enum xlogue_stub): New enum.
22187 (enum xlogue_stub_sets): New enum.
22188 (class xlogue_layout): New class.
22189 (struct ix86_frame): New fields stack_realign_allocate_offset,
22190 stack_realign_offset and outlined_save_offset. Modify comments to
22191 detail stack layout when using out-of-line stubs.
22192 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
22193 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
22194 -mcall-ms2sysv-xlogues.
22195 (stub_managed_regs): New static variable.
22196 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
22197 registers managed by out-of-line stub.
22198 (disable_call_ms2sysv_xlogues): New function.
22199 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
22200 m->call_ms2sysv when appropriate and compute frame layout for
22202 (sp_valid_at, fp_valid_at): New inline functions.
22203 (choose_basereg): New function.
22204 (choose_baseaddr): Add align parameter, use choose_basereg and modify
22206 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
22207 Use align parameter of choose_baseaddr to generated aligned SSE movs
22209 (pro_epilogue_adjust_stack): Modify to track
22210 machine_frame_state::sp_realigned.
22211 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
22212 (ix86_nsaved_sseregs): Likewise.
22213 (ix86_emit_save_regs): Likewise.
22214 (ix86_emit_save_regs_using_mov): Likewise.
22215 (ix86_emit_save_sse_regs_using_mov): Likewise.
22216 (get_scratch_register_on_entry): Likewise.
22217 (gen_frame_set): New function.
22218 (gen_frame_load): Likewise.
22219 (gen_frame_store): Likewise.
22220 (emit_outlined_ms2sysv_save): Likewise.
22221 (emit_outlined_ms2sysv_restore): Likewise.
22222 (ix86_expand_prologue): Modify stack re-alignment code and call
22223 emit_outlined_ms2sysv_save when appropriate.
22224 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
22225 parameter rtx_insn *insn, which allows the function to be used to only
22226 generate the notes.
22227 (ix86_expand_epilogue): Modify validity checks of frame and stack
22228 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
22229 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
22230 * config/i386/predicates.md
22231 (save_multiple): New predicate.
22232 (restore_multiple): Likewise.
22233 * config/i386/sse.md
22234 (save_multiple<mode>): New pattern.
22235 (save_multiple_realign<mode>): Likewise.
22236 (restore_multiple<mode>): Likewise.
22237 (restore_multiple_and_return<mode>): Likewise.
22238 (restore_multiple_leave_return<mode>): Likewise.
22239 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
22241 2017-05-14 Julia Koval <julia.koval@intel.com>
22243 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
22244 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
22245 (__builtin_ia32_xsetbv): New builtins.
22246 * config/i386/i386.c (ix86_expand_special_args_builtin):
22248 (ix86_expand_builtin): Special expand for new intrinsics.
22249 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
22250 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
22251 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
22253 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22255 * cfganal.c (inverted_post_order_compute): Change argument type
22257 * cfganal.h (inverted_post_order_compute): Adjust prototype.
22258 * df-core.c (rest_of_handle_df_initialize): Adjust.
22259 (rest_of_handle_df_finish): Likewise.
22260 (df_analyze_1): Likewise.
22261 (df_analyze): Likewise.
22262 (loop_inverted_post_order_compute): Change argument to be a vec *.
22263 (df_analyze_loop): Adjust.
22264 (df_get_n_blocks): Likewise.
22265 (df_get_postorder): Likewise.
22266 * df.h (struct df_d): Change field to be a vec.
22267 * lcm.c (compute_laterin): Adjust.
22268 (compute_available): Likewise.
22269 * lra-lives.c (lra_create_live_ranges_1): Likewise.
22270 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
22271 * tree-ssa-pre.c (compute_antic): Likewise.
22273 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22275 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
22276 (depth_first_search::depth_first_search): Change structure init
22277 function to this constructor.
22278 (depth_first_search::add_bb): Rename function to this member.
22279 (depth_first_search::execute): Likewise.
22280 (flow_dfs_compute_reverse_finish): Adjust.
22282 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22284 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
22285 (longest_simple_path): Likewise.
22286 * shrink-wrap.c (spread_components): Likewise.
22287 (disqualify_problematic_components): Likewise.
22288 (emit_common_heads_for_components): Likewise.
22289 (emit_common_tails_for_components): Likewise.
22290 (insert_prologue_epilogue_for_components): Likewise.
22292 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22294 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
22297 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22299 * df-core.c (df_set_blocks): Start using auto_bitmap.
22300 (df_compact_blocks): Likewise.
22301 * df-problems.c (df_rd_confluence_n): Likewise.
22302 * df-scan.c (df_insn_rescan_all): Likewise.
22303 (df_process_deferred_rescans): Likewise.
22304 (df_update_entry_block_defs): Likewise.
22305 (df_update_exit_block_uses): Likewise.
22306 (df_entry_block_bitmap_verify): Likewise.
22307 (df_exit_block_bitmap_verify): Likewise.
22308 (df_scan_verify): Likewise.
22309 * lra-constraints.c (lra_constraints): Likewise.
22310 (undo_optional_reloads): Likewise.
22311 (lra_undo_inheritance): Likewise.
22312 * lra-remat.c (calculate_gen_cands): Likewise.
22313 (do_remat): Likewise.
22314 * lra-spills.c (assign_spill_hard_regs): Likewise.
22315 (spill_pseudos): Likewise.
22316 * tree-ssa-pre.c (bitmap_set_and): Likewise.
22317 (bitmap_set_subtract_values): Likewise.
22319 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22321 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
22322 management with auto_bitmap.
22323 (fix_inter_tick): Likewise.
22324 (fix_recovery_deps): Likewise.
22325 * ira.c (add_store_equivs): Likewise.
22326 (find_moveable_pseudos): Likewise.
22327 (split_live_ranges_for_shrink_wrap): Likewise.
22328 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
22329 (rtx_reuse_manager::seen_def_p): Likewise.
22330 (rtx_reuse_manager::set_seen_def): Likewise.
22331 * print-rtl.h (class rtx_reuse_manager): Likewise.
22333 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22335 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
22337 (migrate_btr_def): Likewise.
22338 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
22339 * df-core.c (loop_post_order_compute): Likewise.
22340 (loop_inverted_post_order_compute): Likewise.
22341 * hsa-common.h: Likewise.
22342 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
22343 * init-regs.c (initialize_uninitialized_regs): Likewise.
22344 * ipa-inline.c (resolve_noninline_speculation): Likewise.
22345 (inline_small_functions): Likewise.
22346 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
22347 * ira.c (combine_and_move_insns): Likewise.
22348 (build_insn_chain): Likewise.
22349 * loop-invariant.c (find_invariants): Likewise.
22350 * lower-subreg.c (propagate_pseudo_copies): Likewise.
22351 * predict.c (tree_predict_by_opcode): Likewise.
22352 (predict_paths_leading_to): Likewise.
22353 (predict_paths_leading_to_edge): Likewise.
22354 (estimate_loops_at_level): Likewise.
22355 (estimate_loops): Likewise.
22356 * shrink-wrap.c (try_shrink_wrapping): Likewise.
22357 (spread_components): Likewise.
22358 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
22359 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
22360 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
22361 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
22362 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
22363 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
22364 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
22365 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
22366 (mark_threaded_blocks): Likewise.
22367 (thread_through_all_blocks): Likewise.
22368 * tree-ssa.c (verify_ssa): Likewise.
22369 (execute_update_addresses_taken): Likewise.
22370 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
22372 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22374 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
22376 (post_order_compute): Likewise.
22377 (inverted_post_order_compute): Likewise.
22378 (pre_and_rev_post_order_compute_fn): Likewise.
22380 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22382 * genrecog.c (int_set::int_set): Explicitly construct our
22383 auto_vec base class.
22384 * vec.h (auto_vec::auto_vec): New constructor.
22386 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22388 * bitmap.h (class auto_bitmap): New constructor taking
22389 bitmap_obstack * argument.
22391 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22393 * bitmap.h (class auto_bitmap): Change type of m_bits to
22394 bitmap_head, and adjust ctor / dtor and member operators.
22396 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
22398 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
22399 when returned register mode doesn't match original mode.
22401 2017-05-12 Jeff Law <law@redhat.com>
22402 Jakub Jelinek <jakub@redhat.com>
22404 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
22405 we look for cc setter after the compare-elim changes.
22406 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
22407 within the vector to match what compare-elim now expects.
22408 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
22409 (xorsi3_flags, one_cmplsi2_flags): Likewise.
22411 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
22412 after the compare-elim changes.
22413 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
22414 the vector to match what compare-elim now expects.
22415 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
22416 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
22417 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
22418 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
22419 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
22421 * config/visium/visium.c (single_set_and_flags): Fix where
22422 we look for cc setter after the compare-elim changes.
22423 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
22424 with the vector to match what compare-elim now expects.
22425 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
22426 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
22427 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
22428 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
22429 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
22430 (neg<mode>2_insn_set_overflow): Likewise.
22432 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
22434 PR middle-end/79794
22435 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
22436 maybe_expand_insn call, set ops[0].target. If still set after call,
22437 set alt_rtl. Add extra arg to recursive calls.
22438 (extract_bit_field): Add alt_rtl argument. Pass to
22440 * expmed.h (extract_bit_field): Fix prototype.
22441 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
22442 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
22443 to extract_bit_field_calls.
22444 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
22445 Pass alt_rtl to extract_bit_field calls.
22446 * calls.c (store_unaligned_arguments_into_psuedos)
22447 load_register_parameters): Pass extra NULL to extract_bit_field calls.
22448 * optabs.c (maybe_legitimize_operand): Clear op->target when call
22450 * optabs.h (struct expand_operand): Add target bitfield.
22452 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22454 * compare-elim.c (try_eliminate_compare): Canonicalize
22455 operation with embedded compare to
22456 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
22457 (set (reg) (operation)].
22459 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
22461 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22464 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
22465 cost of adding a carry flag for ADC instruction.
22466 [case MINUS]: Ignore the cost of subtracting a carry flag
22467 for SBB instruction.
22469 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
22471 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
22473 * config/rs6000/bmiintrin.h: New file.
22474 * config/rs6000/bmi2intrin.h: New file.
22475 * config/rs6000/x86intrin.h: New file.
22477 2017-05-12 Jeff Law <law@redhat.com>
22479 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
22482 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
22484 PR middle-end/80707
22485 * tree-cfg.c: Remove cfg edges of unreachable case statements.
22487 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22489 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22490 early expansion of vector divide builtins.
22491 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
22492 builtins identified as having unsigned arguments.
22494 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22496 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
22497 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
22498 expansion of vector logical operations (and, andc, or, xor,
22501 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22503 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
22504 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
22506 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22508 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22509 early GIMPLE expansion of vector multiplies.
22511 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22513 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
22514 TARGET_HAVE_MOVT conditional.
22515 (movt splitter): Likewise.
22517 2017-05-12 Richard Biener <rguenther@suse.de>
22519 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
22520 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22521 Fold all stmts not inplace.
22523 2017-05-12 Richard Biener <rguenther@suse.de>
22525 PR tree-optimization/80713
22526 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
22527 inserted_exprs bit for not removed stmts.
22529 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
22531 PR middle-end/69921
22532 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
22533 parallelized" attribute for parallelized OpenACC kernels.
22534 * omp-offload.c (execute_oacc_device_lower): Use it.
22536 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
22537 Set "oacc kernels" attribute.
22538 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
22539 parameter. Adjust all users.
22540 (oacc_fn_attrib_kernels_p): Remove function.
22541 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
22542 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
22543 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
22544 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
22545 assert "oacc kernels" attribute is set.
22547 2017-05-11 Carl Love <cel@us.ibm.com>
22549 * config/rs6000/rs6000-c: Add support for built-in functions
22550 vector unsigned char vec_popcnt (vector signed char)
22551 vector unsigned char vec_popcnt (vector unsigned char)
22552 vector unsigned short vec_popcnt (vector signed short)
22553 vector unsigned short vec_popcnt (vector unsigned short)
22554 vector unsigned int vec_popcnt (vector signed int)
22555 vector unsigned int vec_popcnt (vector unsigned int)
22556 vector unsigned long long vec_popcnt (vector signed long long)
22557 vector unsigned long long vec_popcnt (vector unsigned long long)
22558 vector signed long long vec_slo (vector signed long long,
22559 vector signed char)
22560 vector signed long long vec_slo (vector signed long long,
22561 vector unsigned char)
22562 vector unsigned long long vec_slo (vector unsigned long long,
22563 vector signed char)
22564 vector unsigned long long vec_slo (vector unsigned long long,
22565 vector unsigned char)
22566 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
22567 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
22568 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
22569 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
22570 * doc/extend.texi: Update the built-in documentation file for the
22571 new built-in functions.
22573 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22575 * attribs.h (sorted_attr_string): Move machine independent
22576 functions for target clone support from the i386 port to common
22577 code. Rename ix86_function_versions to common_function_versions.
22578 Rename make_name to make_unique_name.
22579 (common_function_versions): Likewise.
22580 (make_unique_name): Likewise.
22581 (make_dispatcher_decl): Likewise.
22582 (is_function_default_version): Likewise.
22583 * attribs.c (attr_strcmp): Likewise.
22584 (sorted_attr_string): Likewise.
22585 (common_function_versions): Likewise.
22586 (make_unique_name): Likewise.
22587 (make_dispatcher_decl): Likewise.
22588 (is_function_default_version): Likewise.
22589 * config/i386/i386.c (attr_strcmp): Likewise.
22590 (sorted_attr_string): Likewise.
22591 (ix86_function_versions): Likewise.
22592 (make_name): Likewise.
22593 (make_dispatcher_decl): Likewise.
22594 (is_function_default_version): Likewise.
22595 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
22597 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22600 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22601 Account for direct move costs for vec_construct of integer
22604 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
22607 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
22608 (UNSPEC_STX_ATOMIC): Ditto.
22609 (loaddi_via_sse): New insn.
22610 (storedi_via_sse): Ditto.
22611 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
22612 Update corresponding peephole2 patterns.
22613 (atomic_storedi_fpu): Ditto.
22615 2017-05-11 Julia Koval <julia.koval@intel.com>
22617 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
22618 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
22620 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
22621 (__builtin_ia32_rsqrt14ss_mask): New builtins.
22622 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
22624 2017-05-11 Nathan Sidwell <nathan@acm.org>
22626 * graphite-poly.c: Include dumpfile.h.
22628 * dumpfle.h (dump_function): Declare here ...
22629 * tree-dump.h (dump_function): ... not here.
22630 * dumpfile.c: #include tree-cfg.h.
22631 (dump_function): Move here from ...
22632 * tree-dump.c (dump_function): ... here.
22633 * gimplify.c: #include splay-tree.h, not tree-dump.h.
22634 * graphite-poly.c: Don't include tree-dump.h.
22635 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
22636 * print-tree.c: Likewise.
22637 * stor-layout.c: Likewise.
22638 * tree-nested.c: Likewise.
22640 * dumpfile.c (dump_start): Use TDF_FLAGS.
22641 (dump_enable_all): Fix TDF_KIND check thinko.
22643 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
22645 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22646 array entries to represent two legal parameterizations of the
22647 overloaded __builtin_cmpb function, as represented by the
22648 P6_OV_BUILTIN_CMPB constant.
22649 (altivec_resolve_overloaded_builtin): Add special case handling
22650 for the __builtin_cmpb function, as represented by the
22651 P6_OV_BUILTIN_CMPB constant.
22652 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
22653 (BU_P6_64BIT_2): New macro.
22654 (BU_P6_OVERLOAD_2): New macro
22655 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
22656 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
22657 (CMPB): Add overload support to represent both 32-bit and 64-bit
22658 compare-bytes function.
22659 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
22660 support for TARGET_CMPB.
22661 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
22662 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22663 documentation of the __builtin_cmpb overloaded built-in function.
22665 2017-05-11 Richard Biener <rguenther@suse.de>
22667 PR tree-optimization/80705
22668 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
22669 bases are not vectorizable.
22671 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22673 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
22674 when counting register pressure.
22676 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22678 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
22679 (iv_ca_more_deps): Renamed to ...
22680 (iv_ca_compare_deps): ... this.
22681 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
22683 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22685 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
22687 (determine_group_iv_costs): ... here.
22688 (find_inv_vars_cb): Record inv var if it's not recorded before.
22690 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22692 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
22693 (get_shiftadd_cost): Ditto.
22695 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22697 * tree-ssa-address.c: Include header file.
22698 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
22700 (add_to_parts): Refactor.
22701 (addr_to_parts): New parameter. Update use of move_hint_to_base.
22702 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
22705 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22707 PR tree-optimization/53090
22708 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
22710 (extract_cond_operands): Detect condition with IV on both sides
22711 and return COMP_IV_EXPR_2.
22712 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
22713 (rewrite_use_compare): Simplify by removing call to function
22714 extract_cond_operands.
22716 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22718 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
22719 (extract_cond_operands): Detect condition comparing against non-
22720 invariant bound and return appropriate enum value.
22721 (find_interesting_uses_cond): Update use of extract_cond_operands.
22722 Handle its return value accordingly.
22723 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
22725 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22727 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
22728 nonlinear iv_use computation in loop invariant sensitive way.
22730 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22732 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
22733 (find_iv_candidates): Call relate_compare_use_with_all_cands.
22735 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22737 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
22738 (dump_cand): Support iv_cand.inv_exprs.
22739 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
22741 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
22744 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22746 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
22748 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
22749 as local function. Include necessary header files.
22750 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
22752 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22754 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
22756 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22758 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
22759 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
22760 RSHIFT_EXPR and BIT_NOT_EXPR.
22762 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22764 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
22765 (adjust_setup_cost): New parameter supporting round up adjustment.
22766 (struct address_cost_data): Delete.
22767 (force_expr_to_var_cost): Don't bound cost with spill_cost.
22768 (split_address_cost, ptr_difference_cost): Delete.
22769 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
22770 (struct ainc_cost_data): New struct.
22771 (get_address_cost_ainc): New function.
22772 (get_address_cost, get_computation_cost): Reimplement.
22773 (determine_group_iv_cost_address): Record inv_expr for all uses of
22775 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
22776 (iv_ca_has_deps): Reimplemented to ...
22777 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
22779 (iv_ca_extend): Call iv_ca_more_deps.
22781 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22783 * tree-ssa-address.c (struct mem_address): Move to header file.
22784 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
22785 * tree-ssa-address.h (struct mem_address): Move from C file.
22786 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
22788 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22790 * tree-affine.h (aff_combination_type): New interface.
22791 (aff_combination_zero_p): Remove static.
22792 (aff_combination_const_p): New interface.
22793 (aff_combination_singleton_var_p): New interfaces.
22795 2017-05-11 Richard Biener <rguenther@suse.de>
22797 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22798 Skip unreachable blocks and destinations.
22799 (eliminate): Move stmt removal and fixup ...
22800 (fini_eliminate): ... here. Skip inserted exprs.
22801 (pass_pre::execute): Move fini_pre after fini_eliminate.
22802 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
22803 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
22804 PRE to get rid of dead code that has invalid SSA form and
22805 split critical edges again.
22807 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22809 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22811 2017-05-11 Richard Biener <rguenther@suse.de>
22813 * passes.c (execute_function_todo): Verify loops if they are
22814 said to be up-to-date.
22815 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
22816 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
22818 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
22821 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
22822 handle calling assemble_external ourself.
22825 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
22826 modes with zero size. Enhance comment.
22828 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22830 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
22831 built-ins for vec_xl and vec_xst with short and char pointer
22834 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22836 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
22837 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
22838 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
22839 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
22840 (_mm_maskz_min_round_ss): New intrinsics.
22841 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
22842 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
22843 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
22844 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
22845 (__builtin_ia32_minss_mask_round): New builtins.
22846 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22847 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22848 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
22850 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
22851 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
22853 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
22856 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22858 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
22859 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
22860 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
22861 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
22862 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
22863 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
22864 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
22865 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22866 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22867 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
22868 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
22869 (__builtin_ia32_mulss_mask_round): New builtins.
22870 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22871 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22872 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
22874 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
22875 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22877 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22880 2017-05-10 Julia Koval <julia.koval@intel.com>
22882 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
22883 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
22884 (_mm256_setr_m128i): New intrinsics.
22886 2017-05-10 Julia Koval <julia.koval@intel.com>
22888 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
22889 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
22890 (_mm_maskz_rcp14_ss): New intrinsics.
22891 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
22892 (__builtin_ia32_rcp14ss_mask): New builtins.
22893 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
22895 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
22897 PR tree-optimization/51513
22898 * tree-cfg.c (gimple_seq_unreachable_p): New function.
22899 (assert_unreachable_fallthru_edge_p): Use it.
22900 (group_case_labels_stmt): Likewise.
22901 * tree-cfg.h: Prototype it.
22902 * stmt.c: Include cfghooks.h and tree-cfg.h.
22903 (emit_case_dispatch_table) <gap_label>: New local variable.
22904 Use it to fill dispatch table gaps.
22905 Test for default_label before updating probabilities.
22906 (expand_case) <default_label>: Remove unneeded initialization.
22907 Test for unreachable default case statement and remove its edge.
22908 Set default_label accordingly.
22909 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
22911 2017-05-10 Carl Love <cel@us.ibm.com>
22913 * config/rs6000/rs6000-c: Add support for built-in functions
22914 vector signed char vec_neg (vector signed char)
22915 vector signed short int vec_neg (vector short int)
22916 vector signed int vec_neg (vector signed int)
22917 vector signed long long vec_neg (vector signed long long)
22918 vector float vec_neg (vector float)
22919 vector double vec_neg (vector double)
22920 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
22922 * config/rs6000/altivec.h: Add define for vec_neg
22923 * doc/extend.texi: Update the built-in documentation for the
22924 new built-in functions.
22926 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22928 PR tree-optimization/77644
22929 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
22931 2017-05-10 Nathan Sidwell <nathan@acm.org>
22933 * dumpfile.h (TDI_lang_all): New.
22934 (TDF_KIND): New. Renumber others
22935 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
22937 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
22939 (get_dump_file_name): Adjust suffix generation.
22940 (dump_enable_all): Use TDF_KIND.
22941 * doc/invoke.texi (-fdump-lang-all): Document.
22943 * dumpfile.h: Tabify.
22945 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
22948 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
22949 Move member access before delete.
22951 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
22953 * tree-inline.c (expand_call_inline): Split block at stmt
22956 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
22959 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
22960 are now unused after splitting mov{sf,sd}_hardfloat.
22961 (f32_lr2): Likewise.
22962 (f32_lm): Likewise.
22963 (f32_lm2): Likewise.
22964 (f32_li): Likewise.
22965 (f32_li2): Likewise.
22966 (f32_lv): Likewise.
22967 (f32_sr): Likewise.
22968 (f32_sr2): Likewise.
22969 (f32_sm): Likewise.
22970 (f32_sm2): Likewise.
22971 (f32_si): Likewise.
22972 (f32_si2): Likewise.
22973 (f32_sv): Likewise.
22974 (f32_dm): Likewise.
22975 (f32_vsx): Likewise.
22976 (f32_av): Likewise.
22977 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
22978 For movsf, order stores so the VSX stores occur before the GPR
22979 store which encourages the register allocator to use a traditional
22980 FPR instead of a GPR. For movsd, order the stores so that the GPR
22981 store comes before the VSX stores to allow the power6 to work.
22982 This is due to the power6 not having a 32-bit integer store
22983 instruction from a FPR.
22984 (movsf_hardfloat): Likewise.
22985 (movsd_hardfloat): Likewise.
22987 2017-05-09 Martin Sebor <msebor@redhat.com>
22989 PR translation/80280
22990 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
22993 PR translation/80280
22994 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
22995 data member added in r247778.
22996 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
22998 2017-05-09 Nathan Sidwell <nathan@acm.org>
23000 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
23002 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
23005 2017-05-09 Marek Polacek <polacek@redhat.com>
23007 * doc/invoke.texi: Fix typo.
23009 2017-05-09 Richard Biener <rguenther@suse.de>
23011 * tree-vrp.c (vrp_val_is_max): Adjust comment.
23012 (vrp_val_is_min): Likewise.
23013 (set_value_range_to_value): Likewise.
23014 (set_value_range_to_nonnegative): Likewise.
23015 (gimple_assign_nonzero_p): Likewise.
23016 (gimple_stmt_nonzero_p): Likewise.
23017 (vrp_int_const_binop): Likewise. Remove unreachable case.
23018 (adjust_range_with_scev): Adjust comments.
23019 (compare_range_with_value): Likewise.
23020 (extract_range_from_phi_node): Likewise.
23021 (test_for_singularity): Likewise.
23023 2017-05-09 Richard Biener <rguenther@suse.de>
23025 * tree-vrp.c (get_single_symbol): Add assert that we don't
23026 get overflowed constants as invariant part.
23027 (compare_values_warnv): Add comment before the TREE_NO_WARNING
23028 checks. Use wi::cmp instead of recursing for integer constants.
23029 (compare_values): Just ignore whether we assumed undefined
23030 overflow instead of failing the compare.
23031 (extract_range_for_var_from_comparison_expr): Add comment before the
23032 TREE_NO_WARNING sets.
23033 (test_for_singularity): Likewise.
23034 (extract_range_from_comparison): Do not disable optimization
23035 when we assumed undefined overflow.
23036 (extract_range_basic): Remove init of unused var.
23038 2017-05-09 Richard Biener <rguenther@suse.de>
23040 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
23041 (extract_range_from_multiplicative_op_1): Adjust.
23042 (extract_range_from_binary_expr_1): Use int_const_binop.
23044 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
23047 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
23048 rs6000_store_data_bypass_p in seven define_bypass directives and
23049 in several comments.
23050 * config/rs6000/rs6000-protos.h: Add prototype for
23051 rs6000_store_data_bypass_p function.
23052 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
23053 function implements slightly different (rs6000-specific) semantics
23054 than store_data_bypass_p, returning false rather than aborting
23055 with assertion error when arguments do not satisfy the
23056 requirements of store data bypass.
23057 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
23058 rs6000_store_data_bypass_p.
23060 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
23062 * config/xtensa/xtensa-protos.h
23063 (xtensa_initial_elimination_offset): New declaration.
23064 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
23065 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
23066 macro definition, add case for FRAME_POINTER_REGNUM when
23067 FRAME_GROWS_DOWNWARD.
23068 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
23069 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
23070 xtensa_initial_elimination_offset.
23072 2017-05-08 Nathan Sidwell <nathan@acm.org>
23074 * doc/invoke.texi: Alphabetize -fdump options.
23076 2017-05-08 Martin Sebor <msebor@redhat.com>
23078 PR translation/80280
23079 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
23081 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
23083 * target.def (compute_frame_layout): New optional target hook.
23084 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
23085 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
23086 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
23088 * reload1.c (verify_initial_elim_offsets): Likewise.
23089 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
23090 (use_simple_return_p): Call arm_compute_frame_layout if needed.
23091 (arm_get_frame_offsets): Split up into this ...
23092 (arm_compute_frame_layout): ... and this function.
23094 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
23096 * config/aarch64/constraints.md (Usa): New constraint.
23097 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
23099 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
23101 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
23102 with_multilib_list after it has been checked.
23104 2017-05-08 Richard Biener <rguenther@suse.de>
23106 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
23107 (bitmap_set_subtract_values): Likewise.
23109 2017-05-08 Richard Biener <rguenther@suse.de>
23111 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
23112 (gimple_assign_nonzero): ... this and remove strict_overflow_p
23114 (gimple_stmt_nonzero_warnv_p): Rename to ...
23115 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
23117 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
23118 (extract_range_basic): Adjust, do not disable propagation on
23119 strict overflow sensitive simplification.
23120 (vrp_visit_cond_stmt): Likewise.
23122 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
23124 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
23125 body size unconditionally.
23127 2017-05-07 Jeff Law <law@redhat.com>
23130 2017-05-06 Jeff Law <law@redhat.com>
23131 PR tree-optimization/78496
23132 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23135 PR tree-optimization/78496
23136 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23137 (simplify_stmt_using_ranges): Call it.
23138 (vrp_dom_walker::before_dom_children): Extract equivalences
23139 from an ASSERT_EXPR with an equality comparison against a
23142 2017-05-06 Jeff Law <law@redhat.com>
23144 PR tree-optimization/78496
23145 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23148 PR tree-optimization/78496
23149 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23150 (simplify_stmt_using_ranges): Call it.
23151 (vrp_dom_walker::before_dom_children): Extract equivalences
23152 from an ASSERT_EXPR with an equality comparison against a
23155 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23157 * lra-constraints.c (lra_copy_reg_equiv): New function.
23158 (split_reg): Use it to copy equivalence information from the
23159 original register to the spill register.
23161 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23163 PR rtl-optimization/75964
23164 * simplify-rtx.c (simplify_const_relational_operation): Remove
23165 invalid handling of comparisons of integer ABS.
23167 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
23169 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
23170 initialize to zero.
23171 (init_regs): Remove declaration.
23172 (function_arg_advance_32): Initialize error_p as boolean variable.
23174 2017-05-05 Nathan Sidwell <nathan@acm.org>
23176 * store-motion.c (remove_reachable_equiv_notes): Reformat long
23177 lines. Use for (;;).
23179 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23181 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
23182 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
23183 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
23184 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
23185 VF=2 that require versioning.
23187 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23189 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
23192 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23194 * diagnostic.h (diagnostic_override_option_index): Convert from
23195 macro to inline function.
23197 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23199 * diagnostic.c (last_module_changed_p): New function.
23200 (set_last_module): New function.
23201 (diagnostic_report_current_module): Convert macro usage to
23202 the above functions.
23203 * diagnostic.h (diagnostic_context::last_module): Strengthen
23204 from const line_map * to const line_map_ordinary *.
23205 (diagnostic_last_module_changed): Delete macro.
23206 (diagnostic_set_last_module): Delete macro.
23208 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23210 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
23211 with diagnostic_report_diagnostic.
23212 (diagnostic_n_impl_richloc): Likewise.
23213 * diagnostic.h (report_diagnostic): Delete macro.
23214 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
23215 with diagnostic_report_diagnostic.
23216 * substring-locations.c (format_warning_va): Likewise.
23218 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23220 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
23221 save/restor of format_spec. Move option-printing code to...
23222 (print_option_information): ...this new function, and
23223 reimplement by simply printing to the pretty_printer,
23224 rather than appending to the format string.
23226 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23228 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
23229 handling logic into...
23230 (update_effective_level_from_pragmas): ...this new function.
23232 2017-05-04 Andrew Waterman <andrew@sifive.com>
23234 * config/riscv/riscv.opt (mstrict-align): New option.
23235 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
23236 (SLOW_UNALIGNED_ACCESS): Define.
23237 (riscv_slow_unaligned_access): Declare.
23238 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
23240 (riscv_slow_unaligned_access): New variable.
23241 (rocket_tune_info): Set slow_unaligned_access to true.
23242 (optimize_size_tune_info): Set slow_unaligned_access to false.
23243 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
23244 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
23245 (riscv_option_override): Set riscv_slow_unaligned_access.
23246 * doc/invoke.texi: Add -mstrict-align to RISC-V.
23248 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
23250 * config/riscv/riscv.md: Unify indentation.
23252 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23257 * config/rs6000/rs6000.md (u code attribute): Add FIX and
23259 (extendsi<mode>2): Add support for doing sign extension via
23260 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
23261 don't have ISA 3.0 instructions.
23262 (extendsi<mode>2 splitter): Likewise.
23263 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
23264 generate the normal insns since SImode can now go in vector
23265 registers. Disallow the special UNSPECs needed for previous
23266 machines to hide SImode being used. Add new insns
23267 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
23268 (fix_trunc<mode>si2_stfiwx): Likewise.
23269 (fix_trunc<mode>si2_internal): Likewise.
23270 (fixuns_trunc<mode>si2): Likewise.
23271 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23272 (fctiw<u>z_<mode>_smallint): Likewise.
23273 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
23274 of floating point to 32-bit integer from doing a direct move to
23275 the GPR registers to do a store.
23276 (fctiwz_<mode>): Break long line.
23278 2017-05-05 Bin Cheng <bin.cheng@arm.com>
23280 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
23281 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
23282 (addr_list, addr_offset_valid_p): New.
23283 (split_address_groups): Check offset validity with above function.
23284 (gt-tree-ssa-loop-ivopts.h): Include header file.
23286 2017-05-05 Nathan Sidwell <nathan@acm.org>
23288 * config.gcc (arm*-*-*): Add missing 'fi'.
23290 2017-05-05 Steve Ellcey <sellcey@cavium.com>
23292 * doc/invoke.texi (-fopt-info): Explicitly say order of options
23293 included in -fopt-info does not matter.
23294 * doc/optinfo.texi (-fopt-info): Fix description of default
23295 behavour. Explicitly say order of options included in -fopt-info
23298 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23300 * config.gcc: Allow combinations of aprofile and rmprofile values for
23301 --with-multilib-list.
23302 * config/arm/t-multilib: New file.
23303 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
23304 variables. Remove setting of ISA and floating-point ABI in
23305 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
23306 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
23307 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
23308 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
23310 * config/arm/t-rmprofile: Likewise except for the matches changes.
23311 * doc/install.texi (--with-multilib-list): Document the combination of
23312 aprofile and rmprofile values and warn about pitfalls in doing that.
23314 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23316 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
23317 (movdi_aarch64): Likewise.
23319 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23321 PR tree-optimization/80632
23322 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
23324 (build_arrays): Initialize it for virtual phis.
23325 (fix_phi_nodes): Use it for virtual phis.
23327 PR tree-optimization/80558
23328 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
23329 [x, y] op z into [x op, y op z] for op & or | if conditions
23332 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
23333 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23336 * config/arm/arm.md (use_literal_pool): Remove.
23337 (64-bit immediate split): No longer takes cost into consideration
23338 if arm_disable_literal_pool is enabled.
23339 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
23340 used when arm_disable_literal_pool is enabled.
23341 (arm_max_const_double_inline_cost): Remove use of
23342 arm_disable_literal_pool.
23343 (push_minipool_fix): Add assert.
23344 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
23345 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
23346 (no_literal_pool_sf_immediate): New.
23348 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23350 PR tree-optimization/80613
23351 * tree-ssa-dce.c (propagate_necessity): Remove cases for
23352 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
23354 2017-05-05 Richard Biener <rguenther@suse.de>
23356 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
23358 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
23360 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
23361 of this flag from insn conditions due to removal from r247495.
23363 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23365 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
23367 (arm_early_store_addr_dep_ptr): Likewise.
23368 * config/arm/aarch-common-protos.h
23369 (arm_early_load_addr_dep_ptr): Add prototype.
23370 (arm_early_store_addr_dep_ptr): Likewise.
23371 * config/arm/cortex-a53.md: Add new bypasses.
23373 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23375 * tree.c (next_type_uid): Change type to unsigned.
23376 (type_hash_canon): Decrement back next_type_uid if
23377 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
23378 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
23381 2017-05-04 Martin Sebor <msebor@redhat.com>
23383 * builtins.c: Fix a trivial typo in a comment.
23385 PR middle-end/79234
23386 * builtins.c (check_sizes): Adjust to handle reading past the end.
23387 Avoid printing excessive upper bound of ranges. Use %E to print
23388 tree nodes instead of converting them to %wu.
23389 (expand_builtin_memchr): New function.
23390 (compute_dest_size): Rename...
23391 (compute_objsize): ...to this.
23392 (expand_builtin_memcpy): Adjust.
23393 (expand_builtin_mempcpy): Adjust.
23394 (expand_builtin_strcat): Adjust.
23395 (expand_builtin_strcpy): Adjust.
23396 (check_strncat_sizes): Adjust.
23397 (expand_builtin_strncat): Adjust.
23398 (expand_builtin_strncpy): Adjust and simplify.
23399 (expand_builtin_memset): Adjust.
23400 (expand_builtin_bzero): Adjust.
23401 (expand_builtin_memcmp): Adjust.
23402 (expand_builtin): Handle memcmp.
23403 (maybe_emit_chk_warning): Check strncat just once.
23405 2017-05-04 Martin Sebor <msebor@redhat.com>
23407 PR preprocessor/79214
23408 PR middle-end/79222
23409 PR middle-end/79223
23410 * builtins.c (check_sizes): Add inlining context and issue
23411 warnings even when -Wno-system-headers is set.
23412 (check_strncat_sizes): Same.
23413 (expand_builtin_strncat): Same.
23414 (expand_builtin_memmove): New function.
23415 (expand_builtin_stpncpy): Same.
23416 (expand_builtin): Handle memmove and stpncpy.
23418 2017-05-04 Bin Cheng <bin.cheng@arm.com>
23420 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
23421 which is not used any more.
23423 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23425 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
23427 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23429 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
23430 (cortexa53_tunings): Likewise.
23431 (cortexa57_tunings): Likewise.
23432 (cortexa72_tunings): Likewise.
23433 (cortexa73_tunings): Likewise.
23435 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23437 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
23438 Set loop alignment to 8.
23440 2017-05-04 Martin Sebor <msebor@redhat.com>
23442 PR translation/80280
23443 * builtins.c (expand_builtin_object_size): Add missing quoting to
23444 %D and like directives.
23445 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
23446 (hsa_type_for_tree_type): Same.
23447 (verify_function_arguments): Same.
23448 * symtab.c (symbol_table::change_decl_assembler_name): Same.
23449 * varasm.c (get_section): Same.
23452 2017-05-04 Martin Sebor <msebor@redhat.com>
23454 PR translation/80280
23455 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
23457 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23459 * config/aarch64/aarch64.c (generic_addrcost_table):
23460 Change HI/TI mode setting.
23462 2017-05-04 Martin Jambor <mjambor@suse.cz>
23464 PR tree-optimization/80622
23465 * tree-sra.c (comes_initialized_p): New function.
23466 (build_accesses_from_assign): Only set write lazily when
23467 comes_initialized_p is false.
23468 (analyze_access_subtree): Use comes_initialized_p.
23469 (propagate_subaccesses_across_link): Assert !comes_initialized_p
23470 instead of testing for PARM_DECL.
23472 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23474 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
23475 constraint on operand 0 to allow more general addressing modes.
23476 Adjust output template.
23477 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
23479 * config/aarch64/aarch64-protos.h
23480 (aarch64_address_valid_for_prefetch_p): Declare prototype.
23481 * config/aarch64/constraints.md (Dp): New address constraint.
23482 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
23485 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
23487 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
23488 update use of estimate_ipcp_clone_size_and_time.
23489 (estimate_local_effects): Update use of
23490 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
23491 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
23492 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
23493 Return nonspecialized time.
23495 2017-05-04 Richard Biener <rguenther@suse.de>
23497 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
23498 for the last VUSE which def dominates the PHI. Directly call
23500 (get_continuation_for_phi_1): Remove.
23502 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
23504 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
23505 to explain the use of truncating division. Cap the number of
23506 iterations to the maximum given by nb_iterations_upper_bound,
23509 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23511 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
23512 * configure: Regenerate.
23513 * config.in: Regenerate.
23514 * config/i386/driver-mingw32.c: new file.
23515 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
23516 * config.host: Link driver-mingw32.o on MinGW host.
23517 * doc/install.texi: Document new --enable-mingw-wildcard configure
23520 2017-05-04 Marek Polacek <polacek@redhat.com>
23522 PR tree-optimization/80612
23523 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
23525 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23526 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
23528 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
23529 (movt splitter): Likewise.
23530 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
23531 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
23532 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
23533 block for Thumb-1 with MOVT.
23534 (thumb2_legitimate_address_p): Move code block ...
23535 (can_avoid_literal_pool_for_label_p): ... into this new function.
23536 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
23538 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
23539 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
23540 "M-profile targets with the MOVT instruction".
23542 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23544 * config/arm/arm-builtins.c (arm_init_builtins): Rename
23545 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
23546 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
23548 2017-05-04 Martin Liska <mliska@suse.cz>
23550 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
23551 variable cond_code.
23553 2017-05-04 Richard Biener <rguenther@suse.de>
23555 * tree.c (array_at_struct_end_p): Handle arrays at struct
23556 end with flexarrays more conservatively. Refactor and treat
23557 arrays of arrays or aggregates more strict. Fix
23558 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
23559 * tree.c (array_at_struct_end_p): Adjust prototype.
23560 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
23561 * gimple-fold.c (get_range_strlen): Likewise.
23562 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
23564 2017-05-04 Richard Biener <rguenther@suse.de>
23566 PR tree-optimization/31130
23567 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
23569 (supports_overflow_infinity): Likewise.
23570 (is_negative_overflow_infinity): Likewise.
23571 (is_positive_overflow_infinity): Likewise.
23572 (is_overflow_infinity): Likewise.
23573 (stmt_overflow_infinity): Likewise.
23574 (overflow_infinity_range_p): Likewise.
23575 (usable_range_p): Remove as always returning true.
23576 (make_overflow_infinity): Remove.
23577 (negative_overflow_infinity): Likewise.
23578 (positive_overflow_infinity): Likewise.
23579 (avoid_overflow_infinity): Likewise.
23580 (set_value_range): Adjust accordingly.
23581 (set_value_range_to_nonnegative): Likewise, remove now unused
23582 overflow_infinity arg.
23583 (vrp_operand_equal_p): Adjust.
23584 (update_value_range): Likewise.
23585 (range_int_cst_singleton_p): Likewise.
23586 (operand_less_p): Likewise.
23587 (compare_values_warnv): Likewise.
23588 (extract_range_for_var_from_comparison_expr): Likewise.
23589 (vrp_int_const_binop): Likewise.
23590 (zero_nonzero_bits_from_vr): Likewise.
23591 (extract_range_from_multiplicative_op_1): Likewise.
23592 (extract_range_from_binary_expr_1): Likewise.
23593 (extract_range_from_unary_expr): Likewise.
23594 (extract_range_from_comparison): Likewise.
23595 (extract_range_basic): Likewise.
23596 (adjust_range_with_scev): Likewise.
23597 (compare_ranges): Likewise.
23598 (compare_range_with_value): Likewise.
23599 (dump_value_range): Likewise.
23600 (test_for_singularity): Likewise, remove strict_overflow_p parameter
23602 (simplify_cond_using_ranges): Adjust.
23604 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23606 * brig-builtins.def: Added a builtin for class_f64.
23607 * builtin-types.def: Added a builtin type needed by class_f64.
23609 2017-05-03 Jason Merrill <jason@redhat.com>
23611 * timevar.def: Add TV_CONSTEXPR.
23613 2017-05-03 David Malcolm <dmalcolm@redhat.com>
23615 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
23617 2017-05-03 Martin Jambor <mjambor@suse.cz>
23619 * ipa-prop.c (ipa_update_after_lto_read): Removed.
23620 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
23621 * ipa-cp.c (ipcp_propagate_stage): Do not call
23622 ipa_update_after_lto_read.
23623 * ipa-inline.c (ipa_inline): Likewise.
23625 2017-05-03 Martin Jambor <mjambor@suse.cz>
23627 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
23628 tag. Added a default constructor and a destructor.
23629 (ipa_edge_args_sum_t): New class;
23630 (ipa_edge_args_sum): Declare.
23631 (ipa_edge_args_vector): Remove declaration.
23632 (IPA_EDGE_REF): Use ipa_edge_args_sum.
23633 (ipa_free_edge_args_substructures): Remove declaration.
23634 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
23635 (ipa_edge_args_info_available_for_edge_p): Likewise.
23636 * ipa-prop.c (ipa_edge_args_vector): Removed.
23637 (edge_removal_hook_holder): Likewise.
23638 (edge_duplication_hook_holder): Likewise.
23639 (ipa_edge_args_sum): New variable.
23640 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
23641 ipa_edge_args_vector.
23642 (ipa_free_edge_args_substructures): Likewise.
23643 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
23644 ipa_edge_args_vector.
23645 (ipa_edge_removal_hook): Turned into method
23646 ipa_edge_args_sum_t::remove.
23647 (ipa_edge_duplication_hook): Turned into method
23648 ipa_edge_args_sum_t::duplicate.
23649 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
23650 registering edge hooks.
23651 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
23652 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
23653 ipa_edge_args_sum instead of ipa_edge_args_vector.
23654 * ipa-profile.c (ipa_profile): Likewise.
23656 2017-05-03 Martin Jambor <mjambor@suse.cz>
23658 * symbol-summary.h (function_summary): New method exists.
23659 (function_summary::symtab_removal): Deallocate through release.
23660 (call_summary): New class.
23661 (gt_ggc_mx): New overload.
23662 (gt_pch_nx): Likewise.
23663 (gt_pch_nx): Likewise.
23665 2017-05-03 Jeff Law <law@redhat.com>
23667 PR tree-optimization/78496
23668 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
23669 from simplify_cond_using_ranges. Split off code to walk
23670 backwards through casts into ...
23671 (simplify_cond_using_ranges_2): New function.
23672 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
23673 (execute_vrp): After identifying jump threads, call
23674 simplify_cond_using_ranges_2.
23676 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
23679 * ipa-inline.h (inline_summary): Add ctor.
23680 (create_ggc): Do not use ggc_cleared_alloc.
23682 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
23683 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23685 * gcc.c (handle_braces): Support escaping in switch matching
23687 * doc/invoke.texi (Spec Files): Document it.
23688 Remove superfluous @code markup in items.
23690 2017-05-03 David Malcolm <dmalcolm@redhat.com>
23692 * diagnostic-show-locus.c (struct column_range): New struct.
23693 (get_affected_columns): New function.
23694 (get_printed_columns): New function.
23695 (struct correction): New struct.
23696 (correction::ensure_capacity): New function.
23697 (correction::ensure_terminated): New function.
23698 (struct line_corrections): New struct.
23699 (line_corrections::~line_corrections): New dtor.
23700 (line_corrections::add_hint): New function.
23701 (layout::print_trailing_fixits): Reimplement in terms of the new
23703 (selftest::test_overlapped_fixit_printing): New function.
23704 (selftest::diagnostic_show_locus_c_tests): Call it.
23706 2017-05-03 Nathan Sidwell <nathan@acm.org>
23708 Canonicalize canonical type hashing
23709 * tree.h (type_hash_canon_hash): Declare.
23710 * tree.c (type_hash_list, attribute_hash_list): Move into
23711 type_hash_canon_hash.
23712 (build_type_attribute_qual_variant): Break out hash code calc into
23713 type_hash_canon_hash.
23714 (type_hash_canon_hash): New. Generic type hash computation.
23715 (build_range_type_1, build_array_type_1, build_function_type,
23716 build_method_type_directly, build_offset_type, build_complex_type,
23717 make_vector_type): Call it.
23719 2017-05-03 Richard Biener <rguenther@suse.de>
23721 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
23722 When all DRs have unknown misaligned do not always peel
23723 when there is a store but apply the same costing model as if
23724 there were only loads.
23726 2017-05-03 Richard Biener <rguenther@suse.de>
23729 PR tree-optimization/80492
23730 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23731 compare_base_decls returning dont-know properly.
23733 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
23735 * config/arm/iterators.md (CCSI): New mode iterator.
23736 (arch): New mode attribute.
23737 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
23738 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
23739 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
23740 code iterator for success result mode.
23741 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
23742 the corresponding new insn generators.
23744 2017-05-03 Bin Cheng <bin.cheng@arm.com>
23747 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23748 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23750 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
23752 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
23753 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
23754 (DDR_A): Wrap DDR argument in brackets.
23755 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
23756 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
23757 (DDR_REVERSED_P): Likewise.
23759 2017-05-03 Jakub Jelinek <jakub@redhat.com>
23761 PR tree-optimization/79472
23762 * tree-switch-conversion.c (struct switch_conv_info): Add
23763 contiguous_range and default_case_nonstandard fields.
23764 (collect_switch_conv_info): Compute contiguous_range and
23765 default_case_nonstandard fields, don't clear final_bb if
23766 contiguous_range and only the default case doesn't have the required
23768 (check_all_empty_except_final): Set default_case_nonstandard instead
23769 of failing if contiguous_range and the default case doesn't have empty
23771 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
23772 and only the default case doesn't have the required constants. Skip
23774 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
23775 if default_case_nonstandard.
23776 (build_constructors): Build constant 1 just once. Assert that default
23777 values aren't inserted in between cases if contiguous_range. Skip
23779 (build_arrays): Skip virtual phis.
23780 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
23781 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
23782 Handle virtual phis.
23783 (gen_inbound_check): Handle default_case_nonstandard case.
23784 (process_switch): Adjust check_final_bb caller. Call
23785 gather_default_values with the first non-default case instead of
23786 default case if default_case_nonstandard.
23788 2017-05-02 Nathan Sidwell <nathan@acm.org>
23790 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
23791 check. Fix formatting.
23793 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
23795 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
23796 errors when comparing specialized and unspecialized times.
23798 2017-05-02 David Malcolm <dmalcolm@redhat.com>
23800 * diagnostic-show-locus.c
23801 (layout::should_print_annotation_line_p): Make private.
23802 (layout::print_annotation_line): Make private.
23803 (layout::annotation_line_showed_range_p): Make private.
23804 (layout::show_ruler): Make private.
23805 (layout::print_source_line): Make private. Pass in line and
23806 line_width, rather than calling location_get_source_line. Drop
23808 (layout::print_leading_fixits): New method.
23809 (layout::print_any_fixits): Rename to...
23810 (layout::print_trailing_fixits): ...this, and make private.
23811 Don't print newline fixits.
23812 (diagnostic_show_locus): Move logic for printing one row into...
23813 (layout::print_line): ...this new function. Move the
23814 location_get_source_line call and error-handling from
23815 print_source_line to here. Call print_leading_fixits, and rename
23816 print_any_fixits to print_trailing_fixits.
23817 (selftest::test_fixit_insert_containing_newline): Update now that
23818 newlines are partially supported.
23819 (selftest::test_fixit_insert_containing_newline_2): New test.
23820 (selftest::test_fixit_replace_containing_newline): Update comments.
23821 (selftest::diagnostic_show_locus_c_tests): Call the new test.
23822 * edit-context.c (class added_line): New class.
23823 (class edited_line): Describe newline handling in comment.
23824 (edited_line::actually_edited_p): New method.
23825 (edited_line::print_content): Delete redundant decl.
23826 (edited_line::m_predecessors): New field.
23827 (edited_file::print_content): Call edited_line::print_content.
23828 (edited_file::print_diff): Update to support newlines.
23829 (edited_file::print_diff_hunk): Likewise.
23830 (edited_file::print_run_of_changed_lines): New function.
23831 (edited_file::print_diff_line): Convert to...
23832 (print_diff_line): ...this.
23833 (edited_file::get_effective_line_count): New function.
23834 (edited_line::edited_line): Initialize new field m_predecessors.
23835 (edited_line::~edited_line): Clean up m_predecessors.
23836 (edited_line::apply_fixit): Handle newlines.
23837 (edited_line::get_effective_line_count): New function.
23838 (edited_line::print_content): New function.
23839 (edited_line::print_diff_lines): New function.
23840 (selftest::test_applying_fixits_insert_containing_newline): New
23842 (selftest::test_applying_fixits_replace_containing_newline): New
23844 (selftest::insert_line): New function.
23845 (selftest::test_applying_fixits_multiple_lines): Add example of
23847 (selftest::edit_context_c_tests): Call the new tests.
23849 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23851 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
23852 parameter cand. Update dump information.
23853 (get_computation_cost): Update uses.
23855 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23857 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
23858 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
23859 (get_computation_at, rewrite_use_address): Update use of
23860 get_computation_aff.
23862 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23864 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
23865 (get_computation): Delete.
23866 (get_computation_cost): Implement like get_computation_cost_at.
23867 Use get_computation_at.
23868 (get_computation_cost_at): Delete.
23869 (rewrite_use_nonlinear_expr): Use get_computation_at.
23870 (rewrite_use_compare, remove_unused_ivs): Ditto.
23872 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23874 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
23876 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23878 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
23879 (ivopts_global_cost_for_size): Rename parameter and update uses.
23880 (iv_ca_recount_cost): Update uses.
23881 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
23882 candidates seperately in n_invs and n_cands.
23883 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
23885 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23887 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
23888 (find_inv_vars_cb): New.
23889 (find_depends): Renamed to ...
23890 (find_inv_vars): ... this.
23891 (add_candidate_1, force_var_cost): Call find_inv_vars.
23892 (split_address_cost, determine_group_iv_cost_cond): Ditto.
23894 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23896 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
23897 inv_vars. Add inv_exprs.
23898 (struct iv_cand): Rename depends_on to inv_vars.
23899 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
23900 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
23901 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
23902 (dump_cand): Dump inv_vars.
23903 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
23904 (record_invariant, find_depends, add_candidate_1): Ditto.
23905 (set_group_iv_cost, force_var_cost): Ditto.
23906 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
23907 (get_computation_cost_at, get_computation_cost): Ditto.
23908 (determine_group_iv_cost_generic): Ditto.
23909 (determine_group_iv_cost_address): Ditto.
23910 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
23911 (determine_group_iv_costs): Ditto.
23912 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
23913 (iv_ca_set_remove_invariants): Renamed to ...
23914 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
23915 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
23916 (iv_ca_set_add_invariants): Renamed to ...
23917 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
23918 (iv_ca_set_cp): Use iv_ca_set_add_invs.
23919 (iv_ca_has_deps): Support inv_vars and inv_exprs.
23920 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
23921 (create_new_ivs): Remove useless dump.
23923 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23925 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
23927 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
23928 (iv_ca_set_no_cp, create_new_iv): Ditto.
23930 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23932 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23934 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23936 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
23937 function tree_check2.
23939 2017-05-02 Martin Liska <mliska@suse.cz>
23941 * doc/gcov.texi: Add missing preposition.
23942 * gcov.c (function_info::function_info): Properly fill up
23943 all member variables.
23945 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23947 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
23949 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23951 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
23953 2017-05-02 Martin Liska <mliska@suse.cz>
23956 * lto-streamer-in.c (lto_read_tree_1): Remove
23957 LTO_STREAMER_DEBUG.
23958 * lto-streamer.c (struct tree_hash_entry): Likewise.
23959 (struct tree_entry_hasher): Likewise.
23960 (tree_entry_hasher::hash): Likewise.
23961 (tree_entry_hasher::equal): Likewise.
23962 (lto_streamer_init): Likewise.
23963 (lto_orig_address_map): Likewise.
23964 (lto_orig_address_get): Likewise.
23965 (lto_orig_address_remove): Likewise.
23966 * lto-streamer.h: Likewise.
23967 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
23968 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
23970 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
23972 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
23973 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
23974 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
23975 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
23976 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
23977 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
23978 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
23979 (mm_maskz_sub_ss): New intrinsics.
23980 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23981 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23982 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
23983 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
23984 (__builtin_ia32_subss_mask_round): New builtins.
23985 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23986 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23987 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
23989 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
23990 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23992 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23995 2017-05-02 Martin Jambor <mjambor@suse.cz>
23997 PR tree-optimization/78687
23998 * tree-sra.c (access): New field parent.
23999 (process_subtree_disqualification): New function.
24000 (disqualify_candidate): Call it.
24001 (build_accesses_from_assign): Reset write flag if creating an
24003 (build_access_subtree): Fill in parent field and also prpagate
24004 down grp_write flag.
24005 (create_artificial_child_access): New parameter set_grp_write, set
24006 grp_write to its value.
24007 (propagate_subaccesses_across_link): Also propagate grp_write flag
24009 (propagate_all_subaccesses): Push the closest parent back to work
24010 queue if add_access_to_work_queue returned true.
24012 2017-05-02 Richard Biener <rguenther@suse.de>
24014 * common.opt (fstrict-overflow): Alias negative to fwrapv.
24015 * doc/invoke.texi (fstrict-overflow): Remove all traces of
24016 -fstrict-overflow documentation.
24017 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
24018 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
24019 flag_strict_overflow.
24020 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
24021 * lto-opts.c (lto_write_options): Do not stream it.
24022 * lto-wrapper.c (merge_and_complain): Do not handle it.
24023 * opts.c (default_options_table): Do not set -fstrict-overflow.
24024 (finish_options): Likewise do not clear it when sanitizing.
24025 * simplify-rtx.c (simplify_const_relational_operation): Do not
24026 test flag_strict_overflow.
24028 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24030 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
24031 using enabled attribute.
24032 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
24033 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
24034 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
24035 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
24036 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
24037 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
24038 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
24039 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
24040 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
24041 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
24043 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
24045 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
24047 2017-05-02 Richard Biener <rguenther@suse.de>
24049 PR tree-optimization/80591
24051 2017-04-10 Richard Biener <rguenther@suse.de>
24053 * tree-ssa-structalias.c (find_func_aliases): Properly handle
24056 2017-05-02 Richard Biener <rguenther@suse.de>
24058 PR tree-optimization/80549
24059 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
24060 (cleanup_tree_cfg_noloop): Create forwarders to known loop
24061 headers if they do not have a preheader.
24063 2017-05-02 Martin Liska <mliska@suse.cz>
24066 * common.opt: Fix typo.
24067 * doc/invoke.texi: Likewise.
24069 2017-05-01 Jan Beulich <jbeulich@suse.com>
24071 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
24072 swapping, add (x,x,m,x,n) alternative.
24074 2017-05-01 Nathan Sidwell <nathan@acm.org>
24076 * calls.c (combine_pending_stack_adjustment_and_call): Remove
24077 unnecessary unadjusted_alignment check.
24079 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
24082 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
24084 * gimplify.c (gimplify_cilk_detach): New function.
24085 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
24086 * tree-core.h: Document EXPR_CILK_SPAWN.
24087 * tree.h (EXPR_CILK_SPAWN): Define.
24089 2017-05-01 David Malcolm <dmalcolm@redhat.com>
24091 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
24092 to use new fixit_hint representation, using the "replace" logic.
24093 (get_line_span_for_fixit_hint): Likewise.
24094 (layout::print_any_fixits): Likewise.
24095 (selftest::test_one_liner_many_fixits): Rename to...
24096 (selftest::test_one_liner_many_fixits_1): ...this, and update
24097 comment and expected output to reflect that the multiple fix-it
24098 hints are now consolidated into one insertion.
24099 (selftest::test_one_liner_many_fixits_2): New test.
24100 (selftest::test_diagnostic_show_locus_one_liner): Update for
24102 (selftest::test_fixit_consolidation): Update for fix-it API
24104 * diagnostic.c (print_parseable_fixits): Likewise.
24105 * edit-context.c (edited_line::m_line_events): Convert from
24106 auto_vec <line_event *> to auto_vec <line_event>.
24107 (class line_event): Convert from abstract base class to a concrete
24108 class, taking over the role of replace_event.
24109 (class insert_event): Delete.
24110 (class replace_event): Rename to class line_event. Convert to
24112 (edit_context::add_fixits): Reimplement.
24113 (edit_context::apply_insert): Delete.
24114 (edit_context::apply_replace): Rename to...
24115 (edit_context::apply_fixit): ...this. Convert to half-open range.
24116 (edited_file::apply_insert): Delete.
24117 (edited_file::apply_replace): Rename to...
24118 (edited_file::apply_fixit): ...this.
24119 (edited_line::~edited_line): Drop deletion of events.
24120 (edited_line::apply_insert): Delete.
24121 (edited_line::apply_replace): Rename to...
24122 (edited_line::apply_fixit): ...this. Convert to half-open range.
24123 Update for change to type of m_line_events.
24124 * edit-context.h (edit_context::apply_insert): Delete.
24125 (edit_context::apply_replace): Rename to...
24126 (edit_context::apply_fixit): ...this.
24128 2017-05-01 Martin Sebor <msebor@redhat.com>
24130 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
24133 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
24136 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
24137 __get_cpuid_max returns 0.
24138 (__get_cpuid_count): Ditto.
24140 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
24142 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
24143 replacement expression is another instance of one of its arguments.
24145 2017-05-01 Jakub Jelinek <jakub@redhat.com>
24148 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
24149 check for stack push/pop autoinc.
24150 * config/i386/i386.c (ix86_agi_dependent): Return false
24151 if the only reason why modified_in_p returned true is that
24152 addr is SP based and set_insn is a push or pop.
24154 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24156 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
24159 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24162 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
24163 (account_size_time): Use two predicates - exec_pred and
24165 (evaluate_conditions_for_known_args): Compute both clause and
24167 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
24168 (inline_summary_t::duplicate): Update.
24169 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
24171 (compute_inline_parameters): Likewise.
24172 (estimate_edge_size_and_time): Update caluclation of time.
24173 (estimate_node_size_and_time): Compute both time and nonspecialized
24175 (estimate_ipcp_clone_size_and_time): Update.
24176 (inline_merge_summary): Update.
24177 (do_estimate_edge_time): Update.
24178 (do_estimate_edge_size): Update.
24179 (do_estimate_edge_hints): Update.
24180 (inline_read_section, inline_write_summary): Stream both new predicates.
24181 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
24183 (compute_inlined_call_time): Cleanup.
24184 (big_speedup_p): Update.
24185 (edge_badness): Update.
24186 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
24187 (size_time_entry): Replace predicate by exec_predicate and
24188 nonconst_predicate.
24189 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
24190 (estimate_edge_time): Return also nonspec_time.
24191 (reset_edge_growth_cache): Update.
24193 2017-04-29 Jakub Jelinek <jakub@redhat.com>
24195 PR rtl-optimization/80491
24196 * ifcvt.c (noce_process_if_block): When looking for x setter
24197 with missing else_bb, don't check only the insn right before
24198 cond_earliest, but look for the last insn that x is modified in
24199 within the same bb.
24201 PR rtl-optimization/80491
24202 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
24204 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
24206 PR tree-optimization/80487
24207 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
24209 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24211 PR tree-optimization/79697
24212 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
24213 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
24214 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
24216 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
24217 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
24219 2017-04-28 Martin Sebor <msebor@redhat.com>
24221 PR tree-optimization/80523
24222 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
24223 (init_target_to_host_charmap, target_to_host, target_strtol10): New
24225 (maybe_warn, format_directive, parse_directive): Use new functions.
24226 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
24228 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
24230 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
24232 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
24234 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
24235 target_header_dir): Set correctly.
24236 * configure: Regenerated.
24237 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24238 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24239 instead of SYSTEM_HEADER_DIR.
24241 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
24243 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
24244 (estimate_local_effects): Likewise.
24245 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
24246 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
24247 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
24248 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
24249 do_estimate_edge_time, estimate_edge_time): Likewise.
24250 * ipa-inline-analysis.c (estimate_node_size_and_time,
24251 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
24252 (estimate_time_after_inlining): Remove.
24254 2017-04-28 Martin Liska <mliska@suse.cz>
24256 * doc/gcov.texi: Enhance documentation of gcov.
24258 2017-04-28 Martin Liska <mliska@suse.cz>
24260 * doc/gcov.texi: Sort options in alphabetic order.
24261 * doc/gcov-dump.texi: Likewise.
24262 * doc/gcov-tool.texi: Likewise.
24263 * gcov.c (print_usage): Likewise.
24264 * gcov-dump.c (print_usage): Likewise.
24265 * gcov-tool.c (print_merge_usage_message): Likewise.
24266 (print_rewrite_usage_message): Likewise.
24267 (print_overlap_usage_message): Likewise.
24269 2017-04-28 Martin Liska <mliska@suse.cz>
24271 PR gcov-profile/53915
24272 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
24274 2017-04-28 Martin Liska <mliska@suse.cz>
24276 PR gcov-profile/79891
24277 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
24278 is marked by compiler as living on a line.
24279 (get_cycles_count): Remove usage of the union.
24280 (output_intermediate_file): Likewise.
24281 (find_source): Fix GNU coding style.
24282 (accumulate_line_counts): Remove old non-all block mode.
24283 (output_lines): Remove usage of the union.
24284 * profile.c (output_location): Include all BBs, even if
24285 belonging to a same line (and file) as a previous BB.
24287 2017-04-28 Martin Liska <mliska@suse.cz>
24289 * gcov.c (process_args): Handle new argument 'w'.
24290 (read_graph_file): Assign ID to BBs.
24291 (output_branch_count): Display BB # if verbose flag is set.
24292 (output_lines): Likewise for arcs.
24293 (print_usage): Add '--verbose' option help.
24294 * doc/gcov.texi: Document --verbose (-w) option.
24296 2017-04-28 Martin Liska <mliska@suse.cz>
24298 * gcov.c (struct block_location_info): New struct.
24299 (process_file): Fill up the new structure.
24300 (read_graph_file): Replace usage of encoding by the newly added
24302 (add_line_counts): Likewise.
24303 (accumulate_line_counts): Remove usage of the union.
24304 (function_info::function_info): New function.
24305 (function_info::~function_info): Likewise.
24306 (process_file): Call delete instead of release_function.
24307 (release_function): Release the function.
24308 (release_structures): Call delete instead of release_function.
24309 (solve_flow_graph): Replace usage of num_blocks.
24310 (find_exception_blocks): Likewise.
24311 (output_lines): Fix GNU coding style.
24313 2017-04-28 Martin Liska <mliska@suse.cz>
24316 * coverage.c (coverage_remove_note_file): New function.
24317 * coverage.h: Declare the function.
24318 * toplev.c (finalize): Clean if an error has been seen.
24320 2017-04-28 Martin Liska <mliska@suse.cz>
24322 PR gcov-profile/80031
24323 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
24324 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
24325 * gcov.c (read_graph_file): Read just number of blocks.
24326 * profile.c (branch_prob): Do not stream 0 flags per a basic
24329 2017-04-28 Martin Liska <mliska@suse.cz>
24331 * gcov-dump.c (tag_*): Add new argument to declarations.
24332 (dump_gcov_file): Likewise.
24333 (tag_blocks): Add and use new argument depth.
24334 (tag_arcs): Likewise.
24335 (tag_lines): Likewise.
24336 (tag_counters): Likewise.
24337 (tag_summary): Likewise.
24338 (dump_working_sets): Use depth to do a proper indentation.
24340 2017-04-28 Jakub Jelinek <jakub@redhat.com>
24343 * cgraph.h (symtab_node::debug_symtab): No longer inline.
24344 * symtab.c (symtab_node::debug_symtab): Move definition here.
24346 2017-04-28 Richard Biener <rguenther@suse.de>
24348 * lto-streamer.h (LTO_major_version): Bump to 7.
24350 2017-04-28 Richard Biener <rguenther@suse.de>
24352 * tree-vrp.c (assert_info): New struct.
24353 (add_assert_info): New helper.
24354 (register_edge_assert_for_2): Refactor to add asserts to a vector
24356 (register_edge_assert_for_1): Likewise.
24357 (register_edge_assert_for): Likewise.
24358 (finish_register_edge_assert_for): New helper actually registering
24359 asserts where live on edge.
24360 (find_conditional_asserts): Adjust.
24361 (find_switch_asserts): Likewise.
24362 (evrp_dom_walker::try_find_new_range): Generalize.
24363 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
24365 2017-04-27 Marek Polacek <polacek@redhat.com>
24368 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
24369 arg10 and arg11 to itype.
24371 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
24373 * doc/extend.texi (Object Size Checking): Improve grammar.
24375 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
24378 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
24379 that the logic for permitting reciprocal estimates matches that
24382 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24385 * tree.c (type_cache_hasher::equal): Only compare
24386 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
24387 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
24388 non-aggregate element types.
24389 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
24390 about the flag on ARRAY_TYPEs in the comment, formatting fix.
24392 2017-04-27 Richard Biener <rguenther@suse.de>
24394 PR middle-end/80533
24395 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
24396 stripping ARRAY_REFs from MEM_EXPR make sure we're not
24397 keeping a reference to a trailing array.
24399 2017-04-27 Richard Biener <rguenther@suse.de>
24401 PR middle-end/80539
24402 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
24403 being in loop-closed SSA form conservatively.
24404 (chrec_fold_multiply_poly_poly): Likewise.
24406 2017-04-27 Tamar Christina <tamar.christina@arm.com>
24408 PR middle-end/79665
24409 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
24410 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
24412 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24415 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
24416 (aarch64_function_arg_alignment): Return unsigned int again, but still
24417 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
24418 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24419 Don't emit -Wpsabi note.
24420 (aarch64_function_arg_boundary): Likewise.
24421 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24424 2017-04-26 Nathan Sidwell <nathan@acm.org>
24426 * tree.h (crc32_unsigned_n): Declare.
24427 (crc32_unsigned, crc32_unsigned): Make inline.
24428 * tree.c (crc32_unsigned_bits): Replace with ...
24429 (crc32_unsigned_n): ... this.
24430 (crc32_unsigned, crc32_byte): Remove.
24431 (crc32_string): Remove unnecessary braces.
24433 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24435 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
24436 * ipa-inline-analysis.c (MAX_TIME): Remove.
24437 (account_size_time): Use sreal for time.
24438 (dump_inline_summary): Update.
24439 (estimate_function_body_sizes): Update.
24440 (estimate_edge_size_and_time): Update.
24441 (estimate_calls_size_and_time): Update.
24442 (estimate_node_size_and_time): Update.
24443 (inline_merge_summary): Update.
24444 (inline_update_overall_summary): Update.
24445 (estimate_time_after_inlining): Update.
24446 (inline_read_section): Update.
24447 (inline_write_summary): Update.
24448 * ipa-inline.c (compute_uninlined_call_time): Update.
24449 (compute_inlined_call_time): Update.
24450 (recursive_inlining): Update.
24451 (inline_small_functions): Update.
24452 (dump_overall_stats): Update.
24453 * ipa-inline.h: Include sreal.h.
24454 (size_time_entry): Turn time to sreal.
24455 (inline_summary): Turn self_time nad time to sreal.
24457 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24459 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
24461 (sreal::stream_out, sreal::stream_in): New.
24462 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
24464 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24466 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
24469 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
24472 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
24474 (dimode_scalar_chain::compute_convert_gain): Ditto.
24475 (dimode_scalar_chain::make_vector_copies): Ditto.
24476 (dimode_scalar_chain::convert_reg): Ditto.
24477 (dimode_scalar_chain::convert_insn): Ditto.
24478 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
24479 (VI248_AVX512BW_1): New mode iterator.
24480 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
24481 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
24484 2017-04-25 Martin Sebor <msebor@redhat.com>
24486 PR tree-optimization/80497
24487 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
24488 constants are representable in HOST_WIDE_INT.
24489 (parse_directive): Ditto.
24491 2017-04-25 Martin Sebor <msebor@redhat.com>
24494 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
24495 (new_zero_array): Adjust signature.
24496 (dom_info::dom_init): Used unsigned rather that size_t.
24497 (dom_info::dom_info): Same.
24499 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24500 Jakub Jelinek <jakub@redhat.com>
24503 * config/arm/arm.c: Include gimple.h.
24504 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24505 returns negative, increment ncrn only if it returned positive.
24506 (arm_needs_doubleword_align): Return int instead of bool,
24507 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
24508 members, but if there is any such non-FIELD_DECL
24509 > PARM_BOUNDARY aligned decl, return -1 instead of false.
24510 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24511 returns negative, increment nregs only if it returned positive.
24512 (arm_setup_incoming_varargs): Likewise.
24513 (arm_function_arg_boundary): Emit -Wpsabi note if
24514 arm_needs_doubleword_align returns negative, return
24515 DOUBLEWORD_ALIGNMENT only if it returned positive.
24517 2017-04-25 Marek Polacek <polacek@redhat.com>
24520 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24521 first argument to type.
24523 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
24526 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
24527 type checks to test for compatibility instead of equality.
24529 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24530 Jakub Jelinek <jakub@redhat.com>
24533 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
24535 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
24536 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
24537 the alignment computation, but return their maximum in warn_alignment.
24538 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24539 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
24541 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
24542 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24545 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24547 * config/arc/simdext.md (dmpyh): Fix typo.
24549 2017-04-25 Richard Biener <rguenther@suse.de>
24551 PR tree-optimization/80492
24552 * alias.c (compare_base_decls): Handle registers with asm
24553 specification conservatively.
24554 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24555 compare_base_decls returning dont-know properly.
24557 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24559 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
24560 (legitimate_offset_address_p): New function.
24561 (arc_legitimate_address_p): Use above function.
24563 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24565 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
24567 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24569 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
24570 ACCH registers whenever they are available.
24572 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24574 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
24575 double regs fix when not used.
24577 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24579 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
24581 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
24582 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
24584 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24586 * config/arc/arc.c (arc_output_addsi): Check for h-register class
24587 when emitting short ADD instructions.
24589 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24591 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
24593 (cmpsi_cc_c_insn): Likewise.
24594 (cbranchsi4_scratch): Compute proper instruction length using
24595 compact_hreg_operand.
24596 * config/arc/predicates.md (compact_hreg_operand): New predicate.
24598 2017-04-25 Richard Biener <rguenther@suse.de>
24600 PR middle-end/80509
24601 * passes.c (pass_manager::pass_manager): Initialize
24602 m_name_to_pass_map.
24604 2017-04-25 Richard Biener <rguenther@suse.de>
24606 PR tree-optimization/79201
24607 * tree-ssa-sink.c (statement_sink_location): Handle calls.
24609 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24612 * config/s390/vector.md: Split MEM->GPR vector moves for
24613 non-s_operand addresses.
24615 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24618 * config/s390/predicates.md (reload_const_wide_int_operand): New
24620 * config/s390/s390.md ("movti"): Remove d/P alternative.
24621 ("movti_bigconst"): New pattern definition.
24623 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24626 * s390-protos.h (s390_expand_cs_hqi): Removed.
24627 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
24628 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
24629 modes as well as CCZ1mode and CCZmode.
24630 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
24631 signature of s390_emit_compare_and_swap.
24632 (s390_expand_cs_hqi): Likewise, make static.
24633 (s390_expand_cs_tdsi): Generate an explicit compare before trying
24634 compare-and-swap, in some cases.
24635 (s390_expand_cs): Wrapper function.
24636 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
24638 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
24639 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
24640 patterns for small and large integers. Forbid symref memory operands.
24641 Move expander to s390.c. Require cc register.
24642 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
24643 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
24644 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
24645 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
24646 symref memory operands. Remove CC mode and call s390_match_ccmode
24648 ("atomic_exchange<mode>"): Allow and implement all integer modes.
24650 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24652 * config/s390/s390.md (define_peephole2): New peephole to help
24653 combining the load-and-test pattern with volatile memory.
24655 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24657 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
24658 with CCZmode for TARGET_Z196.
24660 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24662 PR rtl-optimization/80501
24663 * combine.c (make_compound_operation_int): Set subreg_code to SET
24664 even for AND with mask of the sign bit of mode.
24666 PR rtl-optimization/80500
24667 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
24668 sum's initial value.
24670 2017-04-25 Julian Brown <julian@codesourcery.com>
24671 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24673 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
24675 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
24677 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
24679 2017-04-25 Julian Brown <julian@codesourcery.com>
24680 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24682 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
24683 (thunderx2t99_sha): New Reservation.
24685 2017-04-25 Julian Brown <julian@codesourcery.com>
24686 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24688 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
24689 type for 1-element load.
24691 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
24693 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
24695 2017-04-24 Martin Jambor <mjambor@suse.cz>
24697 PR tree-optimization/80293
24698 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
24699 char arrays not totally scalarizable if it is false.
24700 (analyze_all_variable_accesses): Pass correct value in the new
24701 parameter. Add a statistics counter.
24703 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
24705 PR middle-end/79931
24706 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
24708 2017-04-24 Richard Biener <rguenther@suse.de>
24710 PR tree-optimization/80494
24711 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
24712 out for complex types.
24714 2017-04-24 Richard Biener <rguenther@suse.de>
24716 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
24717 * tree-ssa-sccvn.c (print_scc): Print SCC size.
24718 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
24719 (DFS): Adjust and never fail.
24720 (sccvn_dom_walker::fail): Remove.
24721 (sccvn_dom_walker::before_dom_children): Adjust.
24722 (run_scc_vn): Likewise and never fail.
24723 * tree-ssa-pre.c (pass_pre::execute): Adjust.
24724 (pass_fre::execute): Likewise.
24726 2017-04-24 Richard Biener <rguenther@suse.de>
24728 PR tree-optimization/79725
24729 * tree-ssa-sink.c (statement_sink_location): Return whether
24730 failure reason was zero uses. Move that check later.
24731 (sink_code_in_bb): Deal with zero uses by removing the stmt
24734 2017-04-24 Richard Biener <rguenther@suse.de>
24737 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
24738 pointer-based references.
24740 2017-04-24 Richard Biener <rguenther@suse.de>
24743 * pass_manager.h (pass_manager::operator new): Remove.
24744 (pass_manager::operator delete): Likewise.
24745 * passes.c (pass_manager::operator new): Remove.
24746 (pass_manager::operator delete): Likewise.
24747 (pass_manager::pass_manager): Zero individual pass members.
24749 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
24752 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
24753 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
24754 Check "XEXP (src, 1)" operand here.
24755 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
24756 Check "XEXP (src, 1)" operand here.
24757 (dimode_scalar_chain::make_vector_copies): Detect count register
24758 of a shift instruction. Zero extend count register from QImode
24759 to DImode to satisfy vector shift pattern count operand predicate.
24760 Substitute vector shift count operand with a DImode copy.
24761 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
24764 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
24766 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
24767 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
24768 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
24769 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
24770 (UNSPEC_NOREX_MEM): Remove definition.
24772 2017-04-21 Richard Biener <rguenther@suse.de>
24774 PR tree-optimization/79547
24775 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
24776 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
24777 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
24778 without any constraints.
24780 2017-04-21 Richard Biener <rguenther@suse.de>
24782 PR tree-optimization/78847
24783 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
24785 2017-04-21 Richard Biener <rguenther@suse.de>
24787 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
24788 (build_distinct_type_copy): Likewise.
24789 (build_variant_type_copy): Likewise.
24790 * tree.c (build_qualified_type): Pass down mem-stat info.
24791 (build_distinct_type_copy): Likewise.
24792 (build_variant_type_copy): Likewise.
24794 2017-04-21 Richard Biener <rguenther@suse.de>
24796 PR tree-optimization/80237
24797 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
24799 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
24800 for a simplified result.
24802 2016-04-21 Richard Biener <rguenther@suse.de>
24804 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
24805 sth as strict as a simple_iv but a chrec without symbols and an
24806 operand defined in the loop we are peeling (and not some subloop).
24807 (propagate_constants_for_unrolling): Propagate all constants.
24809 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24812 * config/i386/i386.c (print_reg): Remove assert for disalowed
24813 regno values, call output_operand_lossage instead.
24815 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24818 * config/i386/constraints.md (Yc): New register constraint.
24819 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
24820 Use Yc constraint for alternative 2 of operand 0. Remove
24821 preferred_for_speed attribute.
24823 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24825 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
24826 lastprivate clauses in SIMT case.
24828 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
24830 * doc/invoke.texi (-Wextra-semi): Document new warning option.
24832 2017-04-20 Richard Biener <rguenther@suse.de>
24834 PR tree-optimization/57796
24835 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
24836 as N scalar stores.
24837 (vect_model_load_cost): Cost gathers as N scalar loads.
24839 2017-04-20 Richard Biener <rguenther@suse.de>
24841 * ggc-page.c (ggc_allocated_p): Rename to ...
24842 (safe_lookup_page_table_entry): ... this and return the lookup
24844 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
24846 2017-04-20 Richard Biener <rguenther@suse.de>
24848 PR tree-optimization/80453
24849 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
24850 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
24851 from the conditions.
24852 (vn_phi_eq): Pass them down.
24853 (vn_phi_lookup): Record them.
24854 (vn_phi_insert): Likewise.
24856 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
24858 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
24859 uninitialized variable warning to avoid buffer overrun.
24861 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24864 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
24865 is suppressed for '{ 0 }' in C.
24867 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24869 * BASE-VER: Set to 8.0.0.
24871 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
24873 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
24874 priority .init_array and .fini_array section with SECTION_NOTYPE
24877 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24879 PR middle-end/80423
24880 * tree.h (build_array_type): Add typeless_storage default argument.
24881 * tree.c (type_cache_hasher::equal): Also compare
24882 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
24883 (build_array_type): Add typeless_storage argument, set
24884 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
24886 (build_nonshared_array_type): Adjust build_array_type_1 caller.
24887 (build_array_type): Likewise. Add typeless_storage argument.
24889 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24890 Jakub Jelinek <jakub@redhat.com>
24892 PR tree-optimization/80426
24893 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
24894 operation on symbolic operands, also compute the overflow for the
24895 invariant part when the operation degenerates into a negation.
24897 2017-04-19 Jakub Jelinek <jakub@redhat.com>
24900 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
24901 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
24904 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
24906 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
24909 * config/avr/avr.c (tree.h): Include it.
24910 (cgraph.h): Include it.
24911 (avr_encode_section_info): Don't warn for uninitialized progmem
24912 variable if it's just an alias.
24914 2017-04-19 Richard Biener <rguenther@suse.de>
24917 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
24918 when needed by AutoPGO.
24920 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
24923 * doc/lto.texi: Remove an extra 'that'.
24925 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
24927 PR rtl-optimization/80429
24928 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
24929 are only used in debug insns.
24931 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24932 Vladimir Makarov <vmakarov@redhat.com>
24934 * config/sparc/predicates.md (input_operand): Add comment. Return
24935 true for any memory operand when LRA is in progress.
24936 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
24938 2017-04-18 Jeff Law <law@redhat.com>
24941 * mips.md ({return,simple_return}_internal): Do not overwrite
24944 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24946 PR tree-optimization/80443
24947 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
24948 instead of adding 1, subtract -1 and similarly instead of subtracting
24951 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
24953 PR rtl-optimization/80357
24954 * haifa-sched.c (tmp_bitmap): New variable.
24955 (model_recompute): Handle duplicate use records.
24956 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
24957 (free_global_sched_pressure_data): Free it.
24959 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
24962 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24963 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24964 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24965 instead of SYSTEM_HEADER_DIR.
24967 2017-04-18 Jeff Law <law@redhat.com>
24969 PR middle-end/80422
24970 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
24971 predecessors after walking up the insn chain.
24973 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24976 * dwarf2out.c (modified_type_die): Try harder not to emit internal
24977 sizetype type into debug info.
24979 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24982 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
24983 unneeded test for TARGET_UPPER_REGS_SF.
24984 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
24986 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24989 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
24990 instead of gsi_after_labels.
24992 2017-04-18 Jeff Law <law@redhat.com>
24994 * regcprop.c (maybe_mode_change): Avoid creating copies of the
24998 2017-04-13 Jeff Law <law@redhat.com>
24999 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25000 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25002 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
25005 * config/avr/avr.c (intl.h): Include it.
25006 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
25008 2017-04-18 Martin Liska <mliska@suse.cz>
25010 PR gcov-profile/78783
25011 * gcov-tool.c (gcov_output_files): Validate that destination
25012 file is either removed by the tool or by a user.
25014 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25015 Guy Benyei <guybe@mellanox.com>
25017 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
25018 block, and do not negate it, the stored id is already negative.
25020 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
25022 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
25024 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
25027 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
25028 masks of options that should be turned off if the VSX vector
25029 options are turned off.
25030 (OTHER_P8_VECTOR_MASKS): Likewise.
25031 (OTHER_VSX_VECTOR_MASKS): Likewise.
25032 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
25033 rs6000_disable_incompatible_switches to validate no type switches
25035 (rs6000_incompatible_switch): New function to disallow turning on
25036 other vector options if -mno-vsx, -mno-power8-vector, or
25037 -mno-power9-vector are specified.
25039 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25041 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
25043 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25045 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
25046 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
25047 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
25048 (ARG_POINTER_CFA_OFFSET): Likewise.
25050 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
25052 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
25053 conditions to take advantage of various optimizations.
25055 2017-04-13 Jeff Law <law@redhat.com>
25057 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
25058 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
25059 (zero_extendsidi2_dext): Likewise.
25061 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25064 * fold-const.c (fold_ternary_loc): Revert
25065 use op0 instead of fold_convert_loc (loc, type, arg0) part of
25068 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
25070 PR rtl-optimization/80343
25071 * lra-remat.c (update_scratch_ops): Assign original hard reg to
25072 new scratch pseudo.
25074 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
25077 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
25078 to ubsan_encode_value.
25080 2017-04-13 Jeff Law <law@redhat.com>
25082 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
25083 appearing in DEBUG_INSNs.
25085 2017-04-13 Martin Liska <mliska@suse.cz>
25087 PR gcov-profile/80413
25088 * gcov-io.c (gcov_write_string): Copy to buffer just when
25089 allocated size is greater than zero.
25091 2017-04-13 Jakub Jelinek <jakub@redhat.com>
25094 * dwarf2out.c (decls_for_scope): Ignore declarations of
25095 current_function_decl in BLOCK_NONLOCALIZED_VARS.
25097 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
25100 * ipa-visibility.c (non_local_p): Fix typos.
25101 (localize_node): When localizing symbol in same comdat group,
25102 dissolve the group only when we know external symbols are going
25104 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
25106 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25108 PR tree-optimization/79390
25109 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
25110 order does not result in usable sequence, retry with reversed operand
25116 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
25117 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
25118 op0 instead of fold_convert_loc (loc, type, arg0).
25120 2017-04-12 Jeff Law <law@redhat.com>
25122 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
25123 has a delay slot in the generated code.
25125 * config/cris/cris.md (cris_preferred_reload_class): Return
25126 GENNONACR_REGS rather than GENERAL_REGS.
25128 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25131 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
25132 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
25133 signedness of the result type.
25135 2017-04-12 Richard Biener <rguenther@suse.de>
25136 Jeff Law <law@redhat.com>
25138 PR tree-optimization/80359
25139 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
25140 trim stores to TARGET_MEM_REFs.
25142 2017-04-12 Richard Biener <rguenther@suse.de>
25144 PR tree-optimization/79390
25145 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
25146 threading case even more.
25148 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
25151 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
25152 for quad_address_p for TImode, instead of just not indexed_address.
25154 2017-04-12 Richard Biener <rguenther@suse.de>
25155 Bernd Edlinger <bernd.edlinger@hotmail.de>
25157 PR middle-end/79671
25158 * alias.c (component_uses_parent_alias_set_from): Handle
25159 TYPE_TYPELESS_STORAGE.
25160 (get_alias_set): Likewise.
25161 * tree-core.h (tree_type_common): Add typeless_storage flag.
25162 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
25163 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
25164 for types containing members with TYPE_TYPELESS_STORAGE.
25165 (place_field): Likewise.
25166 (layout_type): Likewise for ARRAY_TYPE.
25167 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
25168 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
25169 TYPE_TYPELESS_STORAGE.
25170 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
25172 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25175 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25176 first argument to type.
25178 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25182 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
25183 CONST0_RTX (mode) rather than const0_rtx where appropriate.
25184 (rs6000_expand_binop_builtin): Likewise.
25185 (rs6000_expand_ternop_builtin): Likewise; also add missing
25186 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
25187 vshasigma built-ins.
25188 * doc/extend.texi: Document that vec_xxpermdi's third argument
25189 must be a constant.
25191 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
25193 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
25194 Use shift_const cost parameter when calculating gain of STV shifts.
25196 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
25198 PR rtl-optimization/70478
25199 * lra-constraints.c (process_alt_operands): Check memory for
25200 disfavoring memory insn operand.
25202 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25204 PR middle-end/80100
25205 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
25206 left shift in unsigned HOST_WIDE_INT type.
25208 PR rtl-optimization/80385
25209 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
25210 (not (neg X)) into (plus X -1) for complex or non-integral modes.
25213 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
25214 if they have any depend clauses.
25216 2017-04-11 Martin Liska <mliska@suse.cz>
25219 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
25220 * ipa-split.c (split_function): Create a local comdat symbol
25221 if caller is in a comdat group.
25223 2017-04-11 Martin Liska <mliska@suse.cz>
25226 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
25229 2017-04-11 Martin Sebor <msebor@redhat.com>
25231 PR middle-end/80364
25232 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
25233 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
25235 (directive::set_width, directive::set_precision, format_character):
25237 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
25240 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
25243 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
25244 conflict, set target->arch_name instead of target->cpu_name.
25246 2017-04-11 Richard Biener <rguenther@suse.de>
25248 PR tree-optimization/80374
25249 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
25250 build_zero_cst, remove fold_convertible_p check again.
25252 2017-04-11 Martin Liska <mliska@suse.cz>
25255 * ubsan.c (instrument_object_size): Do not instrument register
25258 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25261 * config/i386/i386-builtin-types.def
25262 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
25263 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
25264 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
25265 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
25266 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
25267 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
25268 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
25269 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
25270 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
25271 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
25272 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
25273 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
25274 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
25275 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
25276 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
25277 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
25278 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
25279 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
25280 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
25281 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
25282 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
25283 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
25284 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
25285 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
25286 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
25287 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
25288 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
25289 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
25290 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
25291 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
25292 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
25293 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
25294 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
25295 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
25296 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
25297 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
25298 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
25299 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
25300 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
25301 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
25302 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
25303 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
25304 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
25305 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
25306 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
25307 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
25309 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
25310 flag to second_arg_count, handle 4 argument function type _COUNT
25311 aliases, handle second_arg_count on second argument rather than last.
25313 2017-04-10 Jeff Law <law@redhat.com>
25315 PR tree-optimization/80374
25316 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
25317 record anything if we can not convert integer_zero_node to the
25320 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25323 * config/rs6000/rs6000.c (rs6000_option_override_internal):
25324 Enhance special handling given to the TARGET_P9_MINMAX option in
25325 relation to certain other options.
25327 2017-04-10 Bin Cheng <bin.cheng@arm.com>
25329 PR tree-optimization/80153
25330 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
25331 remove POINTER_PLUS_EXPR's base part directly, rather than through
25334 2017-04-10 Richard Biener <rguenther@suse.de>
25335 Bin Cheng <bin.cheng@arm.com>
25337 PR tree-optimization/80153
25338 * tree-affine.c (aff_combination_to_tree): Get base pointer from
25339 the first element of pointer type aff_tree. Build result expr in
25341 (add_elt_to_tree): Convert to type unconditionally. Remove other
25342 fold_convert calls.
25343 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
25344 (rewrite_use_nonlinear_expr): Check invariant using iv information.
25346 2017-04-10 Richard Biener <rguenther@suse.de>
25348 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25351 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
25353 PR rtl-optimization/70478
25354 * lra-constraints.c (curr_small_class_check): New.
25355 (update_and_check_small_class_inputs): New.
25356 (process_alt_operands): Update curr_small_class_check. Disfavor
25357 alternative insn memory operands. Check available regs for small
25360 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25363 * config/mips/mips.opt (-mvirt): Update description.
25364 * doc/invoke.texi (-mvirt): Likewise.
25366 2017-04-10 Richard Biener <rguenther@suse.de>
25368 PR middle-end/80362
25369 * fold-const.c (fold_binary_loc): Look at unstripped ops when
25370 looking for NEGATE_EXPR in -A / -B to A / B folding.
25372 2017-04-10 Martin Liska <mliska@suse.cz>
25374 PR gcov-profile/80224
25375 * gcov.c (print_usage): Fix usage string.
25376 (get_gcov_intermediate_filename): Remove.
25377 (output_gcov_file): Use both for normal and intermediate format.
25378 (generate_results): Do not initialize special file for
25379 intermediate format.
25381 2017-04-10 Richard Biener <rguenther@suse.de>
25383 PR tree-optimization/80304
25384 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
25387 2017-04-10 Nathan Sidwell <nathan@acm.org>
25390 * config/rs6000/rs6000.c (rs6000_vector_type): New.
25391 (rs6000_init_builtins): Use it.
25393 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25395 * config/arm/arm.md (<mrc>): Add mode to SET source.
25396 (<mrrc>): Likewise.
25398 2017-04-10 Richard Biener <rguenther@suse.de>
25400 PR middle-end/80344
25401 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
25403 2017-04-10 Jakub Jelinek <jakub@redhat.com>
25406 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
25407 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
25408 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
25409 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
25410 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
25411 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
25412 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
25413 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
25414 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
25415 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
25416 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
25417 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
25418 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
25419 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
25420 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
25421 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
25422 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
25423 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
25424 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
25425 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
25426 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
25427 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
25428 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
25430 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
25432 PR rtl-optimization/70478
25433 * lra-constraints.c: Reverse the last patch.
25435 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
25437 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
25438 Add comment for WCHAR_T.
25440 2017-04-08 Martin Liska <mliska@suse.cz>
25443 2017-04-07 Martin Liska <mliska@suse.cz>
25446 * ipa-split.c (split_function): Add function part to a same comdat
25449 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25452 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
25454 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
25456 * rs6000/rs6000.c (vec_load_pendulum): Rename...
25457 (vec_pairing): ...to this.
25458 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
25459 (rs6000_sched_init): Adjust for name change.
25460 (struct rs6000_sched_context): Likewise.
25461 (rs6000_init_sched_context): Likewise.
25462 (rs6000_set_sched_context): Likewise.
25464 2017-04-07 Jakub Jelinek <jakub@redhat.com>
25470 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
25472 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
25473 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
25474 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
25476 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
25478 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
25480 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25482 PR rtl-optimization/70703
25483 * ira-color.c (update_conflict_hard_regno_costs): Use
25484 int64_t instead of HOST_WIDE_INT.
25486 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25488 PR rtl-optimization/70478
25489 * lra-constraints.c (process_alt_operands): Disfavor alternative
25490 insn memory operands.
25492 2017-04-07 Jeff Law <law@redhat.com>
25494 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
25495 CALL and NOTE_INSN_CALL_ARG_LOCATION.
25497 2017-04-07 Martin Liska <mliska@suse.cz>
25500 * config/aarch64/aarch64.c (aarch64_process_target_attr):
25501 Show error message instead of an ICE.
25503 2017-04-07 Martin Liska <mliska@suse.cz>
25506 * ipa-split.c (split_function): Add function part to a same comdat
25509 2017-04-07 Richard Biener <rguenther@suse.de>
25511 PR middle-end/80341
25512 * tree.c (get_unwidened): Also handle ! for_type case for
25514 * convert.c (do_narrow): Split out from ...
25515 (convert_to_integer_1): ... here. Do not pass final truncation
25516 type to get_unwidened for TRUNC_DIV_EXPR.
25518 2017-04-07 Richard Biener <rguenther@suse.de>
25520 * tree-affine.c (wide_int_ext_for_comb): Take type rather
25522 (aff_combination_const): Adjust.
25523 (aff_combination_scale): Likewise.
25524 (aff_combination_add_elt): Likewise.
25525 (aff_combination_add_cst): Likewise.
25526 (aff_combination_convert): Likewise.
25527 (add_elt_to_tree): Likewise. Remove unused argument.
25528 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
25530 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
25532 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
25534 * config/arm/arm.c (arm_default_short_enums): Use
25535 ARM_DEFAULT_SHORT_ENUMS.
25536 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
25538 2017-04-06 Jakub Jelinek <jakub@redhat.com>
25541 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
25542 members with redundant out-of-class redeclaration.
25544 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25547 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
25548 * config/i386/i386.md (*zero_extendsidi2):
25549 Add (?*x,*x) and (?*v,*v) alternatives.
25551 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25554 * config/i386/i386.c (ix86_expand_builtin)
25555 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
25556 mode from insn data. Convert operands to insn operand mode.
25557 Copy operands that don't satisfy insn predicate to a register.
25559 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
25561 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
25564 2017-04-06 Richard Biener <rguenther@suse.de>
25566 PR tree-optimization/80334
25567 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
25568 preserve alignment of accesses.
25570 2017-04-06 Richard Biener <rguenther@suse.de>
25572 PR tree-optimization/80262
25573 * tree-sra.c (build_ref_for_offset): Preserve address-space
25575 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
25576 Drop useless address-space information on MEM_REF offsets.
25578 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
25580 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
25582 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25584 PR rtl-optimization/70703
25585 * ira-color.c (update_conflict_hard_regno_costs): Use
25586 HOST_WIDE_INT instead of long.
25588 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
25591 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
25592 not defined for x86_64 target. Add -mmmx target option when __SSE2__
25594 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
25595 for x86_64 target. Handle -m3dnowa option.
25597 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25599 PR rtl-optimization/70703
25600 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
25601 (update_conflict_hard_regno_costs): Use long instead of unsigned
25602 arithmetic for cost calculation.
25604 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25605 Bernd Edlinger <bernd.edlinger@hotmail.de>
25608 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
25611 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
25614 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
25615 ptr_mode with Pmode throughout.
25616 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
25617 into probe_stack_range and use DImode.
25619 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
25622 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
25623 call_eh_return is true.
25625 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25627 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
25628 Initialize last_match_fntype_index.
25630 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25633 * tree-nvr.c: Include internal-fn.h.
25634 (pass_return_slot::execute): Ignore internal calls without
25637 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25638 Richard Biener <rguenther@suse.de>
25641 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
25642 captures used multiple times, except for the last use.
25643 * generic-match-head.c: Include gimplify.h.
25645 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25647 PR tree-optimization/79390
25648 * target.h (struct noce_if_info): Declare.
25649 * targhooks.h (default_noce_conversion_profitable_p): Declare.
25650 * target.def (noce_conversion_profitable_p): New target hook.
25651 * ifcvt.h (struct noce_if_info): New type, moved from ...
25652 * ifcvt.c (struct noce_if_info): ... here.
25653 (noce_conversion_profitable_p): Renamed to ...
25654 (default_noce_conversion_profitable_p): ... this. No longer
25656 (noce_try_store_flag_constants, noce_try_addcc,
25657 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
25658 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
25659 instead of noce_conversion_profitable_p.
25660 * config/i386/i386.c: Include ifcvt.h.
25661 (ix86_option_override_internal): Don't override
25662 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
25663 (ix86_noce_conversion_profitable_p): New function.
25664 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
25665 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
25666 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
25667 * doc/tm.texi: Regenerated.
25669 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25671 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
25674 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25677 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
25678 instructions for small multiply cores.
25680 2017-04-04 Jeff Law <law@redhat.com>
25682 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
25684 (mips_expand_vec_perm_const): Initialize elements in orig_perm
25685 that are not set by the loop over the elements.
25687 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25690 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
25691 int mode, convert_modes it to mode as unsigned, otherwise use
25692 lowpart_subreg to mode rather than SImode.
25693 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
25694 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
25695 Use DImode instead of SImode for the shift count operand.
25696 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
25699 2017-04-04 Richard Biener <rguenther@suse.de>
25701 PR middle-end/80281
25702 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
25703 arithmetic done for the negate or the plus. Simplify.
25704 (A - (-B) -> A + B): Likewise.
25705 * fold-const.c (split_tree): Make sure to not negate pointers.
25707 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
25709 PR rtl-optimization/60818
25710 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
25711 a compare of comparisons with the thing compared if this results
25712 in a different machine mode.
25714 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
25716 * alias.c (base_alias_check): Fix typo in comment.
25717 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
25718 * cgraphunit.c (symbol_table::compile): Likewise.
25719 * collect2.c (maybe_run_lto_and_relink): Likewise.
25720 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
25721 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
25722 * config/avr/avr.c (avr_map_op_t): Likewise.
25723 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
25724 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
25725 * config/epiphany/epiphany.md (movcc): Likewise.
25726 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
25727 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
25729 * config/mips/mips.c (mips_save_restore_reg): Likewise.
25730 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
25731 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
25732 * config/sh/sh.c (sh_rtx_costs): Likewise.
25733 * fold-const.c (fold_truth_andor): Likewise.
25734 * genautomata.c (collapse_flag): Likewise.
25735 * gengtype.h (struct type::u::s): Likewise.
25736 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
25737 * input.c (FORMAT_AMOUNT): Likewise.
25738 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
25739 (known_aggs_to_agg_replacement_list): Likewise.
25740 * ipa-inline-analysis.c: Likewise.
25741 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
25742 * ipa-polymorphic-call.c
25743 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
25744 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
25745 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
25747 * modulo-sched.c (apply_reg_moves): Likewise.
25748 * omp-expand.c (build_omp_regions_1): Likewise.
25749 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
25750 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
25751 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
25752 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
25753 * value-prof.c: Likewise.
25754 * var-tracking.c (val_reset): Likewise.
25756 2017-04-03 Richard Biener <rguenther@suse.de>
25758 PR tree-optimization/80275
25759 * fold-const.c (split_address_to_core_and_offset): Handle
25762 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
25764 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
25765 descriptors is at least equal to that of functions.
25767 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
25769 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
25771 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
25774 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
25775 (mov<IMOD4:mode>): New expander.
25776 (*mov<IMOD4:mode>_internal): New insn and split pattern.
25778 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
25780 PR rtl-optimization/79405
25781 * fwprop.c (propagations_left): New variable.
25782 (forward_propagate_into): Decrement it.
25783 (fwprop_init): Initialize it.
25784 (fw_prop): If the variable has reached zero, stop propagating.
25785 (fwprop_addr): Ditto.
25787 2017-03-31 Jakub Jelinek <jakub@redhat.com>
25790 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
25791 a FUNCTION_DECL, pass it as decl instead of origin to
25794 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
25796 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
25799 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
25802 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
25803 TARGET_VSX_SMALL_INTEGER.
25805 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25807 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25808 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
25810 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25812 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
25813 extraction from odd-numbered MSA register.
25815 2017-03-31 Jakub Jelinek <jakub@redhat.com>
25817 PR middle-end/80173
25818 * expmed.c (store_bit_field_1): Don't attempt to create
25819 a word subreg out of hard registers wider than word if they
25820 have HARD_REGNO_NREGS of 1 for their mode.
25822 PR middle-end/80163
25823 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
25824 conversions to integer types wider than word and pointer.
25827 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
25828 (rtx_equal_for_cselib_p): Pass 0 to it.
25829 * cselib.c (cselib_hasher::equal): Likewise.
25830 (rtx_equal_for_cselib_1): Add depth argument. If depth
25831 is 128, don't look up VALUE locs and punt. Increment
25832 depth in recursive calls when walking VALUE locs.
25834 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
25836 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
25837 (make_gcov_file_name): Use the canonical path name for generating
25839 (read_line): Fix handling of files with ascii null bytes.
25841 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
25843 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
25844 to initialise a vector register instead
25845 of using a const_int.
25847 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25849 PR translation/80189
25850 * gimplify.c (omp_default_clause): Use %qs instead of %s in
25851 diagnostic messages.
25853 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
25856 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
25857 (dfp_diex_<mode>): Update mode of operand 1.
25858 * doc/extend.texi (dxex, dxexq): Document change to return type.
25859 (diex, diexq): Document change to argument type.
25861 2017-03-30 Martin Jambor <mjambor@suse.cz>
25864 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
25865 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
25866 it reflects the signature changes performed at the callee side.
25867 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
25868 to cgraph_build_function_type_skip_args.
25869 (build_function_decl_skip_args): Adjust call to the above function.
25871 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25874 * config/i386/sse.md
25875 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
25876 register as dest whenever it is a MEM not rtx_equal_p to the
25877 corresponding dup operand, and when forcing into reg move the
25878 reg into the memory afterwards.
25879 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
25880 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
25881 for the force_reg mode.
25882 (avx512vl_vextractf128<mode>): Use register as dest either
25883 always when a MEM, or when it is a MEM not rtx_equal_p to the
25884 corresponding dup operand, or even not when it is a CONST_VECTOR
25885 depending on the mode and lo vs. hi.
25886 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
25888 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
25889 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
25890 Likewise. Require that operands[2] is even.
25891 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
25892 Remove extraneous parens. Require that operands[2] is a multiple
25894 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
25895 operands[0] is a MEM if <mask_applied>, the predicates/constraints
25896 disallow memory then.
25898 2017-03-30 Richard Biener <rguenther@suse.de>
25900 PR tree-optimization/77498
25901 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
25902 to non-constants over backedges.
25904 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
25906 PR rtl-optimization/80233
25907 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
25908 as last_combined_insn. Do not test for BARRIER_P separately.
25910 2017-03-29 Andreas Schwab <schwab@suse.de>
25913 * calls.c (prepare_call_address): Convert funexp to Pmode before
25914 copying to temp reg.
25916 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25918 PR tree-optimization/80158
25919 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
25920 Handle possible future case of more than one alternate
25922 (replace_rhs_if_not_dup): Likewise.
25923 (replace_one_candidate): Likewise.
25925 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
25927 PR rtl-optimization/80193
25928 * ira.c (ira): Do not check allocation for LRA.
25930 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
25932 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
25933 (nvptx_output_simt_exit): Declare.
25934 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
25935 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
25936 (init_softstack_frame): Move initialization of crtl->is_leaf to...
25937 (nvptx_declare_function_name): ...here. Emit declaration of local
25938 memory space buffer for omp_simt_enter insn.
25939 (nvptx_output_unisimt_switch): New.
25940 (nvptx_output_softstack_switch): New.
25941 (nvptx_output_simt_enter): New.
25942 (nvptx_output_simt_exit): New.
25943 * config/nvptx/nvptx.h (struct machine_function): New fields
25944 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
25945 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
25946 (UNSPECV_SIMT_EXIT): Ditto.
25947 (omp_simt_enter_insn): New insn.
25948 (omp_simt_enter): New expansion.
25949 (omp_simt_exit): New insn.
25950 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
25952 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
25953 (expand_GOMP_SIMT_ENTER_ALLOC): New.
25954 (expand_GOMP_SIMT_EXIT): New.
25955 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
25956 (GOMP_SIMT_ENTER_ALLOC): Ditto.
25957 (GOMP_SIMT_EXIT): Ditto.
25958 * target-insns.def (omp_simt_enter): New insn.
25959 (omp_simt_exit): Ditto.
25960 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
25962 (lower_rec_simd_input_clauses): Implement SIMT privatization.
25963 (lower_rec_input_clauses): Likewise.
25964 (lower_lastprivate_clauses): Handle SIMT privatization.
25966 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
25967 (ompdevlow_adjust_simt_enter): New.
25968 (find_simtpriv_var_op): New.
25969 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
25970 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
25972 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
25973 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
25974 (copy_decl_for_dup_finish): Ditto.
25976 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
25978 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
25981 * config/i386/i386.c (ix86_option_override_internal): Always
25982 allow -mpreferred-stack-boundary=3 for 64-bit targets.
25984 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25986 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
25988 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25990 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
25991 mark new edge's irreducible flag accordign to it.
25992 (vect_do_peeling): Check loop preheader edge's irreducible flag
25993 and pass it to function slpeel_add_loop_guard.
25995 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
25997 PR tree-optimization/80218
25998 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25999 Update block frequencies and counts.
26001 2017-03-28 Richard Biener <rguenther@suse.de>
26003 PR tree-optimization/78644
26004 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
26005 of a simplification result we may not use it at all.
26007 2017-03-28 Richard Biener <rguenther@suse.de>
26010 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
26011 without arguments, generate default definition of a SSA name.
26013 2017-03-28 Richard Biener <rguenther@suse.de>
26015 PR middle-end/80222
26016 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
26017 TYPE_REF_CAN_ALIAS_ALL references.
26018 * fold-const.c (fold_indirect_ref_1): Likewise.
26020 2017-03-28 Martin Liska <mliska@suse.cz>
26023 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
26024 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
26026 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26027 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
26029 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
26030 (EXTRA_SPECS): Define.
26031 (SUBTARGET_EXTRA_SPECS): Likewise.
26032 (SUBTARGET_CPP_SPEC): Likewise.
26033 * config/arc/elf.h (EXTRA_SPECS): Renamed to
26034 SUBTARGET_EXTRA_SPECS.
26035 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
26037 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
26039 * config/arc/simdext.md (vst64_insn): Update pattern.
26040 (vld32wh_insn): Likewise.
26041 (vld32wl_insn): Likewise.
26042 (vld64_insn): Likewise.
26043 (vld32_insn): Likewise.
26045 2017-03-28 Marek Polacek <polacek@redhat.com>
26048 * fold-const.c (fold_comparison): Use protected_set_expr_location
26049 instead of SET_EXPR_LOCATION.
26051 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
26053 * tree.c (add_expr): Avoid name lookup warning.
26055 2017-03-27 Jeff Law <law@redhat.com>
26057 PR tree-optimization/80216
26058 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
26059 function name. Limit recursion depth.
26060 (record_temporary_equivalences): Corresponding changes.
26062 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
26064 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
26067 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26070 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
26072 * cfgcleanup.c (reg_note_cfa_p): New array.
26073 (insns_have_identical_cfa_notes): New function.
26074 (old_insns_match_p): Don't cross-jump in between /f
26075 and non-/f instructions. If both i1 and i2 are frame related,
26076 verify all CFA notes, their order and content.
26078 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
26081 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
26082 HImode and SImode with zero extend to DImode to one insn.
26083 (bswap<mode>2_extenddi): Likewise.
26084 (bswapsi2_extenddi): Likewise.
26085 (bswaphi2_extendsi): Likewise.
26086 (bswaphi2): Combine bswap HImode and SImode into one insn.
26087 Separate memory insns from swapping register.
26088 (bswapsi2): Likewise.
26089 (bswap<mode>2): Likewise.
26090 (bswaphi2_internal): Delete, no longer used.
26091 (bswapsi2_internal): Likewise.
26092 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
26093 store, and gpr<-gpr swap insns.
26094 (bswap<mode>2_store): Likewise.
26095 (bswaphi2_reg): Register only splitter, combine with the splitter.
26096 (bswaphi2 splitter): Likewise.
26097 (bswapsi2_reg): Likewise.
26098 (bswapsi2 splitter): Likewise.
26099 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
26100 the insns into load, store, and register/register insns.
26101 (bswapdi2_ldbrx): Likewise.
26102 (bswapdi2_load): Likewise.
26103 (bswapdi2_store): Likewise.
26104 (bswapdi2_reg): Likewise.
26106 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
26108 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
26109 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
26111 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
26114 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
26116 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26117 special handling for target option conflicts between dform
26118 options (-mpower9-dform, -mpower9-dform-vector,
26119 -mpower9-dform-scalar) and -mno-direct-move.
26121 2017-03-27 Richard Biener <rguenther@suse.de>
26123 PR tree-optimization/80181
26124 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
26126 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26128 * config/arc/predicates.md (move_double_src_operand): Replace the
26129 call to move_double_src_operand with a call to address_operand.
26131 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26133 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
26134 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
26135 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
26137 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26139 * config/arc/predicates.md (long_immediate_loadstore_operand):
26140 Consider scaled addresses cases.
26142 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26144 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
26145 restored when in interrupt.
26146 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
26147 doesn't have delay slot.
26149 2017-03-27 Richard Biener <rguenther@suse.de>
26152 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
26153 inlined thunk clones.
26155 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26158 * asan.c (instrument_derefs): Copy over last operand from
26159 original COMPONENT_REF to the new COMPONENT_REF with
26160 DECL_BIT_FIELD_REPRESENTATIVE.
26161 * ubsan.c (instrument_object_size): Likewise.
26163 2017-03-27 Richard Biener <rguenther@suse.de>
26165 PR tree-optimization/80170
26166 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
26167 sure DR/SCEV didnt fold in constants we do not see when looking
26168 at the reference base alignment.
26170 2017-03-27 Richard Biener <rguenther@suse.de>
26172 PR middle-end/80171
26173 * gimple-fold.c (fold_ctor_reference): Properly guard against
26174 NULL return value from canonicalize_constructor_val.
26176 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
26179 * config/i386/i386.c (ix86_expand_builtin)
26180 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
26181 flags reg setting and flags reg using instructions.
26182 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
26183 clobbering instructions to zero extend op2.
26185 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
26187 * doc/install.texi (Configuration) <--with-aix-soname>:
26188 Update link to AIX ld.
26190 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
26192 PR rtl-optimization/80160
26193 PR rtl-optimization/80159
26194 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
26195 reg_alternate_class into account.
26197 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
26200 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
26201 to consider in curr_insn_transform.
26203 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26205 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
26206 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
26207 and emit_mode_inner.
26209 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26211 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
26212 argument to the overloaded builtin variants. Use the new flag to
26213 deprecate certain builtin variants.
26214 * config/s390/s390-builtin-types.def: Add new builtin types.
26215 * config/s390/s390-builtins.h: Support new flags field for
26216 overloaded builtins.
26217 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
26218 (s390_macro_to_expand): Enable vector float data type.
26219 (s390_cpu_cpp_builtins_internal): Indicate support of the new
26220 builtins by incrementing the __VEC__ version number.
26221 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
26223 (s390_resolve_overloaded_builtin): Emit error messages depending
26224 on the builtin flags.
26225 * config/s390/s390.c (s390_expand_builtin): Support additional
26226 flags argument. Change error message to match the messages
26227 emitted in s390-c.c.
26228 * config/s390/s390.md: New UNSPEC_* constants.
26229 (op_type): Add new instruction types.
26230 * config/s390/vecintrin.h: Add new builtins and test data class
26232 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
26233 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
26234 (VEC_INEXACT, VEC_NOINEXACT): New constants.
26235 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
26236 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
26237 ("vec_mergel<mode>"): V_HW -> VEC_HW.
26239 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
26240 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
26241 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
26242 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
26244 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
26245 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
26246 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
26247 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
26249 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
26250 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
26251 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
26252 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
26253 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
26254 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
26255 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
26257 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
26258 ("vec_scatter_element<V_HW_4:mode>_DI")
26259 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
26260 ("vec_fpint<mode>", "vflls")
26261 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
26262 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
26263 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
26264 ("*vec_cmphe<mode>_cc"): ... these.
26266 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
26267 mode constant instead of magic value.
26269 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26271 * config/s390/s390.c (s390_expand_vec_compare): Support other
26272 vector floating point modes than just V2DF.
26273 (s390_expand_vcond): Likewise.
26274 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
26275 (s390_cannot_change_mode_class): Prevent mode changes between TF
26276 and V1TF in vector registers.
26277 * config/s390/s390.md (DF, SF): New mode attributes.
26278 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
26279 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
26280 SFmode support for VRs.
26281 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
26283 (VFT, VF_HW): New mode iterators.
26284 (vw, sdx): New mode attributes.
26285 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
26286 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
26287 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
26288 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
26289 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
26290 also the new vector floating point modes. Renaming to ...
26292 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
26293 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
26294 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
26295 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
26296 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
26297 ("vec_unordered<mode>"): ... these.
26299 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
26300 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
26301 ("*vec_extendv2df"): New insn definitions.
26303 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26305 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
26306 ("mulditi3_2", "*muldi3_sign"): New patterns.
26307 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
26308 rename the pattern definition.
26310 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26312 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
26314 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
26316 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26318 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
26319 instruction if possible.
26320 * config/s390/vector.md (vec_halfnumelts): New mode
26322 ("*vec_vllezlf<mode>"): New pattern.
26324 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26326 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
26327 ("popcountv4si2", "popcountv2di2"): Rename to ...
26328 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
26329 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
26331 ("popcount<mode>2_vxe"): New pattern.
26333 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26335 * common/config/s390/s390-common.c (processor_flags_table): Add
26337 * config.gcc: Add arch12.
26338 * config/s390/driver-native.c (s390_host_detect_local_cpu):
26339 Default to arch12 for unknown CPU model numbers.
26340 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
26341 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
26342 PROCESSOR_max sanity check.
26343 * config/s390/s390-opts.h (enum processor_type): Add
26345 * config/s390/s390.c (processor_table): Add arch12.
26346 (s390_expand_builtin): Add check for B_VXE flag.
26347 (s390_issue_rate): Add PROCESSOR_ARCH12.
26348 (s390_get_sched_attrmask): Likewise.
26349 (s390_get_unit_mask): Likewise.
26350 (s390_sched_score): Enable z13 scheduling for arch12.
26351 (s390_sched_reorder): Likewise.
26352 (s390_sched_variable_issue): Likewise.
26353 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
26355 (s390_tune_attr): Use z13 scheduling also for arch12.
26356 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
26357 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
26358 (TARGET_VXE_P): New macros.
26359 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
26360 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
26361 * config/s390/s390.opt: Add arch12 as processor_type.
26363 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26365 * config/s390/s390.md
26366 ("fixuns_truncdddi2", "fixuns_trunctddi2")
26367 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
26368 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
26370 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
26371 Rename expanders to ...
26373 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
26374 ("fixuns_truncdddi2_emu"): ... these.
26376 ("fixuns_trunc<mode>si2_emu"): New expander.
26378 ("*fixuns_truncdfdi2_z13"): Rename to ...
26379 ("*fixuns_truncdfdi2_vx"): ... this.
26381 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26383 * config/s390/2964.md: Remove the single element vector compare
26384 instructions which are no longer used.
26385 * config/s390/s390.c (s390_select_ccmode): Remove handling of
26387 (s390_canonicalize_comparison): Remove handling of DFmode
26389 (s390_expand_vec_compare_scalar): Remove function.
26390 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
26391 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
26393 ("*cmp<mode>_ccs"): Add wfcdb instruction.
26395 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26397 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
26399 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
26400 will anyway by matched by mov<mode>_64dfp.
26402 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26404 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
26405 vlef/vstef. Add missing operand to vleif.
26407 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26409 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
26410 pair for all vector types with 64 bit elements.
26411 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
26412 * config/s390/vector.md (V_HW_64): ... here.
26413 (V_128_NOSINGLE): New mode iterator.
26414 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
26415 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
26416 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
26417 ("*vec_load_pairv2di"): Change to ...
26418 ("*vec_load_pair<mode>"): ... this one.
26420 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26422 * config/s390/constraints.md: Add comments.
26423 (jKK): Reject element sizes > 8 bytes.
26424 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
26426 * config/s390/s390.md: Add the s_operand checks formerly in
26427 s390_split_ok_p to various splitters where they are still
26429 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
26430 for 128 bit vectors. Plus two splitters.
26432 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26434 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
26437 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26440 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
26441 error if the boundary argument is not constant.
26443 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26445 PR rtl-optimization/80112
26446 * loop-doloop.c (doloop_condition_get): Don't check condition
26447 if cmp isn't SET with IF_THEN_ELSE src.
26449 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26451 PR tree-optimization/80158
26452 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
26453 replacing a candidate statement, also replace it for the
26454 candidate's alternate interpretation.
26455 (replace_rhs_if_not_dup): Likewise.
26456 (replace_one_candidate): Likewise.
26458 2017-03-24 Richard Biener <rguenther@suse.de>
26460 PR tree-optimization/80167
26461 * graphite-isl-ast-to-gimple.c
26462 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
26464 (translate_isl_ast_to_gimple::get_rename): Likewise.
26466 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26468 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
26469 handling of certain combinations of target options, including the
26470 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
26471 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
26473 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26476 * config/arm/arm.md (*load_multiple): Add reload_completed to
26477 matching condition.
26479 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26480 Richard Biener <rguenth@suse.de>
26482 PR tree-optimization/79908
26483 PR tree-optimization/80136
26484 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26485 been cast away, gimplify_and_add suffices.
26487 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
26489 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
26491 2017-03-23 Richard Biener <rguenther@suse.de>
26493 PR tree-optimization/80032
26494 * gimplify.c (gimple_push_cleanup): Forced unconditional
26495 cleanups still have to go to the conditional_cleanups
26498 2017-03-22 Jakub Jelinek <jakub@redhat.com>
26500 PR tree-optimization/80072
26501 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
26503 (next_operand_entry_id): Change type to unsigned int.
26504 (sort_by_operand_rank): Make sure to return the right return value
26505 even if unsigned fields are bigger than INT_MAX.
26506 (struct oecount): Change cnt and id type to unsigned int.
26507 (oecount_hasher::equal): Formatting fix.
26508 (oecount_cmp): Make sure to return the right return value
26509 even if unsigned fields are bigger than INT_MAX.
26510 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
26513 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
26514 TREE_READONLY on result if writing it more than once.
26517 * doc/invoke.texi (-fsanitize=thread): Document that with
26518 -fnon-call-exceptions atomics are not able to throw
26522 * tsan.c: Include tree-eh.h.
26523 (instrument_builtin_call): Call maybe_clean_eh_stmt or
26524 maybe_clean_or_replace_eh_stmt where needed.
26525 (instrument_memory_accesses): Add cfg_changed argument.
26526 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
26527 if it returned true.
26528 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
26530 PR rtl-optimization/63191
26531 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
26532 wrapper function, moved the whole old content into ...
26533 (ix86_delegitimize_address_1): ... this. New inline function.
26534 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
26535 true as last argument instead of ix86_delegitimize_address.
26537 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26539 * config/aarch64/aarch64.c (generic_branch_cost): Copy
26540 cortexa57_branch_cost.
26542 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26544 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
26546 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26549 * doc/md.texi (Constraints): Document wA constraint.
26550 * config/rs6000/constraints.md (wA): New.
26551 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
26552 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
26553 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
26554 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
26556 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
26559 * gimplify.c (is_oacc_declared): New function.
26560 (oacc_default_clause): Use it to set default flags for acc declared
26561 variables inside parallel regions.
26562 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
26563 declared variables.
26564 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
26565 declare attribute to any decl as necessary.
26567 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
26570 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
26571 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
26572 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
26573 (arm_arch_lpae): This.
26574 * config/arm/arm.c (arm_arch7ve): Rename into ...
26575 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
26576 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
26579 2017-03-22 Martin Liska <mliska@suse.cz>
26582 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
26583 error message instead of an ICE.
26585 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26587 * doc/extend.texi (6.11 Additional Floating Types): Revise.
26589 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
26591 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26593 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26596 2017-03-21 Martin Sebor <msebor@redhat.com>
26598 * doc/extend.texi: Use "cannot" instead of "can't."
26599 * doc/hostconfig.texi: Same.
26600 * doc/install.texi: Same.
26601 * doc/invoke.texi: Same.
26602 * doc/loop.texi: Same.
26603 * doc/md.texi: Same.
26604 * doc/objc.texi: Same.
26605 * doc/rtl.texi: Same.
26606 * doc/tm.texi: Same.
26607 * doc/tm.texi.in: Same.
26608 * doc/trouble.texi: Same.
26610 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
26613 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
26614 (collect_checksum_attributes): Set it.
26615 (die_checksum_ordered): Use it.
26617 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26619 PR tree-optimization/79908
26620 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
26621 change: For a VA_ARG whose LHS has been cast away, use
26622 force_gimple_operand to construct the side effects.
26624 2017-03-21 David Malcolm <dmalcolm@redhat.com>
26626 PR translation/80001
26627 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
26628 more amenable to translation.
26629 (oacc_loop_auto_partitions): Likewise.
26631 2017-03-21 Marek Polacek <polacek@redhat.com>
26632 Martin Sebor <msebor@redhat.com>
26634 PR tree-optimization/80109
26635 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
26636 on INTEGRAL_TYPE_P.
26638 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26639 Segher Boessenkool <segher@kernel.crashing.org>
26642 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
26643 check reg_used_between_p between insn and one of succ or succ2
26644 depending on if succ is artificial insn not inserted into insn
26647 2017-03-21 Martin Liska <mliska@suse.cz>
26649 PR gcov-profile/80081
26650 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
26651 * doc/gcc.texi: Include gcov-dump stuff.
26652 * doc/gcov-dump.texi: New file.
26654 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
26656 PR rtl-optimization/79150
26657 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
26658 conditional jump, if the jump is the last insn of the loop.
26660 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26661 Richard Biener <rguenth@suse.de>
26663 PR tree-optimization/79908
26664 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26665 been cast away, use force_gimple_operand to construct the side
26668 2017-03-21 Martin Liska <mliska@suse.cz>
26670 PR libfortran/79956
26671 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
26674 2017-03-21 Brad Spengler <spender@grsecurity.net>
26677 * plugin.c (htab_hash_plugin): New function.
26678 (add_new_plugin): Use it and adjust.
26679 (parse_plugin_arg_opt): Adjust.
26680 (init_one_plugin): Likewise.
26682 2017-03-21 Richard Biener <rguenther@suse.de>
26684 PR tree-optimization/80032
26685 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
26686 if set force the cleanup to happen unconditionally.
26687 (gimplify_target_expr): Push inserted clobbers with force_uncond
26688 to avoid them being removed by control-dependent DCE.
26690 2017-03-21 Richard Biener <rguenther@suse.de>
26692 PR tree-optimization/80122
26693 * tree-inline.c (copy_bb): Do not expans va-arg packs or
26694 va_arg_pack_len when the inlined call stmt requires pack
26696 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
26698 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26701 * tsan.c (instrument_builtin_call): If the memory model argument
26702 is not a constant, assume it is valid.
26705 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
26708 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
26710 PR rtl-optimization/79910
26711 * combine.c (can_combine_p): Do not allow combining an I0 or I1
26712 if its dest is used by an insn before I2 (other than the combined
26713 insns themselves, which are properly handled already).
26715 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
26718 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26720 * combine.c (record_used_regs): New static function.
26721 (try_combine): Handle situations where there is an additional
26722 instruction between I2 and I3 which needs to have a LOG_LINK
26726 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26728 * combine.c (try_combine): Delete redundant i1 test. Call
26729 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26731 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26734 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
26735 alternatives 13/14.
26737 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26739 PR tree-optimization/80054
26740 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
26741 the optimization if a PHI or any of its arguments is not dominated
26742 by the candidate's basis. Use gphi* rather than gimple* as
26744 (replace_profitable_candidates): Clean up a gimple* variable that
26745 should be a gphi* variable.
26747 2017-03-20 Martin Sebor <msebor@redhat.com>
26750 * doc/extend.texi (attribute constructor): Document present limitation.
26752 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
26755 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
26756 __POWER9_VECTOR__ #ifdef control, change template definition to
26757 use Power9-specific built-in function.
26758 (vec_any_eq): Likewise.
26759 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
26760 to control outcomes from this test.
26761 (vector_ae_<mode>p): For VEC_F modes, likewise.
26763 2017-03-20 Ian Lance Taylor <iant@google.com>
26765 * config/i386/i386.c (ix86_function_regparm): Save an extra
26766 register for -fsplit-stack with DECL_STATIC_CHAIN.
26768 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
26771 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
26772 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
26774 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
26776 * config/riscv/riscv.c (riscv_print_operand): Use "fence
26778 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
26781 2017-03-20 Marek Polacek <polacek@redhat.com>
26784 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
26786 2017-03-20 Richard Biener <rguenther@suse.de>
26788 PR tree-optimization/80113
26789 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
26790 allocate extra SSA name for PHI def.
26791 (add_close_phis_to_outer_loops): Likewise.
26792 (add_close_phis_to_merge_points): Likewise.
26793 (copy_loop_close_phi_args): Likewise.
26794 (copy_cond_phi_nodes): Likewise.
26796 2017-03-20 Martin Liska <mliska@suse.cz>
26798 PR middle-end/79753
26799 * tree-chkp.c (chkp_build_returned_bound): Do not build
26800 returned bounds for a LHS that's not a BOUNDED_P type.
26802 2017-03-20 Martin Liska <mliska@suse.cz>
26806 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
26807 COMPLEX_CST and VECTOR_CST.
26809 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26812 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
26813 target operand. A new splitter adds the clobber statement in case
26814 the target operand is dead anyway.
26816 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
26818 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
26819 to age-old versions of binutils and glibc.
26821 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
26823 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
26825 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26827 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
26829 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26831 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
26832 requirement for binutils 2.13.
26834 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26836 * combine.c (try_combine): Delete redundant i1 test. Call
26837 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26839 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
26841 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
26842 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
26844 <riscv64-*-elf>: Re-arrange section
26845 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
26846 <riscv32-*-linux>: Likewise.
26847 <riscv64-*-elf>: Likewise
26848 <riscv64-*-linux>: Likewise.
26850 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
26853 * aarch64.opt(verbose-cost-dump): Fix typo.
26855 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
26858 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
26859 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
26861 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26863 * reload.c (find_reloads): When reloading a nonoffsettable address,
26864 use RELOAD_OTHER for it and its address reloads.
26866 PR rtl-optimization/79910
26867 * combine.c (record_used_regs): New static function.
26868 (try_combine): Handle situations where there is an additional
26869 instruction between I2 and I3 which needs to have a LOG_LINK
26872 2017-03-17 Jeff Law <law@redhat.com>
26874 PR tree-optimization/71437
26875 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
26876 conditional in the hash table first.
26877 (vrp_dom_walker::before_dom_children): Extract condition from
26878 ASSERT_EXPR. Record condition, its inverion and any implied
26879 conditions as well.
26881 2017-03-17 Marek Polacek <polacek@redhat.com>
26882 Markus Trippelsdorf <markus@trippelsdorf.de>
26884 PR tree-optimization/80079
26885 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
26888 2017-03-17 Richard Biener <rguenther@suse.de>
26890 PR middle-end/80075
26891 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
26892 Properly verify the LHS before the RHS possibly claims to be
26894 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
26897 2017-03-17 Martin Jambor <mjambor@suse.cz>
26899 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
26900 (List of -O2 options): Likewise.
26901 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
26904 2017-03-17 Tom de Vries <tom@codesourcery.com>
26906 * gcov-dump.c (print_usage): Print bug_report_url.
26908 2017-03-17 Richard Biener <rguenther@suse.de>
26910 PR middle-end/80050
26911 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
26912 (parser::peek): Likewise.
26914 2017-03-17 Richard Biener <rguenther@suse.de>
26916 PR tree-optimization/80048
26917 * sese.c (free_sese_info): Properly release rename_map and
26918 copied_bb_map elements.
26920 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
26922 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
26923 Add linked-list forward and backlinks. Insert on
26924 construction, remove on destruction.
26925 (class pass_store_merging): Add m_stores_head field.
26926 (pass_store_merging::terminate_and_process_all_chains):
26927 Iterate over m_stores_head list.
26928 (pass_store_merging::terminate_all_aliasing_chains):
26930 (pass_store_merging::execute): Check for debug stmts first.
26931 Push new chains onto the m_stores_head stack.
26933 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
26936 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
26937 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
26938 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
26940 2017-03-16 Jeff Law <law@redhat.com>
26942 PR tree-optimization/71437
26943 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
26944 member function. Implementation moved into after_dom_children
26945 member function and into the threader's thread_outgoing_edges
26947 (dom_opt_dom_walker::after_dom_children): Simplify by moving
26948 some code into new thread_outgoing_edges.
26949 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
26950 definition. Simplify marker handling (do it here). Assume we always
26951 have the available expression and the const/copies tables.
26952 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
26954 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
26955 * tree-vrp.c (equiv_stack): No longer file scoped.
26956 (vrp_dom_walker): New class.
26957 (vrp_dom_walker::before_dom_children): New member function.
26958 (vrp_dom_walker::after_dom_children): Likewise.
26959 (identify_jump_threads): Setup domwalker. Use it rather than
26960 walking edges in a random order by hand. Simplify setup/finalization.
26961 (finalize_jump_threads): Remove.
26962 (vrp_finalize): Do not call identify_jump_threads here.
26963 (execute_vrp): Do it here instead and call thread_through_all_blocks
26966 PR tree-optimization/71437
26967 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
26969 (simplify_stmt_for_jump_threading): Add basic_block argument. All
26971 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
26972 (dom_opt_dom_walker::thread_across_edge): Remove
26973 handle_dominating_asserts argument. All callers changed.
26974 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
26975 changes. Remove calls to lhs_of_dominating_assert. Other
26976 uses of handle_dominating_asserts turn into unconditional code
26977 (simplify_control_stmt_condition_1): Likewise.
26978 (simplify_control_stmt_condition): Likewise.
26979 (thread_through_normal_block, thread_across_edge): Likewise.
26980 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
26981 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
26982 object if it is not an SSA_NAME.
26983 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
26984 before calling into the VRP specific simplifiers.
26985 (identify_jump_threads): Remove handle_dominating_asserts
26988 2017-03-16 Jakub Jelinek <jakub@redhat.com>
26991 * tree-diagnostic.c (default_tree_printer): No longer static.
26992 * tree-diagnostic.h (default_tree_printer): New prototype.
26994 2017-03-16 Tamar Christina <tamar.christina@arm.com>
26996 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
26997 Change ins into fmov.
26999 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27001 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
27002 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
27003 Use h_con constraint for operand 1.
27004 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
27005 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
27007 2017-03-15 Jeff Law <law@redhat.com>
27009 PR tree-optimization/71437
27010 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
27011 (record_temporary_equivalences): Use it.
27013 PR tree-optimization/71437
27014 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
27015 tree-ssa-scopedtables.
27016 (lookup_avail_expr, build_and_record_new_cond): Likewise.
27017 (record_conditions, record_cond, vuse_eq): Likewise.
27018 (record_edge_info): Adjust to API tweak of record_conditions.
27019 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
27020 (record_temporary_equivalences, optimize_stmt): Likewise.
27021 (eliminate_redundant_computations): Likewise.
27022 (record_equivalences_from_stmt): Likewise.
27023 * tree-ssa-scopedtables.c: Include options.h and params.h.
27024 (vuse_eq): New function, moved from tree-ssa-dom.c
27025 (build_and_record_new_cond): Likewise.
27026 (record_conditions): Likewise. Accept vector of conditions rather
27027 than edge_equivalence structure for first argument.
27028 for the first argument.
27029 (avail_exprs_stack::lookup_avail_expr): New member function, moved
27030 from tree-ssa-dom.c.
27031 (avail_exprs_stack::record_cond): Likewise.
27032 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
27033 from tree-ssa-dom.c.
27034 (avail_exprs_stack): Add new member functions lookup_avail_expr
27036 (record_conditions): Declare.
27038 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
27041 * lra-constraints.c (process_alt_operands): Increase reject for
27042 reloading an input/output operand.
27044 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
27047 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
27048 insns to convert from signed/unsigned char/short to IEEE 128-bit
27050 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
27052 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
27055 * config/i386/i386.c (ix86_vector_duplicate_value): Create
27056 subreg of inner mode for values already in registers.
27058 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
27060 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
27061 iteration reg is used after the loop.
27063 2017-03-14 Martin Sebor <msebor@redhat.com>
27065 PR tree-optimization/79800
27066 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
27067 precision in negative-positive range.
27068 (format_floating): Call non-const overload with adjusted precision.
27070 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
27073 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
27076 2017-03-14 Martin Sebor <msebor@redhat.com>
27078 PR middle-end/80020
27079 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
27080 * builtins.def (aligned_alloc): Use it.
27083 * Makefile.in (GTFILES): Add calls.c.
27084 * calls.c: Include "gt-calls.h".
27086 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
27088 PR rtl-optimization/79728
27089 * regs.h (struct target_regs): New field
27090 x_contains_allocatable_regs_of_mode.
27091 (contains_allocatable_regs_of_mode): New macro.
27092 * reginfo.c (init_reg_sets_1): Initialize it, and change
27093 contains_reg_of_mode so it includes global regs as well.
27094 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
27095 rather than contains_regs_of_mode.
27097 2017-03-14 Martin Liska <mliska@suse.cz>
27099 * doc/invoke.texi: Document options that can't be combined with
27100 -fcheck-pointer-bounds.
27102 2017-03-14 Martin Liska <mliska@suse.cz>
27104 PR middle-end/79831
27105 * doc/invoke.texi (-Wchkp): Document the option.
27107 2017-03-14 Martin Liska <mliska@suse.cz>
27109 * Makefile.in: Install gcov-dump.
27111 2017-03-14 Martin Liska <mliska@suse.cz>
27113 * multiple_target.c (expand_target_clones): Bail out for
27114 an invalid attribute.
27116 2017-03-14 Richard Biener <rguenther@suse.de>
27118 * alias.c (struct alias_set_entry): Pack properly.
27119 * cfgloop.h (struct loop): Likewise.
27120 * cse.c (struct set): Likewise.
27121 * ipa-utils.c (struct searchc_env): Likewise.
27122 * loop-invariant.c (struct invariant): Likewise.
27123 * lra-remat.c (struct cand): Likewise.
27124 * recog.c (struct change_t): Likewise.
27125 * rtl.h (struct address_info): Likewise.
27126 * symbol-summary.h (function_summary): Likewise.
27127 * tree-loop-distribution.c (struct partition): Likewise.
27128 * tree-object-size.c (struct object_size_info): Likewise.
27129 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
27130 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
27131 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
27132 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
27133 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
27134 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
27135 (struct _stmt_vec_info): Likewise.
27137 2017-03-14 Martin Liska <mliska@suse.cz>
27140 * multiple_target.c (create_dispatcher_calls): Check that
27141 a target can create a function dispatcher.
27143 2017-03-14 Martin Liska <mliska@suse.cz>
27146 * multiple_target.c (expand_target_clones): Drop local.local
27147 flag for default implementation.
27149 2017-03-14 Richard Biener <rguenther@suse.de>
27151 PR tree-optimization/80030
27152 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
27154 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
27156 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
27157 gcc_fallthrough() instead of __attribute__((fallthrough));
27159 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27161 * doc/gcc.texi: Remove "up" link to (DIR).
27162 * doc/gccint.texi: Ditto.
27164 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27166 * doc/install.texi (Specific) <avr>: Remove reference to
27169 2017-03-13 Jeff Law <law@redhat.com>
27171 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
27172 attribute rather than comments.
27174 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
27175 match_scratch operand is highest.
27177 2017-03-13 Martin Liska <mliska@suse.cz>
27179 PR middle-end/78339
27180 * ipa-pure-const.c (warn_function_noreturn): If the declarations
27181 is a CHKP clone, use original declaration.
27183 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27185 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
27186 (arc_conditional_register_usage): Use a different allocation order
27187 when optimizing for size.
27188 * common/config/arc/arc-common.c (arc_option_optimization_table):
27189 Section anchors default on when optimizing for size.
27191 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27193 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
27195 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27197 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
27198 * config/arc/arc.md (cpu_facility): Add cd variant.
27199 (*movqi_insn): Add code density variant.
27200 (*movhi_insn): Likewise.
27201 (*movqi_insn): Likewise.
27202 (*addsi3_mixed): Likewise.
27203 (subsi3_insn): Likewise.
27205 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27207 * config/arc/arc.md (movsi_cond_exec): Update constraint.
27209 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27211 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
27212 expressions with MINUS and UNARY ops.
27214 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27217 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
27219 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
27220 between vec_select and vector argument.
27221 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
27222 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
27223 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
27224 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
27225 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
27226 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
27228 2017-03-13 Richard Biener <rguenther@suse.de>
27231 * params.def (vect-max-peeling-for-alignment): Fix typo.
27233 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27235 * doc/install.texi (Specific) <mips-*-*>: Remove description of
27236 issue that only occurred with binutils below 2.18.
27238 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27240 * doc/install.texi (Specific) <cris-axis-elf>: No longer
27241 refer to binutils 2.11/2.12 minimum.
27243 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27245 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
27246 ftp.kernel.org and simplify binutils requirement.
27248 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
27250 * doc/invoke.texi (Warning Options): Fix spelling of link-time
27252 (Optimize Options): Ditto. Also remove redundancy.
27254 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27256 PR translation/79848
27257 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
27259 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
27260 to G_ to avoid double translation.
27262 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27264 PR translation/79923
27265 * auto-profile.c (get_combined_location): Convert leading
27266 character of diagnostics to lower case and remove trailing period.
27267 (read_profile): Likewise for various diagnostics.
27268 * config/arm/arm.c (arm_option_override): Remove trailing period
27269 from various diagnostics.
27270 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
27271 (msp430_expand_delay_cycles): Likewise.
27273 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27276 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
27277 full command-line argument, rather than just "str".
27278 (aarch64_validate_march): Likewise.
27279 (aarch64_validate_mtune): Likewise.
27281 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
27283 PR rtl-optimization/78911
27284 * lra-assigns.c (must_not_spill_p): New function.
27285 (spill_for): Use it.
27287 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27289 PR tree-optimization/79981
27290 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
27291 ATOMIC_COMPARE_EXCHANGE ifn result.
27292 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
27293 IFN_ATOMIC_COMPARE_EXCHANGE.
27295 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27298 * opts.c (parse_sanitizer_options): Add missing question mark to
27299 "did you mean" message.
27301 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27303 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
27305 (VMULEUH_UNS): Likewise.
27306 (VMULOUB_UNS): Likewise.
27307 (VMULOUH_UNS): Likewise.
27308 * config/rs6000/rs6000.c (builtin_function_type): Remove
27309 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
27311 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27314 * read-rtl-function.c (function_reader::read_rtx_operand): Update
27315 x with result of extra_parsing_for_operand_code_0.
27316 (function_reader::extra_parsing_for_operand_code_0): Convert
27317 return type from void to rtx, returning x. When reading
27318 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
27319 larger size containing struct block_symbol.
27321 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
27323 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
27324 -mfloat128-hardware without -m64.
27326 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
27329 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
27330 entries to the case statement that marks unsigned arguments to
27331 overloaded functions.
27333 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27335 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
27336 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
27338 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
27341 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
27342 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
27344 2017-03-10 Martin Liska <mliska@suse.cz>
27348 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
27349 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
27352 2017-03-10 Olivier Hainque <hainque@adacore.com>
27354 * tree-switch-conversion (array_value_type): Start by resetting
27355 candidate type to it's main variant.
27357 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27359 PR rtl-optimization/79909
27360 * combine.c (try_combine): Use simplify_replace_rtx on individual
27361 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
27362 of the whole CALL_INSN_FUNCTION_USAGE.
27364 PR tree-optimization/79972
27365 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
27366 get_range_info on SSA_NAMEs. Formatting fixes.
27368 2017-03-10 Richard Biener <rguenther@suse.de>
27369 Jakub Jelinek <jakub@redhat.com>
27371 PR tree-optimization/77975
27372 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
27373 edge to be constant.
27374 (get_val_for): For constant x return it. Formatting fix.
27375 (loop_niter_by_eval): Avoid pointless looping if the next iteration
27376 would use the same bases as the current one.
27378 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27380 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
27381 instead of vec_select for V1TImode.
27382 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
27384 (VSX_LE_128): Add V1TI to this mode iterator.
27385 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
27386 (*vsx_le_perm_store_<mode>): Likewise.
27387 (pre-reload splitter for VSX stores): Likewise.
27388 (post-reload splitter for VSX stores): Likewise.
27389 (*vsx_xxpermdi2_le_<mode>): Likewise.
27390 (*vsx_lxvd2x2_le_<mode>): Likewise.
27391 (*vsx_stxvd2x2_le_<mode>): Likewise.
27393 2017-03-09 Michael Eager <eager@eagercon.com>
27395 Correct failures with --enable-checking=yes,rtl.
27397 * config/microblaze/microblaze.c (microblaze_expand_shift):
27398 Replace GET_CODE test with CONST_INT_P and INTVAL test with
27399 test for const0_rtx.
27400 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
27401 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
27403 2017-03-09 Richard Biener <rguenther@suse.de>
27405 PR tree-optimization/79977
27406 * graphite-scop-detection.c (scop_detection::merge_sese):
27407 Handle the case of extra exits to blocks dominating the entry.
27409 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
27411 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27414 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
27416 PR rtl-optimization/79949
27417 * lra-constraints.c (process_alt_operands): Check memory when
27418 trying to predict a cycle. Print about the overall increase.
27420 2017-03-09 Richard Biener <rguenther@suse.de>
27422 PR middle-end/79971
27423 * gimple-expr.c (useless_type_conversion_p): Preserve
27424 TYPE_SATURATING for fixed-point types.
27426 2017-03-09 Richard Biener <rguenther@suse.de>
27429 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
27430 alignment of BLKmode params.
27432 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27435 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
27436 (VALL_NO_V2Q): Likewise.
27438 * config/aarch64/aarch64-simd.md
27439 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
27441 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
27442 VALL_NO_V2Q mode iterator.
27443 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
27445 2017-03-09 Martin Liska <mliska@suse.cz>
27447 PR tree-optimization/79631
27448 * tree-chkp-opt.c (chkp_is_constant_addr): Call
27449 tree_int_cst_sign_bit just for INTEGER constants.
27451 2017-03-09 Martin Liska <mliska@suse.cz>
27455 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
27458 2017-03-09 Marek Polacek <polacek@redhat.com>
27461 * tree.c (inchash::add_expr): Handle TREE_VEC.
27463 2017-03-09 Martin Liska <mliska@suse.cz>
27466 (chkp_narrow_size_and_offset): New function.
27467 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
27468 (void chkp_parse_bit_field_ref): New function.
27469 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
27470 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
27472 2017-03-09 Martin Liska <mliska@suse.cz>
27475 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
27476 (chkp_find_bounds_1): Remove gcc_unreachable.
27478 2017-03-09 Jakub Jelinek <jakub@redhat.com>
27481 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
27482 BUILT_IN_SYNC*, determine the access type from the size suffix and
27483 always build a MEM_REF with that type. Handle forgotten
27484 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
27487 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
27488 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
27489 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
27490 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
27491 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
27492 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
27493 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
27494 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
27495 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
27496 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
27497 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
27498 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
27499 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
27500 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
27501 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
27502 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
27503 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
27504 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
27505 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
27506 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
27507 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
27508 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
27509 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
27510 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
27511 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
27512 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
27513 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
27514 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
27515 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
27516 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
27517 definitions outside of __OPTIMIZE__ guarded section.
27520 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
27521 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
27522 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
27523 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
27526 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27528 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
27529 ("vfenez<mode>"): Add missing constraints.
27531 2017-03-08 Martin Sebor <msebor@redhat.com>
27534 * config/nds32/nds32.c (nds32_option_override):
27535 Fix misspelled diagnostic.
27537 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27540 * gimplify.c (gimplify_omp_for): Replace index var in outer
27541 taskloop statement with an artificial variable and add
27542 OMP_CLAUSE_PRIVATE clause for it.
27544 2017-03-08 Richard Biener <rguenther@suse.de>
27546 PR tree-optimization/79955
27547 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
27548 for accesses that are completely outside of the variable.
27550 2017-03-08 Andrew Haley <aph@redhat.com>
27552 PR tree-optimization/79943
27553 * tree-ssa-loop-split.c (compute_new_first_bound): When
27554 calculating the new upper bound, (END-BEG) should be added, not
27557 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27559 * config/avr/avr.md (setmemhi): Make sure match_dup
27560 operand number comes before match_scratch.
27562 2017-03-08 Richard Biener <rguenther@suse.de>
27564 PR tree-optimization/79920
27565 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
27566 with ncopies == 1 to ...
27567 (vect_transform_slp_perm_load): ... here. Properly compute
27568 all element loads by iterating VF times over the group. Do
27569 not handle ncopies (computed in a broken way) in
27570 vect_create_mask_and_perm.
27572 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27575 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
27576 is a uniform vector, use uniform_vector_p return value instead of
27577 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
27579 2017-03-07 Marek Polacek <polacek@redhat.com>
27581 PR middle-end/79809
27582 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
27583 (alloca_call_type): Likewise.
27585 2017-03-07 Martin Liska <mliska@suse.cz>
27587 * gcov.c (process_args): Put comment to correct location.
27589 2017-03-07 Martin Liska <mliska@suse.cz>
27591 PR middle-end/68270
27592 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
27593 Use array_at_struct_end_p instead of DECL_CHAIN (field).
27594 (chkp_narrow_bounds_for_field): Likewise.
27595 (chkp_parse_array_and_component_ref): Pass one more argument to
27598 2017-03-07 Richard Biener <rguenther@suse.de>
27600 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
27603 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
27605 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
27606 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
27608 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27611 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
27612 to end of description.
27613 (PARAM_MAX_STORES_TO_MERGE): Likewise.
27615 2017-03-07 Jakub Jelinek <jakub@redhat.com>
27617 PR rtl-optimization/79901
27618 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
27620 (*avx512f_<code><mode>3<mask_name>): ... this.
27621 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
27622 iterator instead of VI8_AVX2_AVX512BW.
27624 PR rtl-optimization/79901
27625 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
27626 min/max expander, expand it using expand_vec_cond_expr.
27629 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
27632 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27635 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
27636 to void * for PCH reasons.
27637 * dwarf2out.c (output_loc_operands, output_die): Cast
27638 v.val_vec.array to unsigned char *.
27640 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
27643 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
27646 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
27648 PR rtl-optimization/79571
27649 * lra-constraints.c (process_alt_operands): Calculate static
27650 reject and subtract it from overall when only addresses will be
27653 2017-03-06 Julia Koval <julia.koval@intel.com>
27656 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
27657 incoming stack boundary to 128 for 64-bit targets.
27659 2017-03-06 Richard Biener <rguenther@suse.de>
27661 PR tree-optimization/79894
27662 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
27663 to NULL after folding it.
27665 2017-03-06 Richard Biener <rguenther@suse.de>
27667 PR tree-optimization/79824
27668 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
27669 check disabling peeling for gaps.
27671 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
27673 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27674 attributes): Document gettimeofday.
27676 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
27678 * config/s390/s390.c (s390_option_override_internal): Set
27679 PARAM_MIN_VECT_LOOP_BOUND
27681 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
27683 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
27684 * config/s390/s390.md: Likewise.
27686 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27689 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
27690 (<avx2_avx512>_perm<mode>): Rename to ...
27691 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
27693 (<avx512>_perm<mode>_mask): Rename to ...
27694 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
27696 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
27697 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
27698 instead of VI8F_256_512.
27699 (avx512f_perm<mode>): New define_expand.
27700 (avx512f_perm<mode>_mask): Likewise.
27701 (avx512f_perm<mode>_1<mask_name>): New define_insn.
27702 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
27704 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27706 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
27707 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
27709 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
27711 2017-03-06 Martin Liska <mliska@suse.cz>
27714 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
27715 when having a SSA NAME w/o VAR_DECL assigned to it.
27717 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27719 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
27720 msa_dpsub_<su>_d): Fix MODE for vec_select.
27722 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27724 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
27726 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
27728 2017-03-06 Richard Biener <rguenther@suse.de>
27730 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
27731 * plugin.c (register_plugin_info): Likewise.
27732 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
27734 2017-03-05 Jakub Jelinek <jakub@redhat.com>
27736 * config/i386/sse.md (sse_storehps, sse_storelps,
27737 avx_<castmode><avxsizesuffix>_<castmode>,
27738 avx512f_<castmode><avxsizesuffix>_<castmode>,
27739 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
27740 in condition that at least one operand is not a MEM.
27742 2017-03-03 Jakub Jelinek <jakub@redhat.com>
27744 PR middle-end/79805
27745 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
27746 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
27748 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
27749 gimple_call_nothrow_p flag based on whether original builtin can throw.
27750 If it can, emit following stmts on the fallthrough edge.
27751 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
27752 don't create new bb if inserting just debug stmts on the edge, try to
27753 insert them on the fallthru bb or just reset debug stmts.
27755 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
27758 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
27759 restore recog_data (including the operand rtxes inside it) around
27760 the call to get_insn_template.
27762 2017-03-03 Martin Sebor <msebor@redhat.com>
27764 PR tree-optimization/79699
27765 * context.c (context::~context): Free MPFR caches to avoid
27766 a memory leak on program exit.
27768 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27770 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
27771 Use wide_int::ulow () instead of .elt (0).
27773 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
27775 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
27776 (*pushxf): Limit oF constraint to 32bit targets and add oC
27777 constraint for 64bit targets.
27778 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
27779 (*pushdf): Change rmF constraint to rmC.
27781 2017-03-03 Martin Liska <mliska@suse.cz>
27783 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
27784 Remove unused variable.
27786 2017-03-03 Jakub Jelinek <jakub@redhat.com>
27789 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
27790 is a memory operand, increase num_memory.
27791 (ix86_expand_args_builtin): Likewise.
27793 2017-03-03 Jan Hubicka <jh@suse.cz>
27796 * ipa-devirt.c (maybe_record_node): Properly handle
27797 __cxa_pure_virtual visibility.
27799 2017-03-03 Martin Liska <mliska@suse.cz>
27801 PR tree-optimization/79803
27802 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
27804 (pass_loop_prefetch::execute): Disabled optimization if an
27805 assumption about L1 cache size is not met.
27807 2017-03-03 Martin Liska <mliska@suse.cz>
27809 PR rtl-optimization/79574
27810 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
27811 (hash_scan_set): Likewise.
27812 (dump_hash_table): Likewise.
27813 (hoist_code): Likewise.
27815 2017-03-03 Richard Biener <rguenther@suse.de>
27817 * fixed-value.c (fixed_from_string): Restore use of elt (1)
27818 in place of uhigh ().
27819 (fixed_convert_from_real): Likewise.
27821 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
27824 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
27826 2017-03-03 Richard Biener <rguenther@suse.de>
27828 PR middle-end/79818
27829 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
27830 TYPE_OVERFLOW_UNDEFINED check.
27832 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27834 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
27836 (vector_ae_<mode>_p): Likewise.
27837 (vector_nez_<mode>_p): Likewise.
27838 (vector_ne_v2di_p): Likewise.
27839 (vector_ae_v2di_p): Likewise.
27840 (vector_ne_<mode>_p): Likewise.
27841 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
27843 (vsx_tsqrt<mode>2_fe): Likewise.
27845 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
27848 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
27850 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27852 PR rtl-optimization/79780
27853 * cprop.c (one_cprop_pass): When second and further conditional trap
27854 in a single basic block is turned into an unconditional trap, turn it
27855 into a deleted note to avoid RTL verification failures.
27857 2017-03-02 Richard Biener <rguenther@suse.de>
27859 * fold-const.c (const_binop): Use ulow () instead of elt (0).
27861 2017-03-02 Richard Biener <rguenther@suse.de>
27863 PR tree-optimization/79345
27865 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
27866 param and abort the walk, returning -1 if it is hit.
27867 (walk_aliased_vdefs): Take a limit param and pass it on.
27868 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
27869 defaulting to 0 and return a signed int.
27870 * tree-ssa-uninit.c (struct check_defs_data): New struct.
27871 (check_defs): New helper.
27872 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
27873 about uninitialized memory.
27874 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
27875 bogus uninitialized warning.
27876 (fixed_convert_from_real): Likewise.
27878 2017-03-02 Bin Cheng <bin.cheng@arm.com>
27880 PR tree-optimization/66768
27881 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
27882 iv_use if base object can't be determined.
27884 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27886 PR tree-optimization/79345
27887 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
27888 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
27889 (get_pattern_stats): Initialize it.
27890 * genemit.c (gen_expand): Verify match_scratch numbers come after
27891 match_operand/match_dup numbers.
27892 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
27893 match_scratch numbers.
27894 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
27896 * config/s390/s390.md (trunctdsd2): Likewise.
27898 2017-03-02 Richard Biener <rguenther@suse.de>
27900 * wide-int.h (wide_int_storage::operator=): Implement in terms
27903 2017-03-02 Richard Biener <rguenther@suse.de>
27905 PR tree-optimization/79777
27906 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
27907 the to insert expression to sth existing.
27909 2017-03-01 Martin Sebor <msebor@redhat.com>
27911 PR middle-end/79692
27912 * gimple-ssa-sprintf.c
27913 (directive::known_width_and_precision): New function.
27914 (format_integer): Use it.
27915 (get_mpfr_format_length): Consider the full range of precision
27916 when computing %g output with the # flag. Set the likely byte
27917 count to 3 rather than 1 when precision is indeterminate.
27918 (format_floating): Correct the lower bound of precision.
27920 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27922 * doc/invoke.texi: Document default code model for 64-bit Linux.
27924 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27927 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
27928 udiv rather than div since input pattern is unsigned.
27930 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
27932 * config/i386/i386.c (print_reg): Warn for values of
27933 unsupported size in integer register.
27935 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
27938 * config/rs6000/predicates.md (current_file_function_operand): Do
27939 not allow self calls to be local if the function is replaceable.
27941 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
27944 * config/rs6000/altivec.h (vec_ctz and others): Change the
27945 preprocessor macro that controls conditional compilation from
27946 _ARCH_PWR9 to __POWER9_VECTOR__.
27947 (vec_all_ne): Change parameterization of __altivec_scalar_pred
27948 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
27949 control (instead of _ARCH_PWR9 control) so that template
27950 definition uses power9-specific function.
27951 (vec_any_eq): Likewise.
27952 (vec_all_ne): Change macro definition to use a power9-specific
27953 expansion under #ifdef __POWER9_VECTOR__ control (instead of
27954 _ARCH_PWR9 control).
27955 (vec_any_eq) Likewise.
27956 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
27957 expansion for CMPNEF to remove support for xvcmpnesp instruction.
27958 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
27959 support for xvcmpnedp instruction.
27960 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
27961 macro expansion so that Power9 implementation of vec_all_ne does
27962 not use the AltiVec predicate framework.
27963 (VCMPNEH_P): Likewise.
27964 (VCMPNEW_P): Likewise.
27965 (VCMPNED_P): Likewise.
27966 (VCMPNEFP_P): Likewise.
27967 (VCMPNEDP_P): Likewise.
27968 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
27969 implementation of vec_any_eq to not use AltiVec predicate
27971 (VCMPAEH_P): Likewise.
27972 (VCMPAEW_P): Likewise.
27973 (VCMPAED_P): Likewise.
27974 (VCMPAEFP_P): Likewise.
27975 (VCMPAEDP_P): Likewise.
27976 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
27977 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
27978 not use the AltiVec predicate framework.
27979 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
27980 of vec_any_eq to not use AltiVec predicate framework.
27981 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27982 support for predefined __POWER9_VECTOR__ macro to indicate that
27983 Power9 instruction selection is enabled.
27984 (altivec_overloaded_builtins): Remove extraneous
27985 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
27986 function argument types RS6000_BTI_bool_V16QI and
27987 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
27988 entry for overloaded function argument types RS6000_BTI_bool_V4SI
27989 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
27990 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
27991 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
27992 Power9 for implementations of vec_cmpne. Change the signature for
27993 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
27994 (representing vec_all_ne) to remove the previously described first
27995 argument of type RS6000_BTI_INTSI, as this was an artifact of
27996 reliance on the AltiVec predicate framework, which is no longer
27997 used in the implementation of these functions. Add
27998 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
27999 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
28000 since, unlike the AltiVec predicate framework implementation, we
28001 do not share function descriptors between vec_alle and vec_anyeq.
28002 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
28003 set of modes that receive special treatment even when
28004 TARGET_P9_VECTOR is true. The special treatment emits code that
28005 does not depend on Power9 instructions.
28006 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
28007 define_expand to not rely on AltiVec predicate framework.
28008 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28010 (vector_ne_v2di_p): Change this define_expand to not rely on
28011 AltiVec predicate framework.
28012 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
28014 (vector_ne_<mode>_p): Change this define_expand to not rely on
28015 AltiVec predicate framework.
28016 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
28018 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
28019 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
28020 define_insn pattern.
28021 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
28022 define_insn pattern because the xvcmpne<VSs>. instruction is not
28024 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
28025 instruction is not supported.
28027 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28029 * config/nvptx/nvptx.c: Include intl.h.
28031 2017-03-01 Martin Jambor <mjambor@suse.cz>
28034 * ipa-prop.h (ipa_bits): Removed field known.
28035 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
28036 to pointers. Adjusted their comments to warn about their sharing.
28037 (ipcp_transformation_summary): Change bits to a vector of pointers.
28038 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
28039 (ipa_get_ipa_bits_for_value): Declare.
28040 * tree-vrp.h (value_range): Mark as GTY((for_user)).
28041 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
28042 (ipa_bits_hash_table): Likewise.
28043 (ipa_vr_ggc_hash_traits): Likewise.
28044 (ipa_vr_hash_table): Likewise.
28045 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
28046 being pointers and vr_known being removed.
28047 (ipa_set_jf_unknown): Likewise.
28048 (ipa_get_ipa_bits_for_value): New function.
28049 (ipa_set_jfunc_bits): Likewise.
28050 (ipa_get_value_range): New overloaded functions.
28051 (ipa_set_jfunc_vr): Likewise.
28052 (ipa_compute_jump_functions_for_edge): Use the above functions to
28053 construct bits and vr parts of jump functions.
28054 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
28055 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28057 (ipcp_grow_transformations_if_necessary): Also allocate
28058 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
28060 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
28061 them. Fix too long lines.
28062 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
28063 vr_known being removed.
28064 (ipa_read_jump_function): Use new setter functions to construct bits
28065 and vr parts of jump functions or set them to NULL.
28066 (write_ipcp_transformation_info): Adjust for bits being pointers.
28067 (read_ipcp_transformation_info): Likewise.
28068 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
28070 Include gt-ipa-prop.h.
28071 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
28073 (ipcp_store_bits_results): Likewise.
28074 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
28075 Do not write to existing jump functions but use a temporary instead.
28077 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28080 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
28081 attempt to use its first operand as BIT_FIELD_REF base.
28083 2017-03-01 Richard Biener <rguenther@suse.de>
28085 PR middle-end/79721
28086 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
28087 interpolating formula in wrapping arithmetic.
28088 (chrec_apply): Convert chrec_evaluate return value to wanted type.
28090 2017-03-01 Jakub Jelinek <jakub@redhat.com>
28092 PR tree-optimization/79734
28093 * tree-vect-generic.c (expand_vector_condition): Optimize
28094 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
28095 Handle VEC_COND_EXPR where comparison has different inner width from
28096 type's inner width.
28098 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
28100 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
28101 markup, and similar issues. Remove @opindex entries for things
28102 that aren't options. Add missing -mmpy-option entries.
28104 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28106 PR tree-optimization/79737
28107 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
28108 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
28109 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
28110 instead of byte_size. Formatting fix.
28111 (shift_bytes_in_array_right): Formatting fix.
28113 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
28116 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
28117 condition on optimize for the leaf function test.
28119 2017-02-28 Martin Liska <mliska@suse.cz>
28122 * read-rtl-function.c (function_reader::handle_unknown_directive):
28123 Bail out when one uses -flto.
28125 2017-02-28 Martin Liska <mliska@suse.cz>
28127 * common.opt: Replace space with tabular for options of <number>
28129 * config/i386/i386.opt: Show <number> value for
28130 -mlarge-data-threshold.
28131 * opts.c (print_filtered_help): Do not display number in hexadecimal
28134 2017-02-28 Martin Liska <mliska@suse.cz>
28136 * common.opt: Fix --help=option -Q for options which are of
28139 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
28141 * config/i386/i386.c (print_reg): Error out for values
28142 of 8-bit size in invalid integer register.
28144 2017-02-28 Martin Sebor <msebor@redhat.com>
28146 PR tree-optimization/79691
28147 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
28149 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28152 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
28153 gcc_unreachable with output_operand_lossage.
28155 2017-02-28 Richard Biener <rguenther@suse.de>
28157 PR tree-optimization/79740
28158 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
28160 (visit_nary_op): Insert the nary into the hashtable if we
28161 pattern-matched sth.
28162 * tree-ssa-pre.c (eliminate_insert): Robustify.
28164 2017-02-28 Richard Biener <rguenther@suse.de>
28166 PR middle-end/79731
28167 * fold-const.c (decode_field_reference): Reject out-of-bound
28170 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28172 * config/i386/i386.c: Include intl.h.
28173 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
28174 instead of just cond ? "..." : "...".
28175 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
28176 * coverage.c (read_counts_file): Likewise.
28177 * omp-offload.c: Include intl.h.
28178 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
28179 of just cond ? "..." : "...".
28180 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
28181 of just cond ? "..." : "...".
28183 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
28186 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
28188 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
28189 'tune for' CPU name.
28190 * config/arm/arm-cpu-data.h: Regenerated.
28192 2017-02-28 Richard Biener <rguenther@suse.de>
28194 PR tree-optimization/79732
28195 * tree-inline.c (expand_call_inline): Do not shadow var.
28197 2017-02-28 Richard Biener <rguenther@suse.de>
28199 PR tree-optimization/79723
28200 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
28201 address-space properly.
28203 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
28205 * doc/optinfo.texi (Optimization groups): Fix option used for
28207 * doc/invoke.texi (-fopt-info): Document "omp".
28208 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
28209 (OPTGROUP_ALL): Add OPTGROUP_OMP.
28210 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
28211 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
28212 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
28214 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
28216 * dumpfile.c (optgroup_options): Instead of "openmp", associate
28217 OPTGROUP_OMP with "omp".
28219 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
28222 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
28223 for arithmetic shift of unsigned V2DI.
28225 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
28227 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
28228 arc/linux.h headers.
28229 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
28230 (LINK_SPEC): Likewise.
28231 (ARC_TLS_EXTRA_START_SPEC): Likewise.
28232 (EXTRA_SPECS): Likewise.
28233 (STARTFILE_SPEC): Likewise.
28234 (ENDFILE_SPEC): Likewise.
28235 (LIB_SPEC): Likewise.
28236 (TARGET_SDATA_DEFAULT): Likewise.
28237 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
28238 (MULTILIB_DEFAULTS): Likewise.
28239 (DWARF2_UNWIND_INFO): Likewise.
28240 * config/arc/big.h: New file.
28241 * config/arc/elf.h: Likewise.
28242 * config/arc/linux.h: Likewise.
28243 * config/arc/t-uClibc: Remove.
28245 2017-02-27 Bin Cheng <bin.cheng@arm.com>
28247 PR tree-optimization/77536
28248 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
28249 (tree_transform_and_unroll_loop): Use above function to compute the
28250 estimated niter of unrolled loop and use it when scaling profile.
28251 Also use count info rather than frequency if it's non-zero.
28252 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
28253 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
28254 (vect_transform_loop): Call above function.
28256 2017-02-27 Richard Biener <rguenther@suse.de>
28258 PR tree-optimization/45397
28259 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
28260 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
28261 (visit_nary_op): Add pattern matching for CSEing sign-changed
28262 or truncated operations with wider ones.
28264 2017-02-27 Richard Biener <rguenther@suse.de>
28266 PR tree-optimization/79690
28267 * tree-vect-stmts.c (vectorizable_store): Use vector type
28268 built from the DR with address-space.
28270 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
28272 * doc/invoke.texi (Optimize Options): Refine the description
28273 of asan-use-after-return.
28275 2017-02-25 Alan Modra <amodra@gmail.com>
28277 PR rtl-optimization/79584
28278 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
28279 base, not ad->base_term, the reg within base. Remove assertion
28280 that ad->base == ad->base_term. Replace gen_int_mode using
28281 bogus mode with const0_rtx.
28283 2017-02-25 Jakub Jelinek <jakub@redhat.com>
28285 PR middle-end/79396
28286 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
28287 FMA_EXPR like tcc_binary or tcc_unary.
28289 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
28292 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
28294 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
28295 (output_loc_operands): Handle DW_OP_call_ref and
28296 DW_OP_GNU_variable_value.
28297 (struct variable_value_struct): New type.
28298 (struct variable_value_hasher): Likewise.
28299 (variable_value_hash): New variable.
28300 (string_types): Remove.
28301 (copy_loc_descr): New function.
28302 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
28303 (prepend_loc_descr_to_each): New function.
28304 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
28305 instead of add_loc_descr_to_each if the first argument is single
28306 location list and the second has multiple.
28307 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
28308 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
28309 when looking for variable value which doesn't have other location info.
28310 (loc_list_from_tree): Formatting fix.
28311 (gen_array_type_die): Simplify DW_AT_string_length handling.
28312 (adjust_string_types): Remove.
28313 (gen_subprogram_die): Don't call adjust_string_types nor test/set
28314 string_types. Call resolve_variable_values.
28315 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
28316 (resolve_addr_in_expr): Likewise. Add A argument.
28317 (copy_deref_exprloc): Remove deref argument. Adjust for the
28318 original expression being DW_OP_GNU_variable_value with optionally
28319 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
28320 optionally after it.
28321 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
28322 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
28323 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
28324 (variable_value_hasher::hash, variable_value_hasher::equal): New
28326 (resolve_variable_value_in_expr, resolve_variable_value,
28327 resolve_variable_values, note_variable_value_in_expr,
28328 note_variable_value): New functions.
28329 (dwarf2out_early_finish): Call note_variable_value on all toplevel
28332 2017-02-24 Jakub Jelinek <jakub@redhat.com>
28335 * opts.h (handle_generated_option): Add GENERATED_P argument.
28336 * opts-common.c (handle_option): Adjust function comment.
28337 (handle_generated_option): Add GENERATED_P argument, pass it to
28339 (control_warning_option): Pass false to handle_generated_option
28341 * opts.c (maybe_default_option): Pass true to handle_generated_option
28343 * optc-gen.awk: Likewise.
28345 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28347 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
28348 a REG, look at the REG it is a SUBREG of.
28349 (splitter for cmpeqsi_t): Ditto.
28351 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28353 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
28354 the special USEs with the pattern of the insn, not the insn itself.
28356 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
28359 * doc/invoke.texi: Document -mload-store-pairs.
28361 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28362 Sandra Loosemore <sandra@codesourcery.com>
28364 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
28365 argument isn't a CONST_INT.
28366 (nios2_alternate_compare_const): Assert op is a CONST_INT.
28367 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
28368 (nios2_validate_compare): Bypass alternate compare logic if *op2
28369 is not a CONST_INT.
28370 (ldstwm_operation_p): Return false if first_base is not a REG or
28371 if first_offset is not a CONST_INT.
28373 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28375 * config/cris/cris.md: Use correct operand in a define_peephole2.
28377 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28379 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
28381 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28383 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
28384 this_insn if it is an INSN or JUMP_INSN.
28385 (force_offsettable): Look at base, not at addr.
28386 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
28387 on things that aren't necessarily CONST_INTs.
28389 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
28391 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
28392 -mfpmath=sse is the default also for x86-32 targets with SSE2
28393 instruction set when @option{-ffast-math} is enabled
28395 2017-02-24 Jeff Law <law@redhat.com>
28397 PR rtl-optimizatoin/79286
28398 * ira.c (update_equiv_regs): Drop may_trap_p exception to
28401 2017-02-24 Richard Biener <rguenther@suse.de>
28403 PR tree-optimization/79389
28404 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
28407 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
28409 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
28410 function comment to reflect reality.
28411 (loop_exits_before_overflow): Fix typo in function description.
28413 2017-02-24 Richard Biener <rguenther@suse.de>
28415 PR tree-optimization/79389
28416 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
28417 properly that a threading opportunity exists. Detect conditional
28418 copy/constant propagation opportunities.
28420 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
28422 * config/visium/visium.md (type): Add trap.
28423 (b): New mode attribute.
28424 (*btst): Rename into...
28425 (*btst<mode>): ...this and adjust.
28426 (*cbranchsi4_btst_insn): Rename into...
28427 (*cbranch<mode>4_btst_insn): ...this and adjust.
28428 (trap): New define_insn.
28430 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28432 PR tree-optimization/79389
28433 * ifcvt.c (struct noce_if_info): Add rev_cond field.
28434 (noce_reversed_cond_code): New function.
28435 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
28436 reversed_comparison_code. Formatting fix.
28437 (noce_try_store_flag): Test rev_cond != NULL in addition to
28438 reversed_comparison_code.
28439 (noce_try_store_flag_constants): Likewise.
28440 (noce_try_store_flag_mask): Likewise.
28441 (noce_try_addcc): Use rev_cond if non-NULL instead of
28442 reversed_comparison_code.
28443 (noce_try_cmove_arith): Likewise. Formatting fixes.
28444 (noce_try_minmax, noce_try_abs): Clear rev_cond.
28445 (noce_find_if_block): Initialize rev_cond.
28446 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
28447 instead of false as last argument never attempt to reverse it
28450 2017-02-23 Bin Cheng <bin.cheng@arm.com>
28452 PR tree-optimization/79663
28453 * tree-predcom.c (combine_chains): Process refs in reverse order
28454 only for ZERO length chains, and add explaining comment.
28456 2017-02-23 Jeff Law <law@redhat.com>
28458 PR tree-optimization/79578
28459 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
28460 in call to operand_equal_p.
28462 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
28465 * config/i386/cpuid.h: Fix another undefined behavior.
28467 2017-02-23 Richard Biener <rguenther@suse.de>
28469 PR tree-optimization/79683
28470 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
28471 vector types for data-refs.
28473 2017-02-23 Martin Liska <mliska@suse.cz>
28475 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
28477 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28479 PR middle-end/79665
28480 * internal-fn.c (get_range_pos_neg): Moved to ...
28481 * tree.c (get_range_pos_neg): ... here. No longer static.
28482 * tree.h (get_range_pos_neg): New prototype.
28483 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
28484 are known to be in between 0 and signed maximum inclusive, try to
28485 expand both unsigned and signed divmod and use the cheaper one from
28488 2017-02-22 Jeff Law <law@redhat.com>
28490 PR tree-optimization/79578
28491 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
28492 to compare base operands.
28494 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
28497 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
28498 gpc_reg_operand instead of fpr_reg_operand.
28500 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
28502 * config/mips/mips.c (mips_return_in_memory): Force FP
28503 vector types to be returned in memory for o32 ABI.
28505 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28507 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
28508 instead of DW_TAG_member for static data member declarations and don't
28509 set no_linkage_name for static inline data members.
28510 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
28513 2017-02-22 Martin Liska <mliska@suse.cz>
28515 * doc/invoke.texi: Replace inequality signs with square brackets
28518 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
28521 * lra-constraints.c (simplify_operand_subreg): Handle
28522 WORD_REGISTER_OPERATIONS targets.
28524 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28527 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
28528 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
28529 elimination by swapping fld*.
28531 2017-02-22 Richard Biener <rguenther@suse.de>
28533 PR tree-optimization/79673
28534 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
28535 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
28536 irrelevant address-space qualifiers and avoiding a
28537 ADDR_SPACE_CONVERT_EXPR from fold_convert.
28539 2017-02-22 Richard Biener <rguenther@suse.de>
28541 PR tree-optimization/79666
28542 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
28543 to not symbolically negate if that may introduce undefined
28546 2017-02-22 Martin Liska <mliska@suse.cz>
28549 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
28550 * data-streamer-out.c (streamer_write_gcov_count_stream):
28552 * value-prof.c (stream_out_histogram_value): Make assert more
28553 precise based on type of counter.
28555 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
28558 * config/i386/i386.md (standard_x87sse_constant_load splitter):
28559 Use nonimmediate_operand instead of memory_operand for operand 1.
28560 (float-extend standard_x87sse_constant_load splitter): Ditto.
28562 2017-02-21 Jeff Law <law@redhat.com>
28564 PR tree-optimization/79621
28565 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
28566 blocks with edges to themselves.
28568 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28571 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
28572 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
28573 Use gimple_call_builtin_p.
28576 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
28577 on temporarily removed DEBUG_INSNs.
28579 PR tree-optimization/79649
28580 * tree-loop-distribution.c (classify_partition): Give up on
28581 non-generic address space loads/stores.
28583 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
28585 * doc/loop.texi (Loop manipulation): Remove nonexistent
28586 tree_ssa_loop_version from the documentation.
28587 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
28589 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28592 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
28593 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
28594 * config/rs6000/rs6000.c: Include except.h.
28595 (rs6000_expand_split_stack_prologue): Call
28596 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
28598 2017-02-21 Martin Jambor <mjambor@suse.cz>
28601 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
28602 have been analyzed.
28604 2017-02-21 Martin Jambor <mjambor@suse.cz>
28606 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
28607 for backward compatibility only.
28608 * doc/invoke.texi (Option Summary): Remove all references to
28609 -fipa-cp-alignment.
28611 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
28615 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28617 * lra-constraints.c (curr_insn_transform): Handle
28618 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28620 2017-02-21 Martin Liska <mliska@suse.cz>
28622 * config/i386/i386.opt: Replace -masm-dialect with -masm.
28624 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
28626 PR translation/79638
28627 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
28629 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
28632 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
28633 (arm_function_ok_for_sibcall): Return false for an indirect call by
28634 descriptor if all the argument registers are used.
28635 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
28636 alignment of the function.
28638 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28640 PR tree-optimization/61441
28641 * simplify-rtx.c (simplify_const_unary_operation): For
28642 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
28643 the sNaN unmodified.
28645 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28647 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28648 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28649 instead of SYSTEM_HEADER_DIR.
28651 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
28652 Martin Liška <mliska@suse.cz>
28654 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
28655 Fix typos and grammar, use active voice, and clarify.
28657 2017-02-20 Marek Polacek <polacek@redhat.com>
28659 PR middle-end/79537
28660 * gimplify.c (gimplify_expr): Handle unused *&&L;.
28663 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
28665 2017-02-20 Jakub Jelinek <jakub@redhat.com>
28668 * config/i386/i386.c (ix86_expand_builtin): Handle
28669 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
28670 ix86_builtins_isa[fcode].isa as a requirement of those
28671 flags and any other flag in the bitmask.
28672 (ix86_init_mmx_sse_builtins): Use 0 instead of
28673 ~OPTION_MASK_ISA_64BIT as mask.
28674 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
28675 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
28676 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
28677 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
28679 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28682 * lra-constraints.c (split_reg): Check requested split mode
28683 is supported by the register.
28685 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28687 * lra-constraints.c (simplify_operand_subreg): Remove early
28690 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28693 * lra-constraints.c (curr_insn_transform): Tighten condition
28694 for converting SUBREG reloads from OP_OUT to OP_INOUT.
28696 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28699 * lra-constraints.c (curr_insn_transform): Handle
28700 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28702 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
28705 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
28707 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
28709 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
28712 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
28715 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28717 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
28720 2017-02-18 Jakub Jelinek <jakub@redhat.com>
28722 * final.c (last_columnnum, override_columnnum): New variables.
28723 (final_start_function): Set last_columnnum, pass it to begin_prologue
28724 hook and pass 0 to dwarf2out_begin_prologue.
28725 (final_scan_insn): Update override_columnnum. Pass last_columnnum
28726 to source_line debug hook.
28727 (notice_source_line): Compute last_columnnum and for debug_column_info
28728 return true on column changes.
28729 * debug.h (struct gcc_debug_hooks): Add column argument to
28730 source_line and begin_prologue hooks.
28731 (debug_nothing_int_charstar_int_bool): Remove prototype.
28732 (debug_nothing_int_int_charstar,
28733 debug_nothing_int_int_charstar_int_bool): New prototypes.
28734 (dwarf2out_begin_prologue): Add column argument.
28735 * debug.c (do_nothing_debug_hooks): Adjust source_line and
28736 begin_prologue hooks.
28737 (debug_nothing_int_charstar_int_bool): Remove.
28738 (debug_nothing_int_int_charstar,
28739 debug_nothing_int_int_charstar_int_bool): New functions.
28740 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
28741 through to dwarf2out_source_line.
28742 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
28743 (dwarf2out_source_line): Add column argument, emit it if requested.
28744 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
28746 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
28747 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
28748 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
28749 through to dwarf2out_begin_prologue.
28750 (vmsdbgout_source_line): Add column argument, pass it through to
28751 dwarf2out_source_line.
28752 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
28753 dbxout_source_line caller.
28754 (dbxout_source_line): Add column argument.
28756 * common.opt (gno-column-info, gcolumn-info): New options.
28757 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
28758 (check_die): Also test for multiple DW_AT_decl_column attributes.
28759 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
28760 DW_AT_decl_column if requested.
28761 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
28763 (gen_variable_die): Likewise.
28764 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
28765 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
28768 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
28769 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
28770 (ix86_handle_option): Handle OPT_m3dnowa.
28771 * doc/invoke.texi (-m3dnowa): Document.
28772 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
28773 -m3dnowa instead of -m3dnow -march=athlon.
28776 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
28777 instead of gcc_assert for K, r and R code checks. Formatting fixes.
28779 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28782 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
28783 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
28784 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
28785 generator for vsx_xxpermdi_<mode>_be.
28786 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
28787 force big-endian semantics.
28788 (vsx_xxpermdi_<mode>_be): New define_expand with same
28789 implementation as previous version of vsx_xxpermdi_<mode>.
28791 2017-02-17 Jakub Jelinek <jakub@redhat.com>
28793 PR tree-optimization/79327
28794 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
28795 variable, its initialization and use.
28797 2017-02-17 Julia Koval <julia.koval@intel.com>
28799 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
28800 (OPTION_MASK_ISA_PKU_UNSET): New.
28801 (ix86_handle_option): Handle -mrdpid.
28802 * config/i386/cpuid.h (bit_RDPID): New.
28803 * config/i386/driver-i386.c (host_detect_local_cpu):
28804 Detect RDPID feature.
28805 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
28806 * config/i386/i386-c.c (ix86_target_macros_internal):
28808 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
28809 (ix86_valid_target_attribute_inner_p): Add "rdpid".
28810 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
28811 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
28812 * config/i386/i386.md (define_insn "rdpid"): New.
28813 * config/i386/i386.opt Add -mrdpid.
28814 * config/i386/immintrin.h (_rdpid_u32): New.
28816 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
28818 PR rtl-optimization/79541
28819 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
28820 instead of transforming it into USE.
28822 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
28824 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
28825 If HONOR_SNANS (SFmode) force the input to a register.
28826 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
28827 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
28828 an frsp or similar insn.
28830 2017-02-17 Martin Liska <mliska@suse.cz>
28832 PR rtl-optimization/79577
28833 * params.def (selsched-max-sched-times): Increase minimum to 1.
28835 2017-02-17 Martin Liska <mliska@suse.cz>
28837 PR rtl-optimization/79574
28838 * gcse.c (want_to_gcse_p): Prevent integer overflow.
28840 2017-02-17 Martin Liska <mliska@suse.cz>
28842 PR tree-optimization/79529
28843 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
28844 ssa_defined_default_def_p to handle cases which are implicitly
28846 * tree-ssa.c (ssa_defined_default_def_p): New function.
28847 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
28848 which are implicitly defined.
28849 * tree-ssa.h (ssa_defined_default_def_p): Declare.
28851 2017-02-17 Richard Biener <rguenther@suse.de>
28853 PR middle-end/79576
28854 * params.def (max-ssa-name-query-depth): Limit to 10.
28856 2017-02-17 Richard Biener <rguenther@suse.de>
28858 PR tree-optimization/79552
28859 * tree-ssa-structalias.c (visit_loadstore): Properly verify
28862 2017-02-17 Richard Biener <rguenther@suse.de>
28865 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
28867 2017-02-17 Marek Polacek <polacek@redhat.com>
28869 PR middle-end/79536
28870 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
28871 (fold_negate_expr): New wrapper.
28873 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
28875 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
28876 Correct terminology and de-emphasize pre-standard behavior.
28878 2017-02-16 Alan Modra <amodra@gmail.com>
28880 PR rtl-optimization/79286
28881 * ira.c (def_dominates_uses): New function.
28882 (update_equiv_regs): Don't create an equivalence for insns that
28883 may trap where the register def does not dominate the use.
28885 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
28887 PR rtl-optimization/78127
28888 * lra.c (lra): Call lra_eliminate before finish the loop after
28891 2017-02-16 Richard Biener <rguenther@suse.de>
28893 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
28895 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
28896 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
28897 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
28898 (isl_val_int_from_wi): New function.
28899 (extract_affine_gmp): Rename to ...
28900 (extract_affine_wi): ... this, take a widest_int.
28901 (extract_affine_int): Just wrap extract_affine_wi.
28902 (add_param_constraints): Use isl_val_int_from_wi.
28903 (add_loop_constraints): Likewise, and extract_affine_wi.
28905 2017-02-15 Jeff Law <law@redhat.com>
28907 PR middle-end/79521
28908 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
28909 ira_init_register_move_cost_if_necessary.
28911 2017-02-15 Martin Sebor <msebor@redhat.com>
28913 PR middle-end/32003
28914 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
28915 removed in a prior commit.
28917 2017-02-15 Bin Cheng <bin.cheng@arm.com>
28919 PR tree-optimization/79347
28920 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
28921 counters during peeling.
28923 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
28925 * Makefile.in (site.exp): Remove "set ISLVER".
28927 2017-02-15 Jakub Jelinek <jakub@redhat.com>
28930 * real.c (real_from_integer): Call real_convert even for decimal.
28932 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
28935 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
28937 2017-02-14 Andrew Pinski <apinski@cavium.com>
28939 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
28940 cores and change the partno/implementer to be correct.
28941 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
28942 the 'B" as the implementer.
28943 * config/aarch64/aarch64-tune.md: Regenerate.
28945 2017-02-14 Carl Love <cel@us.ibm.com>
28947 * config/rs6000/rs6000.c: Add case statement entry to make the
28948 xvcvuxdsp built-in argument unsigned.
28949 * config/rs6000/vsx.md: Fix the source and return operand types so they
28950 match the instruction definitions from the ISA document. Fix typo
28951 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
28954 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
28957 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
28958 member early_clobber_alts.
28959 * lra-lives.c (reg_early_clobber_p): New.
28960 (process_bb_lives): Use it.
28961 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
28962 (debug_operand_data): Initialize early_clobber_alts.
28963 (setup_operand_alternative): Set up early_clobber_alts.
28964 (collect_non_operand_hard_regs): Ditto. Pass early clobber
28965 alternatives to new_insn_reg.
28966 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
28968 (lra_update_insn_regno_info): Pass the new arg.
28970 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28972 PR middle-end/79505
28973 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
28974 (new_oacc_loop_raw): Don't clear already cleared fields.
28977 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
28978 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
28979 _mm512_prefetch_i64gather_ps): New inline functions and macros.
28981 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
28984 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
28986 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
28989 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
28990 the extra instruction to the right place to store 128-bit constant
28993 2017-02-14 Martin Sebor <msebor@redhat.com>
28995 PR middle-end/79448
28996 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
28997 warning for strings of unknown length.
28999 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
29001 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
29003 2017-02-14 Jeff Law <law@redhat.com>
29006 * ira-costs.c (scan_one_insn): Initialize register move costs
29007 for pseudos seen in USE/CLOBBER insns.
29009 PR tree-optimization/79095
29010 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
29011 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
29012 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
29013 if the operands are known to be not equal, then the resulting range
29015 (intersect_ranges): If the new range is ~[0,0] and the old range is
29016 wide, then prefer ~[0,0].
29017 * tree-vrp.c (overflow_comparison_p_1): New function.
29018 (overflow_comparison_p): New function.
29019 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
29020 if NAME is used in an overflow test.
29021 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
29022 overflow check that can be expressed as an equality test, then adjust
29023 ops to be that equality test.
29025 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29027 * config/s390/s390-builtin-types.def: Remove flags argument.
29028 * config/s390/s390.c (s390_init_builtins): Likewise.
29030 2017-02-14 Martin Liska <mliska@suse.cz>
29032 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
29033 vector. Fix trailing white spaces.
29035 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
29037 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
29040 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29042 PR rtl-optimization/68664
29043 * config/arm/arm.c (arm_sched_can_speculate_insn):
29044 New function. Declare prototype.
29045 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29047 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29049 PR rtl-optimization/68664
29050 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
29052 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
29054 2017-02-14 Amit Pawar <amit.pawar@amd.com>
29056 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
29057 max skip bytes for function, loop and jump.
29059 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
29061 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
29062 ABS_EXPR for gimple dump.
29064 2017-02-14 Jakub Jelinek <jakub@redhat.com>
29067 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
29069 PR tree-optimization/79408
29070 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
29071 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
29072 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
29073 also if rhs1 is INTEGER_CST.
29075 2017-02-14 Richard Biener <rguenther@suse.de>
29077 PR middle-end/79432
29078 * tree-into-ssa.c (insert_phi_nodes): When the function can
29079 have abnormal edges rewrite SSA names with broken use-def
29080 dominance out of SSA and register them for PHI insertion.
29082 2017-02-13 Martin Sebor <msebor@redhat.com>
29084 PR middle-end/79496
29085 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
29086 clearing info.nowrite flag when snprintf size argument is a range.
29088 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29090 * cprop.c (cprop_jump): Add missing space in string literal.
29091 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
29092 (get_constraint_for_component_ref): Likewise.
29093 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
29094 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
29095 * lra-constraints.c (process_alt_operands): Likewise.
29096 * ipa-inline.c (inline_small_functions): Likewise.
29097 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
29098 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29099 * trans-mem.c (diagnose_tm_1_op): Likewise.
29100 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
29101 (grid_parallel_clauses_gridifiable): Likewise.
29103 * config/nvptx/mkoffload.c (process): Add space in between
29106 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
29107 "MOD4_SSE_REGS" and "ALL_REGS".
29109 * spellcheck.c (test_data): Add , in between "foo" and "food".
29111 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29114 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
29115 boundary crossing check and subsequent code generation agree.
29117 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29119 * config/aarch64/aarch64.c (has_memory_op): Delete.
29120 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
29123 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29125 PR rtl-optimization/79388
29126 PR rtl-optimization/79450
29127 * combine.c (distribute_notes): When removing TEM_INSN for which
29128 corresponding dest has last value recorded, invalidate that last
29131 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29133 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
29134 of explicit '@'. Add missing assembly comment marker on branch costs
29137 2017-02-13 Nathan Sidwell <nathan@acm.org>
29139 * gengtype-lex.l (<in_struct>): Add '/'.
29141 2017-02-13 Martin Liska <mliska@suse.cz>
29144 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
29146 2017-02-13 Richard Biener <rguenther@suse.de>
29148 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
29150 * configure: Re-generate.
29151 * config.in: Likewise.
29152 * graphite-dependences.c: Simplify as if
29153 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
29154 * graphite-isl-ast-to-gimple.c: Likewise.
29155 * graphite-optimize-isl.c: Likewise.
29156 * graphite-poly.c: Likewise.
29157 * graphite-sese-to-poly.c: Likewise.
29158 * graphite.h: Likewise.
29159 * toplev.c: Include isl/version.h and use isl_version () for
29160 printing the ISL version.
29161 * doc/install.texi: Update ISL requirement.
29163 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29165 * doc/standards.texi (Standards): Update reference to
29168 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29170 * doc/extend.texi (Named Address Spaces): sourceware.org now
29172 * doc/install.texi (Binaries): Ditto.
29175 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29177 * doc/cpp.texi: Replace "stringify"/"stringification" with C
29178 standard terminology "stringize"/"stringizing" throughout.
29179 * doc/cppinternals.texi: Likewise.
29181 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29183 * doc/extend.texi: Fix some spelling mistakes and typos.
29184 * doc/invoke.texi: Likewise.
29186 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29189 * params.def (inline-min-speedup) Change from 10 to 8.
29191 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29193 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
29196 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29199 * ipa-inline-analysis.c (get_minimal_bb): New function.
29200 (record_modified): Use it.
29201 (remap_edge_change_prob): Handle also ancestor functions.
29203 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
29205 * doc/contrib.texi (Contributors): Remove broken link into
29206 the Mauve CVS repository.
29208 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29210 PR middle-end/79454
29211 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
29212 result computation whenever lhs doesn't have vector mode, not
29213 just when it has BLKmode.
29215 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29217 * doc/makefile.texi (profiledbootstrap): Refer to the
29218 installation instructions only in textual form.
29220 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29223 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
29225 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29227 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
29228 (Specific): Update mingw-w64 reference.
29230 (Specific): Remove broken link to Renesas RX processor.
29232 2017-02-10 Richard Biener <rguenther@suse.de>
29234 * toplev.c (process_options): Do not mention obsolete graphite
29235 options when printing sorry message about missing graphite support.
29236 Mention -floop-nest-optimize.
29238 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
29240 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
29241 (vtst_p16): Likewise.
29242 (vtstq_p8): Likewise.
29243 (vtstq_p16): Likewise.
29245 (vtstq_p64): Likewise.
29246 * config/arm/arm_neon.h (vgetq_lane_p64): New.
29247 (vset_lane_p64): New.
29248 (vsetq_lane_p64): New.
29250 2017-02-10 Jakub Jelinek <jakub@redhat.com>
29252 PR tree-optimization/79411
29253 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
29254 stmt operands are SSA_NAMEs used in abnormal phis.
29255 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
29258 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29261 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
29264 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29266 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
29268 2017-02-09 Jakub Jelinek <jakub@redhat.com>
29270 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
29274 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
29275 not arbitrary TREE_CONSTANT.
29278 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
29279 "omp declare target link" attribute unless is_global_var.
29280 * omp-offload.c (find_link_var_op): Likewise.
29282 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
29283 Chung-Lin Tang <cltang@codesourcery.com>
29285 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
29287 (gimplify_adjust_omp_clauses): Don't delete TILE.
29288 (gimplify_omp_for): Deal with TILE.
29289 * internal-fn.c (expand_GOACC_TILE): New function.
29290 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
29292 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
29293 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
29295 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
29296 avoid DIV for outermost collapse var.
29297 (expand_oacc_for): Insert tile element loop as needed. Adjust.
29298 Remove out of date comments, fix whitespace.
29299 * omp-general.c (omp_extract_for_data): Deal with tiling.
29300 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
29301 adjust OLF_DIM_BASE value.
29302 (struct omp_for_data): Add tiling field.
29303 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
29304 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
29305 for auto loops. Remove default auto determining, moved to
29306 oacc_loop_fixed_partitions.
29307 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
29308 stmts, add e_mask field.
29309 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
29310 (oacc_thread_numbers): Use oacc_dim_call.
29311 (oacc_xform_tile): New.
29312 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
29313 (finish_oacc_loop): Adjust for ifns vector.
29314 (oacc_loop_discover_walk): Append loop abstraction sites to list,
29315 add case for GOACC_TILE fns.
29316 (oacc_loop_xform_loop): Delete.
29317 (oacc_loop_process): Iterate over call list directly, and add
29318 handling for GOACC_TILE fns.
29319 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
29321 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
29322 vector partitioning to outer loops. Assign 2 partitions to loops
29323 when available. Add TILE handling.
29324 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
29325 (execite_oacc_device_lower): Process GOACC_TILE fns,
29326 ignore unknown specs.
29327 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
29328 * tree.c (omp_clause_num_ops): Adjust TILE ops.
29329 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
29331 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
29333 * configure.ac (ACX_BUGURL): Update.
29334 * configure: Regenerate.
29336 2017-02-09 Richard Biener <rguenther@suse.de>
29338 PR tree-optimization/69823
29339 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
29340 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
29342 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29344 * config/arc/arc-c.def: Add __NPS400__ definition.
29345 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
29346 (TARGET_NPS400): Define.
29348 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29350 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
29352 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
29353 pointer, arch_info.
29354 (arc_cpu_types): Fill the arch_info field with a pointer into the
29355 arc_arch_types table.
29356 (arc_selected_cpu): Declare.
29357 * config/arc/arc.c (arc_selected_cpu): Make global.
29358 (arc_selected_arch): Delete.
29359 (arc_base_cpu): Delete.
29360 (arc_override_options): Remove references to deleted variables,
29361 update access to arch information.
29362 (ARC_OPT): Update access to arch information.
29363 (ARC_OPTX): Likewise.
29364 * config/arc/arc.h (arc_base_cpu): Remove declaration.
29365 (TARGET_ARC600): Update access to arch information.
29366 (TARGET_ARC601): Likewise.
29367 (TARGET_ARC700): Likewise.
29368 (TARGET_EM): Likewise.
29369 (TARGET_HS): Likewise.
29370 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
29373 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
29376 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
29377 condition/operands for integer GE/LE/GEU/LEU operations.
29379 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
29381 PR translation/79397
29382 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
29385 2017-02-08 Martin Jambor <mjambor@suse.cz>
29388 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
29389 whether allocation happened.
29390 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
29391 nothing was allocated.
29393 2017-02-08 Jakub Jelinek <jakub@redhat.com>
29395 PR tree-optimization/79408
29396 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
29397 constant, but SSA_NAME with a known integer range, use the minimum
29398 of that range instead of op1 to determine if modulo can be replaced
29399 with its first operand.
29401 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29403 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
29405 2017-02-08 Richard Biener <rguenther@suse.de>
29407 PR tree-optimization/71824
29408 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29409 Check all loops contained in the merged region.
29411 2017-02-07 Andrew Pinski <apinski@cavium.com>
29413 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
29415 2017-02-07 Andrew Pinski <apinski@cavium.com>
29417 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
29418 (thunderxt88): Likewise.
29419 (thunderxt81): Disable LSE and change v8.1 to v8.
29420 (thunderxt83): Likewise.
29422 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29423 Richard Biener <rguenther@suse.de>
29425 PR middle-end/79399
29426 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
29427 type from int to size_t.
29428 * ira-costs.c (struct_costs_size): Change type from int to size_t.
29430 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29432 PR rtl-optimization/79386
29433 * cprop.c (bypass_conditional_jumps): Initialize
29434 bypass_last_basic_block already before splitting bbs after
29435 unconditional traps...
29436 (bypass_conditional_jumps): ... rather than here.
29439 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
29440 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
29441 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
29442 fix -masm=intel patterns.
29444 2017-02-07 Richard Biener <rguenther@suse.de>
29446 PR tree-optimization/79256
29447 PR middle-end/79278
29448 * builtins.c (get_object_alignment_2): Use min_align_of_type
29449 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
29450 and ADJUST_FIELD_ALIGN.
29452 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
29454 * doc/tm.texi: Regenerate.
29455 * stor-layout.c (layout_decl): Adjust.
29456 (update_alignment_for_field): Likewise.
29457 (place_field): Likewise.
29458 (min_align_of_type): Likewise.
29459 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
29460 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
29461 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
29462 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
29463 * config/frv/frv.c (frv_adjust_field_align): Likewise.
29464 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
29465 * config/i386/i386.c (x86_field_alignment): Likewise.
29466 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
29467 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
29468 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
29469 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
29470 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
29471 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
29475 2017-01-30 Richard Biener <rguenther@suse.de>
29477 PR tree-optimization/79256
29478 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29479 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29482 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
29484 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
29485 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
29486 builtins to SImode and emit a zero-extend, if necessary.
29488 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29490 * docs/invoke.texi (RISC-V Options): Alphabetize.
29492 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29494 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
29497 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29499 * config/riscv/riscv.c: New file.
29500 * common/config/riscv/riscv-common.c: Likewise.
29501 * config.gcc: Likewise.
29502 * config/riscv/constraints.md: Likewise.
29503 * config/riscv/elf.h: Likewise.
29504 * config/riscv/generic.md: Likewise.
29505 * config/riscv/linux.h: Likewise.
29506 * config/riscv/multilib-generator: Likewise.
29507 * config/riscv/peephole.md: Likewise.
29508 * config/riscv/pic.md: Likewise.
29509 * config/riscv/predicates.md: Likewise.
29510 * config/riscv/riscv-builtins.c: Likewise.
29511 * config/riscv/riscv-c.c: Likewise.
29512 * config/riscv/riscv-ftypes.def: Likewise.
29513 * config/riscv/riscv-modes.def: Likewise.
29514 * config/riscv/riscv-opts.h: Likewise.
29515 * config/riscv/riscv-protos.h: Likewise.
29516 * config/riscv/riscv.h: Likewise.
29517 * config/riscv/riscv.md: Likewise.
29518 * config/riscv/riscv.opt: Likewise.
29519 * config/riscv/sync.md: Likewise.
29520 * config/riscv/t-elf-multilib: Likewise.
29521 * config/riscv/t-linux: Likewise.
29522 * config/riscv/t-linux-multilib: Likewise.
29523 * config/riscv/t-riscv: Likewise.
29524 * configure.ac: Likewise.
29525 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
29526 Waterman as RISC-V maintainers.
29527 * doc/install.texi: Add RISC-V entries.
29528 * doc/invoke.texi: Add RISC-V options section.
29529 * doc/md.texi: Add RISC-V constraints section.
29530 * configure: Regenerated.
29532 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
29535 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
29536 false values to be constant vectors with all 0 or all 1 bits set.
29537 (vcondu<mode><mode>): Likewise.
29538 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
29540 (fpmask_comparison_operator): Update comment.
29541 (vecint_comparison_operator): New predicate.
29542 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
29543 vector conditionals when the true and false values are constant
29544 vectors with all 0 bits or all 1 bits set.
29546 2017-02-06 Martin Sebor <msebor@redhat.com>
29548 PR tree-optimization/79376
29549 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
29551 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
29553 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
29554 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
29555 to simplify split condition.
29557 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29559 * omp-expand.c (oxpand_omp_atomic_fetch_op,
29560 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
29563 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
29565 PR rtl-optimization/68664
29566 * target.def (can_speculate_insn): New hook.
29567 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
29568 * doc/tm.texi: Regenerate.
29569 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
29570 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
29571 (rs6000_sched_can_speculate_insn): New function.
29573 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29575 PR tree-optimization/79284
29576 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
29577 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
29578 vectorizable_mask_load_store, vectorizable_operation,
29579 vect_is_simple_cond, get_same_sized_vectype): Use it instead
29580 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
29581 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
29582 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
29583 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
29584 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
29585 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
29586 is_gimple_assign (stmt). Replace another such test with
29587 is_gimple_assign (stmt).
29589 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29592 * config/avr/avr.c (rtl-iter.h): Include it.
29593 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
29594 (avr_legitimate_combined_insn): ...and implementation.
29596 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
29598 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
29599 * config/s390/s390.c (s390_const_operand_ok)
29600 (s390_canonicalize_comparison, s390_extract_part)
29601 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
29602 (s390_contiguous_bitmask_p, s390_rtx_costs)
29603 (legitimize_pic_address): Likewise.
29604 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
29605 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
29606 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
29607 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
29608 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
29610 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29612 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
29613 REGNO($0) == REGNO($1).
29615 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29617 * config/s390/linux.h(SIZE_TYPE): Add comment.
29619 2017-02-06 Julian Brown <julian@codesourcery.com>
29620 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29621 Virendra Pathak <virendra.pathak@broadcom.com>
29623 * config/aarch64/aarch64-cores.def: Change the scheduler
29625 * config/aarch64/aarch64.md: Include thunderx2t99.md.
29626 * config/aarch64/thunderx2t99.md: New file.
29628 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29630 * doc/standards.texi (Go Language): Update link to language
29633 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29635 * tree-eh.c (lower_resx): Sanitize profile.
29636 (cleanup_empty_eh_move_lp): Likewise.
29638 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29641 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
29643 * cfgloopmanip.h (loop_version): Update prototype.
29644 * modulo-sched.c (sms_schedule): Update call of loop_version.
29645 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
29646 * tree-parloops.c (gen_parallel_loop): Likewise.
29647 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
29648 * tree-ssa-loop-split.c (split_loop): Likewise.
29649 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
29650 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
29652 2017-02-05 Martin Liska <mliska@suse.cz>
29655 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
29657 (print_operand_address): Initialize a struct to zero.
29659 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29661 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
29662 garbage collector only in textual form.
29664 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29666 * doc/extend.texi (x86 specific memory model extensions for
29667 transactional memory): Simplify a phrase.
29669 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
29672 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
29673 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
29674 (atomic_storedi_1): Likewise.
29676 2017-02-04 Jakub Jelinek <jakub@redhat.com>
29678 PR tree-optimization/79338
29679 * tree-parloops.c (gather_scalar_reductions): Don't call
29680 vect_analyze_loop_form for loop->inner before destroying loop's
29683 2017-02-03 Martin Sebor <msebor@redhat.com>
29685 PR tree-optimization/79327
29686 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
29687 when precision has resulted in leading zeros.
29688 (format_integer): Adjust the likely counter to assume an unknown
29689 argument that may be zero is non-zero.
29691 2017-02-03 Jason Merrill <jason@redhat.com>
29694 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
29695 avoid copying non-taken branch.
29697 2017-02-03 Jakub Jelinek <jakub@redhat.com>
29699 PR tree-optimization/79340
29700 * tree-vect-loop.c (vectorizable_reduction): Release
29701 vec_defs elements after safe_splicing them into other vectors.
29704 PR tree-optimization/79327
29705 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
29706 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
29708 (format_integer): Use wide_int_to_tree instead of build_int_cst
29709 + to_?hwi. If argmin is NULL, just set argmin and argmax to
29710 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
29711 of shortest and longest sequence.
29713 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
29715 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
29716 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
29718 2017-02-03 Walter Lee <walt@tilera.com>
29721 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
29722 after initial stackframe link reg save.
29723 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
29725 2017-02-03 Jakub Jelinek <jakub@redhat.com>
29728 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
29729 wu for stxssp alternative.
29731 2017-02-03 Martin Sebor <msebor@redhat.com>
29733 PR tree-optimization/79352
29734 * gimple-fold.c (get_range_strlen): Add argument.
29735 (get_range_strlen): Change return type to bool.
29736 (get_maxval_strlen): Pass in a dummy argument.
29737 * gimple-fold.h (get_range_strlen): Change return type to bool.
29738 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
29739 * tree.h (array_at_struct_end_p): Add argument.
29740 * tree.c (array_at_struct_end_p): Handle it.
29742 2017-02-03 Martin Liska <mliska@suse.cz>
29745 * multiple_target.c (create_dispatcher_calls): Redirect edge
29746 from a caller of a dispatcher.
29747 (expand_target_clones): Make the clones local.
29748 (ipa_target_clone): Do both target clones and resolvers.
29749 (ipa_dispatcher_calls): Remove the pass.
29750 (pass_dispatcher_calls::gate): Likewise.
29751 (make_pass_dispatcher_calls): Likewise.
29752 * passes.def (pass_target_clone): Put as very first IPA early
29755 2017-02-03 Martin Liska <mliska@suse.cz>
29757 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
29758 in case of a function with ifunc attribute.
29760 2017-02-03 Martin Liska <mliska@suse.cz>
29762 * cgraph.c (cgraph_node::dump): Dump function version info.
29763 * symtab.c (symtab_node::dump_base): Add missing new line.
29765 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
29767 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
29768 (ifcombine_ifandif): Use it.
29770 2017-02-03 Martin Liska <mliska@suse.cz>
29772 * doc/invoke.texi: Document default value for
29773 use-after-scope-direct-emission-threshold.
29775 2017-02-03 Martin Liska <mliska@suse.cz>
29777 PR tree-optimization/79339
29778 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
29779 (format_floating): Likewise.
29781 2017-02-03 Martin Liska <mliska@suse.cz>
29784 * ipa-prop.c (ipa_node_params_t::insert): Remove current
29786 (ipa_node_params_t::remove): Likewise.
29787 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
29788 initialization from removed ipa_node_params_t::insert.
29789 (ipa_node_params::~ipa_node_params): Move from removed
29790 ipa_node_params_t::release.
29791 * symbol-summary.h (symbol_summary::m_released): New member.
29792 Do not release a summary twice. Do not allow to call finalizer
29793 for types of a summary that live in GGC memory.
29795 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29797 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
29800 2017-02-02 Martin Sebor <msebor@redhat.com>
29802 PR middle-end/79275
29803 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
29804 (format_string): Tighten up the range of output for non-constant
29805 strings and correct the expected range for wide non-constant strings.
29807 2017-02-02 Martin Sebor <msebor@redhat.com>
29809 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
29811 PR middle-end/32003
29812 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
29814 (-fdump-tree-@var): Add to index and document how to come up
29815 with pass-specific option and dump file names.
29816 (-fdump-passes): Clarify where to look for output.
29818 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
29820 PR middle-end/77445
29821 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
29822 statistics of the analyzed path; allow threading for speed when
29823 any of BBs along the path are optimized for speed.
29825 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
29827 PR middle-end/78468
29828 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
29829 settings of the virtual registers.
29832 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
29834 * explow.c (get_dynamic_stack_size): Take known alignment of stack
29835 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
29838 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29840 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
29841 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
29843 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29845 * config/s390/s390.md: Add missing comments with the expanded
29847 * config/s390/vector.md: Likewise.
29848 * config/s390/vx-builtins.md: Likewise.
29850 2017-02-02 Jakub Jelinek <jakub@redhat.com>
29853 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
29854 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
29855 conditions on a single line.
29857 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29859 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29860 __S390_VX__ to __VX__.
29862 2017-02-01 Andrew Pinski <apinski@cavium.com>
29864 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
29865 stmt_info to record_stmt_cost.
29866 (vect_get_known_peeling_cost): Pass stmt_info if known to
29868 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
29869 cpu_vector_cost field into
29870 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
29871 field into vec_int_stmt_cost and vec_fp_stmt_cost.
29872 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
29873 splitting of scalar_stmt_cost and vec_stmt_cost.
29874 (thunderx_vector_cost): Likewise.
29875 (cortexa57_vector_cost): LIkewise.
29876 (exynosm1_vector_cost): Likewise.
29877 (xgene1_vector_cost): Likewise.
29878 (thunderx2t99_vector_cost): Improve after the splitting of the two
29880 (aarch64_builtin_vectorization_cost): Update for the splitting of
29881 scalar_stmt_cost and vec_stmt_cost.
29883 2017-02-01 Torvald Riegel <triegel@redhat.com>
29884 Richard Henderson <rth@redhat.com>
29886 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
29887 conditional on existance of a fast atomic load.
29888 * optabs-query.c (can_atomic_load_p): New function.
29889 * optabs-query.h (can_atomic_load_p): Declare it.
29890 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
29891 no fast atomic load is available for the particular size of access.
29892 (expand_atomic_compare_and_swap): Likewise.
29893 (expand_atomic_load): Likewise.
29894 (expand_atomic_store): Likewise.
29895 (expand_atomic_fetch_op): Likewise.
29896 * testsuite/lib/target-supports.exp
29897 (check_effective_target_sync_int_128): Remove x86 because it provides
29898 no fast atomic load.
29899 (check_effective_target_sync_int_128_runtime): Likewise.
29901 2017-02-01 Richard Biener <rguenther@suse.de>
29903 * graphite.c: Include tree-vectorizer.h for find_loop_location.
29904 (graphite_transform_loops): Provide opt-info for optimized nests.
29905 * tree-parloop.c (parallelize_loops): Provide opt-info for
29906 parallelized loops.
29908 2017-02-01 Richard Biener <rguenther@suse.de>
29910 PR middle-end/79315
29911 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
29912 was not set before.
29914 2017-02-01 Richard Biener <rguenther@suse.de>
29916 PR tree-optimization/71824
29917 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29918 Verify the loops are valid in the merged SESE region.
29919 (scop_detection::can_represent_loop_1): Check analyzing the
29920 evolution of the number of iterations in the region succeeds.
29922 2017-01-31 Ian Lance Taylor <iant@golang.org>
29924 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
29925 REG_ARGS_SIZE note to 32-bit push insns and call insn.
29927 2017-01-31 David Malcolm <dmalcolm@redhat.com>
29929 PR preprocessor/79210
29930 * input.c (get_substring_ranges_for_loc): Replace line_width
29931 assertion with error-handling.
29933 2017-01-31 Richard Biener <rguenther@suse.de>
29935 PR tree-optimization/77318
29936 * graphite-sese-to-poly.c (extract_affine): Fix assert.
29937 (create_pw_aff_from_tree): Take loop parameter.
29938 (add_condition_to_pbb): Pass loop of the condition to
29939 create_pw_aff_from_tree.
29941 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29943 * config/s390/s390.c (s390_asan_shadow_offset): New function.
29944 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
29946 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
29950 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
29952 (convert_int_to_float128): Likewise.
29953 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
29954 (convert_int_to_float128): Likewise.
29955 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
29956 (UNSPEC_IEEE128_CONVERT): Likewise.
29957 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
29958 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
29959 Use local variables for IBM extended format.
29960 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
29961 (fix_trunc<mode>si2_fprs): Likewise.
29962 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
29963 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
29964 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
29965 to know that we can now have integers of all sizes in vector
29967 (fix<uns>_<mode>di2_hw): Likewise.
29968 (float<uns>_<mode>si2_hw): Likewise.
29969 (fix_<mode>si2_hw): Likewise.
29970 (fixuns_<mode>si2_hw): Likewise.
29971 (float<uns>_<mode>di2_hw): Likewise.
29972 (float_<mode>di2_hw): Likewise.
29973 (float_<mode>si2_hw): Likewise.
29974 (floatuns_<mode>di2_hw): Likewise.
29975 (floatuns_<mode>si2_hw): Likewise.
29976 (xscvqp<su>wz_<mode>): Delete, no longer used.
29977 (xscvqp<su>dz_<mode>): Likewise.
29978 (xscv<su>dqp_<mode>): Likewise.
29979 (ieee128_mfvsrd_64bit): Likewise.
29980 (ieee128_mfvsrd_32bit): Likewise.
29981 (ieee128_mfvsrwz): Likewise.
29982 (ieee128_mtvsrw): Likewise.
29983 (ieee128_mtvsrd_64bit): Likewise.
29984 (ieee128_mtvsrd_32bit): Likewise.
29986 2017-01-31 Martin Liska <mliska@suse.cz>
29989 * ipa-prop.c (ipa_free_all_node_params): Call release method
29990 instead of ~sumbol_summary to not to trigger double times
29993 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
29995 PR tree-optimization/71691
29996 * bitmap.h (class auto_bitmap): New.
29997 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
29998 is_maybe_undefined instead of ssa_undefined_value_p.
30000 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
30002 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
30003 __S390_ARCH_LEVEL__ to __ARCH__.
30005 2017-01-31 Jakub Jelinek <jakub@redhat.com>
30007 PR tree-optimization/79267
30008 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
30009 if should_remove_lhs_p is true.
30011 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
30014 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
30015 (add_alignment_attribute): New.
30016 (base_type_die): Add alignment attribute.
30017 (subrange_type_die): Likewise.
30018 (modified_type_die): Likewise.
30019 (gen_array_type_die): Likewise.
30020 (gen_descr_array_type_die: Likewise.
30021 (gen_enumeration_type_die): Likewise.
30022 (gen_subprogram_die): Likewise.
30023 (gen_variable_die): Likewise.
30024 (gen_field_die): Likewise.
30025 (gen_ptr_to_mbr_type_die): Likewise.
30026 (gen_struct_or_union_type_die): Likewise.
30027 (gen_subroutine_type_die): Likewise.
30028 (gen_typedef_die): Likewise.
30029 (base_type_cmp): Compare alignment attribute.
30031 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
30034 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
30035 (setb_unsigned) New pattern for setb with CCUNS.
30036 * config/rs6000/rs6000.c (expand_block_compare): Use a different
30037 subfc./subfe sequence to avoid overflow problems. Generate a
30038 shorter sequence with cmpld/setb for power9.
30039 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
30040 for generating subfc. instruction.
30041 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
30042 now uses this instruction.
30044 2017-01-30 Ian Lance Taylor <iant@google.com>
30047 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
30048 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
30050 2017-01-30 Martin Sebor <msebor@redhat.com>
30052 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
30053 Move constant to the right of a relational operator.
30054 (get_mpfr_format_length, format_character, format_string): Ditto.
30055 (should_warn_p, maybe_warn): Same.
30057 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
30059 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
30062 * asan.c (get_translation_unit_decl): Remove function.
30063 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
30065 2017-01-30 Martin Liska <mliska@suse.cz>
30067 PR gcov-profile/79259
30068 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
30069 -fprofile-generate.
30071 2017-01-30 Martin Liska <mliska@suse.cz>
30074 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
30075 Initialize variables with NULL value.
30077 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
30080 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
30082 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
30084 2017-01-30 Richard Biener <rguenther@suse.de>
30086 PR tree-optimization/79276
30087 * tree-vrp.c (process_assert_insertions): Properly adjust common
30088 when removing a duplicate.
30090 2017-01-30 Richard Biener <rguenther@suse.de>
30092 PR tree-optimization/79256
30093 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30094 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30096 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
30098 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
30101 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
30102 ("*r<noxa>sbg_<mode>_sll_bitmask")
30103 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
30104 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
30105 Use contiguous_bitmask_nowrap_operand.
30107 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30110 * config/rs6000/altivec.h (vec_xl): Revise #define.
30111 (vec_xst): Likewise.
30113 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
30115 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
30117 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
30119 PR rtl-optimization/79194
30120 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
30121 traps before call to bypass_conditional_jumps.
30123 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30125 PR tree-optimization/71374
30126 * lra-constraints.c (check_conflict_input_operands): New.
30127 (match_reload): Use it.
30129 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30132 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
30133 account to calculate conflict_set.
30135 2017-01-27 Bin Cheng <bin.cheng@arm.com>
30137 PR rtl-optimization/78559
30138 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
30139 other_insn in combine.
30141 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30143 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
30144 uint16_type_node for BT_UINT16.
30146 2017-01-27 David Malcolm <dmalcolm@redhat.com>
30148 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
30149 "RTL Tests" to menu.
30150 (GIMPLE Tests): New node.
30151 (RTL Tests): New node.
30153 2017-01-27 Richard Biener <rguenther@suse.de>
30155 PR tree-optimization/79245
30156 * tree-loop-distribution.c (distribute_loop): Apply cost
30157 modeling also to detected patterns.
30159 2017-01-27 Richard Biener <rguenther@suse.de>
30161 PR tree-optimization/71433
30162 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
30163 (compare_assert_loc): New function.
30164 (process_assert_insertions): Sort and optimize assert locations
30165 to remove duplicates and push down identical assertions on
30166 edges to their destination block.
30168 2017-01-27 Richard Biener <rguenther@suse.de>
30170 PR tree-optimization/79244
30171 * tree-vrp.c (remove_range_assertions): Forcefully propagate
30172 out SSA names even if abnormal.
30174 2017-01-27 Jakub Jelinek <jakub@redhat.com>
30176 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
30177 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
30178 instead of MPFR_RNDN.
30180 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
30183 * arm.c (arm_option_override): Don't call build_target_option_node
30184 until after doing all option overrides.
30185 (arm_valid_target_attribute_tree): Likewise.
30187 2017-01-27 Martin Liska <mliska@suse.cz>
30189 * doc/invoke.texi (-fprofile-arcs): Document profiling support
30190 for {cd}tors and C++ {cd}tors.
30192 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
30194 * config/s390/s390.md ("*setmem_long_and")
30195 ("*setmem_long_and_31z"): Use zero_extend instead of and.
30197 2017-01-26 Martin Sebor <msebor@redhat.com>
30199 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
30202 2017-01-26 Martin Sebor <msebor@redhat.com>
30204 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
30205 HAVE_DFmode before using XFmode or DFmode.
30206 (parse_directive): Avoid using the z length modifier to avoid
30207 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
30209 PR middle-end/78703
30210 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
30211 to accept adjustment as an array.
30212 (get_int_range): New function.
30213 (struct directive): Make width and prec arrays.
30214 (directive::set_width, directive::set_precision): Call get_int_range.
30215 (format_integer, format_floating): Handle width and precision ranges.
30216 (format_string, parse_directive): Same.
30218 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30221 * dwarf2out.c (generate_skeleton_bottom_up): For children with
30222 comdat_type_p set, just clone them, but keep the children in the
30226 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
30227 which have direct callers with -fvar-tracking-assignments enabled
30229 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
30230 inside of type units.
30232 2017-01-26 Martin Sebor <msebor@redhat.com>
30234 PR middle-end/78703
30235 * gimple-ssa-sprintf.c (struct result_range): Add likely and
30237 (struct format_result): Replace number_chars, number_chars_min,
30238 and number_chars_max with a single member of struct result_range.
30240 (format_result::operator+=): Adjust.
30241 (struct fmtresult): Remove bounded. Handle likely and unlikely
30243 (fmtresult::adjust_for_width_or_precision): New function.
30244 (fmtresult:type_max_digits): New function.
30245 (bytes_remaining): Handle likely and unlikely counters.
30246 (min_bytes_remaining): Remove.
30247 (format_percent): Simplify.
30248 (format_integer, format_floating): Set likely and unlikely counters.
30249 (get_string_length, format_character, format_string): Same.
30250 (format_plain, should_warn_p): New function.
30251 (maybe_warn): Call should_warn_p. Update diagnostic messages
30252 and handle those for all directives, including plain strings.
30253 (format_directive): Handle likely and unlikely counters.
30254 Remove unnecessary quoting from diagnostics. Add an informational
30256 (add_bytes): Remove.
30257 (pass_sprintf_length::compute_format_length): Simplify.
30258 (try_substitute_return_value): Handle likely and unlikely counters.
30260 2017-01-26 Carl Love <cel@us.ibm.com>
30262 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
30263 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
30265 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
30268 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
30269 endianess for subregs into account.
30270 * lra-constraints.c (lra_constraints): Do risky transformations
30271 always on the first iteration.
30272 * lra-lives.c (check_pseudos_live_through_calls): Add arg
30273 last_call_used_reg_set.
30274 (process_bb_lives): Define and use last_call_used_reg_set.
30275 * lra.c (lra): Always continue after lra_constraints on the first
30278 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
30280 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
30282 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
30284 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30286 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
30287 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
30288 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
30289 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
30290 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
30291 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
30292 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
30293 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
30294 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
30296 2017-01-26 Marek Polacek <polacek@redhat.com>
30299 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
30300 for the third operand.
30302 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30304 PR middle-end/79236
30305 * omp-low.c (struct omp_context): Add simt_stmt field.
30306 (scan_omp_for): Return omp_context *.
30307 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
30308 context to the _simt_ SIMD stmt.
30309 (lower_omp_for): For combined SIMD with sibling _simt_
30310 SIMD, make sure to use the same decls in _looptemp_
30311 clauses as in the sibling.
30313 2017-01-26 David Sherwood <david.sherwood@arm.com>
30315 PR middle-end/79212
30316 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
30319 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30322 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
30323 emit fld b; fld a; if possible.
30325 * brig-builtins.def: Update copyright years.
30326 * config/arm/arm_acle_builtins.def: Update copyright years.
30328 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
30331 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
30332 constraint instead of o for the stxsd instruction.
30334 2017-01-25 Carl Love <cel@us.ibm.com>
30336 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
30337 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
30339 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
30341 * doc/invoke.texi (C++ Dialect Options): Fix typo.
30343 2017-01-25 Richard Biener <rguenther@suse.de>
30345 PR tree-optimization/69264
30346 * target.def (vector_alignment_reachable): Improve documentation.
30347 * doc/tm.texi: Regenerate.
30348 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
30350 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
30351 earlier changes with respect to TYPE_USER_ALIGN.
30352 (vector_alignment_reachable_p): Likewise. Improve dumping.
30354 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30357 * config/arm/arm.md (xordi3): Force constant operand into a register
30360 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30362 * doc/invoke.texi (-fstore-merging): Correct default optimization
30363 levels at which it is enabled.
30364 (-O): Move -fstore-merging from list to...
30367 2017-01-25 Richard Biener <rguenther@suse.de>
30370 * omp-expand.c: Include debug.h.
30371 (expand_omp_taskreg): Make sure to generate early debug before
30372 outlining anything from a function.
30373 (expand_omp_target): Likewise.
30374 (grid_expand_target_grid_body): Likewise.
30376 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
30379 * asan.c (get_translation_unit_decl): New function.
30380 (asan_add_global): Extract modules file name from globals
30381 TRANSLATION_UNIT_DECL name.
30383 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
30386 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
30387 for long calls with APCS frame and VFP.
30389 2017-01-24 David Malcolm <dmalcolm@redhat.com>
30391 * cfg.c (original_copy_tables_initialized_p): New function.
30392 * cfg.h (original_copy_tables_initialized_p): New decl.
30393 * cfgrtl.c (relink_block_chain): Guard the call to
30394 free_original_copy_tables with a call to
30395 original_copy_tables_initialized_p.
30396 * cgraph.h (symtab_node::native_rtl_p): New decl.
30397 * cgraphunit.c (symtab_node::native_rtl_p): New function.
30398 (symtab_node::needed_p): Don't assert for early assembly output
30399 for __RTL functions.
30400 (cgraph_node::finalize_function): Set "force_output" for __RTL
30402 (cgraph_node::analyze): Bail out early for __RTL functions.
30403 (analyze_functions): Update assertion to support __RTL functions.
30404 (cgraph_node::expand): Bail out early for __RTL functions.
30405 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
30407 * function.h (struct function): Update comment for field
30409 * gimple-expr.c: Include "tree-pass.h".
30410 (gimple_has_body_p): Return false for __RTL functions.
30411 * Makefile.in (OBJS): Add run-rtl-passes.o.
30412 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
30414 (gcc::pass_manager::get_clean_slate): New accessor.
30415 * passes.c: Include "insn-addr.h".
30416 (should_skip_pass_p): Add logging. Update logic for running
30417 "expand" to be compatible with both __GIMPLE and __RTL. Guard
30418 property-provider override so it is only done for gimple passes.
30420 (skip_pass): New function.
30421 (execute_one_pass): Call skip_pass when skipping passes.
30422 * read-md.c (md_reader::read_char): Support filtering
30423 the input to a subset of line numbers.
30424 (md_reader::md_reader): Initialize fields
30425 m_first_line and m_last_line.
30426 (md_reader::read_file_fragment): New function.
30427 * read-md.h (md_reader::read_file_fragment): New decl.
30428 (md_reader::m_first_line): New field.
30429 (md_reader::m_last_line): New field.
30430 * read-rtl-function.c (function_reader::create_function): Only
30431 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
30432 curr_properties. Set DECL_INITIAL to a dummy block.
30433 (read_rtl_function_body_from_file_range): New function.
30434 * read-rtl-function.h (read_rtl_function_body_from_file_range):
30436 * run-rtl-passes.c: New file.
30437 * run-rtl-passes.h: New file.
30439 2017-01-24 Jeff Law <law@redhat.com>
30441 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
30444 2017-01-24 Bin Cheng <bin.cheng@arm.com>
30446 PR tree-optimization/79159
30447 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
30448 (record_nonwrapping_iv): Improve boundary using above function if no
30449 value range information.
30451 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
30452 Martin Jambor <mjambor@suse.cz>
30454 * brig-builtins.def: New file.
30455 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
30456 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
30457 (DEF_HSAIL_SAT_BUILTIN): Likewise.
30458 (DEF_HSAIL_INTR_BUILTIN): Likewise.
30459 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
30460 * builtin-types.def (BT_INT8): New.
30461 (BT_INT16): Likewise.
30462 (BT_UINT8): Likewise.
30463 (BT_UINT16): Likewise.
30464 (BT_FN_ULONG): Likewise.
30465 (BT_FN_UINT_INT): Likewise.
30466 (BT_FN_UINT_ULONG): Likewise.
30467 (BT_FN_UINT_LONG): Likewise.
30468 (BT_FN_UINT_PTR): Likewise.
30469 (BT_FN_ULONG_PTR): Likewise.
30470 (BT_FN_INT8_FLOAT): Likewise.
30471 (BT_FN_INT16_FLOAT): Likewise.
30472 (BT_FN_UINT32_FLOAT): Likewise.
30473 (BT_FN_UINT16_FLOAT): Likewise.
30474 (BT_FN_UINT8_FLOAT): Likewise.
30475 (BT_FN_UINT64_FLOAT): Likewise.
30476 (BT_FN_UINT16_UINT32): Likewise.
30477 (BT_FN_UINT32_UINT16): Likewise.
30478 (BT_FN_UINT16_UINT16_UINT16): Likewise.
30479 (BT_FN_INT_PTR_INT): Likewise.
30480 (BT_FN_UINT_PTR_UINT): Likewise.
30481 (BT_FN_LONG_PTR_LONG): Likewise.
30482 (BT_FN_ULONG_PTR_ULONG): Likewise.
30483 (BT_FN_VOID_UINT64_UINT64): Likewise.
30484 (BT_FN_UINT8_UINT8_UINT8): Likewise.
30485 (BT_FN_INT8_INT8_INT8): Likewise.
30486 (BT_FN_INT16_INT16_INT16): Likewise.
30487 (BT_FN_INT_INT_INT): Likewise.
30488 (BT_FN_UINT_FLOAT_UINT): Likewise.
30489 (BT_FN_FLOAT_UINT_UINT): Likewise.
30490 (BT_FN_ULONG_UINT_UINT): Likewise.
30491 (BT_FN_ULONG_UINT_PTR): Likewise.
30492 (BT_FN_ULONG_ULONG_ULONG): Likewise.
30493 (BT_FN_UINT_UINT_UINT): Likewise.
30494 (BT_FN_VOID_UINT_PTR): Likewise.
30495 (BT_FN_UINT_UINT_PTR: Likewise.
30496 (BT_FN_UINT32_UINT64_PTR): Likewise.
30497 (BT_FN_INT_INT_UINT_UINT): Likewise.
30498 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
30499 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
30500 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
30501 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
30502 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
30503 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
30504 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
30505 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
30506 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
30507 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
30508 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
30509 * doc/frontends.texi: List BRIG FE.
30510 * doc/install.texi (Testing): Add BRIG tesring requirements.
30511 * doc/invoke.texi (Overall Options): Mention BRIG.
30512 * doc/standards.texi (Standards): Doucment BRIG HSA version.
30514 2017-01-24 Richard Biener <rguenther@suse.de>
30516 PR translation/79208
30517 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
30519 2017-01-24 Martin Jambor <mjambor@suse.cz>
30522 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
30523 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
30524 and known_contexts.
30526 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
30528 PR middle-end/79123
30529 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
30530 casts from signed to unsigned really don't have a range.
30532 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
30534 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
30535 GMP_RNDx for compatiblity.
30537 2017-01-24 Martin Liska <mliska@suse.cz>
30540 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
30541 that would prevent us to call alloca with -1 as argument.
30543 2017-01-24 Jakub Jelinek <jakub@redhat.com>
30545 * dwarf2out.c (output_compilation_unit_header, output_file_names):
30546 Avoid -Wformat-security warning.
30548 2017-01-23 Andrew Pinski <apinski@cavium.com>
30550 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
30553 2017-01-23 Martin Sebor <msebor@redhat.com>
30555 PR middle-end/78703
30556 * gimple-ssa-sprintf.c (warn_level): New global.
30557 (format_integer): Use it here and throughout the rest of the file.
30558 Use the same switch to compute sign as base.
30559 (maybe_warn): New function.
30560 (format_directive): Factor out warnings into maybe_warn.
30561 Add debugging output. Use warn_level.
30562 (add_bytes): Use warn_level.
30563 (pass_sprintf_length::compute_format_length): Add debugging output.
30564 (try_substitute_return_value): Same.
30565 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
30567 PR middle-end/78703
30568 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
30569 (struct fmtresult, format_integer, format_floating): Adjust.
30570 (fmtresult::fmtresult): Set max correctly in two argument ctor.
30571 (get_string_length, format_string,format_directive): Same.
30572 (pass_sprintf_length::compute_format_length): Same.
30573 (try_substitute_return_value): Simplify slightly.
30575 PR middle-end/78703
30576 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
30577 (fmtresult::operator+=): Outlined.
30578 (struct fmtresult): Add ctors.
30579 (struct conversion_spec): Rename...
30580 (struct directive): ...to this. Add and remove data members.
30581 (directive::set_width, directive::set_precision): New functions.
30582 (format_percent): Use fmtresult ctor.
30583 (get_width_and_precision): Remove.
30584 (format_integer): Make naming changes. Avoid computing width and
30586 (format_floating): Same. Adjust indentation.
30587 (format_character, format_none): New functions.
30588 (format_string): Moved character handling to format_character.
30589 (format_directive): Remove arguments, change return type.
30590 (parse_directive): New function.
30591 (pass_sprintf_length::compute_format_length): Move directive
30592 parsing to parse_directive.
30594 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30596 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
30597 (assign_assembler_name_if_needed): ... this.
30598 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
30599 (assign_assembler_name_if_needed): ... this.
30600 (free_lang_data_in_cgraph): Adjust callers.
30601 * cgraphunit.c (cgraph_node::analyze): Likewise.
30602 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
30605 2017-01-23 Richard Biener <rguenther@suse.de>
30607 PR tree-optimization/79088
30608 PR tree-optimization/79188
30609 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
30610 resetting loop bounds after last path deletion. Reset loop
30611 bounds of the target loop, make code match the comments.
30612 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
30613 Make sure loops need no fixups.
30615 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30617 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
30618 exponent support with double type for first argument.
30619 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
30620 type returned by __builtin_vec_extract_sig,
30621 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
30622 functions from "vector int" to "vector unsigned int" or from
30623 "vector long long int" to "vector unsigned long long int".
30624 Changed type returned by __builtin_vec_extract_exp,
30625 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
30626 functions from "vector int" to "vector unsigned int" or from
30627 "vector long long int" to "vector unsigned long long int".
30628 Changed return type of __builtin_vec_test_data_class,
30629 __builtin_vec_test_data_class_sp, and
30630 __builtin_vec_test_data_class_dp from "vector int" to
30631 "vector bool int" or from "vector long long int" to "vector bool
30632 long long int" and changed second argument type from "unsigned
30633 int" to "int". Added new overloaded function forms "vector float
30634 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
30635 "vector float __builtin_vec_insert_exp_sp (vector float, vector
30636 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
30637 double, vector unsigned long long int)" and "vector double
30638 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
30639 long int)". Changed return type of
30640 __builtin_scalar_test_data_class and
30641 __builtin_scalar_test_data_class_sp and
30642 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
30643 int" and changed second argument from "unsigned int" to "int".
30644 Changed type returned by __builtin_scalar_test_neg,
30645 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
30646 from "int" to "bool int". Added new overloaded function form
30647 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
30648 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
30649 exponent double-precision with floating point first argument.
30650 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
30651 documentation of scalar_test_data_class, scalar_test_neg,
30652 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
30653 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
30654 vec_test_data_class built-in functions to reflect refinements in
30655 their type signatures.
30657 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
30659 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
30661 (aarch64_elf_asm_destructor): Likewise.
30663 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
30665 PR rtl-optimization/78634
30666 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
30667 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
30668 * ifcvt.c (noce_try_cmove): Add missing cost check.
30670 PR rtl-optimization/71724
30671 * combine.c (if_then_else_cond): Look for situations where it is
30672 beneficial to undo the work of one of the recursive calls.
30674 2017-01-23 Bin Cheng <bin.cheng@arm.com>
30676 PR tree-optimization/70754
30677 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
30678 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
30679 combined stmt before it if not NULL.
30680 (combine_chains): Process refs reversely and compute dominance point
30683 2017-01-23 Martin Liska <mliska@suse.cz>
30685 PR tree-optimization/79196
30686 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
30687 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
30689 (strlen_optimize_stmt): Call the renamed function.
30691 2017-01-23 Michael Matz <matz@suse.de>
30693 PR tree-optimization/78384
30694 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
30696 2017-01-23 Richard Biener <rguenther@suse.de>
30698 PR tree-optimization/79186
30699 * tree-vrp.c (register_new_assert_for): Make sure we've seen
30700 both incoming edges before moving an assert.
30702 2017-01-23 Martin Jambor <mjambor@suse.cz>
30704 * ipa-prop.c (load_from_param_1): Removed.
30705 (load_from_unmodified_param): Bits from load_from_param_1 put back
30707 (load_from_param): Removed.
30708 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
30709 with stmt. Reverted back to use of load_from_unmodified_param.
30711 2017-01-23 Martin Jambor <mjambor@suse.cz>
30714 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
30715 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
30716 field a pointer to garbage collected vector, mark lattices and
30717 ipcp_orig_node with GTY((skip)).
30718 (ipa_get_param_count): Adjust to descriptors being a pointer.
30719 (ipa_get_param): Likewise.
30720 (ipa_get_type): Likewise.
30721 (ipa_get_param_move_cost): Likewise.
30722 (ipa_set_param_used): Likewise.
30723 (ipa_get_controlled_uses): Likewise.
30724 (ipa_set_controlled_uses): Likewise.
30725 (ipa_is_param_used): Likewise.
30726 (ipa_node_params_t): Move into garbage collector. New methods insert
30728 (ipa_node_params_sum): Annotate wth GTY(()).
30729 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
30731 (ipa_load_from_parm_agg): Adjust declaration.
30732 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
30733 * ipa-profile.c (ipa_profile): Likewise.
30734 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
30735 (ipa_populate_param_decls): Make descriptors parameter garbage
30737 (ipa_dump_param): Adjust to descriptors being a pointer.
30738 (ipa_alloc_node_params): Likewise.
30739 (ipa_initialize_node_params): Likewise.
30740 (load_from_param_1): Make descriptors parameter garbage collected.
30741 (load_from_unmodified_param): Likewise.
30742 (load_from_param): Likewise.
30743 (ipa_load_from_parm_agg): Likewise.
30744 (ipa_node_params::~ipa_node_params): Removed.
30745 (ipa_free_all_node_params): Remove call to delete operator.
30746 (ipa_node_params_t::insert): New.
30747 (ipa_node_params_t::remove): Likewise.
30748 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
30749 copy known_csts and known_contexts vectors.
30750 (ipa_read_node_info): Adjust to descriptors being a pointer.
30751 (ipcp_modif_dom_walker): Make m_descriptors field garbage
30753 (ipcp_transform_function): Make descriptors variable garbage
30756 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
30758 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
30759 * config/i386/avx512dqintrin.h: Ditto.
30760 * config/i386/avx512fintrin.h: Ditto.
30761 * config/i386/i386.c: Handle new builtins.
30762 * config/i386/i386-builtin.def: Add new builtins.
30763 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
30764 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
30766 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30767 Martin Liska <mliska@suse.cz>
30769 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
30770 * asan.c (asan_expand_poison_ifn): Support stores and use
30771 appropriate ASAN report function.
30772 * internal-fn.c (expand_ASAN_POISON_USE): New function.
30773 * internal-fn.def (ASAN_POISON_USE): Declare.
30774 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
30775 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
30776 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
30777 ASAN_POISON calls w/o LHS.
30778 * tree-ssa.c (execute_update_addresses_taken): Create clobber
30779 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
30780 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
30781 * gimplify.c (asan_poison_variables): Add attribute
30782 use_after_scope_memory to variables that really needs to live
30784 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
30785 having the attribute.
30787 2017-01-23 Martin Liska <mliska@suse.cz>
30789 * asan.c (create_asan_shadow_var): New function.
30790 (asan_expand_poison_ifn): Likewise.
30791 * asan.h (asan_expand_poison_ifn): New declaration.
30792 * internal-fn.c (expand_ASAN_POISON): Likewise.
30793 * internal-fn.def (ASAN_POISON): New builtin.
30794 * sanopt.c (pass_sanopt::execute): Expand
30795 asan_expand_poison_ifn.
30796 * tree-inline.c (copy_decl_for_dup_finish): Make function
30798 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
30799 * tree-ssa.c (is_asan_mark_p): New function.
30800 (execute_update_addresses_taken): Rewrite local variables
30801 (identified just by use-after-scope as addressable) into SSA.
30803 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30805 * doc/install.texi (Specific): opensource.apple.com uses https
30806 now. Remove trailing slash.
30808 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30810 * README.Portability: Remove note on an Irix compatibility issue.
30812 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
30814 * gcov.c (INCLUDE_ALGORITHM): Define.
30815 (INCLUDE_VECTOR): Define.
30816 No longer include <vector> and <algorithm> directly.
30818 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
30820 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
30822 * doc/invoke.texi (Code Gen Options): Ditto.
30824 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
30827 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
30829 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
30831 rtl-optimization/79125
30832 * cprop.c (local_cprop_pass): Handle cases where we make an
30833 unconditional trap.
30835 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
30839 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
30840 read from, for big endian.
30842 2017-01-20 Jiong Wang <jiong.wang@arm.com>
30844 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
30845 register pauth builtins for LP64 only.
30847 2017-01-20 Marek Polacek <polacek@redhat.com>
30850 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
30853 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30855 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
30857 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
30858 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
30859 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
30861 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30864 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
30865 in big-endian mode when they are not a single duplicated value.
30867 2017-01-20 Richard Biener <rguenther@suse.de>
30869 * BASE-VER: Bump to 7.0.1.
30871 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30873 * omp-low.c (omplow_simd_context): New struct. Use it...
30874 (lower_rec_simd_input_clauses): ...here and...
30875 (lower_rec_input_clauses): ...here to hold common data. Adjust all
30876 references to idx, lane, max_vf, is_simt.
30878 2017-01-20 Graham Markall <graham.markall@embecosm.com>
30880 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
30883 2017-01-20 Martin Jambor <mjambor@suse.cz>
30885 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
30886 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
30888 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
30889 (GTFILES): Rename hsa.c to hsa-common.c.
30890 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
30891 * hsa-dump.c: Likewise.
30892 * hsa-gen.c: Likewise.
30893 * hsa-regalloc.c: Likewise.
30894 * ipa-hsa.c: Likewise.
30895 * omp-expand.c: Likewise.
30896 * omp-low.c: Likewise.
30897 * toplev.c: Likewise.
30899 2017-01-20 Marek Polacek <polacek@redhat.com>
30902 * doc/invoke.texi: Document -Wduplicated-branches.
30903 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
30904 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
30905 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
30906 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
30907 return 0 only when not OEP_LEXICOGRAPHIC.
30908 (fold_build_cleanup_point_expr): Use the expression
30909 location when building CLEANUP_POINT_EXPR.
30910 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
30911 * tree.c (add_expr): Handle error_mark_node.
30913 2017-01-20 Martin Liska <mliska@suse.cz>
30916 * tree-profile.c (init_ic_make_global_vars): Do not call
30918 (gimple_init_gcov_profiler): Likewise.
30920 2017-01-20 Martin Liska <mliska@suse.cz>
30923 * cgraph.h (maybe_create_reference): Remove argument and
30925 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
30927 * ipa-cp.c (create_specialized_node): Likewise.
30928 * symtab.c (symtab_node::maybe_create_reference): Handle
30929 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
30931 2017-01-20 Martin Liska <mliska@suse.cz>
30933 * read-rtl-function.c (function_reader::create_function): Use
30934 build_decl instread of build_decl_stat.
30936 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
30938 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
30939 * config/i386/avx512dqintrin.h: Ditto.
30940 * config/i386/avx512fintrin.h: Ditto.
30941 * config/i386/i386-builtin-types.def: Add new types.
30942 * config/i386/i386.c: Handle new types.
30943 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
30944 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
30945 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
30946 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
30947 (__builtin_ia32_kshiftridi): New.
30948 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
30950 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
30954 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
30955 define to rs6000_init_stack_protect_guard.
30956 (rs6000_init_stack_protect_guard): New function.
30958 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30959 Yunqiang Su <yunqiang.su@imgtec.com>
30961 * config.gcc (supported_defaults): Add madd4.
30962 (with_madd4): Add validation.
30963 (all_defaults): Add madd4.
30964 * config/mips/mips.opt (mmadd4): New option.
30965 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30967 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30969 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
30970 (ISA_HAS_FUSED_MADD4): Likewise.
30971 * doc/invoke.texi (-mmadd4): Document the new option.
30972 * doc/install.texi (--with-madd4): Document the new option.
30974 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30976 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
30977 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
30978 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
30979 (aarch64_init_pauth_hint_builtins): New.
30980 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
30981 (aarch64_expand_builtin): Expand new builtins.
30983 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30985 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
30986 * combine-stack-adj.c (no_unhandled_cfa): Handle
30987 REG_CFA_TOGGLE_RA_MANGLE.
30988 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
30989 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
30990 info for return address signing.
30991 (aarch64_expand_epilogue): Likewise.
30993 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30995 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
30996 * config/aarch64/aarch64-protos.h
30997 (aarch64_return_address_signing_enabled): New declaration.
30998 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
31000 (aarch64_expand_prologue): Sign return address before it's pushed onto
31002 (aarch64_expand_epilogue): Authenticate return address fetched from
31004 (aarch64_override_options): Sanity check for ILP32 and ISA level.
31005 (aarch64_attributes): New function attributes for "sign-return-address".
31006 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
31007 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
31008 ("*do_return"): Generate combined instructions according to key index.
31009 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
31010 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
31012 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
31013 * config/aarch64/aarch64.opt (msign-return-address=): New.
31014 * doc/extend.texi (AArch64 Function Attributes): Documents
31015 "sign-return-address=".
31016 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
31018 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
31020 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
31021 overall option summary.
31023 2017-01-19 Jiong Wang <jiong.wang@arm.com>
31025 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
31026 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
31027 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
31028 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
31030 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
31032 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
31033 -mpower9-minmax by default for -mcpu=power9.
31034 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
31035 128-bit floating point.
31037 2017-01-20 Alan Modra <amodra@gmail.com>
31039 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
31040 optimizing for size.
31042 2017-01-20 Alan Modra <amodra@gmail.com>
31045 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
31046 for strcmp and strncmp from corresponding builtin decl.
31048 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31050 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
31051 instead of i386/rtems-64.h.
31052 * config/i386/rtems-64.h: Remove.
31054 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
31058 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
31060 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
31062 2017-01-19 Tamar Christina <tamar.christina@arm.com>
31064 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
31065 Change int to HOST_WIDE_INT.
31066 * config/aarch64/aarch64-protos.h
31067 (aarch64_simd_gen_const_vector_dup): Likewise.
31068 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
31070 2017-01-19 David Malcolm <dmalcolm@redhat.com>
31072 * langhooks-def.h (lhd_type_for_size): New decl.
31073 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
31074 * langhooks.c (lhd_type_for_size): New function, taken from
31077 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
31079 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
31080 define_bypass for CR latency.
31081 (power9-cracked-alu): Update bypass latency and remove power9-branch.
31082 (power9-alu2): Add define_bypass for CR latency.
31084 (power9-mul): Update insn latency.
31085 (power9-mul-compare): Update insn latency, bypass latency and remove
31088 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31090 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
31092 * config/aarch64/aarch64.md
31093 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
31094 aarch64_nopcrelative_literal_loads.
31095 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
31097 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
31099 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
31100 TARGET_LOONGSON_3A.
31101 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
31103 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
31106 * config.gcc (supported_defaults): Add lxc1-sxc1.
31107 (with_lxc1_sxc1): Add validation.
31108 (all_defaults): Add lxc1-sxc1.
31109 * config/mips/mips.opt (mlxc1-sxc1): New option.
31110 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31112 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31113 __mips_no_lxc1_sxc1.
31114 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
31115 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
31116 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
31118 2017-01-19 Richard Biener <rguenther@suse.de>
31120 PR tree-optimization/72488
31121 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
31122 sure to restore SSA info.
31123 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
31125 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
31127 PR rtl-optimization/79121
31128 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
31129 of the inner type when shifting an extended value.
31131 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31134 * symtab.c (symtab_node::equal_address_to): Fix comparing of
31135 interposable aliases.
31137 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
31140 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
31141 Use the evmergelohi instruction.
31142 (mov_si<mode>_e500_subreg4_2_le): Likewise.
31143 (mov_sitf_e500_subreg8_2_be): Likewise.
31144 (mov_sitf_e500_subreg12_2_le): Likewise.
31145 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
31146 (mov_si<mode>_e500_subreg4_2_be): Likewise.
31147 (mov_sitf_e500_subreg8_2_le): Likewise.
31148 (mov_sitf_e500_subreg12_2_be): Likewise.
31150 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31152 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
31153 attribute from vecsimple to vecperm.
31154 (altivec_vbpermq2): Likewise.
31156 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31159 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
31161 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31162 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
31163 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
31164 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
31165 case where N arg is SIZE_MAX.
31166 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
31167 (cmpstrsi): Add pattern.
31169 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31171 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31172 __builtin_vec_revb builtins.
31173 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
31174 built-in functions to support generation of the ISA 3.0 XXBR<x>
31175 vector byte reverse instructions.
31176 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
31177 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
31178 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
31179 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
31180 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
31181 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
31182 (P9V_BUILTIN_VEC_REVB): Likewise.
31183 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
31184 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
31185 (p9_xxbrq_v16qi): Likewise.
31186 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
31187 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
31188 (p9_xxbrh_v8hi): Likewise.
31189 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
31190 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
31191 vec_revb built-in functions.
31193 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
31195 PR rtl-optimization/78952
31196 * config/i386/i386.md (any_extract): New code iterator.
31197 (*insvqi_2): Use any_extract for source operand.
31198 (*insvqi_3): Use any_shiftrt for source operand.
31200 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
31202 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
31204 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
31206 2017-01-18 Matthias Klose <doko@ubuntu.com>
31208 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
31210 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31212 * config/rs6000/altivec.h (vec_bperm): Change #define.
31213 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
31214 (altivec_vbpermq2): New define_insn.
31215 (altivec_vbpermd): Likewise.
31216 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
31217 function interface.
31218 (VBPERMD): Likewise.
31219 (VBPERM): New polymorphic function interface.
31220 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
31221 Add entries for P9V_BUILTIN_VEC_VBPERM.
31222 * doc/extend.texi: Add interfaces for vec_bperm.
31224 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31226 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
31227 first letter of error messages.
31228 (s390_resolve_overloaded_builtin): Likewise.
31229 * config/s390/s390.c (s390_expand_builtin): Likewise.
31230 (s390_invalid_arg_for_unprototyped_fn): Likewise.
31231 (s390_valid_target_attribute_inner_p): Likewise.
31232 * config/s390/s390.md ("tabort"): Likewise.
31234 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
31236 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
31237 (ISA_AVOID_DIV_HILO): New macro.
31238 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
31239 (ISA_HAS_DDIV): Likewise.
31241 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31243 * doc/invoke.texi (fabi-version): Correct number of occurrences.
31245 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31247 * doc/invoke.texi (fabi-version): Spelling fix.
31249 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31252 * doc/invoke.texi (fabi-version): Mention mangling fix for
31255 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31258 * doc/invoke.texi (fabi-version): Document discriminator mangling.
31260 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
31263 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
31264 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
31266 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
31268 (stack_protect_test): Ditto.
31269 * config/rs6000/rs6000.opt (mstack-protector-guard=,
31270 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
31272 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
31273 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
31274 -mstack-protector-guard-offset=.
31275 (RS/6000 and PowerPC Options): Ditto.
31277 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31279 * config/i386/i386.h (MASK_CLASS_P): New define.
31280 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
31281 there are no registers from different register sets also when
31282 mask registers are used. Update function comment.
31283 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
31284 to (*k/*r) and (*k/*km) alternatives.
31286 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
31288 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
31289 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
31290 (EH_RETURN_HANDLER_RTX): New define.
31291 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
31292 Force frame pointer in EH return functions.
31293 (aarch64_expand_epilogue): Add barrier for eh_return.
31294 (aarch64_final_eh_return_addr): Remove.
31295 (aarch64_eh_return_handler_rtx): New function.
31296 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
31298 (aarch64_eh_return_handler_rtx): New prototype.
31300 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31302 * config/rs6000/altivec.h (vec_rlmi): New #define.
31303 (vec_vrlnm): Likewise.
31304 (vec_rlnm): Likewise.
31305 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
31306 (UNSPEC_VRLNM): Likewise.
31307 (VIlong): New mode iterator.
31308 (altivec_vrl<VI_char>mi): New define_insn.
31309 (altivec_vrl<VI_char>nm): Likewise.
31310 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
31312 (VRLDNM): Likewise.
31313 (RLNM): New polymorphic function entry.
31314 (VRLWMI): New monomorphic function entry.
31315 (VRLDMI): Likewise.
31316 (RLMI): New polymorphic function entry.
31317 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
31318 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
31319 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
31322 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31325 * dwarf2out.c (field_byte_offset): Restore the
31326 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
31327 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
31328 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
31331 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31334 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
31336 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31339 * dwarf2out.c (add_data_member_location_attribute): For constant
31340 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
31341 instead of DW_AT_data_member_location, DW_AT_bit_offset and
31342 DW_AT_byte_size attributes.
31344 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31346 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
31347 after forcing to constant memory when the code model is medium.
31349 2017-01-17 Julia Koval <julia.koval@intel.com>
31352 * config/i386/avx512fintrin.h
31353 (_mm512_i32gather_ps): Change __addr type to void const*.
31354 (_mm512_mask_i32gather_ps): Ditto.
31355 (_mm512_i32gather_pd): Ditto.
31356 (_mm512_mask_i32gather_pd): Ditto.
31357 (_mm512_i64gather_ps): Ditto.
31358 (_mm512_mask_i64gather_ps): Ditto.
31359 (_mm512_i64gather_pd): Ditto.
31360 (_mm512_mask_i64gather_pd): Ditto.
31361 (_mm512_i32gather_epi32): Ditto.
31362 (_mm512_mask_i32gather_epi32): Ditto.
31363 (_mm512_i32gather_epi64): Ditto.
31364 (_mm512_mask_i32gather_epi64): Ditto.
31365 (_mm512_i64gather_epi32): Ditto.
31366 (_mm512_mask_i64gather_epi32): Ditto.
31367 (_mm512_i64gather_epi64): Ditto.
31368 (_mm512_mask_i64gather_epi64): Ditto.
31369 (_mm512_i32scatter_ps): Change __addr type to void*.
31370 (_mm512_mask_i32scatter_ps): Ditto.
31371 (_mm512_i32scatter_pd): Ditto.
31372 (_mm512_mask_i32scatter_pd): Ditto.
31373 (_mm512_i64scatter_ps): Ditto.
31374 (_mm512_mask_i64scatter_ps): Ditto.
31375 (_mm512_i64scatter_pd): Ditto.
31376 (_mm512_mask_i64scatter_pd): Ditto.
31377 (_mm512_i32scatter_epi32): Ditto.
31378 (_mm512_mask_i32scatter_epi32): Ditto.
31379 (_mm512_i32scatter_epi64): Ditto.
31380 (_mm512_mask_i32scatter_epi64): Ditto.
31381 (_mm512_i64scatter_epi32): Ditto.
31382 (_mm512_mask_i64scatter_epi32): Ditto.
31383 (_mm512_i64scatter_epi64): Ditto.
31384 (_mm512_mask_i64scatter_epi64): Ditto.
31385 * config/i386/avx512pfintrin.h
31386 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
31387 (_mm512_mask_prefetch_i32gather_ps): Ditto.
31388 (_mm512_mask_prefetch_i64gather_pd): Ditto.
31389 (_mm512_mask_prefetch_i64gather_ps): Ditto.
31390 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
31391 (_mm512_prefetch_i32scatter_ps): Ditto.
31392 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
31393 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
31394 (_mm512_prefetch_i64scatter_pd): Ditto.
31395 (_mm512_prefetch_i64scatter_ps): Ditto.
31396 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
31397 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
31398 * config/i386/avx512vlintrin.h
31399 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
31400 (_mm_mmask_i32gather_ps): Ditto.
31401 (_mm256_mmask_i32gather_pd): Ditto.
31402 (_mm_mmask_i32gather_pd): Ditto.
31403 (_mm256_mmask_i64gather_ps): Ditto.
31404 (_mm_mmask_i64gather_ps): Ditto.
31405 (_mm256_mmask_i64gather_pd): Ditto.
31406 (_mm_mmask_i64gather_pd): Ditto.
31407 (_mm256_mmask_i32gather_epi32): Ditto.
31408 (_mm_mmask_i32gather_epi32): Ditto.
31409 (_mm256_mmask_i32gather_epi64): Ditto.
31410 (_mm_mmask_i32gather_epi64): Ditto.
31411 (_mm256_mmask_i64gather_epi32): Ditto.
31412 (_mm_mmask_i64gather_epi32): Ditto.
31413 (_mm256_mmask_i64gather_epi64): Ditto.
31414 (_mm_mmask_i64gather_epi64): Ditto.
31415 (_mm256_i32scatter_ps): Change __addr type to void*.
31416 (_mm256_mask_i32scatter_ps): Ditto.
31417 (_mm_i32scatter_ps): Ditto.
31418 (_mm_mask_i32scatter_ps): Ditto.
31419 (_mm256_i32scatter_pd): Ditto.
31420 (_mm256_mask_i32scatter_pd): Ditto.
31421 (_mm_i32scatter_pd): Ditto.
31422 (_mm_mask_i32scatter_pd): Ditto.
31423 (_mm256_i64scatter_ps): Ditto.
31424 (_mm256_mask_i64scatter_ps): Ditto.
31425 (_mm_i64scatter_ps): Ditto.
31426 (_mm_mask_i64scatter_ps): Ditto.
31427 (_mm256_i64scatter_pd): Ditto.
31428 (_mm256_mask_i64scatter_pd): Ditto.
31429 (_mm_i64scatter_pd): Ditto.
31430 (_mm_mask_i64scatter_pd): Ditto.
31431 (_mm256_i32scatter_epi32): Ditto.
31432 (_mm256_mask_i32scatter_epi32): Ditto.
31433 (_mm_i32scatter_epi32): Ditto.
31434 (_mm_mask_i32scatter_epi32): Ditto.
31435 (_mm256_i32scatter_epi64): Ditto.
31436 (_mm256_mask_i32scatter_epi64): Ditto.
31437 (_mm_i32scatter_epi64): Ditto.
31438 (_mm_mask_i32scatter_epi64): Ditto.
31439 (_mm256_i64scatter_epi32): Ditto.
31440 (_mm256_mask_i64scatter_epi32): Ditto.
31441 (_mm_i64scatter_epi32): Ditto.
31442 (_mm_mask_i64scatter_epi32): Ditto.
31443 (_mm256_i64scatter_epi64): Ditto.
31444 (_mm256_mask_i64scatter_epi64): Ditto.
31445 (_mm_i64scatter_epi64): Ditto.
31446 (_mm_mask_i64scatter_epi64): Ditto.
31447 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
31448 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
31449 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
31450 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
31451 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
31452 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
31453 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
31454 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
31455 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
31456 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
31457 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
31458 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
31459 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
31460 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
31461 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
31462 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
31463 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
31464 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
31465 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
31466 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
31467 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
31468 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
31469 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
31470 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
31471 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
31472 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
31473 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
31474 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
31475 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
31476 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
31477 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
31478 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
31479 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
31480 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
31481 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
31482 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
31483 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
31484 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
31485 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
31486 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
31487 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
31488 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
31489 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
31490 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
31491 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
31492 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
31493 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
31494 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
31495 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
31496 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
31497 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
31498 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
31499 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
31500 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
31501 definitions accordingly.
31503 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
31504 Kuan-Lin Chen <kuanlinchentw@gmail.com>
31507 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
31510 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
31513 * ira-conflicts.c (ira_build_conflicts): Update total conflict
31514 hard regs for inner regno.
31516 2017-01-17 Martin Liska <mliska@suse.cz>
31519 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
31520 assumption and add comment.
31522 2017-01-17 Nathan Sidwell <nathan@acm.org>
31524 * ipa-visibility.c (localize_node): New function, broken out of ...
31525 (function_and_variable_visibility): ... here. Call it.
31527 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31529 PR middle-end/77445
31530 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
31531 correctly set frequency of oudgoing edge.
31532 (duplicate_thread_path): Fix profile updating.
31534 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31537 * configure.ac: Add GCC_BASE_VER.
31538 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
31539 version from BASE-VER file.
31540 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
31541 (gcc.o): Depend on $(BASEVER).
31542 * common.opt (dumpfullversion): New option.
31543 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
31544 * doc/invoke.texi: Document -dumpfullversion.
31545 * doc/install.texi: Document --with-gcc-major-version-only.
31546 * configure: Regenerated.
31548 2017-01-17 Richard Biener <rguenther@suse.de>
31550 PR tree-optimization/71433
31551 * tree-vrp.c (register_new_assert_for): Merge same asserts
31552 on all incoming edges.
31553 (process_assert_insertions_for): Handle insertions at the
31556 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
31558 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
31559 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
31561 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
31564 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
31567 2017-01-17 Alan Modra <amodra@gmail.com>
31570 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
31571 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
31572 symbolic stack limit when pic.
31574 2017-01-16 Martin Sebor <msebor@redhat.com>
31576 PR tree-optimization/78608
31577 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
31579 2017-01-16 Jeff Law <law@redhat.com>
31582 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
31583 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
31584 for several include directories that may be relative to sysroot.
31585 * config/i386/x-mingw32 (gplus_includedir): Define.
31586 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
31587 (native_system_includedir): Likewise.
31588 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
31589 override if TARGET_SYSTEM_ROOT is defined.
31590 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
31592 PR tree-optimization/79090
31593 PR tree-optimization/33562
31594 PR tree-optimization/61912
31595 PR tree-optimization/77485
31596 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
31597 and computed trims into the dump file.
31599 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31601 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
31603 2017-01-16 Jakub Jelinek <jakub@redhat.com>
31606 * gimplify.c (gimplify_init_constructor): If want_value and
31607 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
31611 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
31612 sequence. Formatting fixes.
31613 (doloop_optimize): Formatting fixes.
31616 * gcc.c (debug_level_greater_than_spec_func): New function.
31617 (static_spec_functions): Add debug-level-gt spec function.
31618 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
31620 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31621 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31622 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
31623 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
31624 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
31625 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
31627 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
31629 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
31630 QImode fixups to general and mask registers only.
31632 2017-01-16 Carl Love <cel@us.ibm.com>
31634 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
31635 for built-in functions
31636 vector signed char vec_nabs (vector signed char)
31637 vector signed short vec_nabs (vector signed short)
31638 vector signed int vec_nabs (vector signed int)
31639 vector signed long long vec_nabs (vector signed long long)
31640 vector float vec_nabs (vector float)
31641 vector double vec_nabs (vector double)
31642 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
31644 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
31645 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
31646 * doc/extend.texi: Update the documentation file for the new built-in
31649 2017-01-16 Martin Sebor <msebor@redhat.com>
31651 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
31654 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31656 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
31657 UNSPEC_VSX__XXSPLTD to require special splat handling.
31659 2017-01-16 David Malcolm <dmalcolm@redhat.com>
31662 * system.h: Poison strndup.
31664 2017-01-16 Alan Modra <amodra@gmail.com>
31667 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
31670 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
31672 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
31674 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
31676 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
31677 call recog here. Assert that INSN_CODE (insn) is non-negative.
31679 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
31682 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
31684 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
31685 in the currently scheduled RTL fragment.
31687 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
31689 PR rtl-optimization/78751
31690 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
31693 2017-01-14 Jeff Law <law@redhat.com>
31695 PR tree-optimization/79090
31696 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
31697 variable length stores.
31698 (compute_trims): Delete dead assignment to *trim_tail.
31699 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
31702 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
31704 PR rtl-optimization/78626
31705 PR rtl-optimization/78727
31706 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
31707 of a block, and split such blocks after everything else is finished.
31709 2017-01-14 Alan Modra <amodra@gmail.com>
31712 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
31713 target legitimate_combined_insn.
31714 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
31715 (rs6000_legitimate_combined_insn): New function.
31716 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
31718 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
31719 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
31720 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
31722 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
31724 * doc/frontends.texi (G++ and GCC): Remove references to Java.
31726 2017-01-13 Jeff Law <law@redhat.com>
31728 PR tree-optimization/33562
31729 PR tree-optimization/61912
31730 PR tree-optimization/77485
31731 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
31733 (delete_dead_assignment): Likewise.
31734 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
31735 statement to delete_dead_call and delete_dead_assignment.
31737 2017-01-13 David Malcolm <dmalcolm@redhat.com>
31740 * substring-locations.c (format_warning_va): Strengthen case 1 so
31741 that both endpoints of the substring must be within the format
31742 range for just the substring to be printed.
31744 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
31746 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
31747 * config/i386/i386.c (ix86_target_string): Add missing options
31748 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
31749 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
31750 flags_other and ix86_target_other to flags2_other. Display unknown
31752 (ix86_valid_target_attribute_inner_p): Add missing options and
31753 reorder options by implied ISAs, as in ix86_target_string.
31755 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
31757 * hash-table.h (hash_table::too_empty_p): New function.
31758 (hash_table::expand): Use it.
31759 (hash_table::traverse): Likewise.
31760 (hash_table::empty_slot): Use sizeof (value_type) instead of
31761 sizeof (PTR) to convert bytes to elements. Shrink the table
31762 if the current size is excessive for the current number of
31765 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
31767 * ira-costs.c (record_reg_classes): Break from the inner loop
31768 early once alt_fail is known to be true. Update outer loop
31769 handling accordingly.
31771 2017-01-13 Jeff Law <law@redhat.com>
31773 * tree-ssa-dse.c (decrement_count): New function.
31774 (increment_start_addr, maybe_trim_memstar_call): Likewise.
31775 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
31776 when we know the partially dead statement is a mem* function.
31778 PR tree-optimization/61912
31779 PR tree-optimization/77485
31780 * tree-ssa-dse.c: Include expr.h.
31781 (maybe_trim_constructor_store): New function.
31782 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
31784 PR tree-optimization/33562
31785 PR tree-optimization/61912
31786 PR tree-optimization/77485
31787 * doc/invoke.texi: Document new dse-max-object-size param.
31788 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
31789 * tree-ssa-dse.c: Include params.h.
31790 (dse_store_status): New enum.
31791 (initialize_ao_ref_for_dse): New, partially extracted from
31793 (valid_ao_ref_for_dse, normalize_ref): New.
31794 (setup_live_bytes_from_ref, compute_trims): Likewise.
31795 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
31796 (maybe_trim_partially_dead_store): Likewise.
31797 (maybe_trim_complex_store): Likewise.
31798 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
31799 Track what bytes live from the original store. Return tri-state
31800 for dead, partially dead or live.
31801 (dse_dom_walker): Add constructor, destructor and new private members.
31802 (delete_dead_call, delete_dead_assignment): New extracted from
31804 (dse_optimize_stmt): Make a member of dse_dom_walker.
31805 Use initialize_ao_ref_for_dse.
31807 PR tree-optimization/33562
31808 PR tree-optimization/61912
31809 PR tree-optimization/77485
31810 * sbitmap.h (bitmap_count_bits): Prototype.
31811 (bitmap_clear_range, bitmap_set_range): Likewise.
31812 * sbitmap.c (bitmap_clear_range): New function.
31813 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
31815 2017-01-13 Martin Liska <mliska@suse.cz>
31818 * function.c (set_cfun): Add new argument force.
31819 * function.h (set_cfun): Likewise.
31820 * ipa-inline-transform.c (inline_call): Use the function when
31821 strict alising from is dropped for function we inline to.
31823 2017-01-13 Richard Biener <rguenther@suse.de>
31825 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
31826 for dumping GIMPLE INTEGER_CSTs.
31828 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31830 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
31831 to 201112L since C++17.
31833 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
31836 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
31837 if -fsanitize=kernel-address is present.
31839 2017-01-13 Richard Biener <rguenther@suse.de>
31841 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
31842 as _Literal ( type ) number in case usual suffixes do not
31843 preserve all information.
31845 2017-01-13 Richard Biener <rguenther@suse.de>
31847 PR tree-optimization/77283
31848 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
31849 and ssa-iterators.h.
31850 (is_feasible_trace): Implement a cost model based on joiner
31853 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
31856 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
31857 char or short to __float128/_Float128 directly.
31859 2017-01-12 Martin Sebor <msebor@redhat.com>
31861 to -Wformat-overflow.
31862 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
31863 (min_bytes_remaining): Same.
31864 (get_string_length): Same.
31865 (format_string): Same.
31866 (format_directive): Same.
31868 (pass_sprintf_length::handle_gimple_call): Same.
31870 2017-01-12 Jakub Jelinek <jakub@redhat.com>
31872 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
31873 info.nowrite calls with no lhs that can't throw. Return bool
31874 whether gsi_remove has been called or not.
31875 (pass_sprintf_length::handle_gimple_call): Return bool whether
31876 try_substitute_return_value called gsi_remove. Formatting fix.
31877 (pass_sprintf_length::execute): Don't use gsi_remove if
31878 handle_gimple_call returned true.
31881 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
31882 be removed due to side-effects, don't remove following barrier nor
31883 turn the successor edge into fallthru edge.
31885 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31888 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
31889 element-reversing loads and stores as not swappable.
31891 2017-01-12 Nathan Sidwell <nathan@acm.org>
31892 Nicolai Stange <nicstange@gmail.com>
31894 * combine.c (try_combine): Don't ignore result of overlap checking
31895 loop. Combine overlap & asm check into single loop.
31897 2017-01-12 Richard Biener <rguenther@suse.de>
31899 * tree-pretty-print.c (dump_generic_node): Provide -gimple
31900 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
31902 2017-01-12 Richard Biener <rguenther@suse.de>
31904 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
31905 and TS_TARGET_OPTION directly derive from TS_BASE.
31906 * tree-core.h (tree_optimization_option): Derive from tree_base.
31907 (tree_target_option): Likewise.
31909 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
31911 * config/i386/i386.c (memory_address_length): Increase len
31912 only when rip_relative_addr_p returns false.
31914 2017-01-11 Julia Koval <julia.koval@intel.com>
31916 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
31917 (OPTION_MASK_ISA_SGX_SET): New.
31918 (ix86_handle_option): Handle OPT_msgx.
31919 * config.gcc: Added sgxintrin.h.
31920 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
31921 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
31922 * config/i386/i386.c (ix86_target_string): Add -msgx.
31924 (ix86_option_override_internal): Handle new options.
31925 (ix86_valid_target_attribute_inner_p): Add sgx.
31926 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
31927 * config/i386/i386.opt: Add msgx.
31928 * config/i386/sgxintrin.h: New file.
31929 * config/i386/x86intrin.h: Add sgxintrin.h.
31931 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31934 * fold-const.c (maybe_nonzero_address): Return 1 for function
31936 (tree_single_nonzero_warnv_p): Don't handle function local objects
31940 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
31943 2017-01-11 David Malcolm <dmalcolm@redhat.com>
31946 * opts.c: Include "spellcheck.h"
31947 (struct string_fragment): New struct.
31948 (struct edit_distance_traits<const string_fragment &>): New
31950 (get_closest_sanitizer_option): New function.
31951 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
31953 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31955 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
31957 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
31958 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
31959 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
31960 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
31961 for initial die_offset if dwarf_split_debug_info.
31962 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
31963 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
31965 (output_skeleton_debug_sections): Formatting fix. Use
31966 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
31967 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
31969 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
31971 * config/arm/cortex-a53.md: Add bypasses for
31972 cortex_a53_r2f_cvt.
31973 (cortex_a53_r2f): Only use for transfers.
31974 (cortex_a53_f2r): Likewise.
31975 (cortex_a53_r2f_cvt): Add reservation for conversions.
31976 (cortex_a53_f2r_cvt): Likewise.
31978 2017-01-11 Tamar Christina <tamar.christina@arm.com>
31980 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
31981 to all inlined functions, change static to extern.
31983 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
31986 * config/arm/arm.c (legitimize_pic_address): Handle reference to
31988 (arm_assemble_integer): Likewise.
31990 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
31992 * config.gcc: Use new awk script to check CPU, FPU and architecture
31993 parameters for --with-... options.
31994 * config/arm/parsecpu.awk: New file
31995 * config/arm/arm-cpus.in: New file.
31996 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
31998 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
32000 * config/arm/t-arm: Update dependency rules.
32001 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
32002 of processing .def files.
32003 * config/arm/genopt.sh: Deleted.
32004 * config/arm/gentune.sh: Deleted.
32005 * config/arm/arm-cores.def: Deleted.
32006 * config/arm/arm-arches.def: Deleted.
32007 * config/arm/arm-fpus.def: Deleted.
32008 * config/arm/arm-tune.md: Regenerated.
32009 * config/arm/arm-tables.opt: Regenerated.
32010 * config/arm/arm-cpu.h: New generated file.
32011 * config/arm/arm-cpu-data.h: New generated file.
32012 * config/arm/arm-cpu-cdata.h: New generated file.
32014 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
32017 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
32019 (input_varpool_node): Unpack dynamically_initialized bit.
32021 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
32023 PR rtl-optimization/79032
32024 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
32025 the alignment of the adjusted memory reference against that of MODE,
32026 instead of the alignment of the original memory reference.
32028 2017-01-11 Martin Jambor <mjambor@suse.cz>
32030 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
32032 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
32033 decorated functions.
32035 2017-01-11 Richard Biener <rguenther@suse.de>
32037 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
32038 set range/nonnull info for PHI results. Do not set it on
32039 stmts marked for removal.
32041 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
32043 * expr.c (store_field): In the bitfield case, fetch the return value
32044 from the registers before applying a single big-endian adjustment.
32045 Always do a final load for a BLKmode value not larger than a word.
32047 2017-01-10 David Malcolm <dmalcolm@redhat.com>
32050 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32051 that we correctly handle column numbers greater than
32052 LINE_MAP_MAX_COLUMN_NUMBER.
32054 2017-01-10 Martin Sebor <msebor@redhat.com>
32056 PR middle-end/78245
32057 * gimple-ssa-sprintf.c (get_destination_size): Call
32058 {init,fini}object_sizes.
32059 * tree-object-size.c (addr_object_size): Adjust.
32060 (pass_through_call): Adjust.
32061 (pass_object_sizes::execute): Adjust.
32062 * tree-object-size.h (fini_object_sizes): Declare.
32064 2017-01-10 Martin Sebor <msebor@redhat.com>
32066 PR tree-optimization/78775
32067 * builtins.c (get_size_range): Move...
32068 * calls.c: ...to here.
32069 (alloc_max_size): Accept zero argument.
32070 (operand_signed_p): Remove.
32071 (maybe_warn_alloc_args_overflow): Call get_size_range.
32072 * calls.h (get_size_range): Declare.
32074 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
32076 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
32077 from TI's devices.csv file as of September 2016.
32078 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
32080 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
32082 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
32083 * doc/invoke.texi: Likewise.
32084 * doc/md.texi: Likewise.
32085 * doc/objc.texi: Likewise.
32087 2017-01-10 Joshua Conner <joshconner@google.com>
32089 * config/arm/fuchsia-elf.h: New file.
32090 * config/fuchsia.h: New file.
32091 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
32092 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
32094 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
32096 2016-01-10 Richard Biener <rguenther@suse.de>
32098 PR tree-optimization/79034
32099 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32100 Propagate out degenerate PHIs in the joiner.
32102 2017-01-10 Martin Liska <mliska@suse.cz>
32104 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
32105 (sort_congruence_classes_by_decl_uid): Likewise.
32106 (sort_congruence_class_groups_by_decl_uid): Likewise.
32107 (sem_item_optimizer::merge_classes): Sort class, groups in these
32108 classes and members in the groups by DECL_UID of declarations.
32109 This would make merge operations stable.
32111 2017-01-10 Martin Liska <mliska@suse.cz>
32113 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
32114 usage of m_classes_vec.
32115 (sem_item_optimizer::~sem_item_optimizer): Likewise.
32116 (sem_item_optimizer::get_group_by_hash): Likewise.
32117 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
32118 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
32119 (sem_item_optimizer::verify_classes): Likewise.
32120 (sem_item_optimizer::process_cong_reduction): Likewise.
32121 (sem_item_optimizer::dump_cong_classes): Likewise.
32122 (sem_item_optimizer::merge_classes): Likewise.
32123 * ipa-icf.h (congruence_class_hash): Rename from
32124 congruence_class_group_hash. Remove declaration of m_classes_vec.
32126 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
32128 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
32129 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
32130 * config.gcc: Add avx512vpopcntdqintrin.h.
32131 * config/i386/avx512vpopcntdqintrin.h: New.
32132 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
32133 * config/i386/i386-builtin-types.def: Add new types.
32134 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
32135 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
32136 __builtin_ia32_vpopcountq_v8di_mask): New.
32137 * config/i386/i386-c.c (ix86_target_macros_internal): Define
32138 __AVX512VPOPCNTDQ__.
32139 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
32140 (PTA_AVX512VPOPCNTDQ): Define.
32141 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
32142 TARGET_AVX512VPOPCNTDQ_P): Define.
32143 * config/i386/i386.opt: Add mavx512vpopcntdq.
32144 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
32145 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
32147 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32149 PR middle-end/77484
32150 * predict.def (PRED_CALL): Set to 67.
32152 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
32154 * expr.c (store_field): In the bitfield case, if the value comes from
32155 a function call and is of an aggregate type returned in registers, do
32156 not modify the field mode; extract the value in all cases if the mode
32157 is BLKmode and the size is not larger than a word.
32159 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
32162 * config/i386/cpuid.h: Fix undefined behavior.
32164 2017-01-04 Jeff Law <law@redhat.com>
32166 PR tree-optimization/79007
32167 PR tree-optimization/67955
32168 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
32169 conservative for pt.null when flag_non_call_exceptions is on.
32171 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32173 PR translation/79019
32174 PR translation/79020
32175 * params.def (PARAM_INLINE_MIN_SPEEDUP,
32176 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
32177 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
32179 * config/avr/avr.opt (maccumulate-args): Likewise.
32180 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
32181 * common.opt (freport-bug): Likewise.
32182 * cif-code.def (CIF_FINAL_ERROR): Likewise.
32183 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
32184 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
32185 translatable string.
32186 * config/i386/i386.c (function_value_32): Likewise.
32187 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
32188 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
32190 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
32191 * common/config/msp430/msp430-common.c (msp430_handle_option):
32193 * symtab.c (symtab_node::verify_base): Likewise.
32194 * opts.c (set_debug_level): Likewise.
32195 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
32196 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
32197 missing whitespace to translatable strings.
32198 * config/avr/avr.md (bswapsi2): Fix typo in comment.
32199 * config/sh/superh.h: Likewise.
32200 * config/i386/xopintrin.h: Likewise.
32201 * config/i386/znver1.md: Likewise.
32202 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
32203 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
32204 * double-int.h (struct double_int): Likewise.
32205 * double-int.c (div_and_round_double): Likewise.
32206 * wide-int.cc: Likewise.
32207 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
32208 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
32209 * cfgcleanup.c (crossjumps_occured): Renamed to ...
32210 (crossjumps_occurred): ... this.
32211 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
32214 PR tree-optimization/78899
32215 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
32216 returning bool return struct loop *, NULL for failure and the new
32218 (versionable_outer_loop_p): Don't version outer loop if it has
32219 dont_vectorized bit set.
32220 (tree_if_conversion): When versioning outer loop, ensure
32221 tree_if_conversion is performed also on the inner loop of the
32222 non-vectorizable outer loop copy.
32223 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
32224 LOOP_VECTORIZED in inner loop of the scalar outer loop and
32225 prevent vectorization of it.
32226 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
32227 the outer loop vectorization of the non-scalar version is attempted
32228 before vectorization of the inner loop in scalar version. If
32229 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
32230 vectorization of its inner loop.
32231 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
32232 has 2 inner loops, rename also on edges from bb whose single pred
32233 is outer_loop->header. Fix typo in function comment.
32235 2017-01-09 Martin Sebor <msebor@redhat.com>
32238 * asan.c (asan_emit_stack_protection): Increase local buffer size
32239 to avoid snprintf truncation warning.
32241 2017-01-09 Andrew Pinski <apinski@cavium.com>
32243 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
32244 to reference thunderx2t99 for the tuning structure
32245 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
32247 (thunderx2t99_extra_costs): This.
32248 * config/aarch64/aarch64-tune.md: Regenerate.
32249 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
32250 (vulcan_addrcost_table): This.
32251 (vulcan_regmove_cost): Rename to ...
32252 (thunderx2t99_regmove_cost): This.
32253 (vulcan_vector_cost): Rename to ...
32254 (thunderx2t99_vector_cost): this.
32255 (vulcan_branch_cost): Rename to ...
32256 (thunderx2t99_branch_cost): This.
32257 (vulcan_tunings): Rename to ...
32258 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
32259 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
32261 2017-01-09 Martin Jambor <mjambor@suse.cz>
32265 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
32266 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
32267 (propagate_vr_accross_jump_function): Use the above function for all
32268 value range computations for pass-through jump functions and type
32269 converasion from explicit value range values.
32270 (ipcp_propagate_stage): Do not attempt to deduce types of formal
32271 parameters from TYPE_ARG_TYPES.
32272 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
32273 (ipa_write_node_info): Stream type of the actual argument.
32274 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
32276 2017-01-09 Martin Liska <mliska@suse.cz>
32279 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
32280 (lookup_compiler): Do not show error message with have_E.
32282 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32284 PR tree-optimization/78938
32285 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
32286 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
32287 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
32288 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
32291 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32293 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
32296 2017-01-09 Richard Biener <rguenther@suse.de>
32298 PR tree-optimization/78997
32299 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
32300 name condition properly.
32302 2017-01-09 Richard Biener <rguenther@suse.de>
32305 * dwarf2out.c (is_cxx): New overload with context.
32306 (is_naming_typedef_decl): Use it.
32308 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32310 * invoke.texi (Option Summary): Correct spacing in option lists
32311 and add line breaks to fix over-long lines.
32313 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32315 PR middle-end/17660
32317 * extend.texi (Common Variable Attributes): Add xref to GCC
32318 Internals manual to explain mode attribute keywords.
32320 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32323 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
32324 and Preprocessor Options.
32325 (Options for Linking): Document -pthread here....
32326 (RS/6000 and PowerPC Options): ...not here.
32327 (Solaris 2 Options): ...or here.
32328 * doc/cppopts.texi: Document -pthread.
32330 2017-01-08 Martin Sebor <msebor@redhat.com>
32332 PR middle-end/77708
32333 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
32334 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
32335 New member functions.
32336 (format_directive): Used them.
32338 (pass_sprintf_length::handle_gimple_call): Same.
32339 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
32340 to avoid truncation for any argument.
32341 (extract_affine_mul): Same.
32342 * tree.c (get_file_function_name): Same.
32344 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32346 PR middle-end/77484
32347 * predict.def (PRED_INDIR_CALL): Set to 86.
32349 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32351 PR preprocessor/54124
32352 * doc/cppopts.texi: Reformat -d subtable to list the full name
32353 of the options. Add cross-reference to the docs for the general
32354 compiler -d options.
32355 * doc/invoke.texi (Developer Options): Add cross-reference to the
32356 preprocessor-specific -d option documentation.
32358 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32360 PR preprocessor/13498
32361 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
32362 redudant material, and reflect new command-line options.
32363 (System Headers): Likewise.
32365 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32367 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
32368 -isystem, and -idirafter. Copy-edit.
32369 * doc/cppopts.texi: Copy-edit. Remove contradiction about
32370 default for -ftrack-macro-expansion. Delete obsolete and
32371 badly-formatted implementation details about -fdebug-cpp output.
32372 * doc/cppwarnopts.texi: Copy-edit.
32374 2017-01-07 David Malcolm <dmalcolm@redhat.com>
32377 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32378 that the transition from a max line width >= 1<<10 to narrower
32379 lines works correctly.
32381 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
32383 * doc/options.texi (PerFunction): New.
32384 * opt-functions.awk (switch_flags): Map both Optimization and
32385 PerFunction to CL_OPTIMIZATION.
32386 * opth-gen.awk: Test for PerFunction flag along with
32388 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
32389 it only when the latter is present. Skip those that don't in
32390 the hash function generator.
32391 * common.opt (fvar-tracking): Mark as PerFunction instead of
32393 (fvar-tracking-assignments): Likewise.
32394 (fvar-tracking-assignments-toggle): Likewise.
32395 (fvar-tracking-uninit): Likewise.
32397 2017-01-07 Jakub Jelinek <jakub@redhat.com>
32399 PR translation/79018
32400 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
32403 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
32406 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
32407 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
32408 TARGET_LONG_JUMP_TABLE_OFFSETS.
32409 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
32410 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
32411 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
32412 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
32413 * config/m68k/m68k.md (tablejump expander): Likewise.
32414 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
32415 TARGET_LONG_JUMP_TABLE_OFFSETS.
32416 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
32417 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
32419 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
32420 David Holsgrove <david.holsgrove@xilinx.com>
32422 * common/config/microblaze/microblaze-common.c
32423 (TARGET_EXCEPT_UNWIND_INFO): Remove.
32424 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
32426 * config/microblaze/microblaze.c (microblaze_must_save_register)
32427 (microblaze_expand_epilogue, microblaze_return_addr): Handle
32429 (microblaze_eh_return): New function.
32430 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
32431 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
32432 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
32433 * config/microblaze/microblaze.md (eh_return): New pattern.
32435 2017-01-06 Jakub Jelinek <jakub@redhat.com>
32437 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
32438 GCC_DIAGNOSTIC_STRINGIFY): Define.
32440 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
32442 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32444 * config/arm/arm.md (<mcrr>): New.
32446 * config/arm/arm.c (arm_arch5te): New.
32447 (arm_option_override): Set arm_arch5te.
32448 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
32450 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
32451 (arm_mcrr_qualifiers): ... this. New.
32452 (MRRC_QUALIFIERS): Define to...
32453 (arm_mrrc_qualifiers): ... this. New.
32454 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
32456 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
32457 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
32458 (MRRCI, mrrc, MRRC): New.
32459 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
32460 VUNSPEC_MRRC2): New.
32462 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32464 * config/arm/arm.md (<mcr>): New.
32466 * config/arm/arm.c (arm_coproc_builtin_available): Add
32467 support for mcr, mrc, mcr2 and mrc2.
32468 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
32469 (arm_mcr_qualifiers): ... this. New.
32470 (MRC_QUALIFIERS): Define to ...
32471 (arm_mrc_qualifiers): ... this. New.
32472 (MCR_QUALIFIERS): Define to ...
32473 (arm_mcr_qualifiers): ... this. New.
32474 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
32476 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
32477 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
32478 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
32479 VUNSPEC_MRC2): New.
32481 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32483 * config/arm/arm.md (*ldc): New.
32487 * config/arm/arm.c (arm_coproc_builtin_available): Add
32488 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
32489 (arm_coproc_ldc_stc_legitimate_address): New.
32490 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
32491 'qualifier_const_pointer'.
32492 (LDC_QUALIFIERS): Define to...
32493 (arm_ldc_qualifiers): ... this. New.
32494 (STC_QUALIFIERS): Define to...
32495 (arm_stc_qualifiers): ... this. New.
32496 * config/arm/arm-protos.h
32497 (arm_coproc_ldc_stc_legitimate_address): New.
32498 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
32499 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
32500 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
32501 stc2, stcl, stc2l): New.
32502 * config/arm/constraints.md (Uz): New.
32503 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
32504 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
32505 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
32506 VUNSPEC_STC2L): New.
32508 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32510 * config/arm/arm.md (<cdp>): New.
32511 * config/arm/arm.c (neon_const_bounds): Rename this ...
32512 (arm_const_bounds): ... this.
32513 (arm_coproc_builtin_available): New.
32514 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
32515 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
32516 (CDP_QUALIFIERS): Define to...
32517 (arm_cdp_qualifiers): ... this. New.
32519 (arm_expand_builtin_args): Add case for 6 arguments.
32520 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
32521 (arm_const_bounds): ... this.
32522 (arm_coproc_builtin_available): New.
32523 * config/arm/arm_acle.h (__arm_cdp): New.
32525 * config/arm/arm_acle_builtins.def (cdp): New.
32527 * config/arm/iterators.md (CDPI,CDP,cdp): New.
32528 * config/arm/neon.md: Rename all 'neon_const_bounds' to
32529 'arm_const_bounds'.
32530 * config/arm/types.md (coproc): New.
32531 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
32532 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
32533 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
32534 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
32536 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32538 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
32539 (UBINOP_QUALIFIERS): New.
32541 (acle_builtin_data): New. Change comment.
32542 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
32543 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
32544 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
32545 arm_acle_builtins.def.
32546 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
32547 (arm_init_acle_builtins): New.
32548 (CRC32_BUILTIN): Remove.
32549 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
32550 crc32cb, crc32ch and crc32cw.
32551 (arm_init_crc32_builtins): Remove.
32552 (arm_init_builtins): Use arm_init_acle_builtins rather
32553 than arm_init_crc32_builtins.
32554 (arm_expand_acle_builtin): New.
32555 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
32556 * config/arm/arm_acle_builtins.def: New.
32558 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32560 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
32561 (arm_builtin_datum): ... this.
32562 (arm_init_neon_builtin): Rename to ...
32563 (arm_init_builtin): ... this. Add a new parameters PREFIX
32564 and USE_SIG_IN_NAME.
32565 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
32566 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
32567 'arm_builtin_datum'.
32568 (arm_init_vfp_builtins): Likewise.
32569 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
32570 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
32571 (arm_expand_neon_args): Rename to ...
32572 (arm_expand_builtin_args): ... this. Rename builtin_arg
32573 enum values and differentiate between ARG_BUILTIN_MEMORY
32574 and ARG_BUILTIN_NEON_MEMORY.
32575 (arm_expand_neon_builtin_1): Rename to ...
32576 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
32577 values, arm_expand_builtin_args and add bool parameter NEON.
32578 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
32579 (arm_expand_vfp_builtin): Likewise.
32580 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
32582 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32584 PR middle-end/77484
32585 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
32586 * predict.c (tree_estimate_probability_bb): Reverse direction of
32587 polymorphic call predictor.
32589 2017-01-06 David Malcolm <dmalcolm@redhat.com>
32591 * passes.c (execute_one_pass): Split out pass-skipping logic into...
32592 (determine_pass_name_match): ...this new function and...
32593 (should_skip_pass_p): ...this new function.
32595 2017-01-06 Nathan Sidwell <nathan@acm.org>
32597 * ipa-visibility.c (function_and_variable_visibility): Reformat
32598 comments and long lines. Remove extrneous if.
32599 * symtab.c (symtab_node::make_decl_local): Fix code format.
32600 (symtab_node::set_section_for_node): Fix comment typo.
32602 2017-01-06 Martin Liska <mliska@suse.cz>
32605 * lra-constraints.c: Rename invariant to lra_invariant.
32606 * predict.c (set_even_probabilities): Initialize e to NULL.
32608 2017-01-05 Martin Sebor <msebor@redhat.com>
32610 PR tree-optimization/78910
32611 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
32612 (format_integer): Correct off-by-one error in the handling
32613 of precision with negative numbers in signed conversions..
32615 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
32617 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
32619 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32621 PR tree-optimization/71016
32622 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
32623 factor_out_conditional_conversion. Formatting fix.
32624 (factor_out_conditional_conversion): Add cond_stmt argument.
32625 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
32626 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
32629 2017-01-05 David Malcolm <dmalcolm@redhat.com>
32631 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
32632 read-rtl-function.o, and selftest-rtl.o.
32633 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
32634 (selftest::aarch64_test_loading_full_dump): New function.
32635 (selftest::aarch64_run_selftests): New function.
32636 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
32637 selftest::aarch64_run_selftests.
32638 * config/i386/i386.c
32639 (selftest::ix86_test_loading_dump_fragment_1): New function.
32640 (selftest::ix86_test_loading_call_insn): New function.
32641 (selftest::ix86_test_loading_full_dump): New function.
32642 (selftest::ix86_test_loading_unspec): New function.
32643 (selftest::ix86_run_selftests): Call the new functions.
32644 * emit-rtl.c (maybe_set_max_label_num): New function.
32645 * emit-rtl.h (maybe_set_max_label_num): New decl.
32646 * function.c (instantiate_decls): Guard call to
32647 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
32648 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
32650 * gensupport.c (gen_reader::gen_reader): Pass "false"
32651 for new "compact" param of rtx_reader.
32652 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
32653 rather than an empty string for NULL strings.
32654 * read-md.c: Potentially include config.h rather than bconfig.h.
32655 Wrap include of errors.h with #ifdef GENERATOR_FILE.
32656 (have_error): New global, copied from errors.c.
32657 (md_reader::read_name): Rename to...
32658 (md_reader::read_name_1): ...this, adding "out_loc" param,
32659 and converting "missing name or number" to returning false, rather
32661 (md_reader::read_name): Reimplement in terms of read_name_1.
32662 (md_reader::read_name_or_nil): New function.
32663 (md_reader::read_string): Handle "(nil)" by returning NULL.
32664 (md_reader::md_reader): Add new param "compact".
32665 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
32666 (md_reader::read_file): New method.
32667 * read-md.h (md_reader::md_reader): Add new param "compact".
32668 (md_reader::read_file): New method.
32669 (md_reader::is_compact): New accessor.
32670 (md_reader::read_name): Convert return type from void to file_location.
32671 (md_reader::read_name_or_nil): New decl.
32672 (md_reader::read_name_1): New decl.
32673 (md_reader::m_compact): New field.
32674 (noop_reader::noop_reader): Pass "false" for new "compact" param
32676 (rtx_reader::rtx_reader): Add new "compact" param.
32677 (rtx_reader::read_rtx_operand): Make virtual and convert return
32678 type from void to rtx.
32679 (rtx_reader::read_until): New decl.
32680 (rtx_reader::handle_any_trailing_information): New virtual function.
32681 (rtx_reader::postprocess): New virtual function.
32682 (rtx_reader::finalize_string): New virtual function.
32683 (rtx_reader::m_in_call_function_usage): New field.
32684 (rtx_reader::m_reuse_rtx_by_id): New field.
32685 * read-rtl-function.c: New file.
32686 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
32687 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
32688 (selftest::verify_three_block_rtl_cfg): New decl.
32689 * read-rtl-function.h: New file.
32690 * read-rtl.c: Potentially include config.h rather than bconfig.h.
32691 For host, include function.h, memmodel.h, and emit-rtl.h.
32692 (one_time_initialization): New function.
32693 (struct compact_insn_name): New struct.
32694 (compact_insn_names): New array.
32695 (find_code): Handle insn codes in compact dumps.
32696 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
32697 (bind_subst_iter_and_attr): Likewise.
32698 (add_condition_to_string): Likewise.
32699 (add_condition_to_rtx): Likewise.
32700 (apply_attribute_uses): Likewise.
32701 (add_current_iterators): Likewise.
32702 (apply_iterators): Likewise.
32703 (initialize_iterators): Guard usage of apply_subst_iterator with
32704 #ifdef GENERATOR_FILE.
32705 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
32706 (md_reader::read_mapping): Likewise.
32707 (add_define_attr_for_define_subst): Likewise.
32708 (add_define_subst_attr): Likewise.
32709 (read_subst_mapping): Likewise.
32710 (check_code_iterator): Likewise.
32711 (rtx_reader::read_rtx): Likewise. Move one-time initialization
32713 (one_time_initialization): New function.
32714 (rtx_reader::read_until): New method.
32715 (read_flags): New function.
32716 (parse_reg_note_name): New function.
32717 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
32718 Handle reuse_rtx ids.
32719 Wrap iterator lookup within #ifdef GENERATOR_FILE.
32720 Add parsing support for RTL dumps, mirroring the special-cases in
32721 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
32722 values, and calling handle_any_trailing_information.
32723 (rtx_reader::read_rtx_operand): Convert return type from void
32724 to rtx, returning return_rtx. Handle case 'e'. Call
32725 finalize_string on XSTR and XTMPL fields.
32726 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
32727 "(nil)" values were omitted. Call the postprocess vfunc on the
32729 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
32730 class ctor. Initialize m_in_call_function_usage. Call
32731 one_time_initialization.
32732 * rtl-tests.c (selftest::test_uncond_jump): Call
32733 set_new_first_and_last_insn.
32734 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
32735 * selftest-rtl.c: New file.
32736 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
32737 (selftest::get_insn_by_uid): New decl.
32738 * selftest-run-tests.c (selftest::run_tests): Call
32739 read_rtl_function_c_tests.
32740 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
32741 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
32744 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
32746 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
32747 operands in a special way. Assert that pos+len <= mode precision.
32749 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32751 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
32752 3 argument Alias with unlimited for the negative form.
32753 (fno-vect-cost-model): Removed.
32755 2017-01-05 Martin Liska <mliska@suse.cz>
32757 * hsa-gen.c (gen_hsa_divmod): New function.
32758 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
32760 2017-01-05 Martin Liska <mliska@suse.cz>
32763 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
32766 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32768 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
32769 small constant length operands.
32771 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32773 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
32774 between loop iterations.
32776 2017-01-05 Martin Liska <mliska@suse.cz>
32779 * gimplify.c (gimplify_decl_expr): Compare to
32780 asan_poisoned_variables instread of checking flags.
32781 (gimplify_target_expr): Likewise.
32782 (gimplify_expr): Likewise.
32783 (gimplify_function_tree): Conditionally initialize
32784 asan_poisoned_variables.
32786 2017-01-04 Jeff Law <law@redhat.com>
32788 PR tree-optimizatin/78812
32789 * rtl.h (contains_mem_rtx_p): Prototype.
32790 * ifcvt.c (containts_mem_rtx_p): Move from here to...
32791 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
32792 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
32793 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
32794 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
32796 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32798 * input.c (assert_char_at_range): Default-initialize actual_range.
32800 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32802 * df-scan.c (df_ref_create_structure): Make regno unsigned,
32803 to match the caller.
32805 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32807 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
32808 insns after final jump in test to emit dummy move.
32810 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32812 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
32813 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
32815 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32817 * multiple_target.c (create_dispatcher_calls): Init e_next.
32818 * tree-ssa-loop-split.c (split_loop): Init border.
32819 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
32822 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
32827 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
32828 (altivec_register_operand): Do not return true if the operand
32829 contains a SUBREG mixing SImode and SFmode.
32830 (vsx_register_operand): Likewise.
32831 (vsx_reg_sfsubreg_ok): New predicate.
32832 (vfloat_operand): Do not return true if the operand contains a
32833 SUBREG mixing SImode and SFmode.
32834 (vint_operand): Likewise.
32835 (vlogical_operand): Likewise.
32836 (gpc_reg_operand): Likewise.
32837 (int_reg_operand): Likewise.
32838 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
32839 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
32840 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
32842 (rs6000_emit_move_si_sf_subreg): New helper function.
32843 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
32844 fixup SUBREGs involving SImode and SFmode.
32845 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
32846 numbers for the new peephole2 optimization.
32847 (peephole2 for SFmode unions): New peephole2 to optimize cases in
32848 the GLIBC math library that do AND/IOR/XOR operations on single
32849 precision floating point.
32850 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
32851 target macros to say whether we need to avoid SUBREGs mixing
32853 (TARGET_ALLOW_SF_SUBREG): Likewise.
32854 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
32855 (UNSPEC_SI_FROM_SF): Likewise.
32856 (iorxor): Change spacing.
32857 (and_ior_xor): New iterator for AND, IOR, and XOR.
32858 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
32859 (movdi_from_sf_zero_ext): Likewise.
32860 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
32861 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
32862 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
32863 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
32864 (fms<mode>4): Likewise.
32865 (fnma<mode>4): Likewise.
32866 (fnms<mode>4): Likewise.
32867 (nfma<mode>4): Likewise.
32868 (nfms<mode>4): Likewise.
32870 2017-01-04 Marek Polacek <polacek@redhat.com>
32873 * doc/invoke.texi: Document -Wpointer-compare.
32875 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32877 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
32880 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
32881 descriptions for -gdwarf-5 and emit them as uleb128 instead of
32884 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
32887 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
32888 documentation of the powerpc_popcntb_ok attribute.
32889 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32890 code to issue warning messages if a requested CPU configuration is
32891 not supported by the binary (assembler and loader) toolchain.
32892 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
32893 made to define a built-in function that has been disabled.
32894 (paired_init_builtins): Add assertion to prevent ICE if attempt is
32895 made to define a built-in function that has been disabled.
32896 (altivec_init_builtins): Add comment explaining why definition
32897 of the DST built-in functions is not preceded by an assertion
32898 check. Add assertions to prevent ICE if attempts are made to
32899 define an altivec predicate or an abs* built-in function that has
32901 (htm_init_builtins): Add comment explaining why definition of the
32902 htm built-in functions is not preceded by an assertion check.
32904 2017-01-04 Jeff Law <law@redhat.com>
32906 PR tree-optimizatin/67955
32907 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
32908 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
32909 the points-to solution does not include pt_null. Use DECL_PT_UID
32912 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
32914 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
32915 Use gen_int_mode instead of gen_lopwart for const_int operands.
32917 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32919 PR tree-optimization/71563
32920 * match.pd: Simplify X << Y into X if Y is known to be 0 or
32921 out of range value - has low bits known to be zero.
32923 2017-01-04 Alan Modra <amodra@gmail.com>
32925 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
32926 * configure: Regenerate.
32927 * config.in: Regenerate.
32929 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32932 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
32933 a substring of the message, but strcmp with the whole message. Ifdef
32934 ENABLE_NLS, translate the message first using dgettext.
32936 2017-01-03 Jeff Law <law@redhat.com>
32938 PR tree-optimizatin/78856
32939 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
32940 (mark_threaded_blocks): Remove code to truncate thread paths that
32941 cross multiple loop headers. Instead invalidate the cached loop
32942 iteration information and handle case of a thread path walking
32943 into an irreducible region.
32945 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
32948 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
32949 assertions. Add support for doing the signbit if the IEEE 128-bit
32950 floating point value is in a GPR.
32951 * config/rs6000/rs6000.md (Fsignbit): Delete.
32952 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
32953 Update the length attribute if the value is in a GPR.
32954 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
32955 the sign or zero extension instruction, since the value is always 0/1.
32956 (signbit<mode>2_dm2): Delete using <Fsignbit>.
32959 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
32960 extracting SImode to a GPR register so that we can generate a
32961 store, limit the vector to be in a traditional Altivec register
32962 for the vextuwrx instruction.
32964 2017-01-03 Ian Lance Taylor <iant@google.com>
32966 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
32968 2017-01-03 Martin Sebor <msebor@redhat.com>
32970 PR tree-optimization/78696
32971 * gimple-ssa-sprintf.c (format_floating): Correct handling of
32972 precision. Use MPFR for %f for greater fidelity. Correct handling
32974 (pass_sprintf_length::compute_format_length): Set width and precision
32975 specified by asrerisk to void_node for vararg functions.
32976 (try_substitute_return_value): Adjust dump output.
32978 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
32980 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
32982 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
32984 * doc/invoke.texi (SPARC options): Document -mlra as the default.
32985 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
32986 -mlra/-mno-lra was passed to the compiler.
32988 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
32990 PR rtl-optimization/65618
32991 * emit-rtl.c (try_split): Move initialization of "before" and
32992 "after" to just before the call to emit_insn_after_setloc.
32994 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
32996 * doc/md.texi (Standard Names): Remove reference to Java frontend.
32998 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
33000 * dwarf2out.c (gen_enumeration_type_die): When
33001 -gno-strict-dwarf, add a DW_AT_encoding attribute.
33003 2017-01-03 Jakub Jelinek <jakub@redhat.com>
33005 PR tree-optimization/78965
33006 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
33007 Change first argument from const call_info & to call_info &. For %n
33008 set info.nowrite to false.
33010 PR middle-end/78901
33011 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
33012 possibly throwing calls.
33014 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
33015 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
33016 and fns handling, rather than in a separate case SSA_NAME.
33018 2017-01-02 Jeff Law <law@redhat.com>
33020 * config/darwin-driver.c (darwin_driver_init): Const-correctness
33021 fixes for first_period and second_period variables.
33023 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
33026 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
33027 (*insvqi_1): New insn pattern.
33028 (*insvqi_1_mem_rex64): Ditto.
33029 (*insvqi_2): Ditto.
33030 (*insvqi_3): Rename from *insvqi.
33032 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
33034 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
33036 * doc/cfg.texi (Edges): Remove reference to Java.
33037 (Maintaining the CFG): Ditto.
33039 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33041 PR middle-end/77674
33042 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
33043 transparent aliases.
33045 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
33047 PR middle-end/77484
33048 * predict.def (PRED_CALL): Update hitrate.
33049 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
33050 * predict.c (tree_estimate_probability_bb): Split CALL predictor
33051 into direct/indirect/polymorphic variants.
33053 2017-01-01 Jakub Jelinek <jakub@redhat.com>
33055 Update copyright years.
33057 * gcc.c (process_command): Update copyright notice dates.
33058 * gcov-dump.c (print_version): Ditto.
33059 * gcov.c (print_version): Ditto.
33060 * gcov-tool.c (print_version): Ditto.
33061 * gengtype.c (create_file): Ditto.
33062 * doc/cpp.texi: Bump @copying's copyright year.
33063 * doc/cppinternals.texi: Ditto.
33064 * doc/gcc.texi: Ditto.
33065 * doc/gccint.texi: Ditto.
33066 * doc/gcov.texi: Ditto.
33067 * doc/install.texi: Ditto.
33068 * doc/invoke.texi: Ditto.
33070 Copyright (C) 2017 Free Software Foundation, Inc.
33072 Copying and distribution of this file, with or without modification,
33073 are permitted in any medium without royalty provided the copyright
33074 notice and this notice are preserved.