1 2017-11-02 Jan Hubicka <hubicka@ucw.cz>
3 * x86-tune.def (X86_TUNE_USE_INCDEC): Enable for Haswell+.
5 2017-11-02 Richard Biener <rguenther@suse.de>
7 PR tree-optimization/82795
8 * tree-if-conv.c (predicate_mem_writes): Remove bogus assert.
10 2017-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
12 * acinclude.m4 (gcc_AC_INITFINI_ARRAY): Don't require
14 (gcc_GAS_CHECK_FEATURE): Remove.
15 * configure.ac (ld_vers) <*-*-solaris2*>: Move comments from
16 gcc_AC_INITFINI_ARRAY here. Update for Solaris 11.4 changes.
17 * configure: Regenerate.
19 2017-11-02 Claudiu Zissulescu <claziss@synopsys.com>
21 * config/arc/arc.c (hwloop_optimize): Account for empty
24 2017-11-02 Richard Biener <rguenther@suse.de>
27 * varasm.c (decode_addr_const): Make offset HOST_WIDE_INT.
28 Truncate ARRAY_REF index and element size.
30 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
32 * doc/invoke.texi (RISC-V Options): Use "@minus{}2 GB", not "-2 GB".
34 2017-11-01 Jeff Law <law@redhat.com>
36 * tree-ssa-ccp.c (ccp_folder): New class derived from
37 substitute_and_fold_engine.
38 (ccp_folder::get_value): New member function.
39 (ccp_folder::fold_stmt): Renamed from ccp_fold_stmt.
40 (ccp_fold_stmt): Remove prototype.
41 (ccp_finalize): Call substitute_and_fold from the ccp_class.
42 * tree-ssa-copy.c (copy_folder): New class derived from
43 substitute_and_fold_engine.
44 (copy_folder::get_value): Renamed from get_value.
45 (fini_copy_prop): Call substitute_and_fold from copy_folder class.
46 * tree-vrp.c (vrp_folder): New class derived from
47 substitute_and_fold_engine.
48 (vrp_folder::fold_stmt): Renamed from vrp_fold_stmt.
49 (vrp_folder::get_value): New member function.
50 (vrp_finalize): Call substitute_and_fold from vrp_folder class.
51 (evrp_dom_walker::before_dom_children): Similarly for replace_uses_in.
52 * tree-ssa-propagate.h (substitute_and_fold_engine): New class to
53 provide a class interface to folder/substitute routines.
54 (ssa_prop_fold_stmt_fn): Remove typedef.
55 (ssa_prop_get_value_fn): Likewise.
56 (subsitute_and_fold): Remove prototype.
57 (replace_uses_in): Likewise.
58 * tree-ssa-propagate.c (substitute_and_fold_engine::replace_uses_in):
59 Renamed from replace_uses_in. Call the virtual member function
60 (substitute_and_fold_engine::replace_phi_args_in): Similarly.
61 (substitute_and_fold_dom_walker): Remove initialization of
62 data member entries for calbacks. Add substitute_and_fold_engine
63 member and initialize it.
64 (substitute_and_fold_dom_walker::before_dom_children0: Use the
65 member functions for get_value, replace_phi_args_in c
66 replace_uses_in, and fold_stmt calls.
67 (substitute_and_fold_engine::substitute_and_fold): Renamed from
68 substitute_and_fold. Remove assert. Update ctor call.
70 * tree-ssa-propagate.h (ssa_prop_visit_stmt_fn): Remove typedef.
71 (ssa_prop_visit_phi_fn): Likewise.
72 (class ssa_propagation_engine): New class to provide an interface
74 * tree-ssa-propagate.c (ssa_prop_visit_stmt): Remove file scoped
76 (ssa_prop_visit_phi): Likewise.
77 (ssa_propagation_engine::simulate_stmt): Moved into class.
78 Call visit_phi/visit_stmt from the class rather than via
79 file scoped static variables.
80 (ssa_propagation_engine::simulate_block): Moved into class.
81 (ssa_propagation_engine::process_ssa_edge_worklist): Similarly.
82 (ssa_propagation_engine::ssa_propagate): Similarly. No longer
83 set file scoped statics for the visit_stmt/visit_phi callbacks.
84 * tree-complex.c (complex_propagate): New class derived from
85 ssa_propagation_engine.
86 (complex_propagate::visit_stmt): Renamed from complex_visit_stmt.
87 (complex_propagate::visit_phi): Renamed from complex_visit_phi.
88 (tree_lower_complex): Call ssa_propagate via the complex_propagate
90 * tree-ssa-ccp.c: (ccp_propagate): New class derived from
91 ssa_propagation_engine.
92 (ccp_propagate::visit_phi): Renamed from ccp_visit_phi_node.
93 (ccp_propagate::visit_stmt): Renamed from ccp_visit_stmt.
94 (do_ssa_ccp): Call ssa_propagate from the ccp_propagate class.
95 * tree-ssa-copy.c (copy_prop): New class derived from
96 ssa_propagation_engine.
97 (copy_prop::visit_stmt): Renamed from copy_prop_visit_stmt.
98 (copy_prop::visit_phi): Renamed from copy_prop_visit_phi_node.
99 (execute_copy_prop): Call ssa_propagate from the copy_prop class.
100 * tree-vrp.c (vrp_prop): New class derived from ssa_propagation_engine.
101 (vrp_prop::visit_stmt): Renamed from vrp_visit_stmt.
102 (vrp_prop::visit_phi): Renamed from vrp_visit_phi_node.
103 (execute_vrp): Call ssa_propagate from the vrp_prop class.
105 2017-11-01 Jakub Jelinek <jakub@redhat.com>
107 PR rtl-optimization/82778
108 PR rtl-optimization/82597
109 * compare-elim.c (struct comparison): Add in_a_setter field.
110 (find_comparison_dom_walker::before_dom_children): Remove killed
111 bitmap and df_simulate_find_defs call, instead walk the defs.
112 Compute last_setter and initialize in_a_setter. Merge definitions
113 with first initialization for a few variables.
114 (try_validate_parallel): Use insn_invalid_p instead of
115 recog_memoized. Return insn rather than just the pattern.
116 (try_merge_compare): Fix up comment. Don't uselessly test if
117 in_a is a REG_P. Use cmp->in_a_setter instead of walking UD
119 (execute_compare_elim_after_reload): Remove df_chain_add_problem
122 2017-11-01 Michael Collison <michael.collison@arm.com>
124 PR rtl-optimization/82597
125 * gcc.dg/pr82597.c: New test.
127 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
128 Alan Hayward <alan.hayward@arm.com>
129 David Sherwood <david.sherwood@arm.com>
131 * config/aarch64/aarch64.c (aarch64_rtx_costs): Use
132 aarch64_hard_regno_nregs to get the number of registers
135 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
136 Alan Hayward <alan.hayward@arm.com>
137 David Sherwood <david.sherwood@arm.com>
139 * config/aarch64/constraints.md (Upl): Rename to...
141 * config/aarch64/aarch64.md
142 (*zero_extend<SHORT:mode><GPI:mode>2_aarch64, *addsi3_aarch64_uxtw):
145 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
146 Alan Hayward <alan.hayward@arm.com>
147 David Sherwood <david.sherwood@arm.com>
149 * config/aarch64/aarch64.c (aarch64_add_constant_internal)
150 (aarch64_add_constant, aarch64_add_sp, aarch64_sub_sp): Move
153 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
154 Alan Hayward <alan.hayward@arm.com>
155 David Sherwood <david.sherwood@arm.com>
157 * config/aarch64/aarch64.c (aarch64_evpc_trn, aarch64_evpc_uzp)
158 (aarch64_evpc_zip, aarch64_evpc_ext, aarch64_evpc_rev)
159 (aarch64_evpc_dup): Generate rtl direcly, rather than using
161 (aarch64_expand_vec_perm_const_1): Explicitly check for permutes
163 * config/aarch64/iterators.md: Add a comment above the permute
164 unspecs to say that they are generated directly by
165 aarch64_expand_vec_perm_const.
166 * config/aarch64/aarch64-simd.md: Likewise the permute instructions.
168 2017-11-01 Nathan Sidwell <nathan@acm.org>
170 * tree-dump.c (dequeue_and_dump): Use HAS_DECL_ASSEMBLER_NAME_P.
172 2017-11-01 Palmer Dabbelt <palmer@dabbelt.com>
174 * doc/invoke.texi (RISC-V Options): Explicitly name the medlow
175 and medany code models, and describe what they do.
177 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
179 Revert accidental duplicate:
181 * combine.c (can_change_dest_mode): Reject changes in
182 REGMODE_NATURAL_SIZE.
184 2017-11-01 Segher Boessenkool <segher@kernel.crashing.org>
186 PR rtl-optimization/64682
187 PR rtl-optimization/69567
188 PR rtl-optimization/69737
189 PR rtl-optimization/82683
190 * combine.c (distribute_notes) <REG_DEAD>: If the new I2 sets the same
191 register mentioned in the note, drop the note, unless it came from I3,
192 in which case it should go to I3 again.
194 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
196 * tree-ssa-dse.c (normalize_ref): Check whether the ranges overlap
197 and return false if not.
198 (clear_bytes_written_by, live_bytes_read): Update accordingly.
200 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
202 * tree-ssa-alias.h (ranges_overlap_p): Return false if either
203 range is known to be empty.
205 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
206 Alan Hayward <alan.hayward@arm.com>
207 David Sherwood <david.sherwood@arm.com>
209 * simplify-rtx.c (simplify_const_unary_operation): Use GET_MODE_NUNITS
210 and CONST_VECTOR_NUNITS instead of computing the number of units from
211 the byte sizes of the vector and element.
212 (simplify_binary_operation_1): Likewise.
213 (simplify_const_binary_operation): Likewise.
214 (simplify_ternary_operation): Likewise.
216 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
217 Alan Hayward <alan.hayward@arm.com>
218 David Sherwood <david.sherwood@arm.com>
220 * var-tracking.c (INT_MEM_OFFSET): Replace with...
221 (int_mem_offset): ...this new function.
222 (var_mem_set, var_mem_delete_and_set, var_mem_delete)
223 (find_mem_expr_in_1pdv, dataflow_set_preserve_mem_locs)
224 (same_variable_part_p, use_type, add_stores, vt_get_decl_and_offset):
227 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
228 Alan Hayward <alan.hayward@arm.com>
229 David Sherwood <david.sherwood@arm.com>
231 * lower-subreg.c (interesting_mode_p): New function.
232 (compute_costs, find_decomposable_subregs, decompose_register)
233 (simplify_subreg_concatn, can_decompose_p, resolve_simple_move)
234 (resolve_clobber, dump_choices): Use it.
236 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
237 Alan Hayward <alan.hayward@arm.com>
238 David Sherwood <david.sherwood@arm.com>
240 * rtlhash.c (add_rtx): Use add_hwi for 'w' and add_int for 'i'.
242 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
243 Alan Hayward <alan.hayward@arm.com>
244 David Sherwood <david.sherwood@arm.com>
246 * alias.c (find_base_value, find_base_term): Only process integer
247 truncations. Check the precision rather than the size.
249 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
250 Alan Hayward <alan.hayward@arm.com>
251 David Sherwood <david.sherwood@arm.com>
253 * machmode.h (is_narrower_int_mode): New function
254 * optabs.c (expand_float, expand_fix): Use it.
255 * dwarf2out.c (rotate_loc_descriptor): Likewise.
257 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
258 Alan Hayward <alan.hayward@arm.com>
259 David Sherwood <david.sherwood@arm.com>
261 * rtl.h (narrower_subreg_mode): New function.
262 * ira-color.c (update_costs_from_allocno): Use it.
264 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
265 Alan Hayward <alan.hayward@arm.com>
266 David Sherwood <david.sherwood@arm.com>
268 * optabs-query.h (convert_optab_p): New function, split out from...
269 (convert_optab_handler): ...here.
270 (widening_optab_handler): Delete.
271 (find_widening_optab_handler): Remove permit_non_widening parameter.
272 (find_widening_optab_handler_and_mode): Likewise. Provide an
273 override that operates on mode class wrappers.
274 * optabs-query.c (widening_optab_handler): Delete.
275 (find_widening_optab_handler_and_mode): Remove permit_non_widening
276 parameter. Assert that the two modes are the same class and that
277 the "from" mode is narrower than the "to" mode. Use
278 convert_optab_handler instead of widening_optab_handler.
279 * expmed.c (expmed_mult_highpart_optab): Use convert_optab_handler
280 instead of widening_optab_handler.
281 * expr.c (expand_expr_real_2): Update calls to
282 find_widening_optab_handler.
283 * optabs.c (expand_widen_pattern_expr): Likewise.
284 (expand_binop_directly): Take the insn_code as a parameter.
285 (expand_binop): Only call find_widening_optab_handler for
286 conversion optabs; use optab_handler otherwise. Update calls
287 to find_widening_optab_handler and expand_binop_directly.
288 Use convert_optab_handler instead of widening_optab_handler.
289 * tree-ssa-math-opts.c (convert_mult_to_widen): Update calls to
290 find_widening_optab_handler and use scalar_mode rather than
292 (convert_plusminus_to_widen): Likewise.
294 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
295 Alan Hayward <alan.hayward@arm.com>
296 David Sherwood <david.sherwood@arm.com>
298 * machmode.h (fixed_size_mode): New class.
299 * rtl.h (get_pool_mode): Return fixed_size_mode.
300 * gengtype.c (main): Add fixed_size_mode.
301 * target.def (get_raw_result_mode): Return a fixed_size_mode.
302 (get_raw_arg_mode): Likewise.
303 * doc/tm.texi: Regenerate.
304 * targhooks.h (default_get_reg_raw_mode): Return a fixed_size_mode.
305 * targhooks.c (default_get_reg_raw_mode): Likewise.
306 * config/ia64/ia64.c (ia64_get_reg_raw_mode): Likewise.
307 * config/mips/mips.c (mips_get_reg_raw_mode): Likewise.
308 * config/msp430/msp430.c (msp430_get_raw_arg_mode): Likewise.
309 (msp430_get_raw_result_mode): Likewise.
310 * config/avr/avr-protos.h (regmask): Use as_a <fixed_side_mode>
311 * dbxout.c (dbxout_parms): Require fixed-size modes.
312 * expr.c (copy_blkmode_from_reg, copy_blkmode_to_reg): Likewise.
313 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
314 * omp-low.c (lower_oacc_reductions): Likewise.
315 * simplify-rtx.c (simplify_immed_subreg): Take fixed_size_modes.
316 (simplify_subreg): Update accordingly.
317 * varasm.c (constant_descriptor_rtx::mode): Change to fixed_size_mode.
318 (force_const_mem): Update accordingly. Return NULL_RTX for modes
319 that aren't fixed-size.
320 (get_pool_mode): Return a fixed_size_mode.
321 (output_constant_pool_2): Take a fixed_size_mode.
323 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
324 Alan Hayward <alan.hayward@arm.com>
325 David Sherwood <david.sherwood@arm.com>
327 * doc/rtl.texi (vec_series): Document.
328 (const): Say that the operand can be a vec_series.
329 * rtl.def (VEC_SERIES): New rtx code.
330 * rtl.h (const_vec_series_p_1): Declare.
331 (const_vec_series_p): New function.
332 * emit-rtl.h (gen_const_vec_series): Declare.
333 (gen_vec_series): Likewise.
334 * emit-rtl.c (const_vec_series_p_1, gen_const_vec_series)
335 (gen_vec_series): Likewise.
336 * optabs.c (expand_mult_highpart): Use gen_const_vec_series.
337 * simplify-rtx.c (simplify_unary_operation): Handle negations
339 (simplify_binary_operation_series): New function.
340 (simplify_binary_operation_1): Use it. Handle VEC_SERIES.
341 (test_vector_ops_series): New function.
342 (test_vector_ops): Call it.
343 * config/powerpcspe/altivec.md (altivec_lvsl): Use
344 gen_const_vec_series.
345 (altivec_lvsr): Likewise.
346 * config/rs6000/altivec.md (altivec_lvsl, altivec_lvsr): Likewise.
348 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
349 Alan Hayward <alan.hayward@arm.com>
350 David Sherwood <david.sherwood@arm.com>
352 * doc/rtl.texi (const): Update description of address constants.
353 Say that vector constants are allowed too.
354 * common.md (E, F): Use CONSTANT_P instead of checking for
356 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
357 checking for CONST_VECTOR.
358 * expmed.c (make_tree): Use build_vector_from_val for a CONST
360 * expr.c (expand_expr_real_2): Check for vector modes instead
361 of checking for CONST_VECTOR.
362 * rtl.h (const_vec_p): New function.
363 (const_vec_duplicate_p): Check for a CONST VEC_DUPLICATE.
364 (unwrap_const_vec_duplicate): Handle them here too.
366 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
367 David Malcolm <dmalcolm@redhat.com>
368 Alan Hayward <alan.hayward@arm.com>
369 David Sherwood <david.sherwood@arm.com>
371 * rtl.h (vec_duplicate_p): New function.
372 * selftest-rtl.c (assert_rtx_eq_at): New function.
373 * selftest-rtl.h (ASSERT_RTX_EQ): New macro.
374 (assert_rtx_eq_at): Declare.
375 * selftest.h (selftest::simplify_rtx_c_tests): Declare.
376 * selftest-run-tests.c (selftest::run_tests): Call it.
377 * simplify-rtx.c: Include selftest.h and selftest-rtl.h.
378 (simplify_unary_operation_1): Recursively handle vector duplicates.
379 (simplify_binary_operation_1): Likewise. Handle VEC_SELECTs of
381 (simplify_subreg): Handle subregs of vector duplicates.
382 (make_test_reg, test_vector_ops_duplicate, test_vector_ops)
383 (selftest::simplify_rtx_c_tests): New functions.
385 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
386 Alan Hayward <alan.hayward@arm.com>
387 David Sherwood <david.sherwood@arm.com>
389 * emit-rtl.h (gen_const_vec_duplicate): Declare.
390 (gen_vec_duplicate): Likewise.
391 * emit-rtl.c (gen_const_vec_duplicate_1): New function, split
393 (gen_const_vector): ...here.
394 (gen_const_vec_duplicate, gen_vec_duplicate): New functions.
395 (gen_rtx_CONST_VECTOR): Use gen_const_vec_duplicate for constants
396 whose elements are all equal.
397 * optabs.c (expand_vector_broadcast): Use gen_const_vec_duplicate.
398 * simplify-rtx.c (simplify_const_unary_operation): Likewise.
399 (simplify_relational_operation): Likewise.
400 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
402 (aarch64_simd_dup_constant): Use gen_vec_duplicate.
403 (aarch64_expand_vector_init): Likewise.
404 * config/arm/arm.c (neon_vdup_constant): Likewise.
405 (neon_expand_vector_init): Likewise.
406 (arm_expand_vec_perm): Use gen_const_vec_duplicate.
407 (arm_block_set_unaligned_vect): Likewise.
408 (arm_block_set_aligned_vect): Likewise.
409 * config/arm/neon.md (neon_copysignf<mode>): Likewise.
410 * config/i386/i386.c (ix86_expand_vec_perm): Likewise.
411 (expand_vec_perm_even_odd_pack): Likewise.
412 (ix86_vector_duplicate_value): Use gen_vec_duplicate.
413 * config/i386/sse.md (one_cmpl<mode>2): Use CONSTM1_RTX.
414 * config/ia64/ia64.c (ia64_expand_vecint_compare): Use
415 gen_const_vec_duplicate.
416 * config/ia64/vect.md (addv2sf3, subv2sf3): Use CONST1_RTX.
417 * config/mips/mips.c (mips_gen_const_int_vector): Use
418 gen_const_vec_duplicate.
419 (mips_expand_vector_init): Use CONST0_RTX.
420 * config/powerpcspe/altivec.md (abs<mode>2, nabs<mode>2): Likewise.
421 (define_split): Use gen_const_vec_duplicate.
422 * config/rs6000/altivec.md (abs<mode>2, nabs<mode>2): Use CONST0_RTX.
423 (define_split): Use gen_const_vec_duplicate.
424 * config/s390/vx-builtins.md (vec_genmask<mode>): Likewise.
425 (vec_ctd_s64, vec_ctd_u64, vec_ctsl, vec_ctul): Likewise.
426 * config/spu/spu.c (spu_const): Likewise.
428 2017-11-01 Richard Sandiford <richard.sandiford@linaro.org>
429 Alan Hayward <alan.hayward@arm.com>
430 David Sherwood <david.sherwood@arm.com>
432 * combine.c (can_change_dest_mode): Reject changes in
433 REGMODE_NATURAL_SIZE.
435 2017-10-31 Sandra Loosemore <sandra@codesourcery.com>
437 * configure.ac (--enable-libssp): New.
438 (gcc_cv_libc_provides_ssp): Check for explicit setting before
439 trying to determine target-specific default. Adjust indentation.
440 * configure: Regenerated.
441 * doc/install.texi (Configuration): Expand --disable-libssp
444 2017-10-31 Daniel Santos <daniel.santos@pobox.com>
446 config/i386/i386.c (ix86_expand_epilogue): Correct stack
449 2017-10-31 Martin Jambor <mjambor@suse.cz>
452 * gimple-fold.c (gimple_get_virt_method_for_vtable): Remove assert.
454 2017-10-31 David Malcolm <dmalcolm@redhat.com>
456 * auto-profile.c (autofdo_source_profile::read): Use
457 UNKNOWN_LOCATION rather than 0.
458 * diagnostic-core.h (warning_at_rich_loc): Rename to...
459 (warning_at): ...this overload.
460 (warning_at_rich_loc_n): Rename to...
461 (warning_n): ...this overload.
462 (error_at_rich_loc): Rename to...
463 (error_at): ...this overload.
464 (pedwarn_at_rich_loc): Rename to...
465 (pedwarn): ...this overload.
466 (permerror_at_rich_loc): Rename to...
467 (permerror): ...this overload.
468 (inform_at_rich_loc): Rename to...
469 (inform): ...this overload.
470 * diagnostic.c: (diagnostic_n_impl): Delete location_t-based decl.
471 (diagnostic_n_impl_richloc): Rename to...
472 (diagnostic_n_impl): ...this rich_location *-based decl.
473 (inform_at_rich_loc): Rename to...
474 (inform): ...this, and add an assertion.
475 (inform_n): Update for removal of location_t-based diagnostic_n_impl.
476 (warning_at_rich_loc): Rename to...
477 (warning_at): ...this, and add an assertion.
478 (warning_at_rich_loc_n): Rename to...
479 (warning_n): ...this, and add an assertion.
480 (warning_n): Update location_t-based implementation for removal of
481 location_t-based diagnostic_n_impl.
482 (pedwarn_at_rich_loc): Rename to...
483 (pedwarn): ...this, and add an assertion.
484 (permerror_at_rich_loc): Rename to...
485 (permerror): ...this, and add an assertion.
486 (error_n): Update for removal of location_t-based diagnostic_n_impl.
487 (error_at_rich_loc): Rename to...
488 (error_at): ...this, and add an assertion.
489 * gcc.c (do_spec_1): Use UNKNOWN_LOCATION rather than 0.
490 (driver::do_spec_on_infiles): Likewise.
491 * substring-locations.c (format_warning_va): Update for renaming
492 of inform_at_rich_loc.
494 2017-10-31 Michael Meissner <meissner@linux.vnet.ibm.com>
496 * builtins.def (DEF_FLOATN_BUILTIN): Change most _Float<N> and
497 _Float<N>X built-in functions so that the variant without the
498 "__builtin_" prefix is only enabled for the GNU C and Objective C
499 languages when they are in non-strict ANSI/ISO mode.
500 (DEF_EXT_LIB_FLOATN_NX_BUILTINS): Likewise.
501 * target.def (floatn_builtin_p): Add a target hook to control
502 whether _Float<N> and _Float<N>X built-in functions without the
503 "__builtin_" prefix are enabled, and return true for C and
504 Objective C in the default hook. Include langhooks.h in
506 * targhooks.h (default_floatn_builtin_p): Likewise.
507 * targhooks.c (default_floatn_builtin_p): Likewise.
508 * doc/tm.texi.in (TARGET_FLOATN_BUILTIN_P): Document the
509 floatn_builtin_p target hook.
510 * doc/tm.texi (TARGET_FLOATN_BUILTIN_P): Likewise.
512 2017-10-31 Matthew Fortune <matthew.fortune@imgtec.com>
513 Eric Botcazou <ebotcazou@adacore.com>
515 PR rtl-optimization/81803
516 * lra-constraints.c (curr_insn_transform): Also reload the whole
517 register for a strict subreg no wider than a word if this is for
518 a WORD_REGISTER_OPERATIONS target.
520 2017-10-31 Jason Merrill <jason@redhat.com>
522 * gdbinit.in: Skip over inlines from timevar.h.
524 2017-10-31 Martin Liska <mliska@suse.cz>
526 * doc/gcov.texi: Document new option.
527 * gcov.c (print_usage): Likewise print it.
528 (process_args): Support the argument.
529 (format_count): New function.
530 (format_gcov): Use the function.
532 2017-10-31 Martin Liska <mliska@suse.cz>
534 * gcov.c (struct name_map): do not use typedef.
535 Define operator== and operator<.
536 (name_search): Remove.
538 (main): Do not allocate names.
539 (process_file): Add vertical space.
540 (generate_results): Use std::find.
541 (release_structures): Do not release memory.
542 (find_source): Use std::find.
544 2017-10-31 Martin Liska <mliska@suse.cz>
546 * gcov.c (struct line_info): Remove it's typedef.
547 (line_info::line_info): Add proper ctor.
548 (line_info::has_block): Do not use a typedef.
549 (struct source_info): Do not use typedef.
551 (get_cycles_count): Likewise.
552 (output_intermediate_file): Iterate via vector iterator.
553 (add_line_counts): Use std::vector methods.
554 (accumulate_line_counts): Likewise.
555 (output_lines): Likewise.
557 2017-10-31 Martin Liska <mliska@suse.cz>
559 * gcov.c (struct source_info): Remove typedef.
560 (source_info::source_info): Add proper ctor.
561 (accumulate_line_counts): Use struct, not it's typedef.
562 (output_gcov_file): Likewise.
563 (output_lines): Likewise.
564 (main): Do not allocate an array.
565 (output_intermediate_file): Use size of vector container.
566 (process_file): Resize the vector.
567 (generate_results): Do not preallocate, use newly added vector
569 (release_structures): Do not release sources.
570 (find_source): Use vector methods.
571 (add_line_counts): Do not use typedef.
573 2017-10-31 Martin Liska <mliska@suse.cz>
575 * doc/gcov.texi: Document that.
576 * gcov.c (add_line_counts): Mark lines with a non-executed
578 (output_line_beginning): Handle such lines.
579 (output_lines): Pass new argument.
580 (output_intermediate_file): Print it in intermediate format.
582 2017-10-31 Martin Liska <mliska@suse.cz>
584 * color-macros.h: New file.
585 * diagnostic-color.c: Factor out color related to macros to
587 * doc/gcov.texi: Document -k option.
588 * gcov.c (INCLUDE_STRING): Include string.h.
589 (print_usage): Add -k option.
590 (process_args): Parse it.
591 (pad_count_string): New function.
592 (output_line_beginning): Likewise.
593 (DEFAULT_LINE_START): New macro.
594 (output_lines): Support color output.
596 2017-10-31 Martin Liska <mliska@suse.cz>
598 PR gcov-profile/82633
599 * doc/gcov.texi: Document -fkeep-{static,inline}-functions and
600 their interaction with GCOV infrastructure.
601 * configure.ac: Add -fkeep-{inline,static}-functions to
603 * configure: Regenerate.
605 2017-10-31 Uros Bizjak <ubizjak@gmail.com>
608 * config/alpha/sync.md (fetchop_constr) <and>: Change to "rINM".
610 2017-10-31 Segher Boessenkool <segher@kernel.crashing.org>
613 * config/rs6000/rs6000.md (allocate_stack): Force update interval
614 into a register if it does not fit into an immediate offset field.
616 2017-10-31 Olivier Hainque <hainque@adacore.com>
618 * gcc/Makefile.in (FLAGS_TO_PASS): Pass libsubdir as well.
620 2017-10-31 Julia Koval <julia.koval@intel.com>
622 * config.gcc: Add gfniintrin.h.
623 * config/i386/gfniintrin.h: New.
624 * config/i386/i386-builtin-types.def
625 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
626 __builtin_ia32_vgf2p8affineinvqb_v64qi_mask,
627 __builtin_ia32_vgf2p8affineinvqb_v32qi,
628 __builtin_ia32_vgf2p8affineinvqb_v32qi_mask,
629 __builtin_ia32_vgf2p8affineinvqb_v16qi,
630 __builtin_ia32_vgf2p8affineinvqb_v16qi_mask): New builtins.
631 * config/i386/i386-builtin.def (V64QI_FTYPE_V64QI_V64QI_INT_V64QI_UDI,
632 V32QI_FTYPE_V32QI_V32QI_INT_V32QI_USI,
633 V16QI_FTYPE_V16QI_V16QI_INT_V16QI_UHI,
634 V64QI_FTYPE_V64QI_V64QI_INT): New types.
635 * config/i386/i386.c (ix86_expand_args_builtin): Handle new types.
636 * config/i386/immintrin.h: Include gfniintrin.h.
637 * config/i386/sse.md (vgf2p8affineinvqb_*) New pattern.
639 2017-10-30 Eric Botcazou <ebotcazou@adacore.com>
641 * gcc.c (HAVE_TARGET_EXECUTABLE_SUFFIX): Remove old kludge.
643 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
645 * config/arm/arm.md (ashldi3): Remove shift by 1 expansion.
646 (arm_ashldi3_1bit): Remove pattern.
647 (ashrdi3): Remove shift by 1 expansion.
648 (arm_ashrdi3_1bit): Remove pattern.
649 (lshrdi3): Remove shift by 1 expansion.
650 (arm_lshrdi3_1bit): Remove pattern.
651 * config/arm/arm.c (arm_rtx_costs_internal): Slightly increase
652 cost of ashldi3 by 1.
653 * config/arm/neon.md (ashldi3_neon): Remove shift by 1 expansion.
654 (<shift>di3_neon): Likewise.
656 2017-10-30 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
658 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov): Rename
659 both identically named patterns to (*aarch64_simd_mov<VD:mode>)
660 and (*aarch64_simd_mov<VQ:mode>).
661 (*aarch64_simd_mov<VD:mode>): Change type attribute to match
663 (*aarch64_simd_mov<VQ:mode>): Re-order and change type
664 attributes to match pattern alternative.
666 2017-10-30 Steven Munroe <munroesj@gcc.gnu.org>
668 * config.gcc (powerpc*-*-*): Add emmintrin.h.
669 * config/rs6000/emmintrin.h: New file.
670 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include emmintrin.h.
672 2017-10-30 Wilco Dijkstra <wdijkstr@arm.com>
674 * config/arm/vfp.md (movdi_vfp): Merge changes from movdi_vfp_cortexa8.
675 * (movdi_vfp_cortexa8): Remove pattern.
677 2017-10-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
679 * doc/install.texi (Specific, alpha*-*-*): Remove DEC OSF/1
681 (Specific, alpha*-dec-osf5.1): Remove.
682 (Specific, mips-sgi-irix5): Remove.
683 (Specific, mips-sgi-irix6): Remove.
685 2017-10-30 Jakub Jelinek <jakub@redhat.com>
688 * gimple-ssa-store-merging.c (merged_store_group::apply_stores): Fix
689 arguments to clear_bit_region_be.
691 2017-10-30 Jim Wilson <wilson@tuliptree.org>
693 * gimplify.c: Include memmodel.h.
695 2017-10-30 Martin Jambor <mjambor@suse.cz>
697 * omp-grid.c (grid_attempt_target_gridification): Also insert a
698 condition whether loop should be executed at all.
700 2017-10-30 Will Schmidt <will_schmidt@vnet.ibm.com>
702 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
703 gimple folding of vec_madd() intrinsics.
704 * config/rs6000/altivec.md (mulv8hi3): Rename altivec_vmladduhm to
705 fmav8hi4. (altivec_vmladduhm): Rename to fmav8hi4.
706 * config/rs6000/rs6000-builtin.def: Rename vmladduhm to fmav8hi4
708 2017-10-30 Richard Biener <rguenther@suse.de>
710 PR tree-optimization/82762
712 2017-10-23 Richard Biener <rguenther@suse.de>
714 PR tree-optimization/82129
716 2017-08-01 Richard Biener <rguenther@suse.de>
718 PR tree-optimization/81181
719 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
720 (compute_antic): ... end of iteration here.
722 2017-10-30 Joseph Myers <joseph@codesourcery.com>
724 * doc/invoke.texi (C Dialect Options): Document -std=c17,
725 -std=iso9899:2017 and -std=gnu17.
726 * doc/standards.texi (C Language): Document C17 support.
727 * doc/cpp.texi (Overview): Mention -std=c17.
728 (Standard Predefined Macros): Document C11 and C17 values of
729 __STDC_VERSION__. Do not refer to C99 support as incomplete.
730 * doc/extend.texi (Inline): Do not list individual options for
731 standards newer than C99.
732 * dwarf2out.c (highest_c_language, gen_compile_unit_die): Handle
734 * config/rl78/rl78.c (rl78_option_override): Handle "GNU C17"
737 2017-10-30 Maxim Ostapenko <m.ostapenko@samsung.com>
739 * asan.c (asan_finish_file): Align asan globals array by shadow
742 2017-10-30 Jakub Jelinek <jakub@redhat.com>
745 * gimple-ssa-store-merging.c: Include rtl.h and expr.h.
746 (struct store_immediate_info): Add bitregion_start and bitregion_end
748 (store_immediate_info::store_immediate_info): Add brs and bre
749 arguments and initialize bitregion_{start,end} from those.
750 (struct merged_store_group): Add bitregion_start, bitregion_end,
751 align_base and mask fields. Drop unnecessary struct keyword from
752 struct store_immediate_info. Add do_merge method.
753 (clear_bit_region_be): Use memset instead of loop storing zeros.
754 (merged_store_group::do_merge): New method.
755 (merged_store_group::merge_into): Use do_merge. Allow gaps in between
756 stores as long as the surrounding bitregions have no gaps.
757 (merged_store_group::merge_overlapping): Use do_merge.
758 (merged_store_group::apply_stores): Test that bitregion_{start,end}
759 is byte aligned, rather than requiring that start and width are
760 byte aligned. Drop unnecessary struct keyword from
761 struct store_immediate_info. Allocate and populate also mask array.
762 Make start of the arrays relative to bitregion_start rather than
763 start and size them according to bitregion_{end,start} difference.
764 (struct imm_store_chain_info): Drop unnecessary struct keyword from
765 struct store_immediate_info.
766 (pass_store_merging::gate): Punt if BITS_PER_UNIT or CHAR_BIT is not 8.
767 (pass_store_merging::terminate_all_aliasing_chains): Drop unnecessary
768 struct keyword from struct store_immediate_info.
769 (imm_store_chain_info::coalesce_immediate_stores): Allow gaps in
770 between stores as long as the surrounding bitregions have no gaps.
772 (struct split_store): Add orig non-static data member.
773 (split_store::split_store): Initialize orig to false.
774 (find_constituent_stmts): Return store_immediate_info *, non-NULL
775 if there is exactly a single original stmt. Change stmts argument
776 to pointer from reference, if NULL, don't push anything to it. Add
777 first argument, use it to optimize skipping over orig stmts that
778 are known to be before bitpos already. Simplify.
779 (split_group): Return unsigned int count how many stores are or
780 would be needed rather than a bool. Add allow_unaligned argument.
781 Change split_stores argument from reference to pointer, if NULL,
782 only do a dry run computing how many stores would be produced.
783 Rewritten algorithm to use both alignment and misalign if
784 !allow_unaligned and handle bitfield stores with gaps.
785 (imm_store_chain_info::output_merged_store): Set start_byte_pos
786 from bitregion_start instead of start. Compute allow_unaligned
787 here, if true, do 2 split_group dry runs to compute which one
788 produces fewer stores and prefer aligned if equal. Punt if
789 new count is bigger or equal than original before emitting any
790 statements, rather than during that. Remove no longer needed
791 new_ssa_names tracking. Replace num_stmts with
792 split_stores.length (). Use 32-bit stack allocated entries
793 in split_stores auto_vec. Try to reuse original store lhs/rhs1
794 if possible. Handle bitfields with gaps.
795 (pass_store_merging::execute): Ignore bitsize == 0 stores.
796 Compute bitregion_{start,end} for the stores and construct
797 store_immediate_info with that. Formatting fixes.
799 2017-10-30 Uros Bizjak <ubizjak@gmail.com>
802 * config/i386/i386.c (legitimate_pic_address_disp_p): Allow
803 UNSPEC_DTPOFF and UNSPEC_NTPOFF with SImode immediate offset.
805 2017-10-29 Jim Wilson <wilson@tuliptree.org>
807 * gimplify.c: Include tm_p.h.
809 * common.opt (gcoff): Re-add as ignored option.
810 (gcoff1, gcoff2, gcoff3): Likewise.
812 * Makefile.in (OBJS): Delete sdbout.o.
813 (GTFILES): Delete $(srcdir)/sdbout.c.
814 * debug.h: Delete sdb_debug_hooks.
815 * final.c: Delete sdbout.h include.
816 (final_scan_insn): Delete SDB_DEBUG check.
817 (rest_of_clean_state): Likewise.
818 * output.h: Delete sdb_begin_function_line.
821 * toplev.c: Delete sdbout.h include.
822 (process_options): Delete SDB_DEBUG check.
823 * tree-core.h (tree_type_common): Delete pointer field of
825 * tree.c (copy_node): Clear TYPE_SYMTAB_DIE instead of
827 * tree.h (TYPE_SYMTAB_POINTER): Delete.
828 (TYPE_SYMTAB_IS_POINTER): Delete.
829 (TYPE_SYMTAB_IS_DIE): Renumber.
830 * xcoffout.c: Refer to former sdbout.c file.
831 (xcoffout_begin_prologue): Use past tense for sdbout.c reference.
833 * doc/install.texi (--with-stabs): Delete COFF and ECOFF info.
834 * doc/invoke.texi (SEEALSO): Delete adb and sdb references.
835 (Debugging Options): Delete -gcoff.
836 (-gstabs): Delete SDB reference.
838 (-gcoff@var{level}): Delete.
839 * doc/passes.texi (Debugging information output): Delete SDB and
841 * doc/tm.texi: Regenerate.
842 * doc/tm.texi.in (DWARF_CIE_DATA_ALIGNMENT): Delete SDB from xref.
843 (SDB and DWARF): Change node name to DWARF and delete SDB and COFF
845 (DEBUGGER_AUTO_OFFSET): Delete COFF and SDB references.
846 (PREFERRED_DEBUGGING_TYPE): Delete SDB_DEBUG and -gcoff references.
847 (SDB_DEBUGGING_INFO): Delete.
848 (PUT_SDB_@dots{}, SDB_DELIM, SDB_ALLOW_UNKNOWN_REFERENCES)
849 SDB_ALLOW_FORWARD_REFERENCES, SDB_OUTPUT_SOURCE_LINE): Delete.
850 * target.def (output_source_filename): Delete COFF reference.
852 * common.opt (gcoff): Delete.
853 (gxcoff+): Update Negative chain.
854 * defaults.h: Delete all references to SDB_DEBUGGING_INFO and
856 * dwarf2out.c (gen_array_type_die): Change SDB to debuggers.
857 * flag-types.h (enum debug_info_type): Delete SDB_DEBUG.
858 * function.c (number_blocks): Delete SDB_DEBUGGING_INFO, SDB_DEBUG,
860 (expand_function_start): Change sdb reference to past tense.
861 (expand_function_end): Change sdb reference to past tense.
862 * gcc.c (cpp_unique_options): Delete gcoff3 reference.
863 * opts.c (debug_type_names): Delete coff entry.
864 (common_handle_option): Delete OPT_gcoff case.
865 * system.h (SDB_DEBUG, SDB_DEBUGGING_INFO): Poison.
867 * config/dbxcoff.h (PREFERRED_DEBUGGING_TYPE): Set to DBX_DEBUG.
868 * config/cris/cris.h: Delete SDB reference in comment.
869 * config/i386/cygming.h: Don't define SDB_DEBUGGING_INFO.
870 (ASM_DECLARE_FUNCTION_NAME): Delete SDB reference from comment.
871 * config/i386/gas.h: Don't define SDB_DEBUGGING_INFO.
872 * config/i386/i386.c (svr4_dbx_register_map): Change SDB references
874 (ix86_expand_prologue): Likewise.
875 * config/i386/winnt.c (i386_pe_start_function): Don't check SDB_DEBUG.
876 * config/ia64/ia64.h: Likewise.
877 * config/m68k/m68kelf.h (DBX_REGISTER_NUMBER): Delete SDB reference.
878 * config/mips/mips.h (SUBTARGET_ASM_DEBUGGING_SPEC): Delete gcoff*
880 * config/mmix/mmix.h: Likewise.
881 * config/nds32/nds32.c: Likewise.
882 * config/stormy/storym16.h: Likewise.
883 * config/visium/visium.h: Likewise.
884 * config/vx-common.h (SDB_DEBUGGING_INFO): Delete undef.
886 2017-10-28 Sandra Loosemore <sandra@codesourcery.com>
888 * config/nios2/nios2.h (FRAME_GROWS_DOWNWARD): Define to 1.
889 * config/nios2/nios2.c (nios2_initial_elimination_offset): Make
890 FRAME_POINTER_REGNUM point at high end of local var area.
892 2017-10-27 Eric Botcazou <ebotcazou@adacore.com>
894 * bb-reorder.c (find_traces_1_round): Fix off-by-one index.
895 Move comment around. Do not reset best_edge for a copiable
896 destination if the copy would cause a partition change.
897 (better_edge_p): Remove redundant check.
899 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
901 * config/i386/i386-protos.h (ix86_fp_compare_mode): Remove prototype.
903 2017-10-27 Michael Meissner <meissner@linux.vnet.ibm.com>
905 * builtins.c (CASE_MATHFN_FLOATN): New helper macro to add cases
906 for math functions that have _Float<N> and _Float<N>X variants.
907 (mathfn_built_in_2): Add support for math functions that have
908 _Float<N> and _Float<N>X variants.
909 (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro.
910 (expand_builtin_mathfn_ternary): Add support for fma with
911 _Float<N> and _Float<N>X variants.
912 (expand_builtin): Likewise.
913 (fold_builtin_3): Likewise.
914 * builtins.def (DEF_EXT_LIB_FLOATN_NX_BUILTINS): New macro to
915 create math function _Float<N> and _Float<N>X variants as external
917 (BUILT_IN_COPYSIGN _Float<N> and _Float<N>X variants) Use
918 DEF_EXT_LIB_FLOATN_NX_BUILTINS to make built-in functions using
919 the __builtin_ prefix and if not strict ansi, without the prefix.
920 (BUILT_IN_FABS _Float<N> and _Float<N>X variants): Likewise.
921 (BUILT_IN_FMA _Float<N> and _Float<N>X variants): Likewise.
922 (BUILT_IN_FMAX _Float<N> and _Float<N>X variants): Likewise.
923 (BUILT_IN_FMIN _Float<N> and _Float<N>X variants): Likewise.
924 (BUILT_IN_NAN _Float<N> and _Float<N>X variants): Likewise.
925 (BUILT_IN_SQRT _Float<N> and _Float<N>X variants): Likewise.
926 * builtin-types.def (BT_FN_FLOAT16_FLOAT16_FLOAT16_FLOAT16): New
927 function signatures for fma _Float<N> and _Float<N>X variants.
928 (BT_FN_FLOAT32_FLOAT32_FLOAT32_FLOAT32): Likewise.
929 (BT_FN_FLOAT64_FLOAT64_FLOAT64_FLOAT64): Likewise.
930 (BT_FN_FLOAT128_FLOAT128_FLOAT128_FLOAT128): Likewise.
931 (BT_FN_FLOAT32X_FLOAT32X_FLOAT32X_FLOAT32X): Likewise.
932 (BT_FN_FLOAT64X_FLOAT64X_FLOAT64X_FLOAT64X): Likewise.
933 (BT_FN_FLOAT128X_FLOAT128X_FLOAT128X_FLOAT128X): Likewise.
934 * gencfn-macros.c (print_case_cfn): Add support for math functions
935 that have _Float<N> and _Float<N>X variants.
936 (print_define_operator_list): Likewise.
937 (fltfn_suffixes): Likewise.
939 * internal-fn.def (DEF_INTERNAL_FLT_FLOATN_FN): New helper macro
940 for math functions that have _Float<N> and _Float<N>X variants.
941 (SQRT): Add support for sqrt, copysign, fmin and fmax _Float<N>
942 and _Float<N>X variants.
943 (COPYSIGN): Likewise.
946 * fold-const.c (tree_call_nonnegative_warnv_p): Add support for
947 copysign, fma, fmax, fmin, and sqrt _Float<N> and _Float<N>X
949 (integer_valued_read_call_p): Likewise.
950 * fold-const-call.c (fold_const_call_ss): Likewise.
951 (fold_const_call_sss): Add support for copysign, fmin, and fmax
952 _Float<N> and _Float<N>X variants.
953 (fold_const_call_ssss): Add support for fma _Float<N> and
955 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Add
956 support for copysign and fma _Float<N> and _Float<N>X variants.
957 (backprop::process_builtin_call_use): Likewise.
958 * tree-call-cdce.c (can_test_argument_range); Add support for
959 sqrt _Float<N> and _Float<N>X variants.
960 (edom_only_function): Likewise.
961 (get_no_error_domain): Likewise.
962 * tree-ssa-math-opts.c (internal_fn_reciprocal): Likewise.
963 * tree-ssa-reassoc.c (attempt_builtin_copysign): Add support for
964 copysign _Float<N> and _Float<N>X variants.
965 * config/rs6000/rs6000-builtin.def (SQRTF128): Delete, this is now
966 handled by machine independent code.
968 * doc/cpp.texi (Common Predefined Macros): Document defining
969 __FP_FAST_FMAF<N> and __FP_FAST_FMAF<N>X if the backend supports
970 fma _Float<N> and _Float<N>X variants.
972 2017-10-27 Uros Bizjak <ubizjak@gmail.com>
975 * config/i386/i386-modes.def (CCFPU): Remove definition.
976 * config/i386/i386.c (put_condition_mode): Remove CCFPU mode handling.
977 (ix86_cc_modes_compatible): Ditto.
978 (ix86_expand_carry_flag_compare): Ditto.
979 (ix86_expand_int_movcc): Ditto.
980 (ix86_expand_int_addcc): Ditto.
981 (ix86_reverse_condition): Ditto.
982 (ix86_unordered_fp_compare): Rename from ix86_fp_compare_mode.
983 Return true/false for unordered/ordered fp comparisons.
984 (ix86_cc_mode): Always return CCFPmode for float mode comparisons.
985 (ix86_prepare_fp_compare_args): Update for rename.
986 (ix86_expand_fp_compare): Update for rename. Generate unordered
987 compare RTXes wrapped with UNSPEC_NOTRAP unspec.
988 (ix86_expand_sse_compare_and_jump): Ditto.
989 * config/i386/predicates.md (fcmov_comparison_operator):
990 Remove CCFPU mode handling.
991 (ix86_comparison_operator): Ditto.
992 (ix86_carry_flag_operator): Ditto.
993 * config/i386/i386.md (UNSPEC_NOTRAP): New unspec.
994 (*cmpu<mode>_i387): Wrap compare RTX with UNSPEC_NOTRAP unspec.
995 (*cmpu<mode>_cc_i387): Ditto.
996 (FPCMP): Remove mode iterator.
997 (unord): Remove mode attribute.
998 (unord_subst): New define_subst transformation
999 (unord): New define_subst attribute.
1001 (*cmpi<unord><MODEF:mode>): Rewrite using unord_subst transformation.
1002 (*cmpi<unord>xf_i387): Ditto.
1003 * config/i386/sse.md (<sse>_<unord>comi<round_saeonly_name>): Merge
1004 from <sse>_comi<round_saeonly_name> and <sse>_ucomi<round_saeonly_name>
1005 using unord_subst transformation.
1006 * config/i386/subst.md (SUBST_A): Remove CCFP and CCFPU modes.
1007 (round_saeonly): Also handle CCFP mode.
1008 * reg-stack.c (subst_stack_regs_pat): Handle UNSPEC_NOTRAP unspec.
1009 Remove UNSPEC_SAHF unspec handling.
1011 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1013 * x86-tune.def (X86_TUNE_INTER_UNIT_MOVES_TO_VEC): Disable for Zen.
1015 2017-10-27 Jeff Law <law@redhat.com>
1017 * gimple-ssa-sprintf.c: Include domwalk.h.
1018 (class sprintf_dom_walker): New class, derived from dom_walker.
1019 (sprintf_dom_walker::before_dom_children): New function.
1020 (struct call_info): Moved into sprintf_dom_walker class
1021 (compute_formath_length, handle_gimple_call): Likewise.
1022 (sprintf_length::execute): Call the dominator walker rather
1023 than walking the statements.
1025 * tree-vrp.c (check_all_array_refs): Do not use wi->info to smuggle
1026 gimple statement locations.
1027 (check_array_bounds): Corresponding changes. Get the statement's
1028 location directly from wi->stmt.
1030 2017-10-27 Palmer Dabbelt <palmer@dabbelt.com>
1033 * doc/invoke.texi (RISC-V) <-mabi>: Correct and improve.
1035 2017-10-27 Jan Hubicka <hubicka@ucw.cz>
1037 * config/i386/x86-tune.def (X86_TUNE_PARTIAL_REG_DEPENDENCY,
1038 X86_TUNE_MOVX): Disable for Haswell and newer CPUs.
1040 2017-10-27 Jakub Jelinek <jakub@redhat.com>
1043 * config/i386/i386-protos.h (maybe_get_pool_constant): Removed.
1044 * config/i386/i386.c (maybe_get_pool_constant): Removed.
1045 (ix86_split_to_parts): Use avoid_constant_pool_reference instead of
1046 maybe_get_pool_constant.
1047 * config/i386/predicates.md (zero_extended_scalar_load_operand):
1050 2017-10-27 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1052 * doc/install.texi (Specific, i?86-*-solaris2.10): Simplify gas
1053 2.26 caveat. Update gas and gld versions.
1054 (Specific, *-*-solaris2*): Update binutils version. Remove caveat
1057 2017-10-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1059 * cgraph.h (set_malloc_flag): Declare.
1060 * cgraph.c (set_malloc_flag_1): New function.
1061 (set_malloc_flag): Likewise.
1062 * ipa-fnsummary.h (ipa_call_summary): Add new field is_return_callee.
1063 * ipa-fnsummary.c (ipa_call_summary::reset): Set is_return_callee to
1065 (read_ipa_call_summary): Add support for reading is_return_callee.
1066 (write_ipa_call_summary): Stream is_return_callee.
1067 * ipa-inline.c (ipa_inline): Remove call to ipa_free_fn_summary.
1068 * ipa-pure-const.c: Add headers ssa.h, alloc-pool.h, symbol-summary.h,
1069 ipa-prop.h, ipa-fnsummary.h.
1070 (pure_const_names): Change to static.
1071 (malloc_state_e): Define.
1072 (malloc_state_names): Define.
1073 (funct_state_d): Add field malloc_state.
1074 (varying_state): Set malloc_state to STATE_MALLOC_BOTTOM.
1075 (check_retval_uses): New function.
1076 (malloc_candidate_p): Likewise.
1077 (analyze_function): Add support for malloc attribute.
1078 (pure_const_write_summary): Stream malloc_state.
1079 (pure_const_read_summary): Add support for reading malloc_state.
1080 (dump_malloc_lattice): New function.
1081 (propagate_malloc): New function.
1082 (warn_function_malloc): New function.
1083 (ipa_pure_const::execute): Call propagate_malloc and
1084 ipa_free_fn_summary.
1085 (pass_local_pure_const::execute): Add support for malloc attribute.
1086 * ssa-iterators.h (RETURN_FROM_IMM_USE_STMT): New macro.
1087 * doc/invoke.texi: Document Wsuggest-attribute=malloc.
1089 2017-10-27 Martin Liska <mliska@suse.cz>
1091 PR gcov-profile/82457
1092 * doc/invoke.texi: Document that one needs a non-strict ISO mode
1093 for fork-like functions to be properly instrumented.
1095 2017-10-27 Richard Biener <rguenther@suse.de>
1098 * tree-eh.c (pass_lower_eh_dispatch::execute): Free dominator
1099 info when we redirected EH.
1101 2017-10-26 Michael Collison <michael.collison@arm.com>
1103 * config/aarch64/aarch64.md(<optab>_trunc><vf><GPI:mode>2):
1105 (<optab>_trunchf<GPI:mode>2: New pattern.
1106 (<optab>_trunc<vgp><GPI:mode>2: New pattern.
1107 * config/aarch64/iterators.md (wv): New mode attribute.
1108 (vf, VF): New mode attributes.
1109 (vgp, VGP): New mode attributes.
1110 (s): Update attribute with SImode and DImode prefixes.
1112 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1114 * config/nios2/constraints.md ("S"): Match r0rel_constant_p too.
1115 * config/nios2/nios2-protos.h (r0rel_constant_p): Declare.
1116 * config/nios2/nios2.c: (nios2_r0rel_sec_regex): New.
1117 (nios2_option_overide): Initialize it. Don't allow R0-relative
1118 addressing with PIC.
1119 (nios2_rtx_costs): Handle r0rel_constant_p like gprel_constant_p.
1120 (nios2_symbolic_constant_p): Likewise.
1121 (nios2_legitimate_address_p): Likewise.
1122 (nios2_r0rel_section_name_p): New.
1123 (nios2_symbol_ref_in_r0rel_data_p): New.
1124 (nios2_emit_move_sequence): Handle r0rel_constant_p.
1125 (r0rel_constant_p): New.
1126 (nios2_print_operand_address): Handle r0rel_constant_p.
1127 (nios2_cdx_narrow_form_p): Likewise.
1128 * config/nios2/nios2.opt (mr0rel-sec=): New option.
1129 * doc/invoke.texi (Option Summary): Add -mr0rel-sec.
1130 (Nios II Options): Document -mr0rel-sec.
1132 2017-10-26 Sandra Loosemore <sandra@codesourcery.com>
1134 * config/nios2/nios2.c: Include xregex.h.
1135 (nios2_gprel_sec_regex): New.
1136 (nios2_option_overide): Initialize it. Don't allow GP-relative
1137 addressing with PIC.
1138 (nios2_small_section_name_p): Check for regex match.
1139 * config/nios2/nios2.opt (mgprel-sec=): New option.
1140 * doc/invoke.texi (Option Summary): Add -mgprel-sec.
1141 (Nios II Options): Document -mgprel-sec.
1143 2017-10-26 Jim Wilson <wilson@tuliptree.org>
1145 * doc/invoke.texi (-fdebug-prefix-map): Expand documentation.
1147 2017-10-26 Tom de Vries <tom@codesourcery.com>
1149 PR tree-optimization/82707
1150 * gimple.c (gimple_copy): Fix unsharing of
1151 GIMPLE_OMP_{SINGLE,TARGET,TEAMS}.
1153 2017-10-26 Olga Makhotina <olga.makhotina@intel.com>
1155 * config/i386/avx512fintrin.h (_mm512_cmpeq_pd_mask,
1156 _mm512_cmple_pd_mask, _mm512_cmplt_pd_mask,
1157 _mm512_cmpneq_pd_mask, _mm512_cmpnle_pd_mask,
1158 _mm512_cmpnlt_pd_mask, _mm512_cmpord_pd_mask,
1159 _mm512_cmpunord_pd_mask, _mm512_mask_cmpeq_pd_mask,
1160 _mm512_mask_cmple_pd_mask, _mm512_mask_cmplt_pd_mask,
1161 _mm512_mask_cmpneq_pd_mask, _mm512_mask_cmpnle_pd_mask,
1162 _mm512_mask_cmpnlt_pd_mask, _mm512_mask_cmpord_pd_mask,
1163 _mm512_mask_cmpunord_pd_mask, _mm512_cmpeq_ps_mask,
1164 _mm512_cmple_ps_mask, _mm512_cmplt_ps_mask,
1165 _mm512_cmpneq_ps_mask, _mm512_cmpnle_ps_mask,
1166 _mm512_cmpnlt_ps_mask, _mm512_cmpord_ps_mask,
1167 _mm512_cmpunord_ps_mask, _mm512_mask_cmpeq_ps_mask,
1168 _mm512_mask_cmple_ps_mask, _mm512_mask_cmplt_ps_mask,
1169 _mm512_mask_cmpneq_ps_mask, _mm512_mask_cmpnle_ps_mask,
1170 _mm512_mask_cmpnlt_ps_mask, _mm512_mask_cmpord_ps_mask,
1171 _mm512_mask_cmpunord_ps_mask): New intrinsics.
1173 2017-10-26 Michael Meissner <meissner@linux.vnet.ibm.com>
1175 * config/rs6000/aix.h (TARGET_IEEEQUAD_DEFAULT): Set long double
1177 * config/rs6000/darwin.h (TARGET_IEEEQUAD_DEFAULT): Likewise.
1178 * config/rs6000/rs6000.opt (-mabi=ieeelongdouble): Move the
1179 warning to rs6000.c. Remove the Undocumented flag, since it has
1181 (-mabi=ibmlongdouble): Likewise.
1182 * config/rs6000/rs6000.c (TARGET_IEEEQUAD_DEFAULT): If it is not
1183 already set, set the default format for long double.
1184 (rs6000_debug_reg_global): Print whether long double is IBM or
1186 (rs6000_option_override_internal): Rework setting long double
1187 format. Only warn if the user is changing the long double default
1188 and they did not use -Wno-psabi.
1189 * doc/invoke.texi (PowerPC options): Update the documentation for
1190 -mabi=ieeelongdouble and -mabi=ibmlongdouble.
1192 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1193 Alan Hayward <alan.hayward@arm.com>
1194 David Sherwood <david.sherwood@arm.com>
1196 * rtl.h (wider_subreg_mode): New function.
1197 * ira.h (ira_sort_regnos_for_alter_reg): Take a machine_mode *
1198 rather than an unsigned int *.
1199 * ira-color.c (regno_max_ref_width): Replace with...
1200 (regno_max_ref_mode): ...this new variable.
1201 (coalesced_pseudo_reg_slot_compare): Update accordingly.
1202 Use wider_subreg_mode.
1203 (ira_sort_regnos_for_alter_reg): Likewise. Take a machine_mode *
1204 rather than an unsigned int *.
1205 * lra-constraints.c (uses_hard_regs_p): Use wider_subreg_mode.
1206 (process_alt_operands): Likewise.
1207 (invariant_p): Likewise.
1208 * lra-spills.c (assign_mem_slot): Likewise.
1209 (add_pseudo_to_slot): Likewise.
1210 * lra.c (collect_non_operand_hard_regs): Likewise.
1211 (add_regs_to_insn_regno_info): Likewise.
1212 * reload1.c (regno_max_ref_width): Replace with...
1213 (regno_max_ref_mode): ...this new variable.
1214 (reload): Update accordingly. Update call to
1215 ira_sort_regnos_for_alter_reg.
1216 (alter_reg): Update to use regno_max_ref_mode. Call wider_subreg_mode.
1217 (init_eliminable_invariants): Update to use regno_max_ref_mode.
1218 (scan_paradoxical_subregs): Likewise.
1220 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1222 * config/aarch64/aarch64.h (EXIT_IGNORE_STACK): Set if alloca is used.
1223 (aarch64_frame): Add emit_frame_chain boolean.
1224 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1225 Move eh_return case to aarch64_layout_frame.
1226 (aarch64_layout_frame): Initialize emit_frame_chain.
1227 (aarch64_expand_prologue): Use emit_frame_chain.
1229 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1231 * config/aarch64/aarch64.c (aarch64_layout_frame):
1232 Ensure LR is always stored at the bottom of the callee-saves.
1233 Remove rarely used frame layout which saves callee-saves at top of
1234 frame, so the store of LR can be used as a valid probe in all cases.
1236 2017-10-26 Wilco Dijkstra <wdijkstr@arm.com>
1238 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
1239 Improve unaligned TImode/TFmode base/offset split.
1241 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1242 Alan Hayward <alan.hayward@arm.com>
1243 David Sherwood <david.sherwood@arm.com>
1245 * caller-save.c (mark_referenced_regs): Use read_modify_subreg_p.
1246 * combine.c (find_single_use_1): Likewise.
1247 (expand_field_assignment): Likewise.
1248 (move_deaths): Likewise.
1249 * lra-constraints.c (simplify_operand_subreg): Likewise.
1250 (curr_insn_transform): Likewise.
1251 * lra.c (collect_non_operand_hard_regs): Likewise.
1252 (add_regs_to_insn_regno_info): Likewise.
1253 * rtlanal.c (reg_referenced_p): Likewise.
1254 (covers_regno_no_parallel_p): Likewise.
1256 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1258 * wide-int-print.cc (print_hex): Loop based on extract_uhwi.
1259 Don't print any bits outside the precision of the value.
1260 * wide-int.cc (test_printing): Add some new tests.
1262 2017-10-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
1264 * configure.ac (gcc_cv_as_ix86_xbrace_comment): Check if assembler
1265 supports -xbrace_comment option.
1266 * configure: Regenerate.
1267 * config.in: Regenerate.
1268 * config/i386/sol2.h (ASM_XBRACE_COMMENT_SPEC): Define.
1269 (ASM_CPU_SPEC): Use it.
1271 2017-10-26 Richard Sandiford <richard.sandiford@linaro.org>
1273 * target.def (static_rtx_alignment): New hook.
1274 * targhooks.h (default_static_rtx_alignment): Declare.
1275 * targhooks.c (default_static_rtx_alignment): New function.
1276 * doc/tm.texi.in (TARGET_STATIC_RTX_ALIGNMENT): New hook.
1277 * doc/tm.texi: Regenerate.
1278 * varasm.c (force_const_mem): Use targetm.static_rtx_alignment
1279 instead of targetm.constant_alignment. Remove call to
1281 * config/cris/cris.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1282 (cris_preferred_mininum_alignment): New function, split out from...
1283 (cris_constant_alignment): ...here.
1284 (cris_static_rtx_alignment): New function.
1285 * config/i386/i386.c (ix86_static_rtx_alignment): New function,
1287 (ix86_constant_alignment): ...here.
1288 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1289 * config/mmix/mmix.c (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1290 (mmix_static_rtx_alignment): New function.
1291 * config/spu/spu.c (spu_static_rtx_alignment): New function.
1292 (TARGET_STATIC_RTX_ALIGNMENT): Redefine.
1294 2017-10-26 Tamar Christina <tamar.christina@arm.com>
1297 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2):
1298 Add flag_trapping_math and flag_fp_int_builtin_inexact.
1300 2017-10-25 Palmer Dabbelt <palmer@dabbelt.com>
1302 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Define.
1303 * config/riscv/pic.md (local_pic_load): Rename to local_pic_load_s,
1304 mark as a sign-extending load.
1305 (local_pic_load_u): Define.
1307 2017-10-25 Eric Botcazou <ebotcazou@adacore.com>
1310 * fold-const.c (operand_equal_for_comparison_p): Also return true
1311 if ARG0 is a simple variant of ARG1 with narrower precision.
1312 (fold_ternary_loc): Always pass unstripped operands to the predicate.
1314 2017-10-25 Jan Hubicka <hubicka@ucw.cz>
1316 * i386.c (ix86_builtin_vectorization_cost): Compute scatter/gather
1318 * i386.h (processor_costs): Add gather_static, gather_per_elt,
1319 scatter_static, scatter_per_elt.
1320 * x86-tune-costs.h: Add new cost entries.
1322 2017-10-25 Richard Biener <rguenther@suse.de>
1324 * tree-ssa-sccvn.h (vn_eliminate): Declare.
1325 * tree-ssa-pre.c (class eliminate_dom_walker, eliminate,
1326 class pass_fre): Move to ...
1327 * tree-ssa-sccvn.c (class eliminate_dom_walker, vn_eliminate,
1328 class pass_fre): ... here and adjust for statistics.
1330 2017-10-25 Jakub Jelinek <jakub@redhat.com>
1333 * attribs.c (attribute_value_equal): Use omp_declare_simd_clauses_equal
1334 for comparison of OMP_CLAUSEs regardless of flag_openmp{,_simd}.
1335 (duplicate_one_attribute, copy_attributes_to_builtin): New functions.
1336 * attribs.h (duplicate_one_attribute, copy_attributes_to_builtin): New
1339 2017-10-25 Richard Biener <rguenther@suse.de>
1341 * tree-ssa-pre.c (need_eh_cleanup, need_ab_cleanup, el_to_remove,
1342 el_to_fixup, el_todo, el_avail, el_avail_stack, eliminate_avail,
1343 eliminate_push_avail, eliminate_insert): Move inside...
1344 (class eliminate_dom_walker): ... this class in preparation
1346 (fini_eliminate): Remove by merging with ...
1347 (eliminate): ... this function. Adjust for class changes.
1348 (pass_pre::execute): Remove fini_eliminate call.
1349 (pass_fre::execute): Likewise.
1351 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1354 * config/i386/sse.md (UNSPEC_VPERMI2, UNSPEC_VPERMI2_MASK): Remove.
1355 (VPERMI2, VPERMI2I): New mode iterators.
1356 (<avx512>_vpermi2var<mode>3_maskz): Remove 3 define_expand patterns.
1357 (<avx512>_vpermi2var<mode>3<sd_maskz_name>): Remove 3 define_insn
1359 (<avx512>_vpermi2var<mode>3_mask): New define_expand using VPERMI2
1360 mode iterator. Remove 3 old define_insn patterns.
1361 (*<avx512>_vpermi2var<mode>3_mask): 2 new define_insn patterns.
1362 (<avx512>_vpermt2var<mode>3_maskz): Adjust 1 define_expand to use
1363 VPERMI2 mode iterator, remove the other two expanders.
1364 (<avx512>_vpermt2var<mode>3<sd_maskz_name>): Adjust 1 define_insn
1365 to use VPERMI2 mode iterator, add another alternative for vpermi2*
1366 instructions, remove the other two patterns.
1367 (<avx512>_vpermt2var<mode>3_mask): Adjust 1 define_insn to use VPERMI2
1368 mode iterator, remove the other two patterns.
1369 * config/i386/i386.c (ix86_expand_vec_perm_vpermi2): Renamed to ...
1370 (ix86_expand_vec_perm_vpermt2): ... this. Swap mask and op0
1371 arguments, use gen_*vpermt2* expanders instead of gen_*vpermi2*
1372 and adjust argument order accordingly.
1373 (ix86_expand_vec_perm): Adjust caller.
1374 (expand_vec_perm_1): Likewise.
1375 (expand_vec_perm_vpermi2_vpshub2): Rename to ...
1376 (expand_vec_perm_vpermt2_vpshub2): ... this.
1377 (ix86_expand_vec_perm_const_1): Adjust caller.
1378 (ix86_vectorize_vec_perm_const_ok): Adjust comments.
1381 * config/i386/sse.md (VIMAX_AVX2): Remove V4TImode.
1382 (VIMAX_AVX2_AVX512BW, VIMAX_AVX512VL): New mode iterators.
1383 (vec_shl_<mode>): Remove unused expander.
1384 (avx512bw_<shift_insn><mode>3): New define_insn.
1385 (<sse2_avx2>_ashl<mode>3, <sse2_avx2>_lshr<mode>3): Replaced by ...
1386 (<sse2_avx2>_<shift_insn><mode>3): ... this. New define_insn.
1388 2017-10-24 Paolo Carlini <paolo.carlini@oracle.com>
1391 * doc/invoke.texi ([Wbuiltin-declaration-mismatch]): Extend
1394 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1396 PR rtl-optimization/82396
1397 * gcc/haifa-sched.c (ready_sort_real): Remove qsort workaround.
1398 (autopref_multipass_init): Simplify initialization.
1399 (autopref_rank_data): Simplify sort order.
1400 * gcc/sched-int.h (autopref_multipass_data_): Remove
1401 multi_mem_insn_p, min_offset and max_offset.
1403 2017-10-24 Wilco Dijkstra <wdijkstr@arm.com>
1406 * config/aarch64/aarch64.c (aarch64_frame_pointer_required)
1407 Check special value of flag_omit_frame_pointer.
1408 (aarch64_can_eliminate): Likewise.
1409 (aarch64_override_options_after_change_1): Simplify handling of
1410 -fomit-frame-pointer and -fomit-leaf-frame-pointer.
1412 2017-10-24 Richard Biener <rguenther@suse.de>
1414 PR tree-optimization/82697
1415 * tree-ssa-phiopt.c (cond_store_replacement): Use alias-set
1416 zero for conditional load and unconditional store.
1418 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1420 * doc/install.texi: Document bootstrap-cet.
1422 2017-10-24 H.J. Lu <hongjiu.lu@intel.com>
1425 * config/i386/i386.c (rest_of_insert_endbranch): Don't insert
1426 ENDBR instruction at function entrance if function is only
1429 2017-10-24 Jakub Jelinek <jakub@redhat.com>
1432 * config/i386/i386.md (addcarry<mode>, subborrow<mode>): Change
1433 patterns to better describe from which operation the CF is computed.
1434 (addcarry<mode>_0, subborrow<mode>_0): New patterns.
1435 * config/i386/i386.c (ix86_expand_builtin) <case handlecarry>: Pass
1436 one LTU with [DT]Imode and another one with [SD]Imode. If arg0
1437 is 0, use _0 suffixed expanders instead of emitting a comparison
1440 2017-10-06 Sergey Shalnov <Sergey.Shalnov@intel.com>
1442 * config/i386/i386.md(*movsf_internal, *movdf_internal):
1443 Avoid 512-bit AVX modes for TARGET_PREFER_AVX256.
1445 2017-10-24 Eric Botcazou <ebotcazou@adacore.com>
1448 * tree-outof-ssa.h (always_initialized_rtx_for_ssa_name_p): Delete.
1449 * expr.c (expand_expr_real_1) <expand_decl_rtl>: Revert latest change.
1450 * loop-iv.c (iv_get_reaching_def): Likewise.
1451 * cfgexpand.c (expand_one_ssa_partition): Initialize the RTX if the
1452 variable is promoted and the partition contains undefined values.
1454 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1456 * config/nios2/nios2.c (nios2_rtx_costs): Make costs better
1458 (nios2_address_cost): Define.
1459 (nios2_legitimize_address): Recognize (exp + constant) directly.
1460 (TARGET_ADDRESS_COST): Define.
1462 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1464 * config/nios2/nios2-protos.h (nios2_large_constant_p): Declare.
1465 (nios2_symbolic_memory_operand_p): Declare.
1466 (nios2_split_large_constant): Declare.
1467 (nios2_split_symbolic_memory_operand): Declare.
1468 * config/nios2/nios2.c: Adjust includes.
1469 (nios2_symbolic_constant_allowed): New.
1470 (nios2_symbolic_constant_p): New.
1471 (nios2_plus_symbolic_constant_p): New.
1472 (nios2_valid_addr_expr_p): Recognize addresses involving
1474 (nios2_legitimate_address_p): Likewise, also LO_SUM.
1475 (nios2_symbolic_memory_operand_p): New.
1476 (nios2_large_constant_p): New.
1477 (nios2_split_large_constant): New.
1478 (nios2_split_plus_large_constant): New.
1479 (nios2_split_symbolic_memory_operand): New.
1480 (nios2_legitimize_address): Code refactoring. Handle addresses
1481 involving symbolic constants.
1482 (nios2_emit_move_sequence): Likewise.
1483 (nios2_print_operand): Improve error output.
1484 (nios2_print_operand_address): Handle LO_SUM.
1485 (nios2_cdx_narrow_form_p): Likewise.
1486 * config/nios2/nios2.md (movqi_internal): Add splitter for memory
1487 operands involving symbolic constants.
1488 (movhi_internal, movsi_internal): Likewise.
1489 (zero_extendhisi2, zero_extendqi<mode>2): Likewise.
1490 (extendhisi2, extendqi<mode>2): Likewise.
1492 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1494 * tree-pass.h (PROP_rtl_split_insns): Define.
1495 * recog.c (pass_data_split_all_insns): Provide PROP_rtl_split_insns.
1497 2017-10-23 Sandra Loosemore <sandra@codesourcery.com>
1499 * config/nios2/nios2.c (TARGET_LRA_P): Don't override.
1501 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1504 * target.def (const_not_ok_for_debug_p): Default to
1505 default_const_not_ok_for_debug_p instead of hook_bool_rtx_false.
1506 * targhooks.h (default_const_not_ok_for_debug_p): New declaration.
1507 * targhooks.c (default_const_not_ok_for_debug_p): New function.
1508 * dwarf2out.c (const_ok_for_output_1): Only reject UNSPECs for
1509 which targetm.const_not_ok_for_debug_p returned true.
1510 * config/arm/arm.c (arm_const_not_ok_for_debug_p): Return true
1512 * config/powerpcspe/powerpcspe.c (rs6000_const_not_ok_for_debug_p):
1514 * config/rs6000/rs6000.c (rs6000_const_not_ok_for_debug_p): Likewise.
1515 * config/i386/i386.c (ix86_delegitimize_address_1): Don't delegitimize
1516 UNSPEC_GOTOFF with addend into addend - _GLOBAL_OFFSET_TABLE_ + symbol
1518 (ix86_const_not_ok_for_debug_p): New function.
1519 (i386_asm_output_addr_const_extra): Handle UNSPEC_GOTOFF.
1520 (TARGET_CONST_NOT_OK_FOR_DEBUG_P): Redefine.
1522 2017-10-23 David Malcolm <dmalcolm@redhat.com>
1525 * system.h: Conditionally include "unique-ptr.h" if
1526 INCLUDE_UNIQUE_PTR is defined.
1527 * unique-ptr-tests.cc: Remove include of "unique-ptr.h" in favor
1528 of defining INCLUDE_UNIQUE_PTR before including "system.h".
1530 2017-10-23 Sebastian Perta <sebastian.perta@renesas.com>
1532 * config/rl78/rl78.md: New define_expand "subdi3".
1534 2017-10-23 H.J. Lu <hongjiu.lu@intel.com>
1537 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Skip
1538 DF_REF_INSN if DF_REF_INSN_INFO is false.
1540 2017-10-23 Jan Hubicka <hubicka@ucw.cz>
1542 * i386.c (dimode_scalar_chain::compute_convert_gain): Use
1543 xmm_move instead of sse_move.
1544 (sse_store_index): New function.
1545 (ix86_register_move_cost): Be more sensible about mismatch stall;
1546 model AVX moves correctly; make difference between sse->integer and
1548 (ix86_builtin_vectorization_cost): Model correctly aligned and unaligned
1549 moves; make difference between SSE and AVX.
1550 * i386.h (processor_costs): Remove sse_move; add xmm_move, ymm_move
1551 and zmm_move. Increase size of sse load and store tables;
1552 add unaligned load and store tables; add ssemmx_to_integer.
1553 * x86-tune-costs.h: Update all entries according to real
1554 move latencies from Agner Fog's manual and chip documentation.
1556 2017-10-23 Jakub Jelinek <jakub@redhat.com>
1559 * config/i386/predicates.md (x86_64_dwzext_immediate_operand): New.
1560 * config/i386/constraints.md (Wf): New constraint.
1561 * config/i386/i386.md (UNSPEC_SBB): New unspec.
1562 (cmp<dwi>_doubleword): Removed.
1563 (sub<mode>3_carry_ccc, *sub<mode>3_carry_ccc_1): New patterns.
1564 (sub<mode>3_carry_ccgz): Use unspec instead of compare.
1565 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>: Don't
1566 expand with cmp<dwi>_doubleword. For LTU and GEU use
1567 sub<mode>3_carry_ccc instead of sub<mode>3_carry_ccgz and use CCCmode.
1569 * common.opt (gcolumn-info): Enable by default.
1570 * doc/invoke.texi (gcolumn-info): Document new default.
1572 2017-10-23 Richard Biener <rguenther@suse.de>
1574 PR tree-optimization/82672
1575 * graphite-isl-ast-to-gimple.c (graphite_copy_stmts_from_block):
1576 Fold the stmt if we propagated into it.
1578 2017-10-23 Richard Biener <rguenther@suse.de>
1580 * tree-ssa-pre.c (bitmap_remove_from_set): Rename to...
1581 (bitmap_remove_expr_from_set): ... this. All callers call this
1582 for non-constant values.
1583 (bitmap_set_subtract): Rename to...
1584 (bitmap_set_subtract_expressions): ... this. Adjust and
1586 (bitmap_set_contains_value): Remove superfluous check.
1587 (bitmap_set_replace_value): Inline into single caller ...
1588 (bitmap_value_replace_in_set): ... here and simplify.
1589 (dependent_clean): Merge into ...
1590 (clean): ... this using an overload. Adjust.
1591 (prune_clobbered_mems): Adjust.
1592 (compute_antic_aux): Likewise.
1593 (compute_partial_antic_aux): Likewise.
1595 2017-10-23 Richard Biener <rguenther@suse.de>
1597 PR tree-optimization/82129
1599 2017-08-01 Richard Biener <rguenther@suse.de>
1601 PR tree-optimization/81181
1602 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
1603 (compute_antic): ... end of iteration here.
1605 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1607 * target.def (starting_frame_offset): New hook.
1608 * doc/tm.texi (STARTING_FRAME_OFFSET): Remove in favor of...
1609 (TARGET_STARTING_FRAME_OFFSET): ...this new hook.
1610 * doc/tm.texi.in: Regenerate.
1611 * hooks.h (hook_hwi_void_0): Declare.
1612 * hooks.c (hook_hwi_void_0): New function.
1613 * doc/rtl.texi: Refer to TARGET_STARTING_FRAME_OFFSET instead of
1614 STARTING_FRAME_OFFSET.
1615 * builtins.c (expand_builtin_setjmp_receiver): Likewise.
1616 * reload1.c (reload): Likewise.
1617 * cfgexpand.c (expand_used_vars): Use targetm.starting_frame_offset
1618 instead of STARTING_FRAME_OFFSET.
1619 * function.c (try_fit_stack_local): Likewise.
1620 (assign_stack_local_1): Likewise
1621 (instantiate_virtual_regs): Likewise.
1622 * rtlanal.c (rtx_addr_can_trap_p_1): Likewise.
1623 * config/avr/avr.md (nonlocal_goto_receiver): Likewise.
1624 * config/aarch64/aarch64.h (STARTING_FRAME_OFFSET): Delete.
1625 * config/alpha/alpha.h (STARTING_FRAME_OFFSET): Likewise.
1626 * config/arc/arc.h (STARTING_FRAME_OFFSET): Likewise.
1627 * config/arm/arm.h (STARTING_FRAME_OFFSET): Likewise.
1628 * config/bfin/bfin.h (STARTING_FRAME_OFFSET): Likewise.
1629 * config/c6x/c6x.h (STARTING_FRAME_OFFSET): Likewise.
1630 * config/cr16/cr16.h (STARTING_FRAME_OFFSET): Likewise.
1631 * config/cris/cris.h (STARTING_FRAME_OFFSET): Likewise.
1632 * config/fr30/fr30.h (STARTING_FRAME_OFFSET): Likewise.
1633 * config/frv/frv.h (STARTING_FRAME_OFFSET): Likewise.
1634 * config/ft32/ft32.h (STARTING_FRAME_OFFSET): Likewise.
1635 * config/h8300/h8300.h (STARTING_FRAME_OFFSET): Likewise.
1636 * config/i386/i386.h (STARTING_FRAME_OFFSET): Likewise.
1637 * config/ia64/ia64.h (STARTING_FRAME_OFFSET): Likewise.
1638 * config/m32c/m32c.h (STARTING_FRAME_OFFSET): Likewise.
1639 * config/m68k/m68k.h (STARTING_FRAME_OFFSET): Likewise.
1640 * config/mcore/mcore.h (STARTING_FRAME_OFFSET): Likewise.
1641 * config/mn10300/mn10300.h (STARTING_FRAME_OFFSET): Likewise.
1642 * config/moxie/moxie.h (STARTING_FRAME_OFFSET): Likewise.
1643 * config/msp430/msp430.h (STARTING_FRAME_OFFSET): Likewise.
1644 * config/nds32/nds32.h (STARTING_FRAME_OFFSET): Likewise.
1645 * config/nios2/nios2.h (STARTING_FRAME_OFFSET): Likewise.
1646 * config/nvptx/nvptx.h (STARTING_FRAME_OFFSET): Likewise.
1647 * config/pdp11/pdp11.h (STARTING_FRAME_OFFSET): Likewise.
1648 * config/riscv/riscv.h (STARTING_FRAME_OFFSET): Likewise.
1649 * config/rl78/rl78.h (STARTING_FRAME_OFFSET): Likewise.
1650 * config/rx/rx.h (STARTING_FRAME_OFFSET): Likewise.
1651 * config/s390/s390.h (STARTING_FRAME_OFFSET): Likewise.
1652 * config/sh/sh.h (STARTING_FRAME_OFFSET): Likewise.
1653 * config/sparc/sparc.c (sparc_compute_frame_size): Likewise.
1654 * config/sparc/sparc.h (STARTING_FRAME_OFFSET): Likewise.
1655 * config/spu/spu.h (STARTING_FRAME_OFFSET): Likewise.
1656 * config/stormy16/stormy16.h (STARTING_FRAME_OFFSET): Likewise.
1657 * config/tilegx/tilegx.h (STARTING_FRAME_OFFSET): Likewise.
1658 * config/tilepro/tilepro.h (STARTING_FRAME_OFFSET): Likewise.
1659 * config/v850/v850.h (STARTING_FRAME_OFFSET): Likewise.
1660 * config/visium/visium.h (STARTING_FRAME_OFFSET): Likewise.
1661 * config/avr/avr.h (STARTING_FRAME_OFFSET): Likewise.
1662 * config/avr/avr-protos.h (avr_starting_frame_offset): Likewise.
1663 * config/avr/avr.c (avr_starting_frame_offset): Make static and
1664 return a HOST_WIDE_INT.
1665 (avr_builtin_setjmp_frame_value): Use it instead of
1666 STARTING_FRAME_OFFSET.
1667 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1668 * config/epiphany/epiphany.h (STARTING_FRAME_OFFSET): Delete.
1669 * config/epiphany/epiphany.c (epiphany_starting_frame_offset):
1671 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1672 * config/iq2000/iq2000.h (STARTING_FRAME_OFFSET): Delete.
1673 * config/iq2000/iq2000.c (iq2000_starting_frame_offset): New function.
1674 (TARGET_CONSTANT_ALIGNMENT): Redefine.
1675 * config/lm32/lm32.h (STARTING_FRAME_OFFSET): Delete.
1676 * config/lm32/lm32.c (lm32_starting_frame_offset): New function.
1677 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1678 * config/m32r/m32r.h (STARTING_FRAME_OFFSET): Delete.
1679 * config/m32r/m32r.c (m32r_starting_frame_offset): New function.
1680 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1681 * config/microblaze/microblaze.h (STARTING_FRAME_OFFSET): Delete.
1682 * config/microblaze/microblaze.c (microblaze_starting_frame_offset):
1684 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1685 * config/mips/mips.h (STARTING_FRAME_OFFSET): Delete.
1686 * config/mips/mips.c (mips_compute_frame_info): Refer to
1687 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
1688 (mips_starting_frame_offset): New function.
1689 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1690 * config/mmix/mmix.h (STARTING_FRAME_OFFSET): Delete.
1691 * config/mmix/mmix-protos.h (mmix_starting_frame_offset): Delete.
1692 * config/mmix/mmix.c (mmix_starting_frame_offset): Make static
1693 and return a HOST_WIDE_INT.
1694 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1695 (mmix_initial_elimination_offset): Refer to
1696 TARGET_STARTING_FRAME_OFFSET instead of STARTING_FRAME_OFFSET.
1697 * config/pa/pa.h (STARTING_FRAME_OFFSET): Delete.
1698 * config/pa/pa.c (pa_starting_frame_offset): New function.
1699 (pa_compute_frame_size): Use it instead of STARTING_FRAME_OFFSET.
1700 (pa_expand_prologue): Likewise.
1701 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1702 * config/powerpcspe/aix.h (STARTING_FRAME_OFFSET): Split out
1703 !FRAME_GROWS_DOWNWARD handling to...
1704 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1705 * config/powerpcspe/darwin.h (STARTING_FRAME_OFFSET): Split out
1706 !FRAME_GROWS_DOWNWARD handling to...
1707 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1708 * config/powerpcspe/powerpcspe.h (STARTING_FRAME_OFFSET): Split out
1709 !FRAME_GROWS_DOWNWARD handling to...
1710 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1711 * config/powerpcspe/powerpcspe.c (TARGET_STARTING_FRAME_OFFSET):
1713 (rs6000_starting_frame_offset): New function.
1714 * config/rs6000/aix.h (STARTING_FRAME_OFFSET): Split out
1715 !FRAME_GROWS_DOWNWARD handling to...
1716 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1717 * config/rs6000/darwin.h (STARTING_FRAME_OFFSET): Split out
1718 !FRAME_GROWS_DOWNWARD handling to...
1719 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1720 * config/rs6000/rs6000.h (STARTING_FRAME_OFFSET): Split out
1721 !FRAME_GROWS_DOWNWARD handling to...
1722 (RS6000_STARTING_FRAME_OFFSET): ...this new macro.
1723 * config/rs6000/rs6000.c (TARGET_STARTING_FRAME_OFFSET): Refine.
1724 (rs6000_starting_frame_offset): New function.
1725 * config/vax/elf.h (STARTING_FRAME_OFFSET): Delete.
1726 * config/vax/vax.h (STARTING_FRAME_OFFSET): Delete.
1727 * config/vax/vax.c (vax_starting_frame_offset): New function.
1728 (vax_expand_prologue): Use it instead of STARTING_FRAME_OFFSET.
1729 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1730 * config/xtensa/xtensa.h (STARTING_FRAME_OFFSET): Delete.
1731 * config/xtensa/xtensa.c (xtensa_starting_frame_offset): New function.
1732 (TARGET_STARTING_FRAME_OFFSET): Redefine.
1733 * system.h (STARTING_FRAME_OFFSET): Poison.
1735 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1737 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
1738 SCALAR_TYPE_MODE instead of TYPE_MODE.
1740 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1741 Alan Hayward <alan.hayward@arm.com>
1742 David Sherwood <david.sherwood@arm.com>
1744 * dwarf2out.c (loc_list_from_tree_1): Use SCALAR_INT_TYPE_MODE
1746 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1747 Alan Hayward <alan.hayward@arm.com>
1748 David Sherwood <david.sherwood@arm.com>
1750 * expmed.c (expand_shift_1): Use scalar_mode for scalar_mode.
1752 2017-10-23 Richard Biener <rguenther@suse.de>
1754 PR tree-optimization/82129
1755 * tree-ssa-pre.c (bitmap_set_and): Remove.
1756 (compute_antic_aux): Compute ANTIC_OUT intersection in a way
1757 canonicalizing expressions in the set to those with lowest
1758 ID rather than taking that from the first edge.
1760 2017-10-23 Richard Sandiford <richard.sandiford@linaro.org>
1762 * combine.c (rtx_equal_for_field_assignment_p): Use
1763 byte_lowpart_offset.
1765 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1766 Alan Hayward <alan.hayward@arm.com>
1767 David Sherwood <david.sherwood@arm.com>
1769 * internal-fn.c (expand_direct_optab_fn): Don't assign directly
1770 to a SUBREG_PROMOTED_VAR.
1772 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1773 Alan Hayward <alan.hayward@arm.com>
1774 David Sherwood <david.sherwood@arm.com>
1776 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_PRECISION.
1777 (expand_debug_source_expr): Likewise.
1778 * combine.c (combine_simplify_rtx): Likewise.
1779 * cse.c (fold_rtx): Likewise.
1780 * optabs.c (expand_float): Likewise.
1781 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
1782 (simplify_binary_operation_1): Likewise.
1784 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1785 Alan Hayward <alan.hayward@arm.com>
1786 David Sherwood <david.sherwood@arm.com>
1788 * combine.c (simplify_comparison): Use HWI_COMPUTABLE_MODE_P.
1789 (record_promoted_value): Likewise.
1790 * expr.c (expand_expr_real_2): Likewise.
1791 * ree.c (update_reg_equal_equiv_notes): Likewise.
1792 (combine_set_extension): Likewise.
1793 * rtlanal.c (low_bitmask_len): Likewise.
1794 * simplify-rtx.c (neg_const_int): Likewise.
1795 (simplify_binary_operation_1): Likewise.
1797 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1798 Alan Hayward <alan.hayward@arm.com>
1799 David Sherwood <david.sherwood@arm.com>
1801 * lra-spills.c (assign_mem_slot): Use subreg_size_lowpart_offset.
1802 * regcprop.c (maybe_mode_change): Likewise.
1803 * reload1.c (alter_reg): Likewise.
1805 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1807 * inchash.h (inchash::hash::add_wide_int): New function.
1808 * lto-streamer-out.c (hash_tree): Use it.
1810 2017-10-22 Richard Sandiford <richard.sandiford@linaro.org>
1812 * inchash.h (inchash::hash::add_wide_int): Rename to...
1813 (inchash::hash::add_hwi): ...this.
1814 * ipa-devirt.c (hash_odr_vtable): Update accordingly.
1815 (polymorphic_call_target_hasher::hash): Likewise.
1816 * ipa-icf.c (sem_function::get_hash, sem_function::init): Likewise.
1817 (sem_item::add_expr, sem_item::add_type, sem_variable::get_hash)
1818 (sem_item_optimizer::update_hash_by_addr_refs): Likewise.
1819 * lto-streamer-out.c (hash_tree): Likewise.
1820 * optc-save-gen.awk: Likewise.
1821 * tree.c (add_expr): Likewise.
1823 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1826 * config/i386/i386.c (ix86_fp_compare_mode): Return CCFPmode
1827 for ordered inequality comparisons even with TARGET_IEEE_FP.
1829 2017-10-22 Uros Bizjak <ubizjak@gmail.com>
1832 * config/i386/i386.md (cmp<dwi>_doubleword): New pattern.
1833 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
1834 Expand with cmp<dwi>_doubleword.
1836 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1838 * extend.texi: Add x86 specific to 'nocf_check' attribute.
1839 List CET intrinsics.
1840 * invoke.texi: Add -mcet, -mibt, -mshstk options. Add x86
1841 specific to -fcf-protection option.
1843 2017-10-21 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1845 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET): New.
1846 (OPTION_MASK_ISA_SHSTK_SET): Likewise.
1847 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
1848 (OPTION_MASK_ISA_SHSTK_UNSET): Likewise.
1849 (ix86_handle_option): Add -mibt, -mshstk, -mcet handling.
1850 * config.gcc (extra_headers): Add cetintrin.h for x86 targets.
1851 (extra_objs): Add cet.o for Linux/x86 targets.
1852 (tmake_file): Add i386/t-cet for Linux/x86 targets.
1853 * config/i386/cet.c: New file.
1854 * config/i386/cetintrin.h: Likewise.
1855 * config/i386/t-cet: Likewise.
1856 * config/i386/cpuid.h (bit_SHSTK): New.
1857 (bit_IBT): Likewise.
1858 * config/i386/driver-i386.c (host_detect_local_cpu): Detect and
1859 pass IBT and SHSTK bits.
1860 * config/i386/i386-builtin-types.def
1861 (VOID_FTYPE_UNSIGNED_PVOID): New.
1862 (VOID_FTYPE_UINT64_PVOID): Likewise.
1863 * config/i386/i386-builtin.def: Add CET intrinsics.
1864 * config/i386/i386-c.c (ix86_target_macros_internal): Add
1865 OPTION_MASK_ISA_IBT, OPTION_MASK_ISA_SHSTK handling.
1866 * config/i386/i386-passes.def: Add pass_insert_endbranch pass.
1867 * config/i386/i386-protos.h (make_pass_insert_endbranch): New
1869 * config/i386/i386.c (rest_of_insert_endbranch): New.
1870 (pass_data_insert_endbranch): Likewise.
1871 (pass_insert_endbranch): Likewise.
1872 (make_pass_insert_endbranch): Likewise.
1873 (ix86_notrack_prefixed_insn_p): Likewise.
1874 (ix86_target_string): Add -mibt, -mshstk flags.
1875 (ix86_option_override_internal): Add flag_cf_protection
1877 (ix86_valid_target_attribute_inner_p): Set OPT_mibt, OPT_mshstk.
1878 (ix86_print_operand): Add 'notrack' prefix output.
1879 (ix86_init_mmx_sse_builtins): Add CET intrinsics.
1880 (ix86_expand_builtin): Expand CET intrinsics.
1881 (x86_output_mi_thunk): Add 'endbranch' instruction.
1882 * config/i386/i386.h (TARGET_IBT): New.
1883 (TARGET_IBT_P): Likewise.
1884 (TARGET_SHSTK): Likewise.
1885 (TARGET_SHSTK_P): Likewise.
1886 * config/i386/i386.md (unspecv): Add UNSPECV_NOP_RDSSP,
1887 UNSPECV_INCSSP, UNSPECV_SAVEPREVSSP, UNSPECV_RSTORSSP,
1888 UNSPECV_WRSS, UNSPECV_WRUSS, UNSPECV_SETSSBSY, UNSPECV_CLRSSBSY.
1889 (builtin_setjmp_setup): New pattern.
1890 (builtin_longjmp): Likewise.
1891 (rdssp<mode>): Likewise.
1892 (incssp<mode>): Likewise.
1893 (saveprevssp): Likewise.
1894 (rstorssp): Likewise.
1895 (wrss<mode>): Likewise.
1896 (wruss<mode>): Likewise.
1897 (setssbsy): Likewise.
1898 (clrssbsy): Likewise.
1899 (nop_endbr): Likewise.
1900 * config/i386/i386.opt: Add -mcet, -mibt, -mshstk and -mcet-switch
1902 * config/i386/immintrin.h: Include <cetintrin.h>.
1903 * config/i386/linux-common.h
1904 (file_end_indicate_exec_stack_and_cet): New prototype.
1905 (TARGET_ASM_FILE_END): New.
1907 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1909 * i386.c (ix86_builtin_vectorization_cost): Use existing rtx_cost
1910 latencies instead of having separate table; make difference between
1911 integer and float costs.
1912 * i386.h (processor_costs): Remove scalar_stmt_cost,
1913 scalar_load_cost, scalar_store_cost, vec_stmt_cost, vec_to_scalar_cost,
1914 scalar_to_vec_cost, vec_align_load_cost, vec_unalign_load_cost,
1916 * x86-tune-costs.h: Remove entries which has been removed in
1917 procesor_costs from all tables; make cond_taken_branch_cost
1918 and cond_not_taken_branch_cost COST_N_INSNS based.
1920 2017-10-20 Jan Hubicka <hubicka@ucw.cz>
1922 * x86-tune-costs.h (intel_cost, generic_cost): Fix move costs.
1924 2017-10-20 Jakub Jelinek <jakub@redhat.com>
1926 * config/i386/i386.md (isa): Remove fma_avx512f.
1927 * config/i386/sse.md (<avx512>_fmadd_<mode>_mask<round_name>,
1928 <avx512>_fmadd_<mode>_mask3<round_name>,
1929 <avx512>_fmsub_<mode>_mask<round_name>,
1930 <avx512>_fmsub_<mode>_mask3<round_name>,
1931 <avx512>_fnmadd_<mode>_mask<round_name>,
1932 <avx512>_fnmadd_<mode>_mask3<round_name>,
1933 <avx512>_fnmsub_<mode>_mask<round_name>,
1934 <avx512>_fnmsub_<mode>_mask3<round_name>,
1935 <avx512>_fmaddsub_<mode>_mask<round_name>,
1936 <avx512>_fmaddsub_<mode>_mask3<round_name>,
1937 <avx512>_fmsubadd_<mode>_mask<round_name>,
1938 <avx512>_fmsubadd_<mode>_mask3<round_name>): Remove isa attribute.
1939 (*vec_widen_umult_even_v16si<mask_name>,
1940 *vec_widen_smult_even_v16si<mask_name>): Likewise.
1941 (<mask_codefor>avx512bw_dbpsadbw<mode><mask_name>): Likewise.
1943 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1945 * extend.texi: Add 'nocf_check' documentation.
1946 * gimple.texi: Add second parameter to
1947 gimple_build_call_from_tree.
1948 * invoke.texi: Add -fcf-protection documentation.
1949 * rtl.texi: Add REG_CALL_NOTRACK documenation.
1951 2017-10-20 Richard Biener <rguenther@suse.de>
1953 PR tree-optimization/82473
1954 * tree-vect-loop.c (vectorizable_reduction): Properly get at
1955 the largest input type.
1957 2017-10-20 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
1959 * c-attribs.c (handle_nocf_check_attribute): New function.
1960 (c_common_attribute_table): Add 'nocf_check' handling.
1961 * gimple-parser.c: Add second argument NULL to
1962 gimple_build_call_from_tree.
1963 * attrib.c (comp_type_attributes): Check nocf_check attribute.
1964 * cfgexpand.c (expand_call_stmt): Set REG_CALL_NOCF_CHECK for
1966 * combine.c (distribute_notes): Add REG_CALL_NOCF_CHECK handling.
1967 * common.opt: Add fcf-protection flag.
1968 * emit-rtl.c (try_split): Add REG_CALL_NOCF_CHECK handling.
1969 * flag-types.h: Add enum cf_protection_level.
1970 * gimple.c (gimple_build_call_from_tree): Add second parameter.
1971 Add 'nocf_check' attribute propagation to gimple call.
1972 * gimple.h (gf_mask): Add GF_CALL_NOCF_CHECK.
1973 (gimple_build_call_from_tree): Update prototype.
1974 (gimple_call_nocf_check_p): New function.
1975 (gimple_call_set_nocf_check): Likewise.
1976 * gimplify.c: Add second argument to gimple_build_call_from_tree.
1977 * ipa-icf.c: Add nocf_check attribute in statement hash.
1978 * recog.c (peep2_attempt): Add REG_CALL_NOCF_CHECK handling.
1979 * reg-notes.def: Add REG_NOTE (CALL_NOCF_CHECK).
1980 * toplev.c (process_options): Add flag_cf_protection handling.
1982 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
1984 * x86-tune-costs.h (core_cost): Fix div, move and sqrt latencies.
1986 2017-10-20 Richard Biener <rguenther@suse.de>
1988 PR tree-optimization/82603
1989 * tree-if-conv.c (predicate_mem_writes): Make sure to only
1990 remove false predicated stores.
1992 2017-10-20 Richard Biener <rguenther@suse.de>
1994 * graphite-isl-ast-to-gimple.c
1995 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block):
1996 Remove return value and simplify, dump copied stmt after lhs
1998 (translate_isl_ast_to_gimple::translate_isl_ast_node_user):
1999 Reduce dump verbosity.
2000 (gsi_insert_earliest): Likewise.
2001 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Adjust.
2002 * graphite.c (print_global_statistics): Adjust dumping.
2003 (print_graphite_scop_statistics): Likewise.
2004 (print_graphite_statistics): Do not dump loops here.
2005 (graphite_transform_loops): But here.
2007 2017-10-20 Nicolas Roche <roche@adacore.com>
2009 * configure.ac (ACX_PROG_GNAT): Append "libgnat" to include search dir.
2010 * configure: Regenerate.
2012 2017-10-20 Jakub Jelinek <jakub@redhat.com>
2015 * tree-cfg.c (pass_warn_function_return::execute): In noreturn
2016 functions when optimizing replace GIMPLE_RETURN stmts with
2017 calls to __builtin_unreachable ().
2020 * config/gnu-user.h (LIBTSAN_EARLY_SPEC): Add libtsan_preinit.o
2021 for -fsanitize=thread link of executables.
2022 (LIBLSAN_EARLY_SPEC): Add liblsan_preinit.o for -fsanitize=leak
2023 link of executables.
2026 * config/i386/sse.md (VI248_AVX2, VI248_AVX512BW, VI248_AVX512BW_2):
2028 (<shift_insn><mode>3<mask_name>): Change the last of the 3
2029 define_insns for logical vector shifts to use VI248_AVX512BW
2030 iterator instead of VI48_AVX512, remove <mask_mode512bit_condition>
2031 condition, useless isa and prefix attributes. Change the first
2032 2 of these define_insns to ...
2033 (<mask_codefor><shift_insn><mode>3<mask_name>): ... this, new
2034 define_insn for avx512vl.
2035 (<shift_insn><mode>3): ... and this, new define_insn without
2036 masking for non-avx512vl.
2039 * config/i386/sse.md (*andnot<mode>3,
2040 <mask_codefor><code><mode>3<mask_name>, *<code><mode>3): Split
2041 (=v,v,vm) alternative into (=x,x,xm) and (=v,v,vm), for 128-bit
2042 and 256-bit vectors, the (=x,x,xm) alternative and when mask is
2043 not applied use empty suffix even for TARGET_AVX512VL.
2044 * config/i386/subst.md (mask_prefix3, mask_prefix4): When mask
2045 is applied, supply evex,evex or evex,evex,evex instead of just
2048 2017-10-20 Julia Koval <julia.koval@intel.com>
2050 * common/config/i386/i386-common.c (OPTION_MASK_ISA_GFNI_SET,
2051 (OPTION_MASK_ISA_GFNI_UNSET): New.
2052 (ix86_handle_option): Handle OPT_mgfni.
2053 * config/i386/cpuid.h (bit_GFNI): New.
2054 * config/i386/driver-i386.c (host_detect_local_cpu): Detect gfni.
2055 * config/i386/i386-c.c (ix86_target_macros_internal): Define __GFNI__.
2056 * config/i386/i386.c (ix86_target_string): Add -mgfni.
2057 (ix86_valid_target_attribute_inner_p): Add OPT_mgfni.
2058 * config/i386/i386.h (TARGET_GFNI, TARGET_GFNI_P): New.
2059 * config/i386/i386.opt: Add mgfni.
2061 2017-10-20 Orlando Arias <oarias@knights.ucf.edu>
2063 * config/msp430/msp430.c (msp430_option_override): Disable
2064 -fdelete-null-pointer-checks.
2065 * doc/invoke.text (-fdelete-null-pointer-checks): Document that.
2067 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2069 * x86-tune-costs.h (generic_cost, core_cost): Correct costs
2070 of x87 and SSE instructions.
2072 2017-10-19 Jan Hubicka <hubicka@ucw.cz>
2074 * asan.c (create_cond_insert_point): Do not update edge count.
2075 * auto-profile.c (afdo_propagate_edge): Update for edge count removal.
2076 (afdo_propagate_circuit): Likewise.
2077 (afdo_calculate_branch_prob): Likewise.
2078 (afdo_annotate_cfg): Likewise.
2079 * basic-block.h (struct edge_def): Remove count.
2080 (edge_def::count): New accessor.
2081 * bb-reorder.c (rotate_loop): Update.
2082 (find_traces_1_round): Update.
2083 (connect_traces): Update.
2084 (sanitize_hot_paths): Update.
2085 * cfg.c (unchecked_make_edge): Update.
2086 (make_single_succ_edge): Update.
2087 (check_bb_profile): Update.
2088 (dump_edge_info): Update.
2089 (update_bb_profile_for_threading): Update.
2090 (scale_bbs_frequencies_int): Update.
2091 (scale_bbs_frequencies_gcov_type): Update.
2092 (scale_bbs_frequencies_profile_count): Update.
2093 (scale_bbs_frequencies): Update.
2094 * cfganal.c (connect_infinite_loops_to_exit): Update.
2095 * cfgbuild.c (compute_outgoing_frequencies): Update.
2096 (find_many_sub_basic_blocks): Update.
2097 * cfgcleanup.c (try_forward_edges): Update.
2098 (try_crossjump_to_edge): Update
2099 * cfgexpand.c (expand_gimple_cond): Update
2100 (expand_gimple_tailcall): Update
2101 (construct_exit_block): Update
2102 * cfghooks.c (verify_flow_info): Update
2103 (redirect_edge_succ_nodup): Update
2104 (split_edge): Update
2105 (make_forwarder_block): Update
2106 (duplicate_block): Update
2107 (account_profile_record): Update
2108 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
2109 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
2110 * cfgloopmanip.c (scale_loop_profile): Update.
2112 (lv_adjust_loop_entry_edge): Update.
2113 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
2114 (force_nonfallthru_and_redirect): Update.
2115 (purge_dead_edges): Update.
2116 (rtl_flow_call_edges_add): Update.
2117 * cgraphunit.c (init_lowered_empty_function): Update.
2118 (cgraph_node::expand_thunk): Update.
2119 * gimple-pretty-print.c (dump_probability): Update.
2120 (dump_edge_probability): Update.
2121 * gimple-ssa-isolate-paths.c (isolate_path): Update.
2122 * haifa-sched.c (sched_create_recovery_edges): Update.
2123 * hsa-gen.c (convert_switch_statements): Update.
2124 * ifcvt.c (dead_or_predicable): Update.
2125 * ipa-inline-transform.c (inline_transform): Update.
2126 * ipa-split.c (split_function): Update.
2127 * ipa-utils.c (ipa_merge_profiles): Update.
2128 * loop-doloop.c (add_test): Update.
2129 * loop-unroll.c (unroll_loop_runtime_iterations): Update.
2130 * lto-streamer-in.c (input_cfg): Update.
2131 (input_function): Update.
2132 * lto-streamer-out.c (output_cfg): Update.
2133 * modulo-sched.c (sms_schedule): Update.
2134 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
2135 * predict.c (maybe_hot_edge_p): Update.
2136 (unlikely_executed_edge_p): Update.
2137 (probably_never_executed_edge_p): Update.
2138 (dump_prediction): Update.
2139 (drop_profile): Update.
2140 (propagate_unlikely_bbs_forward): Update.
2141 (determine_unlikely_bbs): Update.
2142 (force_edge_cold): Update.
2143 * profile.c (compute_branch_probabilities): Update.
2144 * reg-stack.c (better_edge): Update.
2145 * shrink-wrap.c (handle_simple_exit): Update.
2146 * tracer.c (better_p): Update.
2147 * trans-mem.c (expand_transaction): Update.
2148 (split_bb_make_tm_edge): Update.
2149 * tree-call-cdce.c: Update.
2150 * tree-cfg.c (gimple_find_sub_bbs): Update.
2151 (gimple_split_edge): Update.
2152 (gimple_duplicate_sese_region): Update.
2153 (gimple_duplicate_sese_tail): Update.
2154 (gimple_flow_call_edges_add): Update.
2155 (insert_cond_bb): Update.
2156 (execute_fixup_cfg): Update.
2157 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update.
2158 * tree-complex.c (expand_complex_div_wide): Update.
2159 * tree-eh.c (lower_resx): Update.
2160 (unsplit_eh): Update.
2161 (cleanup_empty_eh_move_lp): Update.
2162 * tree-inline.c (copy_edges_for_bb): Update.
2163 (freqs_to_counts): Update.
2164 (copy_cfg_body): Update.
2165 * tree-ssa-dce.c (remove_dead_stmt): Update.
2166 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
2167 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
2168 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
2169 (unloop_loops): Update.
2170 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2171 * tree-ssa-loop-split.c (connect_loops): Update.
2172 (split_loop): Update.
2173 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
2174 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
2175 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
2176 * tree-ssa-reassoc.c (branch_fixup): Update.
2177 * tree-ssa-tail-merge.c (replace_block_by): Update.
2178 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
2179 (compute_path_counts): Update.
2180 (update_profile): Update.
2181 (recompute_probabilities): Update.
2182 (update_joiner_offpath_counts): Update.
2183 (estimated_freqs_path): Update.
2184 (freqs_to_counts_path): Update.
2185 (clear_counts_path): Update.
2186 (ssa_fix_duplicate_block_edges): Update.
2187 (duplicate_thread_path): Update.
2188 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
2189 (case_bit_test_cmp): Update.
2190 (collect_switch_conv_info): Update.
2191 (gen_inbound_check): Update.
2192 (do_jump_if_equal): Update.
2193 (emit_cmp_and_jump_insns): Update.
2194 * tree-tailcall.c (decrease_profile): Update.
2195 (eliminate_tail_call): Update.
2196 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
2197 (vect_do_peeling): Update.
2198 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2199 * ubsan.c (ubsan_expand_null_ifn): Update.
2200 (ubsan_expand_ptr_ifn): Update.
2201 * value-prof.c (gimple_divmod_fixed_value): Update.
2202 (gimple_mod_pow2): Update.
2203 (gimple_mod_subtract): Update.
2204 (gimple_ic): Update.
2205 (gimple_stringop_fixed_value): Update.
2207 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2210 * config/i386/i386.md (sub to cmp): New peephole2 pattern.
2212 2017-10-19 Alexander Monakov <amonakov@ispras.ru>
2214 PR rtl-optimization/82395
2215 * ira-color.c (allocno_priority_compare_func): Fix comparison step
2216 based on non_spilled_static_chain_regno_p.
2218 2017-10-19 Uros Bizjak <ubizjak@gmail.com>
2220 * config/i386/i386.c (output_387_binary_op): Rewrite SSE part.
2221 (ix86_emit_mode_set): Rewrite insn mnemonic construction.
2222 (ix86_prepare_fp_compare_args): Redefine is_sse as bool.
2224 2017-10-19 Martin Sebor <msebor@redhat.com>
2226 PR tree-optimization/82596
2227 * tree.c (array_at_struct_end_p): Handle STRING_CST.
2229 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2231 * asan.c (handle_builtin_alloca): Deal with all alloca variants.
2232 (get_mem_refs_of_builtin_call): Likewise.
2233 * builtins.c (expand_builtin_apply): Adjust call to
2234 allocate_dynamic_stack_space.
2235 (expand_builtin_alloca): For __builtin_alloca_with_align_and_max, pass
2236 the third argument to allocate_dynamic_stack_space, otherwise -1.
2237 (expand_builtin): Deal with all alloca variants.
2238 (is_inexpensive_builtin): Likewise.
2239 * builtins.def (BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX): New.
2240 * calls.c (special_function_p): Deal with all alloca variants.
2241 (initialize_argument_information): Adjust call to
2242 allocate_dynamic_stack_space.
2243 (expand_call): Likewise.
2244 * cfgexpand.c (expand_call_stmt): Deal with all alloca variants.
2245 * doc/extend.texi (Built-ins): Add __builtin_alloca_with_align_and_max
2246 * explow.c (allocate_dynamic_stack_space): Add MAX_SIZE parameter and
2247 use it for the stack usage computation.
2248 * explow.h (allocate_dynamic_stack_space): Adjust prototype.
2249 * function.c (gimplify_parameters): Call build_alloca_call_expr.
2250 * gimple-ssa-warn-alloca.c (alloca_call_type): Simplify control flow.
2251 Take into account 3rd argument of __builtin_alloca_with_align_and_max.
2252 (in_loop_p): Remove first argument and useless check.
2253 (pass_walloca::execute): Remove useless test and adjust call to above.
2254 * gimple.c (gimple_build_call_from_tree): Deal with all alloc variants
2255 * gimplify.c (gimplify_vla_decl): Call build_alloca_call_expr.
2256 (gimplify_call_expr): Deal with all alloca variants.
2257 * hsa-gen.c (gen_hsa_alloca): Likewise.
2258 (gen_hsa_insns_for_call): Likewise.
2259 * ipa-pure-const.c (special_builtin_state): Likewise.
2260 * tree-chkp.c (chkp_build_returned_bound): Likewise.
2261 * tree-object-size.c (alloc_object_size): Likewise.
2262 * tree-ssa-alias.c (ref_maybe_used_by_call_p_1): Likewise.
2263 (call_may_clobber_ref_p_1): Likewise.
2264 * tree-ssa-ccp.c (evaluate_stmt): Likewise.
2265 (ccp_fold_stmt): Likewise.
2266 (optimize_stack_restore): Likewise.
2267 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
2268 (mark_all_reaching_defs_necessary_1): Likewise.
2269 (propagate_necessity): Likewise.
2270 (eliminate_unnecessary_stmts): Likewise.
2271 * tree.c (build_common_builtin_nodes): Build
2272 BUILT_IN_ALLOCA_WITH_ALIGN_AND_MAX.
2273 (build_alloca_call_expr): New function.
2274 * tree.h (ALLOCA_FUNCTION_CODE_P): New macro.
2275 (CASE_BUILT_IN_ALLOCA): Likewise.
2276 (build_alloca_call_expr): Declare.
2277 * varasm.c (incorporeal_function_p): Deal with all alloca variants.
2279 2017-10-19 Eric Botcazou <ebotcazou@adacore.com>
2282 * dwarf2out.c (new_die_raw): New static inline function.
2283 (new_die): Use it to create the DIE.
2284 (add_AT_external_die_ref): Likewise.
2285 (clone_die): Likewise.
2286 (clone_as_declaration): Likewise.
2287 (dwarf2out_vms_debug_main_pointer): Likewise.
2288 (base_type_die): Likewise. Remove early return for corner cases.
2289 Do not call add_pubtype on the DIE here.
2290 (is_base_type): Remove ERROR_MARK and return 0 for VOID_TYPE.
2291 (modified_type_die): Adjust the lookup for reverse order DIEs. Skip
2292 typedefs for base types with DW_AT_endianity. Make sure a DIE with
2293 native order exists for base types, attach the DIE manually and call
2294 add_pubtype on it. Do not equate a reverse order DIE to the type.
2296 2017-10-19 Richard Earnshaw <rearnsha@arm.com>
2298 * config/arm/arm.c (align_ok_ldrd_strd): New function.
2299 (mem_ok_for_ldrd_strd): New parameter align. Extract the alignment of
2301 (gen_operands_ldrd_strd): Validate the alignment of the accesses.
2303 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2305 * flag-types.h (enum sanitize_code): Add SANITIZE_BUILTIN. Or
2306 SANITIZE_BUILTIN into SANITIZE_UNDEFINED.
2307 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN,
2308 BUILT_IN_UBSAN_HANDLE_INVALID_BUILTIN_ABORT): New builtins.
2309 * opts.c (sanitizer_opts): Add builtin.
2310 * ubsan.c (instrument_builtin): New function.
2311 (pass_ubsan::execute): Call it.
2312 (pass_ubsan::gate): Enable even for SANITIZE_BUILTIN.
2313 * doc/invoke.texi: Document -fsanitize=builtin.
2315 * ubsan.c (ubsan_expand_null_ifn): Use _v1 suffixed type mismatch
2316 builtins, store max (log2 (align), 0) into uchar field instead of
2317 align into uptr field.
2318 (ubsan_expand_objsize_ifn): Use _v1 suffixed type mismatch builtins,
2319 store uchar 0 field instead of uptr 0 field.
2320 (instrument_nonnull_return): Use _v1 suffixed nonnull return builtin,
2321 instead of passing one address of struct with 2 locations pass
2322 two addresses of structs with 1 location each.
2323 * sanitizer.def (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH,
2324 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_ABORT,
2325 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN,
2326 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_ABORT): Removed.
2327 (BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1,
2328 BUILT_IN_UBSAN_HANDLE_TYPE_MISMATCH_V1_ABORT,
2329 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1,
2330 BUILT_IN_UBSAN_HANDLE_NONNULL_RETURN_V1_ABORT): New builtins.
2332 2017-10-19 Martin Liska <mliska@suse.cz>
2335 * file-find.c (remove_prefix): Remove.
2336 * file-find.h (remove_prefix): Likewise.
2337 * gcc-ar.c: Remove smartness of lookup.
2339 2017-10-19 Segher Boessenkool <segher@kernel.crashing.org>
2341 * config/rs6000/rs6000.md (*call_indirect_aix<mode>,
2342 *call_value_indirect_aix<mode>, *call_indirect_elfv2<mode>,
2343 *call_value_indirect_elfv2<mode>): Add correct mode to the unspec.
2345 2017-10-19 Jakub Jelinek <jakub@redhat.com>
2348 * config/i386/i386.md (setcc + movzbl to xor + setcc): New peephole2.
2349 (setcc + and to xor + setcc): New peephole2.
2351 2017-10-19 Tom de Vries <tom@codesourcery.com>
2353 * doc/sourcebuild.texi (Test Directives, Variants of
2354 dg-require-support): Add dg-require-stack-size.
2356 2017-10-19 Martin Liska <mliska@suse.cz>
2359 * gimplify.c (gimplify_decl_expr): Do not instrument variables
2360 that have a large alignment.
2361 (gimplify_target_expr): Likewise.
2363 2017-10-18 Segher Boessenkool <segher@kernel.crashing.org>
2365 PR rtl-optimization/82602
2366 * ira.c (rtx_moveable_p): Return false for volatile asm.
2368 2017-10-18 Uros Bizjak <ubizjak@gmail.com>
2371 * config/i386/i386-modes.def (CCGZ): New CC mode.
2372 * config/i386/i386.md (sub<mode>3_carry_ccgz): New insn pattern.
2373 * config/i386/predicates.md (ix86_comparison_operator):
2375 * config/i386/i386.c (ix86_expand_branch) <case E_TImode>:
2376 Emulate LE, LEU, GT, GTU, LT, LTU, GE and GEU double-word comparisons
2377 with double-word subtraction.
2378 (put_condition_code): Handle CCGZmode.
2380 2017-10-18 Aldy Hernandez <aldyh@redhat.com>
2382 * wide-int.cc (debug (const wide_int &)): New.
2383 (debug (const wide_int *)): New.
2384 (debug (const widest_int &)): New.
2385 (debug (const widest_int *)): New.
2387 2017-10-18 Vladimir Makarov <vmakarov@redhat.com>
2390 * lra-constraints.c (curr_insn_transform): Use non-input operand
2391 instead of output one for matched reload.
2393 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2395 * tree-loop-distribution.c (INCLUDE_ALGORITHM): New header file.
2396 (tree-ssa-loop-ivopts.h): New header file.
2397 (struct builtin_info): New fields.
2398 (classify_builtin_1): Compute and record base and offset parts for
2399 memset builtin partition by calling strip_offset.
2400 (offset_cmp, fuse_memset_builtins): New functions.
2401 (finalize_partitions): Fuse adjacent memset partitions by calling
2403 * tree-ssa-loop-ivopts.c (strip_offset): Delete static declaration.
2404 Expose the interface.
2405 * tree-ssa-loop-ivopts.h (strip_offset): New declaration.
2407 2017-10-18 Bin Cheng <bin.cheng@arm.com>
2409 PR tree-optimization/82574
2410 * tree-loop-distribution.c (find_single_drs): New parameter. Check
2411 that data reference must be executed exactly once per iteration
2412 against the outermost loop in nest.
2413 (classify_partition): Update call to above function.
2415 2017-10-18 Richard Biener <rguenther@suse.de>
2417 PR tree-optimization/82591
2418 * graphite.c (graphite_transform_loops): Move code gen message
2420 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2421 Here. Handle scop_to_isl_ast failing.
2422 (scop_to_isl_ast): Limit the number of ISL operations.
2424 2017-10-18 Richard Biener <rguenther@suse.de>
2426 * graphite-isl-ast-to-gimple.c
2427 (translate_isl_ast_to_gimple::set_rename): Simplify.
2428 (translate_isl_ast_to_gimple::set_rename_for_each_def): Inline...
2429 (graphite_copy_stmts_from_block): ... here.
2430 (copy_bb_and_scalar_dependences): Simplify.
2431 (add_parameters_to_ivs_params): Canonicalize.
2432 (generate_entry_out_of_ssa_copies): Simplify.
2433 * graphite-sese-to-poly.c (extract_affine_name): Simplify
2434 by passing in ISL dimension.
2435 (parameter_index_in_region_1): Rename to ...
2436 (parameter_index_in_region): ... this.
2437 (extract_affine): Adjust assert, pass down parameter index.
2438 (add_param_constraints): Use range-info when available.
2439 (build_scop_context): Adjust.
2440 * sese.c (new_sese_info): Adjust.
2441 (free_sese_info): Likewise.
2442 * sese.h (bb_map_t, rename_map_t, phi_rename, init_back_edge_pair_t):
2443 Remove unused typedefs.
2444 (struct sese_info_t): Simplify rename_map, remove incomplete_phis.
2446 2017-10-18 Martin Liska <mliska@suse.cz>
2448 * combine.c (simplify_compare_const): Add gcc_fallthrough.
2450 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2452 * config/s390/s390.c (s390_bb_fallthru_entry_likely): New function.
2453 (s390_sched_init): Do not reset s390_sched_state if we entered the
2454 current basic block via a fallthru edge and all others are unlikely.
2456 2017-10-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
2458 * config/s390/s390.c (NUM_SIDES): New variable.
2459 (LONGRUNNING_THRESHOLD): New variable.
2460 (LATENCY_FACTOR): New variable.
2461 (s390_sched_score): Decrease score for long-running instructions on
2463 (s390_sched_variable_issue): Perform bookkeeping for long-running
2466 2017-10-18 Richard Biener <rguenther@suse.de>
2468 * graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
2469 Simplify with removal of the parameter rename map.
2470 (set_rename): Likewise.
2471 (should_copy_to_new_region): Likewise.
2472 (graphite_copy_stmts_from_block): Likewise.
2473 (copy_bb_and_scalar_dependences): Remove initialization of
2474 unused copied_bb_map.
2476 (copy_internal_parameters): Likewise.
2477 (graphite_regenerate_ast_isl): Do not call copy_internal_parameters.
2478 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
2479 Use INTEGRAL_TYPE_P.
2480 (parameter_index_in_region_1): Rename to ...
2481 (assign_parameter_index_in_region): ... this. Assert we have
2482 a parameter we handle.
2483 (scan_tree_for_params): Adjust.
2484 * sese.h (parameter_rename_map_t): Remove.
2485 (struct sese_info_t): Remove unused parameter_rename_map and
2486 copied_bb_map members.
2487 * sese.c (new_sese_info): Adjust.
2488 (free_sese_info): Likewise.
2490 2017-10-18 Martin Liska <mliska@suse.cz>
2493 * asan.c (asan_expand_poison_ifn): Do not put gimple stmt
2494 on an abnormal edge.
2496 2017-10-18 Sebastian Huber <sebastian.huber@embedded-brains.de>
2498 * doc/invoke.texi (ffunction-sections and fdata-sections):
2501 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2503 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Bail out only if
2504 the use statement can throw internally.
2506 2017-10-17 Eric Botcazou <ebotcazou@adacore.com>
2508 * config/visium/visium.c (visium_select_cc_mode): Return CCmode for
2509 any RTX present on the RHS of a SET.
2510 * compare-elim.c (try_eliminate_compare): Restore comment.
2512 2017-10-17 Jakub Jelinek <jakub@redhat.com>
2514 * langhooks.h (struct lang_hooks): Document that tree_size langhook
2515 may be also called on tcc_type nodes.
2516 * langhooks.c (lhd_tree_size): Likewise.
2518 2017-10-17 David Malcolm <dmalcolm@redhat.com>
2520 * gimple-ssa-sprintf.c (fmtwarn): Update for changed signature of
2521 format_warning_at_substring.
2522 (maybe_warn): Convert source_range * param to a location_t. Pass
2523 UNKNOWN_LOCATION rather than NULL to fmtwarn.
2524 (format_directive): Remove code to extract source_ranges and
2525 source_range * in favor of just a location_t.
2526 (parse_directive): Pass UNKNOWN_LOCATION rather than NULL to
2528 * substring-locations.c (format_warning_va): Convert
2529 source_range * param to a location_t.
2530 (format_warning_at_substring): Likewise.
2531 * substring-locations.h (format_warning_va): Likewise.
2532 (format_warning_at_substring): Likewise.
2534 2017-10-17 Jan Hubicka <hubicka@ucw.cz>
2536 * target.h (enum vect_cost_for_stmt): Add vec_gather_load and
2538 * tree-vect-stmts.c (record_stmt_cost): Make difference between normal
2539 and scatter/gather ops.
2541 * aarch64/aarch64.c (aarch64_builtin_vectorization_cost): Add
2542 vec_gather_load and vec_scatter_store.
2543 * arm/arm.c (arm_builtin_vectorization_cost): Likewise.
2544 * powerpcspe/powerpcspe.c (rs6000_builtin_vectorization_cost): Likewise.
2545 * rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Likewise.
2546 * s390/s390.c (s390_builtin_vectorization_cost): Likewise.
2547 * spu/spu.c (spu_builtin_vectorization_cost): Likewise.
2548 * i386/i386.c (x86_builtin_vectorization_cost): Likewise.
2550 2017-10-17 Uros Bizjak <ubizjak@gmail.com>
2552 * reg-stack.c (compare_for_stack_reg): Add bool argument.
2553 Detect FTST instruction and handle its register pops. Only pop
2554 second operand if can_pop_second_op is true.
2555 (subst_stack_regs_pat) <case COMPARE>: Detect FCOMI instruction to
2556 set can_pop_second_op to false in the compare_for_stack_reg call.
2558 * config/i386/i386.md (*cmpi<FPCMP:unord><MODEF:mode>): Only call
2559 output_fp_compare for stack register operands.
2560 * config/i386/i386.c (output_fp_compare): Do not output SSE compare
2561 instructions here. Do not emit stack register pops here. Assert
2562 that FCOMPP pops next to top stack register. Rewrite function.
2564 2017-10-17 Nathan Sidwell <nathan@acm.org>
2567 * alias.c (compare_base_decls): Check HAS_DECL_ASSEMBLER_NAME_P,
2568 use DECL_ASSEMBLER_NAME_RAW.
2571 * tree.c (tree_code_size): Reformat. Punt to lang hook for unknown
2574 2017-10-17 Qing Zhao <qing.zhao@oracle.com>
2575 Wilco Dijkstra <wilco.dijkstra@arm.com>
2577 * builtins.c (expand_builtin_update_setjmp_buf): Add a
2578 converstion to Pmode from the buf_addr.
2580 2017-10-17 Richard Biener <rguenther@suse.de>
2582 * graphite-dependences.c (scop_get_reads_and_writes): Change
2583 output parameters to references.
2585 2017-10-17 Jackson Woodruff <jackson.woodruff@arm.com>
2587 PR 71026/tree-optimization
2588 * fold-const.c (distribute_real_division): Removed.
2589 (fold_binary_loc): Remove calls to distribute_real_divison.
2591 2017-10-17 Richard Biener <rguenther@suse.de>
2593 * graphite-scop-detection.c
2594 (scop_detection::stmt_has_simple_data_refs_p): Always use
2595 the full nest as region.
2596 (try_generate_gimple_bb): Likewise.
2597 * sese.c (scalar_evolution_in_region): Simplify now that
2598 SCEV can handle instantiation in regions.
2599 * tree-scalar-evolution.c (instantiate_scev_name): Also instantiate
2600 in the non-loop part of a function if requested.
2602 2017-10-17 Richard Biener <rguenther@suse.de>
2604 PR tree-optimization/82563
2605 * graphite-isl-ast-to-gimple.c (generate_entry_out_of_ssa_copies):
2607 (graphite_regenerate_ast_isl): Call it.
2608 * graphite-scop-detection.c (build_scops): Remove entry edge split.
2610 2017-10-17 Jakub Jelinek <jakub@redhat.com>
2612 PR tree-optimization/82549
2613 * fold-const.c (optimize_bit_field_compare, fold_truth_andor_1):
2614 Formatting fixes. Instead of calling make_bit_field_ref with negative
2617 2017-10-17 Olga Makhotina <olga.makhotina@intel.com>
2619 * config/i386/avx512dqintrin.h (_mm_mask_reduce_sd,
2620 _mm_maskz_reduce_sd, _mm_mask_reduce_ss,=20
2621 _mm_maskz_reduce_ss): New.
2622 * config/i386/i386-builtin.def (__builtin_ia32_reducesd_mask,
2623 __builtin_ia32_reducess_mask): Ditto..
2624 (__builtin_ia32_reducesd, __builtin_ia32_reducess): Remove.
2625 * config/i386/sse.md (reduces<mode>): Renamed to ...
2626 (reduces<mode><mask_scalar_name>): ... this.
2627 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0|%0, %1, %2, %3}):
2629 (vreduce<ssescalarmodesuffix>\t{%3, %2, %1, %0<mask_scalar_operand4>|
2630 %0<mask_scalar_operand4>, %1, %2, %3}): ... this.
2632 2017-10-16 David Malcolm <dmalcolm@redhat.com>
2634 * Makefile.in (OBJS): Add unique-ptr-tests.o.
2635 * selftest-run-tests.c (selftest::run_tests): Call
2636 selftest::unique_ptr_tests_cc_tests.
2637 * selftest.h (selftest::unique_ptr_tests_cc_tests): New decl.
2638 * unique-ptr-tests.cc: New file.
2640 2017-10-16 Vladimir Makarov <vmakarov@redhat.com>
2643 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
2645 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
2646 (make_hard_regno_born, make_hard_regno_dead): Update
2647 bb_killed_pseudos and bb_gen_pseudos for fixed regs.
2649 2017-10-16 Jeff Law <law@redhat.com>
2651 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
2653 2017-10-16 Jan Hubicka <hubicka@ucw.cz>
2655 * x86-tune-costs.h (znver1_cost): Fix move cost tables.
2657 2017-10-16 Olivier Hainque <hainque@adacore.com>
2659 * gcc/config.gcc (powerpc*-*-*spe*): Pick 8548 as the default
2660 with_cpu if we were configured for an e500v2 target cpu name.
2662 2017-10-16 Thomas Preud'homme <thomas.preudhomme@arm.com>
2664 * config/arm/arm-cpus.in (cortex-m33): Add nodsp option.
2665 * doc/invoke.texi: Document +nodsp as a valid extension for
2668 2017-10-16 Martin Liska <mliska@suse.cz>
2670 * sbitmap.c (bitmap_bit_in_range_p_checking): New function.
2671 (test_set_range): Likewise.
2672 (test_range_functions): Rename to ...
2673 (test_bit_in_range): ... this.
2674 (sbitmap_c_tests): Add new test.
2676 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2678 * config/aarch64/arm_neon.h (vdot_u32, vdotq_u32, vdot_s32, vdotq_s32):
2680 (vdot_lane_u32, vdot_laneq_u32, vdotq_lane_u32, vdotq_laneq_u32): New.
2681 (vdot_lane_s32, vdot_laneq_s32, vdotq_lane_s32, vdotq_laneq_s32): New.
2683 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2685 * config/aarch64/aarch64-builtins.c
2686 (aarch64_types_quadopu_lane_qualifiers): New.
2687 (TYPES_QUADOPU_LANE): New.
2688 * config/aarch64/aarch64-simd.md (aarch64_<sur>dot<vsi2qi>): New.
2689 (<sur>dot_prod<vsi2qi>, aarch64_<sur>dot_lane<vsi2qi>): New.
2690 (aarch64_<sur>dot_laneq<vsi2qi>): New.
2691 * config/aarch64/aarch64-simd-builtins.def (sdot, udot): New.
2692 (sdot_lane, udot_lane, sdot_laneq, udot_laneq): New.
2693 * config/aarch64/iterators.md (sur): Add UNSPEC_SDOT, UNSPEC_UDOT.
2694 (Vdottype, DOTPROD): New.
2695 (sur): Add SDOT and UDOT.
2697 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2699 * config/aarch64/aarch64.h (AARCH64_FL_DOTPROD): New.
2700 (AARCH64_ISA_DOTPROD, TARGET_DOTPROD): New.
2701 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
2703 * config/aarch64/aarch64-option-extensions.def (dotprod): New.
2704 * config/aarch64/aarch64-cores.def (cortex-a55, cortex-a75):
2705 Enable TARGET_DOTPROD.
2706 (cortex-a75.cortex-a55): Likewise.
2707 * doc/invoke.texi (aarch64-feature-modifiers): Document dotprod.
2709 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2711 * config/arm/arm-builtins.c (arm_unsigned_uternop_qualifiers): New.
2712 (UTERNOP_QUALIFIERS, arm_umac_lane_qualifiers, UMAC_LANE_QUALIFIERS):
2714 * config/arm/arm_neon_builtins.def (sdot, udot, sdot_lane, udot_lane):
2716 * config/arm/iterators.md (DOTPROD, VSI2QI, vsi2qi): New.
2717 (UNSPEC_DOT_S, UNSPEC_DOT_U, opsuffix): New.
2718 * config/arm/neon.md (neon_<sup>dot<vsi2qi>): New.
2719 (neon_<sup>dot_lane<vsi2qi>, <sup>dot_prod<vsi2qi>): New.
2720 * config/arm/types.md (neon_dot, neon_dot_q): New.
2721 * config/arm/unspecs.md (sup): Add UNSPEC_DOT_S, UNSPEC_DOT_U.
2723 2017-10-16 Tamar Christina <tamar.christina@arm.com>
2725 * config/arm/arm.h (TARGET_DOTPROD): New.
2726 * config/arm/arm.c (arm_arch_dotprod): New.
2727 (arm_option_reconfigure_globals): Add arm_arch_dotprod.
2728 * config/arm/arm-c.c (__ARM_FEATURE_DOTPROD): New.
2729 * config/arm/arm-cpus.in (armv8.2-a): Enabled +dotprod.
2730 (feature dotprod, group dotprod, ALL_SIMD_INTERNAL): New.
2731 (ALL_FPU_INTERNAL): Use ALL_SIMD_INTERNAL.
2732 * config/arm/t-multilib (v8_2_a_simd_variants): Add dotprod.
2733 * doc/invoke.texi (armv8.2-a): Document dotprod
2735 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2737 * i386.c (ix86_vec_cost): New function.
2738 (ix86_rtx_costs): Handle vector operations better.
2739 * i386.h (struct processor_costs): Add sse_op, fmasd, fmass.
2740 * x86-tune-costs.h: Add new costs to all tables.
2742 2017-10-14 Jan Hubicka <hubicka@ucw.cz>
2744 * i386.c (ix86_rtx_costs): Make difference between x87 and SSE
2746 * i386.h (struct processor_costs): Add addss, mulss, mulsd, divss,
2747 divsd, sqrtss and sqrtsd
2748 * x86-tune-costs.h: Add new entries to all costs.
2749 (znver1_cost): Fix to match real instruction latencies.
2751 2017-10-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2752 Michael Collison <michael.collison@arm.com>
2754 * compare-elim.c: Include emit-rtl.h.
2755 (can_merge_compare_into_arith): New function.
2756 (try_validate_parallel): Likewise.
2757 (try_merge_compare): Likewise.
2758 (try_eliminate_compare): Call the above when no previous clobber
2760 (execute_compare_elim_after_reload): Add DF_UD_CHAIN and DF_DU_CHAIN
2763 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2767 * tree-ssa-phiopt.c (value_replacement): Comment fix. Handle
2768 up to 2 preparation statements for ASSIGN in MIDDLE_BB.
2772 * tree-ssa-forwprop.c (simplify_rotate): Allow def_arg1[N]
2773 to be any operand_equal_p operands. For & (B - 1) require
2774 B to be power of 2. Recognize
2775 (X << (Y & (B - 1))) | (X >> ((-Y) & (B - 1))) and similar patterns.
2777 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2780 * optabs.c (expand_memory_blockage): Fix call of
2781 targetm.have_memory_blockage.
2783 2017-10-14 Jakub Jelinek <jakub@redhat.com>
2786 * config.gcc (*-*-solaris2*, i[34567]86-*-cygwin*,
2787 x86_64-*-cygwin*, i[34567]86-*-mingw* | x86_64-*-mingw*): Append
2788 objects to extra_objs instead of overwriting it.
2790 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2792 * config/i386/sync.md (FILD_ATOMIC/FIST_ATOMIC FP load peephole2):
2793 Use any_fp_register_operand as operand[3] predicate. Simplify
2794 equality test for operands[2] and operands[4] memory location.
2795 (LDX_ATOMIC/STX_ATOMIC FP load peephole2): Ditto.
2796 (FILD_ATOMIC/FIST_ATOMIC FP load peephole2 with mem blockage): New.
2797 (LDX_ATOMIC/LDX_ATOMIC FP load peephole2 with mem blockage): Ditto.
2798 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2): Use
2799 any_fp_register_operand as operand[1] predicate. Simplify
2800 equality test for operands[0] and operands[3] memory location.
2801 (LDX_ATOMIC/STX_ATOMIC FP store peephole2): Ditto.
2802 (FILD_ATOMIC/FIST_ATOMIC FP store peephole2 with mem blockage): New.
2803 (LDX_ATOMIC/LDX_ATOMIC FP storepeephole2 with mem blockage): Ditto.
2805 2017-10-14 Uros Bizjak <ubizjak@gmail.com>
2807 * target-insns.def: Add memory_blockage.
2808 * optabs.c (expand_memory_blockage): New function.
2809 (expand_asm_memory_barrier): Rename ...
2810 (expand_asm_memory_blockage): ... to this.
2811 (expand_mem_thread_fence): Call expand_memory_blockage
2812 instead of expand_asm_memory_barrier.
2813 (expand_mem_singnal_fence): Ditto.
2814 (expand_atomic_load): Ditto.
2815 (expand_atomic_store): Ditto.
2816 * doc/md.texi (Standard Pattern Names For Generation):
2817 Document memory_blockage instruction pattern.
2819 2017-10-13 Sebastian Perta <sebastian.perta@renesas.com>
2821 * config/rl78/rl78.c (rl78_emit_libcall): New function.
2822 * config/rl78/rl78-protos.h (rl78_emit_libcall): New function.
2823 * config/rl78/rl78.md: New define_expand "adddi3".
2825 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2827 * cfghooks.c (verify_flow_info): Disable check that all probabilities
2830 2017-10-13 Jeff Law <law@redhat.com>
2832 * tree-ssa-reassoc.c (reassociate_bb): Clarify code slighly.
2834 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2837 * internal-fn.c (expand_mul_overflow): If both operands have
2838 the same highpart of -1 or 0 and the topmost bit of lowpart
2839 is different, overflow is if res <= 0 rather than res < 0.
2841 2017-10-13 Pat Haugen <pthaugen@us.ibm.com>
2843 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost): Remove
2844 TARGET_P9_VECTOR code for unaligned_load case.
2846 2017-10-13 Jan Hubicka <hubicka@ucw.cz>
2848 * cfghooks.c (verify_flow_info): Check that edge probabilities are set.
2850 2017-10-13 Nathan Sidwell <nathan@acm.org>
2852 * tree-core.h (tree_contains_struct): Make bool.
2853 * tree.c (tree_contains_struct): Likewise.
2854 * tree.h (MARK_TS_BASE): Remove do ... while (0) idiom.
2855 (MARK_TS_TYPED, MARK_TS_COMMON, MARK_TS_TYPE_COMMON,
2856 MARK_TS_TYPE_WITH_LANG_SPECIFIC, MARK_TS_DECL_MINIMAL,
2857 MARK_TS_DECL_COMMON, MARK_TS_DECL_WRTL, MARK_TS_DECL_WITH_VIS,
2858 MARK_TS_DECL_NON_COMMON): Likewise, use comma operator.
2860 2017-10-13 Richard Biener <rguenther@suse.de>
2862 * graphite-isl-ast-to-gimple.c
2863 (translate_isl_ast_to_gimple::get_rename_from_scev): Remove unused
2864 parameters and dominance check.
2865 (translate_isl_ast_to_gimple::graphite_copy_stmts_from_block): Adjust.
2866 (translate_isl_ast_to_gimple::copy_bb_and_scalar_dependences): Likewise.
2867 (translate_isl_ast_to_gimple::graphite_regenerate_ast_isl):
2868 Do not update SSA form here or do intermediate IL verification.
2869 * graphite.c: Include tree-ssa.h and tree-into-ssa.h.
2870 (graphite_initialize): Remove check on the number of loops in
2871 the function and inline into graphite_transform_loops.
2872 (graphite_finalize): Inline into graphite_transform_loops.
2873 (graphite_transform_loops): Perform SSA update and IL verification
2875 * params.def (PARAM_GRAPHITE_MIN_LOOPS_PER_FUNCTION): Remove.
2877 2017-10-13 Richard Biener <rguenther@suse.de>
2879 * graphite-isl-ast-to-gimple.c (max_mode_int_precision,
2880 graphite_expression_type_precision): Avoid global constructor
2882 (translate_isl_ast_to_gimple::translate_isl_ast_to_gimple): Here.
2883 (translate_isl_ast_to_gimple::graphite_expr_type): Add type member.
2884 (translate_isl_ast_to_gimple::translate_isl_ast_node_for): Use it.
2885 (translate_isl_ast_to_gimple::build_iv_mapping): Likewise.
2886 (translate_isl_ast_to_gimple::graphite_create_new_guard): Likewise.
2887 * graphite-sese-to-poly.c (build_original_schedule): Return nothing.
2889 2017-10-13 H.J. Lu <hongjiu.lu@intel.com>
2892 * config/i386/i386.h (ix86_red_zone_size): New.
2893 * config/i386/i386.md (push peephole2s): Replace
2894 "!ix86_using_red_zone ()" with "ix86_red_zone_size == 0".
2896 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2897 Alan Hayward <alan.hayward@arm.com>
2898 David Sherwood <david.sherwood@arm.com>
2900 * combine.c (can_change_dest_mode): Reject changes in
2901 REGMODE_NATURAL_SIZE.
2903 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2904 Alan Hayward <alan.hayward@arm.com>
2905 David Sherwood <david.sherwood@arm.com>
2907 * cfgexpand.c (expand_debug_expr): Use GET_MODE_UNIT_BITSIZE.
2908 (expand_debug_source_expr): Likewise.
2909 * combine.c (combine_simplify_rtx): Likewise.
2910 * cse.c (fold_rtx): Likewise.
2911 * fwprop.c (canonicalize_address): Likewise.
2912 * targhooks.c (default_shift_truncation_mask): Likewise.
2914 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2915 Alan Hayward <alan.hayward@arm.com>
2916 David Sherwood <david.sherwood@arm.com>
2918 * optabs.c (add_equal_note): Use GET_MODE_UNIT_SIZE.
2919 (widened_mode): Likewise.
2920 (expand_unop): Likewise.
2921 * ree.c (transform_ifelse): Likewise.
2922 (merge_def_and_ext): Likewise.
2923 (combine_reaching_defs): Likewise.
2924 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
2926 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2927 Alan Hayward <alan.hayward@arm.com>
2928 David Sherwood <david.sherwood@arm.com>
2930 * caller-save.c (replace_reg_with_saved_mem): Use byte_lowpart_offset.
2931 * combine.c (gen_lowpart_for_combine): Likewise.
2932 * dwarf2out.c (rtl_for_decl_location): Likewise.
2933 * final.c (alter_subreg): Likewise.
2934 * rtlhooks.c (gen_lowpart_general): Likewise.
2935 (gen_lowpart_if_possible): Likewise.
2937 2017-10-13 Richard Sandiford <richard.sandiford@linaro.org>
2938 Alan Hayward <alan.hayward@arm.com>
2939 David Sherwood <david.sherwood@arm.com>
2941 * calls.c (expand_call): Use subreg_lowpart_offset.
2942 * cse.c (cse_insn): Likewise.
2943 * regcprop.c (copy_value): Likewise.
2944 (copyprop_hardreg_forward_1): Likewise.
2946 2017-10-13 Jakub Jelinek <jakub@redhat.com>
2949 * config/i386/i386.md (addqi_ext_1, andqi_ext_1,
2950 *andqi_ext_1_cc, *<code>qi_ext_1, *xorqi_ext_1_cc): Change
2951 =Q constraints to +Q and into insn condition add check
2952 that operands[0] and operands[1] are equal.
2953 (*addqi_ext_2, *andqi_ext_2, *<code>qi_ext_2): Change
2954 =Q constraints to +Q and into insn condition add check
2955 that operands[0] is equal to either operands[1] or operands[2].
2958 * fold-const.c (fold_binary_loc) <bit_rotate>: Code cleanups,
2959 instead of handling MINUS_EXPR twice (once for each argument),
2960 canonicalize operand order and handle just once, use rtype where
2961 possible. Handle (A << B) | (A >> (-B & (Z - 1))).
2964 * config/i386/ia32intrin.h (__rold, __rord, __rolq, __rorq): Allow
2965 any values of __C while still being pattern recognizable as a simple
2968 2017-10-13 Richard Biener <rguenther@suse.de>
2970 PR tree-optimization/82451
2972 2017-10-02 Richard Biener <rguenther@suse.de>
2974 PR tree-optimization/82355
2975 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
2976 a mapping for the enclosing loop but avoid generating one for
2978 (copy_bb_and_scalar_dependences): Remove premature codegen
2979 error on PHIs in blocks duplicated into multiple places.
2980 * graphite-scop-detection.c
2981 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
2982 in the region use it as loop and nest to analyze the DR in.
2983 (try_generate_gimple_bb): Likewise.
2984 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
2985 (add_loop_constraints): For blocks in a loop not in the region
2986 create a dimension with a single iteration.
2987 * sese.h (gbb_loop_at_index): Remove assert.
2989 * cfgloop.c (loop_preheader_edge): For the loop tree root
2990 return the single successor of the entry block.
2991 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
2992 Reset the SCEV hashtable and niters.
2993 * graphite-scop-detection.c
2994 (scop_detection::graphite_can_represent_scev): Add SCOP parameter,
2995 assert that we only have POLYNOMIAL_CHREC that vary in loops
2996 contained in the region.
2997 (scop_detection::graphite_can_represent_expr): Adjust.
2998 (scop_detection::stmt_has_simple_data_refs_p): For loops
2999 not in the region set loop to NULL. The nest is now the
3000 entry edge to the region.
3001 (try_generate_gimple_bb): Likewise.
3002 * sese.c (scalar_evolution_in_region): Adjust for
3003 instantiate_scev change.
3004 * tree-data-ref.h (graphite_find_data_references_in_stmt):
3005 Make nest parameter the edge into the region.
3006 (create_data_ref): Likewise.
3007 * tree-data-ref.c (dr_analyze_indices): Make nest parameter an
3008 entry edge into a region and adjust instantiate_scev calls.
3009 (create_data_ref): Likewise.
3010 (graphite_find_data_references_in_stmt): Likewise.
3011 (find_data_references_in_stmt): Pass the loop preheader edge
3012 from the nest argument.
3013 * tree-scalar-evolution.h (instantiate_scev): Make instantiate_below
3014 parameter the edge into the region.
3015 (instantiate_parameters): Use the loop preheader edge as entry.
3016 * tree-scalar-evolution.c (analyze_scalar_evolution): Handle
3018 (get_instantiated_value_entry): Make instantiate_below parameter
3019 the edge into the region.
3020 (instantiate_scev_name): Likewise. Adjust dominance checks,
3021 when we cannot use loop-based instantiation instantiate by
3022 walking use-def chains.
3023 (instantiate_scev_poly): Adjust.
3024 (instantiate_scev_binary): Likewise.
3025 (instantiate_scev_convert): Likewise.
3026 (instantiate_scev_not): Likewise.
3027 (instantiate_array_ref): Remove.
3028 (instantiate_scev_3): Likewise.
3029 (instantiate_scev_2): Likewise.
3030 (instantiate_scev_1): Likewise.
3031 (instantiate_scev_r): Do not blindly handle N-operand trees.
3032 Do not instantiate array-refs. Handle all constants and invariants.
3033 (instantiate_scev): Make instantiate_below parameter
3034 the edge into the region.
3035 (resolve_mixers): Use the loop preheader edge for the region
3036 parameter to instantiate_scev_r.
3037 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Adjust.
3039 2017-10-13 Richard Biener <rguenther@suse.de>
3041 PR tree-optimization/82525
3042 * graphite-isl-ast-to-gimple.c
3043 (translate_isl_ast_to_gimple::widest_int_from_isl_expr_int): Split
3045 (translate_isl_ast_to_gimple::gcc_expression_from_isl_expr_int): Here.
3046 Fail code generation when we cannot represent the isl integer.
3047 (binary_op_to_tree): Elide modulo operations that are no-ops
3048 in the type we code generate. Remove now superfluous code
3051 2017-10-13 Richard Biener <rguenther@suse.de>
3053 * graphite-scop-detection.c (loop_ivs_can_be_represented): Remove.
3054 (scop_detection::harmful_loop_in_region): Remove premature
3055 IV type restriction.
3056 (scop_detection::graphite_can_represent_scev): We can handle
3057 pointer IVs just fine.
3059 2017-10-13 Alan Modra <amodra@gmail.com>
3061 * doc/extend.texi (Extended Asm <Clobbers>): Rename to
3062 "Clobbers and Scratch Registers". Add paragraph on
3063 alternative to clobbers for scratch registers and OpenBLAS
3066 2017-10-13 Alan Modra <amodra@gmail.com>
3068 * doc/extend.texi (Clobbers): Correct vax example. Delete old
3069 example of a memory input for a string of known length. Move
3070 commentary out of table. Add a number of new examples
3071 covering array memory inputs.
3073 2017-10-12 Martin Liska <mliska@suse.cz>
3075 PR tree-optimization/82493
3076 * sbitmap.c (bitmap_bit_in_range_p): Fix the implementation.
3077 (test_range_functions): New function.
3078 (sbitmap_c_tests): Likewise.
3079 * selftest-run-tests.c (selftest::run_tests): Run new tests.
3080 * selftest.h (sbitmap_c_tests): New function.
3082 * tree-ssa-dse.c (live_bytes_read): Fix thinko.
3084 2017-10-12 Michael Meissner <meissner@linux.vnet.ibm.com>
3086 * config/rs6000/amo.h: Fix spacing issue.
3088 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3091 * config/i386/i386.md (*ashl<mode>3_mask_1,
3092 *<shift_insn><mode>3_mask_1, *<rotate_insn><mode>3_mask_1,
3093 *<btsc><mode>_mask_1, *btr<mode>_mask_1): New define_insn_and_split
3096 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3098 * profile-count.h (safe_scale_64bit): Fix GCC4.x path.
3099 (profile_probability): Set max_probability
3100 to (uint32_t) 1 << (n_bits - 2) and update accessors to avoid overlfows
3102 * profile-count.c (profile_probability::differs_from_p): Do not
3103 rely on max_probaiblity == 10000
3105 2017-10-12 Jeff Law <law@redhat.com>
3107 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject ao_refs with
3110 2017-10-12 Martin Sebor <msebor@redhat.com>
3114 * cgraphunit.c (maybe_diag_incompatible_alias): New function.
3115 (handle_alias_pairs): Call it.
3116 * common.opt (-Wattribute-alias): New option.
3117 * doc/extend.texi (ifunc attribute): Discuss C++ specifics.
3118 * doc/invoke.texi (-Wattribute-alias): Document.
3120 2017-10-12 Vladimir Makarov <vmakarov@redhat.com>
3123 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3125 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3127 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3128 (make_hard_regno_born, make_hard_regno_dead): Update
3129 bb_killed_pseudos and bb_gen_pseudos.
3131 2017-10-12 Jan Hubicka <hubicka@ucw.cz>
3133 * config/i386/x86-tune-sched.c (ix86_adjust_cost): Fix Zen support.
3135 2017-10-12 Uros Bizjak <ubizjak@gmail.com>
3137 * config/alpha/alpha.c (alpha_split_conditional_move):
3138 Use std::swap instead of manually swapping.
3139 (alpha_stdarg_optimize_hook): Ditto.
3140 (alpha_canonicalize_comparison): Ditto.
3142 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3144 * tree-loop-distribution.c (struct builtin_info): New struct.
3145 (struct partition): Refactor fields into struct builtin_info.
3146 (partition_free): Free struct builtin_info.
3147 (build_size_arg_loc, build_addr_arg_loc): Delete.
3148 (generate_memset_builtin, generate_memcpy_builtin): Get memory range
3149 information from struct builtin_info.
3150 (find_single_drs): New function refactored from classify_partition.
3151 Also moved builtin validity checks to this function.
3152 (compute_access_range, alloc_builtin): New functions.
3153 (classify_builtin_st, classify_builtin_ldst): New functions.
3154 (classify_partition): Refactor code into functions find_single_drs,
3155 classify_builtin_st and classify_builtin_ldst.
3156 (distribute_loop): Don't do runtime alias check when distributing
3158 (find_seed_stmts_for_distribution): New function.
3159 (pass_loop_distribution::execute): Refactor code finding seed
3160 stmts into above function. Support distribution for the innermost
3161 two-level loop nest. Adjust dump information.
3163 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3165 * tree-loop-distribution.c: Adjust the general comment.
3166 (NUM_PARTITION_THRESHOLD): New macro.
3167 (ssa_name_has_uses_outside_loop_p): Support loop nest distribution.
3168 (classify_partition): Skip builtin pattern of loop nest's inner loop.
3169 (merge_dep_scc_partitions): New parameter ignore_alias_p and use it
3170 in call to build_partition_graph.
3171 (finalize_partitions): New parameter. Make loop distribution more
3172 conservative by fusing more partitions.
3173 (distribute_loop): Don't do runtime alias check in case of loop nest
3175 (find_seed_stmts_for_distribution): New function.
3176 (prepare_perfect_loop_nest): New function.
3177 (pass_loop_distribution::execute): Refactor code finding seed stmts
3178 and loop nest into above functions. Support loop nest distribution.
3179 Adjust dump information accordingly.
3181 2017-10-12 Bin Cheng <bin.cheng@arm.com>
3183 * tree-loop-distribution.c (break_alias_scc_partitions): Add comment
3184 and set PTYPE_SEQUENTIAL for merged partition.
3186 2017-10-12 Richard Biener <rguenther@suse.de>
3188 PR tree-optimization/69728
3190 2017-09-19 Richard Biener <rguenther@suse.de>
3192 PR tree-optimization/69728
3193 * graphite-sese-to-poly.c (schedule_error): New global.
3194 (add_loop_schedule): Handle empty domain by failing the
3196 (build_original_schedule): Handle schedule_error.
3198 * graphite-sese-to-poly.c (add_loop_schedule): Handle empty
3199 domain by returning an unchanged schedule.
3201 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3203 * genrecog.c (validate_pattern): For VEC_SELECT verify that
3204 CONST_INT selectors are 0 to GET_MODE_NUNITS (imode) - 1.
3206 2017-10-12 Aldy Hernandez <aldyh@redhat.com>
3208 * Makefile.in (TAGS): Merge all the *.def files into one pattern.
3211 2017-10-12 Jakub Jelinek <jakub@redhat.com>
3214 * expr.c (store_field): Don't optimize away bitsize == 0 store
3215 from CALL_EXPR with addressable return type.
3217 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3219 * config/rs6000/rs6000.h (TARGET_ISEL64): Delete.
3220 * config/rs6000/rs6000.md (sel): Delete mode attribute.
3221 (mov<mode>cc, isel_signed_<mode>, isel_unsigned_<mode>,
3222 *isel_reversed_signed_<mode>, *isel_reversed_unsigned_<mode>): Use
3223 TARGET_ISEL instead of TARGET_ISEL<sel>.
3225 2017-10-11 David Edelsohn <dje.gcc@gmail.com>
3227 * config/rs6000/rs6000.c
3228 (rs6000_xcoff_asm_output_aligned_decl_common): Test for NULL decl.
3230 2017-10-11 Segher Boessenkool <segher@kernel.crashing.org>
3232 * config/rs6000/predicates.md (zero_constant, all_ones_constant):
3234 (reg_or_cint_operand): Fix comment.
3235 (reg_or_zero_operand): New predicate.
3236 * config/rs6000/rs6000-protos.h (output_isel): Delete.
3237 * config/rs6000/rs6000.c (output_isel): Delete.
3238 * config/rs6000/rs6000.md (isel_signed_<mode>): Use reg_or_zero_operand
3239 instead of reg_or_cint_operand. Output instruction directly (not via
3241 (isel_unsigned_<mode>): Ditto.
3242 (*isel_reversed_signed_<mode>): Use reg_or_zero_operand instead of
3243 gpc_reg_operand. Add an instruction alternative for this. Output
3244 instruction directly.
3245 (*isel_reversed_unsigned_<mode>): Ditto.
3247 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3249 * config/i386/i386.c (ix86_canonicalize_comparison): New function.
3250 (TARGET_CANONICALIZE_COMPARISON): Define.
3252 2017-10-11 Qing Zhao <qing.zhao@oracle.com>
3255 * config/aarch64/aarch64.c (aarch64_load_symref_appropriately):
3256 Check whether the dest is REG before adding REG_EQUIV note.
3258 2017-10-11 Vladimir Makarov <vmakarov@redhat.com>
3261 * lra.c (collect_non_operand_hard_regs): Don't ignore operator
3263 * lra-lives.c (bb_killed_pseudos, bb_gen_pseudos): Move up.
3264 (make_hard_regno_born, make_hard_regno_dead): Update
3265 bb_killed_pseudos and bb_gen_pseudos.
3267 2017-10-11 Nathan Sidwell <nathan@acm.org>
3269 * incpath.h (enum incpath_kind): Name enum, prefix values.
3270 (add_path, add_cpp_dir_path, get_added_cpp_dirs): Use incpath_kind.
3271 * incpath.c (heads, tails): Use INC_MAX.
3272 (add_env_var_paths, add_standard_paths): Use incpath_kind.
3273 (merge_include_chains, split_quote_chain,
3274 register_include_chains): Update incpath_kind names.
3275 (add_cpp_dir_path, add_path, get_added_cpp_dirs): Use incpath_kind.
3276 * config/darwin-c.c (add_system_framework_path): Update incpath_kind
3278 (add_framework_path, darwin_register_objc_includes): Likewise.
3279 * config/vms/vms-c.c (vms_c_register_includes): Likewise.
3281 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3283 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
3284 Do not use float_operator operator predicate.
3285 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
3286 * config/i386/predicates.md (float_operator): Remove predicate.
3288 2017-10-11 Uros Bizjak <ubizjak@gmail.com>
3290 * config/i386/i386.md (*jcc<mode>_0_i387): Remove insn pattern.
3291 (*jccxf_i387): Ditto.
3292 (*jcc<mode>_i387): Ditto.
3293 (*jccu<mode>_i387): Ditto.
3294 (*jcc<X87MODEF:mode>_<SWI24:mode>_i387): Ditto.
3295 (*jcc_*_i387 splitters): Remove.
3296 * config/i386/i386-protos.h (ix86_split_fp_branch): Remove prototype.
3297 * config/i386/i386.c (ix86_split_fp_branch): Remove.
3298 * config/i386/predicates.md (ix86_swapped_fp_comparison_operator):
3301 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3303 * profile-count.h (slow_safe_scale_64bit): New function.
3304 (safe_scale_64bit): New inline.
3305 (profile_count::max_safe_multiplier): Remove; use safe_scale_64bit.
3306 * profile-count.c: Include wide-int.h
3307 (slow_safe_scale_64bit): New.
3309 2017-10-11 Nathan Sidwell <nathan@acm.org>
3311 * tree.h (DECL_ASSEMBLER_NAME_SET_P): Don't check
3312 HAS_DECL_ASSEMBLER_NAME_P.
3313 * gimple-expr.c (gimple_decl_printable_name: Check
3314 HAS_DECL_ASSEMBLER_NAME_P too.
3315 * ipa-utils.h (type_in_anonymous_namespace_p): Check
3316 DECL_ASSEMBLER_NAME_SET_P of TYPE_NAME.
3317 (odr_type_p): No need to assert TYPE_NAME is a TYPE_DECL.
3318 * passes.c (rest_of_decl_compilation): Check
3319 HAS_DECL_ASSEMBLER_NAME_P too.
3320 * recog.c (verify_changes): Likewise.
3321 * tree-pretty-print.c (dump_decl_name): Likewise.
3322 * tree-ssa-structalias.c (alias_get_name): Likewise. Reimplement.
3324 * tree.h (DECL_ASSEMBLER_NAME_RAW): New.
3325 (SET_DECL_ASSEMBLER_NAME): Use it.
3326 (DECL_ASSEMBLER_NAME_SET_P): Likewise.
3327 (COPY_DECL_ASSEMBLER_NAME): Likewise.
3328 * tree.c (decl_assembler_name): Use DECL_ASSEMBLER_NAME_RAW.
3330 2017-10-11 Jan Hubicka <hubicka@ucw.cz>
3332 * config.gcc (i386, x86_64): Add extra objects.
3333 * config/i386/i386-protos.h (ix86_rip_relative_addr_p): Declare.
3334 (ix86_min_insn_size): Declare.
3335 (ix86_issue_rate): Declare.
3336 (ix86_adjust_cost): Declare.
3337 (ia32_multipass_dfa_lookahead): Declare.
3338 (ix86_macro_fusion_p): Declare.
3339 (ix86_macro_fusion_pair_p): Declare.
3340 (ix86_bd_has_dispatch): Declare.
3341 (ix86_bd_do_dispatch): Declare.
3342 (ix86_core2i7_init_hooks): Declare.
3343 (ix86_atom_sched_reorder): Declare.
3344 * config/i386/i386.c Move all CPU cost tables to x86-tune-costs.h.
3345 (COSTS_N_BYTES): Move to x86-tune-costs.h.
3346 (DUMMY_STRINGOP_ALGS):Move to x86-tune-costs.h.
3347 (rip_relative_addr_p): Rename to ...
3348 (ix86_rip_relative_addr_p): ... this one; export.
3349 (memory_address_length): Update.
3350 (ix86_issue_rate): Move to x86-tune-sched.c.
3351 (ix86_flags_dependent): Move to x86-tune-sched.c.
3352 (ix86_agi_dependent): Move to x86-tune-sched.c.
3353 (exact_dependency_1): Move to x86-tune-sched.c.
3354 (exact_store_load_dependency): Move to x86-tune-sched.c.
3355 (ix86_adjust_cost): Move to x86-tune-sched.c.
3356 (ia32_multipass_dfa_lookahead): Move to x86-tune-sched.c.
3357 (ix86_macro_fusion_p): Move to x86-tune-sched.c.
3358 (ix86_macro_fusion_pair_p): Move to x86-tune-sched.c.
3359 (do_reorder_for_imul): Move to x86-tune-sched-atom.c.
3360 (swap_top_of_ready_list): Move to x86-tune-sched-atom.c.
3361 (ix86_sched_reorder): Move to x86-tune-sched-atom.c.
3362 (core2i7_first_cycle_multipass_init): Move to x86-tune-sched-core.c.
3363 (core2i7_dfa_post_advance_cycle): Move to x86-tune-sched-core.c.
3364 (min_insn_size): Rename to ...
3365 (ix86_min_insn_size): ... this one; export.
3366 (core2i7_first_cycle_multipass_begin): Move to x86-tune-sched-core.c.
3367 (core2i7_first_cycle_multipass_issue): Move to x86-tune-sched-core.c.
3368 (core2i7_first_cycle_multipass_backtrack): Move to
3369 x86-tune-sched-core.c.
3370 (core2i7_first_cycle_multipass_end): Move to x86-tune-sched-core.c.
3371 (core2i7_first_cycle_multipass_fini): Move to x86-tune-sched-core.c.
3372 (ix86_sched_init_global): Break up logic to ix86_core2i7_init_hooks.
3373 (ix86_avoid_jump_mispredicts): Update.
3374 (TARGET_SCHED_DISPATCH): Move to ix86-tune-sched-bd.c.
3375 (TARGET_SCHED_DISPATCH_DO): Move to ix86-tune-sched-bd.c.
3376 (TARGET_SCHED_REORDER): Move to ix86-tune-sched-bd.c.
3377 (DISPATCH_WINDOW_SIZE): Move to ix86-tune-sched-bd.c.
3378 (MAX_DISPATCH_WINDOWS): Move to ix86-tune-sched-bd.c.
3379 (MAX_INSN): Move to ix86-tune-sched-bd.c.
3380 (MAX_IMM): Move to ix86-tune-sched-bd.c.
3381 (MAX_IMM_SIZE): Move to ix86-tune-sched-bd.c.
3382 (MAX_IMM_32): Move to ix86-tune-sched-bd.c.
3383 (MAX_IMM_64): Move to ix86-tune-sched-bd.c.
3384 (MAX_LOAD): Move to ix86-tune-sched-bd.c.
3385 (MAX_STORE): Move to ix86-tune-sched-bd.c.
3386 (BIG): Move to ix86-tune-sched-bd.c.
3387 (enum dispatch_group): Move to ix86-tune-sched-bd.c.
3388 (enum insn_path): Move to ix86-tune-sched-bd.c.
3389 (get_mem_group): Move to ix86-tune-sched-bd.c.
3390 (is_cmp): Move to ix86-tune-sched-bd.c.
3391 (dispatch_violation): Move to ix86-tune-sched-bd.c.
3392 (is_branch): Move to ix86-tune-sched-bd.c.
3393 (is_prefetch): Move to ix86-tune-sched-bd.c.
3394 (init_window): Move to ix86-tune-sched-bd.c.
3395 (allocate_window): Move to ix86-tune-sched-bd.c.
3396 (init_dispatch_sched): Move to ix86-tune-sched-bd.c.
3397 (is_end_basic_block): Move to ix86-tune-sched-bd.c.
3398 (process_end_window): Move to ix86-tune-sched-bd.c.
3399 (allocate_next_window): Move to ix86-tune-sched-bd.c.
3400 (find_constant): Move to ix86-tune-sched-bd.c.
3401 (get_num_immediates): Move to ix86-tune-sched-bd.c.
3402 (has_immediate): Move to ix86-tune-sched-bd.c.
3403 (get_insn_path): Move to ix86-tune-sched-bd.c.
3404 (get_insn_group): Move to ix86-tune-sched-bd.c.
3405 (count_num_restricted): Move to ix86-tune-sched-bd.c.
3406 (fits_dispatch_window): Move to ix86-tune-sched-bd.c.
3407 (add_insn_window): Move to ix86-tune-sched-bd.c.
3408 (add_to_dispatch_window): Move to ix86-tune-sched-bd.c.
3409 (debug_dispatch_window_file): Move to ix86-tune-sched-bd.c.
3410 (debug_dispatch_window): Move to ix86-tune-sched-bd.c.
3411 (debug_insn_dispatch_info_file): Move to ix86-tune-sched-bd.c.
3412 (debug_ready_dispatch): Move to ix86-tune-sched-bd.c.
3413 (do_dispatch): Move to ix86-tune-sched-bd.c.
3414 (has_dispatch): Move to ix86-tune-sched-bd.c.
3415 * config/i386/t-i386: Add new object files.
3416 * config/i386/x86-tune-costs.h: New file.
3417 * config/i386/x86-tune-sched-atom.c: New file.
3418 * config/i386/x86-tune-sched-bd.c: New file.
3419 * config/i386/x86-tune-sched-core.c: New file.
3420 * config/i386/x86-tune-sched.c: New file.
3422 2017-10-11 Liu Hao <lh_mouse@126.com>
3424 * pretty-print.c [_WIN32] (colorize_init): Remove. Use
3425 the generic version below instead.
3426 (should_colorize): Recognize Windows consoles as terminals
3428 * pretty-print.c [__MINGW32__] (write_all): New function.
3429 [__MINGW32__] (find_esc_head): Likewise.
3430 [__MINGW32__] (find_esc_terminator): Likewise.
3431 [__MINGW32__] (eat_esc_sequence): Likewise.
3432 [__MINGW32__] (mingw_ansi_fputs): New function that handles
3434 (pp_write_text_to_stream): Use mingw_ansi_fputs instead of fputs
3437 2017-10-11 Richard Biener <rguenther@suse.de>
3439 * tree-ssa-loop-niter.c (infer_loop_bounds_from_pointer_arith):
3440 Properly call analyze_scalar_evolution with the loop of the stmt.
3442 2017-10-11 Richard Biener <rguenther@suse.de>
3444 * tree.def (POLYNOMIAL_CHREC): Remove CHREC_VARIABLE tree operand.
3445 * tree-core.h (tree_base): Add chrec_var union member.
3446 * tree.h (CHREC_VAR): Remove.
3447 (CHREC_LEFT, CHREC_RIGHT, CHREC_VARIABLE): Adjust.
3448 * tree-chrec.h (build_polynomial_chrec): Adjust.
3449 * tree-chrec.c (reset_evolution_in_loop): Use build_polynomial_chrec.
3450 * tree-pretty-print.c (dump_generic_node): Use CHREC_VARIABLE.
3452 2017-10-11 Marc Glisse <marc.glisse@inria.fr>
3454 * fold-const.c (fold_binary_loc) [X +- Y CMP X]: Move ...
3455 * match.pd: ... here.
3456 ((T) X == (T) Y): Relax condition.
3458 2017-10-11 Bin Cheng <bin.cheng@arm.com>
3460 PR tree-optimization/82472
3461 * tree-loop-distribution.c (sort_partitions_by_post_order): Refine
3463 (break_alias_scc_partitions): Update postorder number.
3465 2017-10-11 Martin Liska <mliska@suse.cz>
3468 * opts.c (parse_no_sanitize_attribute): Do not use error_value
3470 * opts.h (parse_no_sanitize_attribute): Remove last argument.
3472 2017-10-11 Martin Liska <mliska@suse.cz>
3474 * print-rtl.c (print_insn): Move declaration of idbuf
3475 to same scope as name.
3477 2017-10-11 Martin Liska <mliska@suse.cz>
3482 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3485 2017-10-11 Martin Liska <mliska@suse.cz>
3488 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
3491 2017-10-11 Adhemerval Zanella <adhemerval.zanella@linaro.org>
3493 * config.gcc (default_gnu_indirect_function): Default to yes for
3494 arm*-*-linux* with glibc.
3496 2017-10-11 Richard Biener <rguenther@suse.de>
3498 * tree-scalar-evolution.c (get_scalar_evolution): Handle
3499 default-defs and types we do not want to analyze.
3500 (interpret_loop_phi): Replace unreachable code with an assert.
3501 (compute_scalar_evolution_in_loop): Remove and inline ...
3502 (analyze_scalar_evolution_1): ... here, replacing condition with
3503 what makes the intent clearer. Remove handling of cases
3504 get_scalar_evolution now handles.
3506 2017-10-10 Jim Wilson <wilson@tuliptree.org>
3508 PR rtl-optimization/81434
3509 * haifa-sched.c (prune_ready_list): Init min_cost_group to 0. Update
3510 comment for main loop. In sched_group_found if, also add checks for
3511 pass and min_cost_group.
3513 2017-10-10 Segher Boessenkool <segher@kernel.crashing.org>
3515 * config/rs6000/rs6000.c (TARGET_INSN_COST): New.
3516 (rs6000_insn_cost): New function.
3517 * config/rs6000/rs6000.md (cost): New attribute.
3519 2017-10-10 Jakub Jelinek <jakub@redhat.com>
3520 H.J. Lu <hongjiu.lu@intel.com>
3524 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Add
3525 OPTION_MASK_ISA_MMX for __builtin_ia32_maskmovq,
3526 __builtin_ia32_vec_ext_v4hi and __builtin_ia32_vec_set_v4hi.
3527 (ix86_expand_builtin): Treat OPTION_MASK_ISA_MMX similarly
3528 to OPTION_MASK_ISA_AVX512VL - builtins that have both
3529 OPTION_MASK_ISA_MMX and some other bit set require both
3530 mmx and the ISAs without the mmx bit.
3531 * config/i386/i386-builtin.def (__builtin_ia32_cvtps2pi,
3532 __builtin_ia32_cvttps2pi, __builtin_ia32_cvtpi2ps,
3533 __builtin_ia32_pavgb, __builtin_ia32_pavgw, __builtin_ia32_pmulhuw,
3534 __builtin_ia32_pmaxub, __builtin_ia32_pmaxsw, __builtin_ia32_pminub,
3535 __builtin_ia32_pminsw, __builtin_ia32_psadbw, __builtin_ia32_pmovmskb,
3536 __builtin_ia32_pshufw, __builtin_ia32_cvtpd2pi,
3537 __builtin_ia32_cvttpd2pi, __builtin_ia32_cvtpi2pd,
3538 __builtin_ia32_pmuludq, __builtin_ia32_pabsb, __builtin_ia32_pabsw,
3539 __builtin_ia32_pabsd, __builtin_ia32_phaddw, __builtin_ia32_phaddd,
3540 __builtin_ia32_phaddsw, __builtin_ia32_phsubw, __builtin_ia32_phsubd,
3541 __builtin_ia32_phsubsw, __builtin_ia32_pmaddubsw,
3542 __builtin_ia32_pmulhrsw, __builtin_ia32_pshufb, __builtin_ia32_psignb,
3543 __builtin_ia32_psignw, __builtin_ia32_psignd, __builtin_ia32_movntq,
3544 __builtin_ia32_paddq, __builtin_ia32_psubq, __builtin_ia32_palignr):
3545 Add OPTION_MASK_ISA_MMX.
3547 2017-10-10 Andreas Tobler <andreast@gcc.gnu.org>
3549 * config.gcc (armv7*-*-freebsd*): New target.
3550 (armv6*-*-freebsd*): Remove obsolete TARGET_FREEBSD_ARMv6 define.
3552 2017-10-10 Jan Hubicka <hubicka@ucw.cz>
3554 * x86-tune.def (X86_TUNE_AVOID_FALSE_DEP_FOR_BMI,
3555 X86_TUNE_ADJUST_UNROLL, X86_TUNE_ONE_IF_CONV_INSN): Move to right
3558 2017-10-10 Richard Sandiford <richard.sandiford@linaro.org>
3560 * wide-int.h (wide_int_ref_storage): Make host_dependent_precision
3561 a template parameter.
3562 (WIDE_INT_REF_FOR): Update accordingly.
3563 * tree.h (wi::int_traits <const_tree>): Delete.
3564 (wi::tree_to_widest_ref, wi::tree_to_offset_ref): New typedefs.
3565 (wi::to_widest, wi::to_offset): Use them. Expand commentary.
3566 (wi::tree_to_wide_ref): New typedef.
3567 (wi::to_wide): New function.
3568 * calls.c (get_size_range): Use wi::to_wide when operating on
3570 * cgraph.c (cgraph_node::create_thunk): Likewise.
3571 * config/i386/i386.c (ix86_data_alignment): Likewise.
3572 (ix86_local_alignment): Likewise.
3573 * dbxout.c (stabstr_O): Likewise.
3574 * dwarf2out.c (add_scalar_info, gen_enumeration_type_die): Likewise.
3575 * expr.c (const_vector_from_tree): Likewise.
3576 * fold-const-call.c (host_size_t_cst_p, fold_const_call_1): Likewise.
3577 * fold-const.c (may_negate_without_overflow_p, negate_expr_p)
3578 (fold_negate_expr_1, int_const_binop_1, const_binop)
3579 (fold_convert_const_int_from_real, optimize_bit_field_compare)
3580 (all_ones_mask_p, sign_bit_p, unextend, extract_muldiv_1)
3581 (fold_div_compare, fold_single_bit_test, fold_plusminus_mult_expr)
3582 (pointer_may_wrap_p, expr_not_equal_to, fold_binary_loc)
3583 (fold_ternary_loc, multiple_of_p, fold_negate_const, fold_abs_const)
3584 (fold_not_const, round_up_loc): Likewise.
3585 * gimple-fold.c (gimple_fold_indirect_ref): Likewise.
3586 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Likewise.
3587 (alloca_call_type): Likewise.
3588 * gimple.c (preprocess_case_label_vec_for_gimple): Likewise.
3589 * godump.c (go_output_typedef): Likewise.
3590 * graphite-sese-to-poly.c (tree_int_to_gmp): Likewise.
3591 * internal-fn.c (get_min_precision): Likewise.
3592 * ipa-cp.c (ipcp_store_vr_results): Likewise.
3593 * ipa-polymorphic-call.c
3594 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Likewise.
3595 * ipa-prop.c (ipa_print_node_jump_functions_for_edge): Likewise.
3596 (ipa_modify_call_arguments): Likewise.
3597 * match.pd: Likewise.
3598 * omp-low.c (scan_omp_1_op, lower_omp_ordered_clauses): Likewise.
3599 * print-tree.c (print_node_brief, print_node): Likewise.
3600 * stmt.c (expand_case): Likewise.
3601 * stor-layout.c (layout_type): Likewise.
3602 * tree-affine.c (tree_to_aff_combination): Likewise.
3603 * tree-cfg.c (group_case_labels_stmt): Likewise.
3604 * tree-data-ref.c (dr_analyze_indices): Likewise.
3605 (prune_runtime_alias_test_list): Likewise.
3606 * tree-dump.c (dequeue_and_dump): Likewise.
3607 * tree-inline.c (remap_gimple_op_r, copy_tree_body_r): Likewise.
3608 * tree-predcom.c (is_inv_store_elimination_chain): Likewise.
3609 * tree-pretty-print.c (dump_generic_node): Likewise.
3610 * tree-scalar-evolution.c (iv_can_overflow_p): Likewise.
3611 (simple_iv_with_niters): Likewise.
3612 * tree-ssa-address.c (addr_for_mem_ref): Likewise.
3613 * tree-ssa-ccp.c (ccp_finalize, evaluate_stmt): Likewise.
3614 * tree-ssa-loop-ivopts.c (constant_multiple_of): Likewise.
3615 * tree-ssa-loop-niter.c (split_to_var_and_offset)
3616 (refine_value_range_using_guard, number_of_iterations_ne_max)
3617 (number_of_iterations_lt_to_ne, number_of_iterations_lt)
3618 (get_cst_init_from_scev, record_nonwrapping_iv)
3619 (scev_var_range_cant_overflow): Likewise.
3620 * tree-ssa-phiopt.c (minmax_replacement): Likewise.
3621 * tree-ssa-pre.c (compute_avail): Likewise.
3622 * tree-ssa-sccvn.c (vn_reference_fold_indirect): Likewise.
3623 (vn_reference_maybe_forwprop_address, valueized_wider_op): Likewise.
3624 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Likewise.
3625 * tree-ssa-uninit.c (is_pred_expr_subset_of): Likewise.
3626 * tree-ssanames.c (set_nonzero_bits, get_nonzero_bits): Likewise.
3627 * tree-switch-conversion.c (collect_switch_conv_info, array_value_type)
3628 (dump_case_nodes, try_switch_expansion): Likewise.
3629 * tree-vect-loop-manip.c (vect_gen_vector_loop_niters): Likewise.
3630 (vect_do_peeling): Likewise.
3631 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
3632 * tree-vect-stmts.c (vectorizable_load): Likewise.
3633 * tree-vrp.c (compare_values_warnv, vrp_int_const_binop): Likewise.
3634 (zero_nonzero_bits_from_vr, ranges_from_anti_range): Likewise.
3635 (extract_range_from_binary_expr_1, adjust_range_with_scev): Likewise.
3636 (overflow_comparison_p_1, register_edge_assert_for_2): Likewise.
3637 (is_masked_range_test, find_switch_asserts, maybe_set_nonzero_bits)
3638 (vrp_evaluate_conditional_warnv_with_ops, intersect_ranges): Likewise.
3639 (range_fits_type_p, two_valued_val_range_p, vrp_finalize): Likewise.
3640 (evrp_dom_walker::before_dom_children): Likewise.
3641 * tree.c (cache_integer_cst, real_value_from_int_cst, integer_zerop)
3642 (integer_all_onesp, integer_pow2p, integer_nonzerop, tree_log2)
3643 (tree_floor_log2, tree_ctz, mem_ref_offset, tree_int_cst_sign_bit)
3644 (tree_int_cst_sgn, get_unwidened, int_fits_type_p): Likewise.
3645 (get_type_static_bounds, num_ending_zeros, drop_tree_overflow)
3646 (get_range_pos_neg): Likewise.
3647 * ubsan.c (ubsan_expand_ptr_ifn): Likewise.
3648 * config/darwin.c (darwin_mergeable_constant_section): Likewise.
3649 * config/aarch64/aarch64.c (aapcs_vfp_sub_candidate): Likewise.
3650 * config/arm/arm.c (aapcs_vfp_sub_candidate): Likewise.
3651 * config/avr/avr.c (avr_fold_builtin): Likewise.
3652 * config/bfin/bfin.c (bfin_local_alignment): Likewise.
3653 * config/msp430/msp430.c (msp430_attr): Likewise.
3654 * config/nds32/nds32.c (nds32_insert_attributes): Likewise.
3655 * config/powerpcspe/powerpcspe-c.c
3656 (altivec_resolve_overloaded_builtin): Likewise.
3657 * config/powerpcspe/powerpcspe.c (rs6000_aggregate_candidate)
3658 (rs6000_expand_ternop_builtin): Likewise.
3659 * config/rs6000/rs6000-c.c
3660 (altivec_resolve_overloaded_builtin): Likewise.
3661 * config/rs6000/rs6000.c (rs6000_aggregate_candidate): Likewise.
3662 (rs6000_expand_ternop_builtin): Likewise.
3663 * config/s390/s390.c (s390_handle_hotpatch_attribute): Likewise.
3665 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3667 * tree-vect-loop-manip.c (rename_variables_in_bb): Rename PHI nodes
3668 when copying loop nest with only one inner loop.
3670 2017-10-10 Richard Biener <rguenther@suse.de>
3672 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Avoid compacting
3673 blocks if SCEV is active.
3674 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Remove
3676 (analyze_scalar_evolution): Handle cached evolutions the obvious way.
3677 (scev_initialize): Assert we are not yet initialized.
3679 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3681 * tree-loop-distribution.c (generate_loops_for_partition): Remove
3682 inner loop's exit stmt by making it always exit the loop, otherwise
3683 we would generate an infinite empty loop.
3685 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3687 * tree-vect-loop-manip.c (slpeel_tree_duplicate_loop_to_edge_cfg): Skip
3688 renaming variables in new preheader if it's deleted.
3690 2017-10-10 Bin Cheng <bin.cheng@arm.com>
3692 * tree-loop-distribution.c (struct partition): Remove unused field
3693 loops of the structure.
3694 (partition_alloc, partition_free): Ditto.
3695 (build_rdg_partition_for_vertex): Ditto.
3697 2017-10-09 Jeff Law <law@redhat.com>
3699 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): Fix
3700 return type to match prototype and documentation.
3702 2010-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3704 * config/rs6000/rs6000.c (processor_costs): Move to ...
3705 * config/rs6000/rs6000.h: ... here.
3706 (rs6000_cost): Declare.
3708 2017-10-09 Eric Botcazou <ebotcazou@adacore.com>
3710 * except.c (setjmp_fn): New global variable.
3711 (init_eh): Initialize it if DONT_USE_BUILTIN_SETJMP is defined.
3712 (sjlj_emit_function_enter): Call it instead of BUILTIN_SETJMP
3713 if DONT_USE_BUILTIN_SETJMP is defined.
3715 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3717 * target.def (insn_cost): New hook.
3718 * doc/tm.texi.in (TARGET_INSN_COST): New hook.
3719 * doc/tm.texi: Regenerate.
3720 * rtlanal.c (insn_cost): Use the new hook.
3722 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3724 * combine.c (combine_validate_cost): Compute the new insn_cost,
3725 not just pattern_cost.
3726 (try_combine): Adjust comment.
3728 2017-10-09 Segher Boessenkool <segher@kernel.crashing.org>
3730 * cfgrtl.c (rtl_account_profile_record): Replace insn_rtx_cost with
3732 * combine.c (uid_insn_cost): Adjust comment.
3733 (combine_validate_cost): Adjust comment. Use pattern_cost instead
3735 (combine_instructions): Use insn_cost instead of insn_rtx_cost.
3736 * dse.c (find_shift_sequence): Ditto.
3737 * ifcvt.c (cheap_bb_rtx_cost_p): Ditto.
3738 (bb_valid_for_noce_process_p): Use pattern_cost.
3739 * rtl.h (insn_rtx_cost): Delete.
3740 (pattern_cost): New prototype.
3741 (insn_cost): New prototype.
3742 * rtlanal.c (insn_rtx_cost): Rename to...
3743 (pattern_cost): ... this.
3746 2017-10-09 Uros Bizjak <ubizjak@gmail.com>
3748 * config/i386/i386.md (*jcc_2): Remove insn pattern.
3749 (*jcc<mode>_0_r_i387): Ditto.
3750 (*jccxf_r_i387): Ditto.
3751 (*jcc<mode>_r_i387): Ditto.
3752 (*jccu<mode>_r_i387): Ditto.
3753 (*jcc<X87MODEF:mode>_<SWI24:mode>_r_i387): Ditto.
3754 (*jcc): Rename from *jcc_1.
3756 2017-10-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3758 * config/rs6000/rs6000-p8swap.c (rs6000_analyze_swaps): Process
3759 deferred rescans after the lvx/stvx recombination pre-pass.
3761 2017-10-09 Michael Meissner <meissner@linux.vnet.ibm.com>
3763 * config/rs6000/amo.h: New include file to provide ISA 3.0 atomic
3764 memory operation instruction support.
3765 * config.gcc (powerpc*-*-*): Include amo.h as an extra header.
3766 (rs6000-ibm-aix[789]*): Likewise.
3767 * doc/extend.texi (PowerPC Atomic Memory Operation Functions):
3768 Document new functions.
3770 2017-10-09 Richard Biener <rguenther@suse.de>
3772 PR tree-optimization/82397
3773 * tree-data-ref.c (data_ref_compare_tree): Make sure to return
3774 equality only for semantically equal trees.
3776 2017-10-09 Richard Biener <rguenther@suse.de>
3778 PR tree-optimization/82449
3779 * sese.c (scev_analyzable_p): Check whether the SCEV is linear.
3780 * tree-chrec.h (evolution_function_is_constant_p): Adjust to
3781 allow constant addresses.
3782 * tree-chrec.c (scev_is_linear_expression): Constant evolutions
3785 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3787 * config/s390/s390-builtins.def (vec_nabs, vec_vfi): Fix builtin
3790 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3793 * config/s390/vecintrin.h (vec_madd, vec_msub): Fix macro
3796 2017-10-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3799 * config/s390/s390-builtins.def (vec_sqrt): Fix builtin flags.
3801 2017-10-09 Jakub Jelinek <jakub@redhat.com>
3804 * config/s390/s390-builtins.def (s390_vec_xor_flt_a,
3805 s390_vec_xor_flt_b, s390_vec_xor_flt_c): New.
3807 2017-10-09 Richard Sandiford <richard.sandiford@linaro.org>
3809 * wide-int.h (WI_BINARY_OPERATOR_RESULT): New macro.
3810 (WI_BINARY_PREDICATE_RESULT): Likewise.
3811 (wi::binary_traits::operator_result): New type.
3812 (wi::binary_traits::predicate_result): Likewise.
3813 (generic_wide_int::operator~, unary generic_wide_int::operator-)
3814 (generic_wide_int::operator==, generic_wide_int::operator!=)
3815 (generic_wide_int::operator&, generic_wide_int::and_not)
3816 (generic_wide_int::operator|, generic_wide_int::or_not)
3817 (generic_wide_int::operator^, generic_wide_int::operator+
3818 (binary generic_wide_int::operator-, generic_wide_int::operator*):
3820 (operator~, unary operator-, operator==, operator!=, operator&)
3821 (operator|, operator^, operator+, binary operator-, operator*): New
3823 * expr.c (get_inner_reference): Use wi::bit_and_not.
3824 * fold-const.c (fold_binary_loc): Likewise.
3825 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Likewise.
3826 * tree-ssa-ccp.c (get_value_from_alignment): Likewise.
3827 (bit_value_binop): Likewise.
3828 * tree-ssa-math-opts.c (find_bswap_or_nop_load): Likewise.
3829 * tree-vrp.c (zero_nonzero_bits_from_vr): Likewise.
3830 (extract_range_from_binary_expr_1): Likewise.
3831 (masked_increment): Likewise.
3832 (simplify_bit_ops_using_ranges): Likewise.
3834 2017-10-09 Martin Jambor <mjambor@suse.cz>
3837 * hsa-common.h (hsa_op_with_type): New method extend_int_to_32bit.
3838 * hsa-gen.c (hsa_extend_inttype_to_32bit): New function.
3839 (hsa_type_for_scalar_tree_type): Use it. Always force min32int for
3841 (hsa_fixup_mov_insn_type): New function.
3842 (hsa_op_with_type::get_in_type): Use it.
3843 (hsa_build_append_simple_mov): Likewise. Allow sub-32bit
3844 immediates in an assert.
3845 (hsa_op_with_type::extend_int_to_32bit): New method.
3846 (gen_hsa_insns_for_bitfield): Fixup instruction and intermediary
3847 types. Convert to dest type if necessary.
3848 (gen_hsa_insns_for_bitfield_load): Fixup load type if necessary.
3849 (reg_for_gimple_ssa): Pass false as min32int to
3850 hsa_type_for_scalar_tree_type.
3851 (gen_hsa_addr): Fixup type when creating addresable temporary.
3852 (gen_hsa_cmp_insn_from_gimple): Extend operands if necessary.
3853 (gen_hsa_unary_operation): Extend operands and convert to dest type if
3854 necessary. Call hsa_fixup_mov_insn_type.
3855 (gen_hsa_binary_operation): Changed operand types to hsa_op_with_type,
3856 extend operands and convert to dest type if necessary.
3857 (gen_hsa_insns_for_operation_assignment): Extend operands and convert
3858 to dest type if necessary.
3859 (set_output_in_type): Call hsa_fixup_mov_insn_type. Just ude dest
3860 if conversion nt necessary and size matches.
3861 (gen_hsa_insns_for_load): Call hsa_fixup_mov_insn_type, convert
3862 to dest type if necessary.
3863 (gen_hsa_insns_for_store): Call hsa_fixup_mov_insn_type.
3864 (gen_hsa_insns_for_switch_stmt): Likewise. Also extend operands if
3866 (gen_hsa_clrsb): Likewise.
3867 (gen_hsa_ffs): Likewise.
3868 (gen_hsa_divmod): Extend operands and convert to dest type if
3870 (gen_hsa_atomic_for_builtin): Change type of op to hsa_op_with_type.
3872 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3874 * config/rs6000/rs6000.md (conditional branch): Clean up formatting.
3875 Remove empty default arguments. Use a brace block as output
3877 (conditional return): Ditto.
3879 (indirect_jump): Ditto. Use b%T0 instead of bctr/blr.
3880 (tablejump, tablejumpsi, tablejumpdi, *tablejump<mode>_internal1):
3882 (group_ending_nop): Ditto.
3883 (doloop_end): Ditto.
3884 (ctr<mode>, ctr<mode>_internal1, ctr<mode>_internal2): Ditto.
3885 (splitters for those): Ditto.
3887 2017-10-08 Segher Boessenkool <segher@kernel.crashing.org>
3889 * config/rs6000/rs6000-string.c (expand_strncmp_align_check): Invert
3890 a conditional jump (and the compare for it) so that pc_rtx is the
3892 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Adjust
3893 for the deleted and renamed ctr<mode>_internal[234] patterns.
3894 * config/rs6000/rs6000.md: Delete second conditional branch pattern.
3895 Delete second conditional return pattern.
3896 (ctr<mode>_internal2): Delete this second bdnz pattern.
3897 (ctr<mode>_internal3): Rename to ctr<mode>_internal2.
3898 (ctr<mode>_internal4): Delete this second bdz pattern.
3900 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3902 * tree-outof-ssa.h (ssaexpand): Add partitions_for_undefined_values.
3903 (always_initialized_rtx_for_ssa_name_p): New predicate.
3904 * tree-outof-ssa.c (remove_ssa_form): Initialize new field of SA.
3905 (finish_out_of_ssa): Free new field of SA.
3906 * tree-ssa-coalesce.h (get_undefined_value_partitions): Declare.
3907 * tree-ssa-coalesce.c: Include tree-ssa.h.
3908 (get_parm_default_def_partitions): Remove extern keyword.
3909 (get_undefined_value_partitions): New function.
3910 * expr.c (expand_expr_real_1) <expand_decl_rtl>: For a SSA_NAME, do
3911 not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
3913 * loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
3915 2017-10-08 Eric Botcazou <ebotcazou@adacore.com>
3917 * builtins.def (BUILT_IN_SETJMP): Revert latest change.
3919 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3921 * config/i386/i386.c (ix86_expand_set_or_movmem): Disable 512bit loops
3922 for targets that preffer 128bit.
3924 2017-10-08 Jan Hubicka <hubicka@ucw.cz>
3926 * config/i386/i386.c (has_dispatch): Disable for Ryzen.
3928 2017-10-08 Olivier Hainque <hainque@adacore.com>
3930 * config/arm/arm.c (arm_set_return_address): Use MEM_VOLATILE_P
3931 on the target mem instead of RTX_FRAME_RELATED_P on the insn to
3933 (thumb_set_return_address): Likewise.
3935 2017-10-08 Olivier Hainque <hainque@adacore.com>
3937 * common/config/arm/arm-common.c (arm_except_unwind_info):
3938 Handle DWARF2_UNWIND_INFO.
3940 2017-10-07 Michael Collison <michael.collison@arm.com>
3942 * config/aarch64/aarch64.md (*aarch64_reg_<optab>_minus<mode>3):
3945 2017-10-07 Eric Botcazou <ebotcazou@adacore.com>
3947 * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
3948 of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
3949 * except.c (sjlj_emit_function_enter): If DONT_USE_BUILTIN_SETJMP is
3950 defined, force the creation of a new block for a dispatch label.
3952 2017-10-07 Jan Hubicka <hubicka@ucw.cz>
3954 * invoke.texi (Wsuggest-attribute=cold): Document.
3955 * common.opt (Wsuggest-attribute=cold): New
3956 * ipa-pure-const.c (warn_function_cold): New function.
3957 * predict.c (compute_function_frequency): Use it.
3958 * predict.h (warn_function_cold): Declare.
3960 2017-10-06 Jan Hubicka <hubicka@ucw.cz>
3962 * tree-switch-conversion.c (do_jump_if_equal, emit_cmp_and_jump_insns):
3965 2017-10-06 Martin Liska <mliska@suse.cz>
3967 * sanopt.c (struct sanopt_tree_triplet_hash): Remove inline
3968 keyword for member functions.
3969 (struct sanopt_tree_couple): New struct.
3970 (struct sanopt_tree_couple_hash): New function.
3971 (struct sanopt_ctx): Add new hash_map.
3972 (has_dominating_ubsan_ptr_check): New function.
3973 (record_ubsan_ptr_check_stmt): Likewise.
3974 (maybe_optimize_ubsan_ptr_ifn): Likewise.
3975 (sanopt_optimize_walker): Handle IFN_UBSAN_PTR.
3976 (pass_sanopt::execute): Handle also SANITIZE_POINTER_OVERFLOW.
3978 2017-10-06 Sudakshina Das <sudi.das@arm.com>
3981 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Only call
3982 aarch64_simd_valid_immediate on CONST_VECTORs.
3983 (aarch64_reg_or_bic_imm): Likewise.
3985 2017-10-06 Wilco Dijkstra <wdijkstr@arm.com>
3987 PR rtl-optimization/82396
3988 * haifa-sched.c (ready_sort_real): Disable qsort checking.
3990 2017-10-06 Sebastian Pop <sebpop@gmail.com>
3992 * graphite-dependences.c (scop_get_reads): Move code to...
3993 (scop_get_must_writes): Move code to...
3994 (scop_get_may_writes): Move code to...
3995 (scop_get_reads_and_writes): ... here.
3996 (scop_get_dependences): Call scop_get_reads_and_writes.
3998 2017-10-06 Jakub Jelinek <jakub@redhat.com>
4000 PR tree-optimization/82434
4001 * fold-const.h (can_native_encode_type_p,
4002 can_native_encode_string_p): Remove.
4003 * fold-const.c (native_encode_int): Formatting fixes. If ptr is NULL,
4004 don't encode anything, just return what would be otherwise returned.
4005 (native_encode_fixed, native_encode_complex, native_encode_vector):
4007 (native_encode_string): Likewise. Inline by hand
4008 can_native_encode_string_p.
4009 (can_native_encode_type_p): Remove.
4010 (can_native_encode_string_p): Remove.
4011 * tree-vect-stmts.c (vectorizable_store): Instead of testing just
4012 STRING_CSTs using can_native_encode_string_p, test all
4013 CONSTANT_CLASS_P values using native_encode_expr with NULL ptr.
4014 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Remove last
4015 argument from native_encode_expr.
4016 (rhs_valid_for_store_merging_p): Use native_encode_expr with NULL ptr.
4017 (pass_store_merging::execute): Don't unnecessarily look for 3 stmts,
4020 2017-10-06 Richard Biener <rguenther@suse.de>
4022 PR tree-optimization/82397
4023 * tree-vect-data-refs.c (dr_group_sort_cmp): Do not use
4024 operand_equal_p but rely on data_ref_compare_tree for detecting
4026 (vect_analyze_data_ref_accesses): Use data_ref_compare_tree
4027 to match up with dr_group_sort_cmp.
4029 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4032 * config/s390/s390-builtins.def (s390_vfi): Define new overloaded
4034 * config/s390/s390-builtin-types.def: Regenerate.
4036 2017-10-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4039 * config/s390/s390-builtin-types.def: Regenerate.
4040 * config/s390/s390-builtins.def (s390_vfmaxdb_4, s390_vfmindb_4):
4041 Change flag from B_VXE to B_VX.
4042 (s390_vec_min_dbl): Remove B_VXE flag.
4044 2017-10-06 Richard Biener <rguenther@suse.de>
4046 * graphite-isl-ast-to-gimple.c: Include ssa.h and tree-ssa.h.
4047 (translate_isl_ast_to_gimple::translate_pending_phi_nodes,
4048 translate_isl_ast_to_gimple::is_valid_rename,
4049 translate_isl_ast_to_gimple::get_rename,
4050 translate_isl_ast_to_gimple::get_def_bb_for_const,
4051 translate_isl_ast_to_gimple::get_new_name,
4052 translate_isl_ast_to_gimple::collect_all_ssa_names,
4053 translate_isl_ast_to_gimple::copy_loop_phi_args,
4054 translate_isl_ast_to_gimple::collect_all_ssa_names,
4055 translate_isl_ast_to_gimple::copy_loop_phi_args,
4056 translate_isl_ast_to_gimple::copy_loop_phi_nodes,
4057 translate_isl_ast_to_gimple::add_close_phis_to_merge_points,
4058 translate_isl_ast_to_gimple::add_close_phis_to_outer_loops,
4059 translate_isl_ast_to_gimple::copy_loop_close_phi_args,
4060 translate_isl_ast_to_gimple::copy_loop_close_phi_nodes,
4061 translate_isl_ast_to_gimple::copy_cond_phi_args,
4062 translate_isl_ast_to_gimple::copy_cond_phi_nodes,
4063 translate_isl_ast_to_gimple::edge_for_new_close_phis,
4064 translate_isl_ast_to_gimple::add_phi_arg_for_new_expr,
4065 translate_isl_ast_to_gimple::rename_uses,
4066 translate_isl_ast_to_gimple::rename_all_uses): Remove.
4067 (translate_isl_ast_to_gimple::get_rename_from_scev): Simplify.
4068 (set_rename_for_each_def): Likewise.
4069 (graphite_copy_stmts_from_block): Handle debug stmt resetting
4070 here. Handle rewriting SCEV analyzable uses here.
4071 (copy_bb_and_scalar_dependences): Generate code for PHI
4073 (graphite_regenerate_ast_isl): Adjust.
4074 * graphite-scop-detection.c (trivially_empty_bb_p): Move to sese.[ch].
4075 (add_write, add_read): New functions.
4076 (build_cross_bb_scalars_def): Use it and simplify.
4077 (build_cross_bb_scalars_use): Likewise.
4078 (graphite_find_cross_bb_scalar_vars): Inline into...
4079 (try_generate_gimple_bb): ...here. Add dependences for PHIs,
4080 simulating out-of-SSA. Compute liveout and add dependencies.
4081 (build_scops): Force an empty entry block.
4082 * sese.h (sese_info_t::liveout, sese_info_t::debug_liveout): New
4084 (sese_build_liveouts): Declare.
4085 (sese_trivially_empty_bb_p): Likewise.
4086 * sese.c (sese_build_liveouts_bb): Properly handle PHIs,
4087 compute liveout and debug_liveout.
4088 (sese_bad_liveouts_use): Remove.
4089 (sese_reset_debug_liveouts_bb): Likewise.
4090 (sese_reset_debug_liveouts): Rewrite in terms of debug_liveout.
4091 (sese_build_liveouts): Build liveout and debug_liveout and store
4093 (new_sese_info): Adjust.
4094 (free_sese_info): Likewise.
4095 (sese_insert_phis_for_liveouts): Reset debug stmts from here,
4096 do not build liveout here.
4097 (move_sese_in_condition): Adjust region entry.
4098 (scev_analyzable_p): Match up with chrec_apply requirements.
4099 (sese_trivially_empty_bb_p): New.
4100 * tree-into-ssa.c (get_reaching_def): Properly support generating
4101 default-defs for incremental rewrite of anonymous names.
4103 2017-10-06 Richard Biener <rguenther@suse.de>
4105 * graphite-sese-to-poly.c (extract_affine): For casts increasing
4106 precision do not perform modulo reduction.
4108 2017-10-06 Richard Biener <rguenther@suse.de>
4110 PR tree-optimization/82436
4111 * tree-vect-slp.c (vect_supported_load_permutation_p): More
4112 conservatively choose the vectorization factor when checking
4113 whether we can perform the required load permutation.
4114 (vect_transform_slp_perm_load): Assert when we may not fail.
4116 2017-10-05 Segher Boessenkool <segher@kernel.crashing.org>
4118 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Correct error
4119 message for incompatible -msdata=* and -mcall-* options.
4121 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4123 * config/i386/i386.c (ia32_multipass_dfa_lookahead): Default to issue
4124 rate for post-reload scheduling.
4126 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4128 * doc/sourcebuild.texi (vect_sizes_16B_8B, vect_sizes_32B_16B): New.
4130 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4132 * config/i386/i386.c (znver1_cost): Set branch_cost to 3 (instead of 2)
4133 to improve monte carlo in scimark.
4135 2017-10-05 Jan Hubicka <hubicka@ucw.cz>
4137 * config/i386/i386.c (ix86_size_cost, i386_cost, i486_cost,
4138 pentium_cost, lakemont_cost, pentiumpro_cost, geode_cost, k6_cost,
4139 athlon_cost, k8_cost, amdfam10_cost, btver1_cost, btver2_cost,
4140 pentium4_cost, nocona_cost): Set reassociation width to 1.
4141 (bdver1_cost, bdver2_cost, bdver3_cost, bdver4_cost): Set reassociation
4142 width to 2 for fp operations and 1 otherwise.
4143 (znver1_cost): Set scalar reassoc width to 4 and vector to 3 and 6
4145 (atom_cost): Set reassociation width to 2.
4146 (slm_cost, generic_cost): Set fp reassociation width
4147 to 2 and 1 otherwise.
4148 (intel_cost): Set fp reassociation width to 4 and 1 otherwise.
4149 (core_cost): Set fp reassociation width to 4 and vector to 2.
4150 (ix86_reassociation_width): Rewrite using cost table; special case
4151 plus/minus on Zen; honor X86_TUNE_SSE_SPLIT_REGS
4152 and TARGET_AVX128_OPTIMAL.
4153 * config/i386/i386.h (processor_costs): Add
4154 reassoc_int, reassoc_fp, reassoc_vec_int, reassoc_vec_fp.
4155 (TARGET_VECTOR_PARALLEL_EXECUTION, TARGET_REASSOC_INT_TO_PARALLEL,
4156 TARGET_REASSOC_FP_TO_PARALLEL): Remove.
4157 * x86-tune.def (X86_TUNE_REASSOC_INT_TO_PARALLEL): Remove.
4158 (X86_TUNE_REASSOC_FP_TO_PARALLEL): Remove.
4159 (X86_TUNE_VECTOR_PARALLEL_EXECUTION): Remove.
4161 2017-10-05 Nathan Sidwell <nathan@acm.org>
4163 * doc/invoke.texi (Wparentheses): Document C++ MVP behaviour.
4165 2017-10-05 Tamar Christina <tamar.christina@arm.com>
4167 * config/arm/arm.c (arm_test_fpu_data): New.
4168 (arm_run_selftests): Call arm_test_fpu_data.
4170 2017-10-04 Nathan Sidwell <nathan@acm.org>
4172 * toplev.c (toplev::main): Remove excess parens on pretty_printer
4174 * caller-save.c (insert_save): Remove excess parens on TO_SAVE parm.
4176 2017-10-04 Sudakshina Das <sudi.das@arm.com>
4178 * config/aarch64/aarch64-protos.h (enum simd_immediate_check): New
4179 check type for aarch64_simd_valid_immediate.
4180 (aarch64_output_simd_mov_immediate): Update prototype.
4181 (aarch64_simd_valid_immediate): Update prototype.
4182 * config/aarch64/aarch64-simd.md (orr<mode>3): modified pattern to add
4183 support for ORR-immediate.
4184 (and<mode>3): modified pattern to add support for BIC-immediate.
4185 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Function
4186 now checks for valid immediate for BIC and ORR based on new enum
4188 (aarch64_output_simd_mov_immediate): Function now used to output
4189 BIC/ORR imm as well based on new enum argument.
4190 * config/aarch64/constraints.md (Do): New vector immediate constraint.
4192 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): New predicate.
4193 (aarch64_reg_or_bic_imm): Likewise.
4195 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4197 * config/s390/vx-builtins.md ("vec_mergeh<mode>")
4198 ("vec_mergel<mode>"): Change mode iterator to V_128_NOSINGLE.
4200 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4204 PR rtl-optimization/82396
4205 * haifa-sched.c (autopref_multipass_init): Simplify
4207 (autopref_rank_data): Simplify sort order.
4208 * sched-int.h (autopref_multipass_data_): Remove
4209 multi_mem_insn_p, min_offset and max_offset.
4211 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4213 * doc/sourcebuild.texi: Document vect_peeling_profitable.
4215 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4217 * doc/sourcebuild.texi: Document vect_intdouble_cvt and
4220 2017-10-04 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4222 * doc/sourcebuild.texi: Document vect_long_mult.
4224 2017-10-04 Richard Sandiford <richard.sandiford@linaro.org>
4226 PR tree-optimization/82413
4227 * fold-const.c (build_range_check): Use widest_int when comparing
4228 the maximum ETYPE value with HIGH.
4230 2017-10-04 Wilco Dijkstra <wdijkstr@arm.com>
4232 PR rtl-optimization/82396
4233 * haifa-sched.c (autopref_multipass_init): Simplify
4235 (autopref_rank_data): Simplify sort order.
4236 * sched-int.h (autopref_multipass_data_): Remove
4237 multi_mem_insn_p, min_offset and max_offset.
4239 2017-10-04 Jakub Jelinek <jakub@redhat.com>
4241 PR tree-optimization/82381
4242 * tree-ssa-reassoc.c (sort_by_operand_rank): Check for different
4243 oeN->rank first. Return 1 or -1 if one op is SSA_NAME and the other
4246 PR tree-optimization/82374
4247 * omp-low.c (create_omp_child_function): Copy DECL_ATTRIBUTES,
4248 DECL_FUNCTION_SPECIFIC_OPTIMIZATION,
4249 DECL_FUNCTION_SPECIFIC_TARGET and DECL_FUNCTION_VERSIONED from
4250 current_function_decl to the new decl.
4252 2017-10-03 Michael Meissner <meissner@linux.vnet.ibm.com>
4254 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_2_HW): Define new
4255 helper macro for IEEE float128 hardware built-in functions.
4256 (SQRTF128_ODD): Add built-in functions with the round-to-odd
4258 (TRUNCF128_ODD): Likewise.
4259 (ADDF128_ODD): Likewise.
4260 (SUBF128_ODD): Likewise.
4261 (MULF128_ODD): Likewise.
4262 (DIVF128_ODD): Likewise.
4263 (FMAF128_ODD): Likewise.
4264 * config/rs6000/rs6000.md (UNSPEC_ROUND_TO_ODD): Rename to
4265 UNSPEC_TRUNC_ROUND_TO_ODD.
4266 (UNSPEC_TRUNC_ROUND_TO_ODD): Likewise.
4267 (UNSPEC_ADD_ROUND_TO_ODD): New unspec codes for the IEEE 128-bit
4268 floating point round to odd instructions.
4269 (UNSPEC_SUB_ROUND_TO_ODD): Likewise.
4270 (UNSPEC_MUL_ROUND_TO_ODD): Likewise.
4271 (UNSPEC_DIV_ROUND_TO_ODD): Likewise.
4272 (UNSPEC_FMA_ROUND_TO_ODD): Likewise.
4273 (UNSPEC_SQRT_ROUND_TO_ODD): Likewise.
4274 (trunc<mode>sf2_hw): Change the truncate with round to odd
4275 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD.
4276 (add<mode>3_odd): Add insns for IEEE 128-bit floating point round
4277 to odd hardware instructions.
4278 (sub<mode>3_odd): Likewise.
4279 (mul<mode>3_odd): Likewise.
4280 (div<mode>3_odd): Likewise.
4281 (sqrt<mode>2_odd): Likewise.
4282 (fma<mode>4_odd): Likewise.
4283 (fms<mode>4_odd): Likewise.
4284 (nfma<mode>4_odd): Likewise.
4285 (nfms<mode>4_odd): Likewise.
4286 (trunc<mode>df2_odd): Change the truncate with round to odd
4287 expansion to use UNSPEC_TRUNC_ROUND_TO_ODD. Add a generator
4289 * doc/extend.texi (PowerPC built-in functions): Update documentation
4290 for existing IEEE float128-bit built-in functions. Add built-in
4291 functions that generate the IEEE 128-bit floating point round to
4294 2017-10-03 Segher Boessenkool <segher@kernel.crashing.org>
4296 PR rtl-optimization/77729
4297 * simplify-rtx.c (simplify_binary_operation_1): Delete the (X&C1)|C2
4298 to (X&(C1&~C2))|C2 transformations.
4300 2017-10-03 Martin Jambor <mjambor@suse.cz>
4302 PR tree-optimization/82363
4303 * tree-sra.c (propagate_subaccesses_across_link): In unrecoverable
4304 mismatch, mark lacc written regardless of racc.
4306 2017-10-03 Jakub Jelinek <jakub@redhat.com>
4308 PR tree-optimization/82381
4309 * tree-ssa-reassoc.c (sort_by_operand_rank): Don't check
4310 stmt_to_insert nor wheather SSA_NAMEs are default defs.
4311 Return 1 or -1 if one of bba and bbb is NULL. If bb_rank is equal,
4312 fallthrough into reassoc_stmt_dominates_stmt_p.
4315 * combine.c (combine_instructions): Don't combine in unreachable
4318 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
4321 * config/rs6000/rs6000.c (rs6000_option_override_internal): Rewrite
4322 function to not use the have_cpu variable. Do not set cpu_index,
4323 rs6000_cpu_index or rs6000_tune_index if we end up using TARGET_DEFAULT
4325 (rs6000_valid_attribute_p): Remove duplicate initializations of
4326 old_optimize and func_optimize.
4327 (rs6000_pragma_target_parse): Call rs6000_activate_target_options ().
4328 (rs6000_activate_target_options): Make global.
4329 * config/rs6000/rs6000-protos.h (rs6000_activate_target_options): Add
4332 2017-10-02 Jakub Jelinek <jakub@redhat.com>
4334 * tree-dfa.c (get_ref_base_and_extent): Set *pmax_size to -1
4335 if *poffset + *pmax_size overflows in HOST_WIDE_INT.
4336 Set *poffset to 0 and *psize and *pmax_size to -1 if
4337 *poffset + *psize overflows in HOST_WIDE_INT.
4339 PR tree-optimization/82387
4340 PR tree-optimization/82388
4341 PR tree-optimization/82389
4342 * tree-ssa-dse.c (dse_classify_store): Test byte_tracking_enabled
4343 instead of live_bytes non-NULL.
4345 2017-10-02 Georg-Johann Lay <avr@gjlay.de>
4348 * confg/avr/avr.md (*iorhi3.ashift8-ext.zerox): Add "r,r,0"
4351 2017-10-02 Richard Biener <rguenther@suse.de>
4353 * graphite-isl-ast-to-gimple.c (set_codegen_error): With
4354 -fchecking and --param graphite-allow-codegen-errors=0 ICE.
4355 * params.def (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): New param.
4357 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4359 * tree.h (wi::int_traits <const_tree>::decompose): Assert that the
4360 requested precision matches the type's.
4361 * calls.c (alloc_max_size): Calculate the new candidate size as
4362 a widest_int and use wi::to_widest when comparing it with the
4363 current candidate size.
4364 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Compare with
4365 zero rather than integer_zero_node.
4366 * match.pd: Check for a no-op conversion before using wi::add
4367 rather than after. Use tree_to_uhwi when summing small shift
4368 counts into an unsigned int.
4370 2017-10-02 Richard Sandiford <richard.sandiford@linaro.org>
4371 Alan Hayward <alan.hayward@arm.com>
4372 David Sherwood <david.sherwood@arm.com>
4375 * config/aarch64/aarch64.h (POINTER_AND_FP_REGS): New reg class.
4376 (REG_CLASS_NAMES, REG_CLASS_CONTENTS): Update accordingly.
4377 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
4378 POINTER_AND_FP_REGS.
4380 2017-10-02 Richard Biener <rguenther@suse.de>
4382 PR tree-optimization/82355
4383 * graphite-isl-ast-to-gimple.c (build_iv_mapping): Also build
4384 a mapping for the enclosing loop but avoid generating one for
4386 (copy_bb_and_scalar_dependences): Remove premature codegen
4387 error on PHIs in blocks duplicated into multiple places.
4388 * graphite-scop-detection.c
4389 (scop_detection::stmt_has_simple_data_refs_p): For a loop not
4390 in the region use it as loop and nest to analyze the DR in.
4391 (try_generate_gimple_bb): Likewise.
4392 * graphite-sese-to-poly.c (extract_affine_chrec): Adjust.
4393 (add_loop_constraints): For blocks in a loop not in the region
4394 create a dimension with a single iteration.
4395 * sese.h (gbb_loop_at_index): Remove assert.
4397 2017-10-01 Kevin Buettner <kevinb@redhat.com>
4399 * omp-expand.c (adjust_context_scope): New function.
4400 (expand_parallel_call): Call adjust_context_scope.
4402 2017-10-01 Jeff Law <law@redhat.com>
4404 * tree-ssa-dom.c (optimize_stmt): Make this a method within the
4405 dom_opt_dom_walker class with direct access to private members.
4406 Add comments. Call test_for_singularity.
4407 (dom_opt_dom_walker::before_dom_children): Corresponding changes.
4408 (dom_opt_dom_walker::after_dom_children): Do not lazily initialize
4409 m_dummy_cond anymore.
4410 (class dom_opt_dom_walker): Initialize m_dummy_cond member in the
4412 (pass_dominator:execute): Build the dummy_cond here and pass it
4413 to the dom_opt_dom_walker ctor.
4414 (test_for_singularity): New function.
4416 2017-09-30 Krister Walfridsson <krister.walfridsson@gmail.com>
4417 Maya Rashish <coypu@sdf.org>
4419 * config.gcc (*-*-netbsd*): New variable nbsd_tm_file containing
4420 netbsd.h, netbsd-stdint.h, and netbsd-elf.h.
4421 (alpha*-*-netbsd*) Use nbsd_tm_file.
4422 (arm*-*-netbsdelf*) Likewise.
4423 (i[34567]86-*-netbsdelf*) Likewise.
4424 (x86_64-*-netbsd*) Likewise.
4425 (mips*-*-netbsd*) Likewise.
4426 (powerpc-*-netbsd*) Likewise.
4427 (sh*-*-netbsd*) Likewise.
4428 (sparc-*-netbsdelf*) Likewise.
4429 (sparc64-*-netbsd*) Likewise.
4430 (m68k*-*-netbsdelf*) Use nbsd_tm_file and add CHAR_FAST8/SHORT_FAST16
4432 (vax-*-netbsdelf*) Likewise.
4433 * config/netbsd-stdint.h (INT_FAST8_TYPE): Check CHAR_FAST8.
4434 (UINT_FAST8_TYPE) Likewise.
4435 (INT_FAST16_TYPE) Check CHAR_FAST16.
4436 (UINT_FAST16_TYPE) Likewise.
4438 2017-09-30 Jakub Jelinek <jakub@redhat.com>
4441 * config/i386/i386.md
4442 (TARGET_USE_8BIT_IDIV zext divmodsi4 splitter): New define_split.
4443 (divmodsi4_zext_1, divmodsi4_zext_2, *divmodsi4_zext_1,
4444 *divmodsi4_zext_2): New define_insn_and_split.
4445 (*divmodsi4_noext_zext_1, *divmodsi4_noext_zext_2): New define_insn.
4446 (TARGET_USE_8BIT_IDIV zext udivmodsi4 splitter): New define_split.
4447 (udivmodsi4_zext_1, udivmodsi4_zext_2, *udivmodsi4_zext_1,
4448 *udivmodsi4_zext_2, *udivmodsi4_pow2_zext_1, *udivmodsi4_pow2_zext_2):
4449 New define_insn_and_split.
4450 (*udivmodsi4_noext_zext_1, *udivmodsi4_noext_zext_2): New define_insn.
4451 * config/i386/i386.c (ix86_split_idivmod): Handle operands[0] or
4452 operands[1] having DImode when mode is SImode.
4454 * config/i386/i386.c (ix86_split_idivmod): Use mode instead of
4455 always SImode for DIV and MOD in REG_EQUAL notes.
4457 2017-09-29 Yury Gribov <tetra2005@gmail.com>
4460 * match.pd: Fix handling of NaNs in pattern.
4462 2017-09-29 Jeff Law <law@redhat.com>
4464 * sbitmap.c (bitmap_bit_in_range_p): New function.
4465 * sbitmap.h (bitmap_bit_in_range_p): Prototype.
4466 * tree-ssa-dse.c (live_bytes_read): New function.
4467 (dse_classify_store): Ignore reads of dead bytes.
4469 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Fix
4470 typos and whitespace errors.
4471 * config/i386/predicates.md (address_no_seg_operand): Likewise.
4472 * config/s390/s390.c (s390_emit_prologue): Likewise.
4474 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4477 * ira-costs.c (scan_one_insn): Don't take into account PIC equiv
4478 with a symbol for LRA.
4480 2017-09-29 Vladimir Makarov <vmakarov@redhat.com>
4482 PR rtl-optimization/82338
4483 * lra-constraints.c (inherit_in_ebb): Check usage_insns check.
4485 2017-09-29 Alexander Monakov <amonakov@ispras.ru>
4487 * genmodes.c (calc_wider_mode): Suppress qsort macro.
4488 * system.h [CHECKING_P] (qsort): Redirect to qsort_chk.
4489 (qsort_chk): Declare.
4490 * vec.c [CHECKING_P] (qsort_chk_error): New static function.
4491 (qsort_chk): New function.
4493 2017-09-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4495 PR tree-optimization/82337
4496 * gimple-ssa-strength-reduction.c (find_phi_def): Don't record a
4497 phi definition if the PHI result appears in an abnormal PHI.
4498 (find_basis_for_base_expr): Don't record a basis if the LHS of the
4499 basis appears in an abnormal PHI.
4501 2017-09-29 Richard Biener <rguenther@suse.de>
4503 * graphite-isl-ast-to-gimple.c
4504 (translate_isl_ast_to_gimple::set_codegen_error): New function.
4505 (binary_op_to_tree): Use it.
4506 (get_rename_from_scev): Likewise.
4507 (copy_loop_phi_nodes): Likewise.
4508 (copy_bb_and_scalar_dependences): Likewise.
4509 (translate_pending_phi_nodes): Likewise.
4511 2017-09-29 Jakub Jelinek <jakub@redhat.com>
4514 * config/i386/i386.md (*movdi_internal peephole2): New -Os peephole
4515 for movabsq $(i32 << shift), r64.
4517 2017-09-28 Uros Bizjak <ubizjak@gmail.com>
4519 * config/i386/i386.c (ix86_print_operand_address_as): Do not check
4520 index when encoding %esp as %rsp to avoid 0x67 prefix.
4522 2017-09-28 Sergey Shalnov <Sergey.Shalnov@intel.com>
4524 * config/i386/i386.md (*movsf_internal, *movdf_internal):
4525 Return 256-bit AVX modes for TARGET_PREFER_AVX256.
4527 2017-09-28 Thomas Preud'homme <thomas.preudhomme@arm.com>
4529 * config/arm/arm.c (arm_option_override): Forbid ARMv8-M Security
4530 Extensions with more than 16 double VFP registers.
4531 (cmse_nonsecure_entry_clear_before_return): Remove second entry of
4532 to_clear_mask and all code related to it. Replace the remaining
4533 entry by a sbitmap and adapt code accordingly.
4535 2017-09-28 Henry Linjamäki <henry.linjamaki@parmance.com>
4537 * brig-builtins.def: Change pure attributes to const.
4539 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4541 * config.gcc (default_gnu_indirect_function): Default to yes for
4542 sparc*-*-linux* with glibc.
4544 2017-09-28 Joseph Myers <joseph@codesourcery.com>
4546 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor)
4547 (aarch64_elf_asm_destructor): Pass SECTION_NOTYPE to get_section
4548 when creating .init_array and .fini_array sections with priority
4551 2017-09-27 Christophe Lyon <christophe.lyon@linaro.org>
4554 * config/aarch64/aarch64.c
4555 (aarch64_builtin_support_vector_misalignment): Always return false
4556 when misalignment is unknown.
4558 2017-09-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
4560 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Revise
4561 this function to return false if the definition used by the swap
4562 instruction is artificial, or if the memory address from which the
4563 constant value is loaded is not represented by a base address held
4564 in a register or if the base address register is a frame or stack
4565 pointer. Additionally, return false if the base address of the
4566 loaded constant is a SYMBOL_REF but is not considered to be a
4568 (replace_swapped_load_constant): New function.
4569 (rs6000_analyze_swaps): Add a new pass to replace a swap of a
4570 loaded constant vector with a load of a swapped constant vector.
4572 2017-09-27 Carl Love <cel@us.ibm.com>
4574 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
4575 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
4576 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
4579 2017-09-27 Alexander Monakov <amonakov@ispras.ru>
4581 * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
4582 first, always call autopref_rank_data otherwise.
4584 2017-09-27 Richard Biener <rguenther@suse.de>
4586 * graphite-scop-detection.c (find_scop_parameters): Move
4587 loop bound handling ...
4588 (gather_bbs::before_dom_children): ... here, avoiding the need
4589 to build scop_info->loop_nest.
4590 (record_loop_in_sese): Remove.
4591 * sese.h (sese_info_t::loop_nest): Remove.
4592 * sese.c (new_sese_info): Do not allocate loop_nest.
4593 (free_sese_info): Do not free loop_nest.
4595 2017-09-27 Jakub Jelinek <jakub@redhat.com>
4598 * gimplify.c (gimplify_modify_expr): Don't optimize away zero sized
4599 lhs from calls if the lhs has addressable type.
4601 2017-09-27 Richard Biener <rguenther@suse.de>
4603 * graphite.h (scop::max_alias_set): New member.
4604 * graphite-scop-detection.c: Remove references to non-existing
4605 --param in comments.
4606 (build_alias_sets): Record the maximum alias set used for drs.
4607 (build_scops): Support zero as unlimited for
4608 --param graphite-max-arrays-per-scop.
4609 * graphite-sese-to-poly.c (add_scalar_version_numbers): Remove
4611 (build_poly_sr_1): ... here. Compute alias set based on the
4612 maximum alias set used for drs rather than
4613 PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP
4615 2017-09-27 Richard Biener <rguenther@suse.de>
4617 * graphite-optimize-isl.c (get_schedule_for_node_st): Allow
4618 --param loop-block-tile-size=0 to disable tiling.
4620 2017-09-27 Richard Biener <rguenther@suse.de>
4622 * doc/invoke.texi (graphite-max-bbs-per-function): Remove.
4623 (graphite-max-nb-scop-params): Document special value zero.
4624 * domwalk.h (dom_walker::STOP): New symbolical constant.
4625 (dom_walker::dom_walker): Add optional parameter for bb to
4627 (dom_walker::~dom_walker): Declare.
4628 (dom_walker::before_dom_children): Document STOP return value.
4629 (dom_walker::m_user_bb_to_rpo): New member.
4630 (dom_walker::m_bb_to_rpo): Likewise.
4631 * domwalk.c (dom_walker::dom_walker): Compute bb to RPO
4632 mapping here if not provided by the user.
4633 (dom_walker::~dom_walker): Free bb to RPO mapping if not
4634 provided by the user.
4635 (dom_walker::STOP): Define.
4636 (dom_walker::walk): Do not compute bb to RPO mapping here.
4637 Support STOP return value from before_dom_children to stop
4639 * graphite-optimize-isl.c (optimize_isl): If the schedule
4640 is the same still generate code if -fgraphite-identity
4641 or -floop-parallelize-all are given.
4642 * graphite-scop-detection.c: Include cfganal.h.
4643 (gather_bbs::gather_bbs): Get and pass through bb to RPO
4645 (gather_bbs::before_dom_children): Return STOP for BBs
4647 (build_scops): Compute bb to RPO mapping and pass it to
4648 the domwalk. Treat --param graphite-max-nb-scop-params=0
4649 as not limiting the number of params.
4650 * graphite.c (graphite_initialize): Remove limit on the
4651 number of basic-blocks in a function.
4652 * params.def (PARAM_GRAPHITE_MAX_BBS_PER_FUNCTION): Remove.
4653 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Adjust to documented
4654 default value of 10.
4656 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4658 * config/rs6000/vsx.md (peephole for optimizing move SF to GPR):
4659 Adjust code to eliminate needing to do the shift right 32-bits
4660 operation after XSCVDPSPN.
4662 2017-09-26 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
4664 * match.pd ((X / Y) == 0 -> X < Y): New pattern.
4665 ((X / Y) != 0 -> X >= Y): Likewise.
4667 2017-09-26 Carl Love <cel@us.ibm.com>
4669 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_XL_LEN_R,
4670 P9V_BUILTIN_VEC_XST_LEN_R): Add support for builtins
4671 vector unsigned char vec_xl_len_r (unsigned char *, size_t);
4672 void vec_xst_len_r (vector unsigned char, unsigned char *, size_t);
4673 * config/rs6000/altivec.h (vec_xl_len_r, vec_xst_len_r): Add defines.
4674 * config/rs6000/rs6000-builtin.def (XL_LEN_R, XST_LEN_R): Add
4675 definitions and overloading.
4676 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case
4677 statement for P9V_BUILTIN_XST_LEN_R.
4678 (altivec_init_builtins): Add def_builtin for P9V_BUILTIN_STXVLL.
4679 * config/rs6000/vsx.md (lxvll, stxvll, xl_len_r, xst_len_r): Add
4680 define_expand and define_insn for the instructions and builtins.
4681 * doc/extend.texi: Update the built-in documentation file for the new
4683 * config/rs6000/altivec.md (altivec_lvsl_reg, altivec_lvsr_reg): Add
4684 define_insn for the instructions
4686 2017-09-26 Krister Walfridsson <krister.walfridsson@gmail.com>
4689 * gcc/config/netbsd-protos.h: New file.
4690 * gcc/config/netbsd.c: New file.
4691 * gcc/config/netbsd.h (SUBTARGET_INIT_BUILTINS): Define.
4692 * gcc/config/t-netbsd: New file.
4693 * gcc/config.gcc (tm_p_file): Add netbsd-protos.h.
4694 (tmake_file) Add t-netbsd.
4695 (extra_objs) Add netbsd.o.
4697 2017-09-26 Janus Weil <janus@gcc.gnu.org>
4701 * doc/sourcebuild.texi: Document fortran_real_10 and fortran_real_16.
4703 2017-09-26 Michael Meissner <meissner@linux.vnet.ibm.com>
4705 * config/rs6000/rs6000.md (extendsi<mode>2): Add a splitter to do
4706 sign extension from a vector register to a GPR by doing a 32-bit
4707 direct move and then an EXTSW.
4708 (extendsi<mode>2 splitter): Likewise.
4709 (movsi_from_sf): Adjust code to eliminate doing a 32-bit shift
4710 right or vector extract after doing XSCVDPSPN. Use
4711 zero_extendsidi2 instead of p8_mfvsrd_4_disf to move the value to
4713 (movdi_from_sf_zero_ext): Likewise.
4714 (reload_gpr_from_vsxsf): Likewise.
4715 (p8_mfvsrd_4_disf): Delete, no longer used.
4716 (movsi_from_df): Optimize converting a DFmode to a SFmode, and
4717 then needing to move the SFmode to a GPR to use the XSCVDPSP
4718 instruction instead of FRSP and XSCVDPSPN.
4719 * config/rs6000/vsx.md (vsx_xscvspdp_scalar2): Move insn so that
4720 it is adjacent to the other XSCVSPDP insns.
4721 (vsx_xscvdpsp_scalar): Use "ww" constraint instead of "f" to allow
4722 SFmode to be in traditional Altivec registers.
4723 (vsx_xscvdpspn): Eliminate useless alternative constraint.
4724 (vsx_xscvspdpn): Likewise.
4725 (vsx_xscvspdpn_scalar): Likewise.
4727 2017-09-26 Martin Jambor <mjambor@suse.cz>
4729 * tree-sra.c (compare_access_positions): Put integral types first,
4730 stabilize sorting of integral types, remove conditions putting
4731 non-full-precision integers last.
4732 (sort_and_splice_var_accesses): Disable scalarization if a
4733 non-integert would be represented by a non-full-precision integer.
4735 2017-09-26 Joseph Myers <joseph@codesourcery.com>
4737 * config/microblaze/linux.h (TARGET_ASM_FILE_END): Likewise.
4738 * config/pa/pa.h (NEED_INDICATE_EXEC_STACK): Likewise.
4739 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Likewise.
4740 * config/pa/pa.c (pa_hpux_file_end): Rename to pa_file_end.
4741 Define unconditionally, with [ASM_OUTPUT_EXTERNAL_REAL]
4742 conditionals inside the function instead of around it. Call
4743 file_end_indicate_exec_stack if NEED_INDICATE_EXEC_STACK.
4744 (TARGET_ASM_FILE_END): Define unconditionally to pa_file_end.
4746 2017-09-26 Richard Biener <rguenther@suse.de>
4748 * graphite-scop-detection.c (scop_detection::build_scop_depth): Rewrite,
4750 (scop_detection::build_scop_breadth): ... this. Removed.
4751 (scop_detection::loop_is_valid_in_scop): Fold into single caller.
4752 (scop_detection::harmful_stmt_in_bb): Likewise.
4753 (scop_detection::graphite_can_represent_stmt): Likewise.
4754 (scop_detection::loop_body_is_valid_scop): Likewise. Remove recursion.
4755 (scop_detection::can_represent_loop): Remove recursion, fold in ...
4756 (scop_detection::can_represent_loop_1): ... this. Removed.
4757 (scop_detection::harmful_loop_in_region): Simplify after inlining
4758 the above and remove more quadraticness.
4759 (build_scops): Adjust.
4760 * tree-data-ref.c (loop_nest_has_data_refs): Remove pointless
4763 2017-09-26 Jakub Jelinek <jakub@redhat.com>
4766 * config/i386/i386.c (ix86_print_operand_address_as): Only test
4767 REGNO (base) == SP_REG if base is a REG.
4770 * tree-ssa-reassoc.c (update_range_test): Dump r->exp each time
4771 if it is different SSA_NAME.
4772 (optimize_range_tests_cmp_bitwise): New function.
4773 (optimize_range_tests): Call it.
4775 2017-09-26 Richard Biener <rguenther@suse.de>
4777 PR tree-optimization/82321
4778 * graphite.c (canonicalize_loop_closed_ssa): Properly check
4779 for the def being inside the loop.
4781 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4783 * config/s390/vx-builtins.md ("vmslg"): Add missing operand in
4785 * config/s390/s390-builtins.def: Fix constraint on op4.
4787 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4789 * config/s390/s390.c (s390_expand_vec_compare): Use the new mode
4790 independent expanders.
4791 * config/s390/vector.md ("vec_cmpuneq", "vec_cmpltgt")
4792 ("vec_ordered", "vec_unordered"): New expanders.
4794 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4796 * config/s390/s390.c (s390_preferred_simd_mode): Return V4SFmode
4799 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4801 * config/s390/vector.md ("vec_unpacks_low_v16qi"): Rename to
4802 vec_unpacks_lo_v16qi.
4803 ("vec_unpacku_low_v16qi"): Rename to vec_unpacku_lo_v16qi.
4805 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4807 * config/s390/vector.md ("vec_unpacks_lo_v4sf")
4808 ("vec_unpacks_hi_v4sf", "vec_unpacks_lo_v2df")
4809 ("vec_unpacks_hi_v2df", "vec_pack_trunc_v2df"): New expanders.
4811 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4813 * config/s390/predicates.md ("const_shift_by_byte_operand"): New
4815 * config/s390/vector.md ("*vec_srb<mode>"): Change modes to V_128
4817 ("*vec_slb<mode>"): New insn pattern.
4818 ("vec_shr_<mode>"): New expander.
4819 * config/s390/vx-builtins.md ("vec_slb<mode>"): Turn into expander
4820 and force the shift count operand to V16QImode.
4821 ("vec_srb<mode>"): Set shift count mode to V16QI.
4823 2017-09-26 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4825 * config/s390/vector.md ("vec_widen_umult_lo_<mode>")
4826 ("vec_widen_umult_hi_<mode>", "vec_widen_smult_lo_<mode>")
4827 ("vec_widen_smult_hi_<mode>"): New expander definitions.
4829 2017-09-26 Richard Earnshaw <rearnsha@arm.com>
4832 * config/arm/arm.h (DRIVER_SELF_SPECS): Separate sub-rules with commas.
4834 2017-09-26 Richard Biener <rguenther@suse.de>
4836 PR tree-optimization/82320
4837 * tree-ssa-sccvn.c (set_ssa_val_to): Changing undef to undef
4840 2017-09-25 Jeff Law <law@redhat.com>
4842 * config/rs6000/rs6000-protos.h (output_probe_stack_range): Update
4843 prototype for new argument.
4844 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack_1): New function,
4845 mostly extracted from rs6000_emit_allocate_stack.
4846 (rs6000_emit_probe_stack_range_stack_clash): New function.
4847 (rs6000_emit_allocate_stack): Call
4848 rs6000_emit_probe_stack_range_stack_clash as needed.
4849 (rs6000_emit_probe_stack_range): Add additional argument
4850 to call to gen_probe_stack_range{si,di}.
4851 (output_probe_stack_range): New.
4852 (output_probe_stack_range_1): Renamed from output_probe_stack_range.
4853 (output_probe_stack_range_stack_clash): New.
4854 (rs6000_emit_prologue): Emit notes into dump file as requested.
4855 * rs6000.md (allocate_stack): Handle -fstack-clash-protection.
4856 (probe_stack_range<P:mode>): Operand 0 is now early-clobbered.
4857 Add additional operand and pass it to output_probe_stack_range.
4859 2017-09-25 Bin Cheng <bin.cheng@arm.com>
4861 PR tree-optimization/82163
4862 * tree-ssa-loop-manip.h (verify_loop_closed_ssa): New parameter.
4863 (checking_verify_loop_closed_ssa): New parameter.
4864 * tree-ssa-loop-manip.c (check_loop_closed_ssa_use): Delete.
4865 (check_loop_closed_ssa_stmt): Delete.
4866 (check_loop_closed_ssa_def, check_loop_closed_ssa_bb): New functions.
4867 (verify_loop_closed_ssa): Check loop closed ssa form for LOOP.
4868 (tree_transform_and_unroll_loop): Check loop closed ssa form only for
4871 2017-09-25 Pekka Jaaskelainen <pekka@parmance.com>
4873 * brig-builtins.def: Treat HSAIL barrier builtins as
4874 setjmp/longjump style functions.
4876 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
4878 * target.def (constant_alignment): New hook.
4879 * defaults.h (CONSTANT_ALIGNMENT): Delete.
4880 * doc/tm.texi.in (CONSTANT_ALIGNMENT): Replace with...
4881 (TARGET_CONSTANT_ALIGNMENT): ...this new hook.
4882 * doc/tm.texi: Regenerate.
4883 * targhooks.h (default_constant_alignment): Declare.
4884 (constant_alignment_word_strings): Likewise.
4885 * targhooks.c (default_constant_alignment): New function.
4886 (constant_alignment_word_strings): Likewise.
4887 * builtins.c (get_object_alignment_2): Use targetm.constant_alignment
4888 instead of CONSTANT_ALIGNMENT.
4889 * varasm.c (align_variable, get_variable_align, build_constant_desc)
4890 (force_const_mem): Likewise.
4891 * config/aarch64/aarch64.h (CONSTANT_ALIGNMENT): Delete.
4892 * config/aarch64/aarch64.c (aarch64_constant_alignment): New function.
4893 (aarch64_classify_address): Call it instead of CONSTANT_ALIGNMENT.
4894 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4895 * config/alpha/alpha.h (CONSTANT_ALIGNMENT): Delete commented-out
4897 * config/arc/arc.h (CONSTANT_ALIGNMENT): Delete.
4898 * config/arc/arc.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4899 constant_alignment_word_strings.
4900 * config/arm/arm.h (CONSTANT_ALIGNMENT_FACTOR): Delete.
4901 (CONSTANT_ALIGNMENT): Likewise.
4902 * config/arm/arm.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4903 (arm_constant_alignment): New function.
4904 * config/bfin/bfin.h (CONSTANT_ALIGNMENT): Delete.
4905 * config/bfin/bfin.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4906 constant_alignment_word_strings.
4907 * config/cr16/cr16.h (CONSTANT_ALIGNMENT): Delete.
4908 * config/cr16/cr16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4909 constant_alignment_word_strings.
4910 * config/cris/cris.h (CONSTANT_ALIGNMENT): Delete.
4911 * config/cris/cris.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4912 (cris_constant_alignment): New function.
4913 * config/epiphany/epiphany.h (CONSTANT_ALIGNMENT): Delete.
4914 * config/epiphany/epiphany.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4915 (epiphany_constant_alignment): New function.
4916 * config/fr30/fr30.h (CONSTANT_ALIGNMENT): Delete.
4917 * config/fr30/fr30.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4918 constant_alignment_word_strings.
4919 * config/frv/frv.h (CONSTANT_ALIGNMENT): Delete.
4920 * config/frv/frv.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4921 constant_alignment_word_strings.
4922 * config/ft32/ft32.h (CONSTANT_ALIGNMENT): Delete.
4923 * config/ft32/ft32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4924 constant_alignment_word_strings.
4925 * config/i386/i386.h (CONSTANT_ALIGNMENT): Delete.
4926 * config/i386/i386-protos.h (ix86_constant_alignment): Delete.
4927 * config/i386/i386.c (ix86_constant_alignment): Make static.
4928 Use the same interface as the target hook.
4929 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4930 * config/ia64/ia64.h (CONSTANT_ALIGNMENT): Delete.
4931 * config/ia64/ia64.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4932 constant_alignment_word_strings.
4933 * config/iq2000/iq2000.h (CONSTANT_ALIGNMENT): Delete.
4934 * config/iq2000/iq2000.c (iq2000_constant_alignment): New function.
4935 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4936 * config/lm32/lm32.h (CONSTANT_ALIGNMENT): Delete.
4937 * config/lm32/lm32.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4938 constant_alignment_word_strings.
4939 * config/m32r/m32r.h (CONSTANT_ALIGNMENT): Delete.
4940 * config/m32r/m32r.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4941 constant_alignment_word_strings.
4942 * config/mcore/mcore.h (CONSTANT_ALIGNMENT): Delete.
4943 * config/mcore/mcore.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4944 constant_alignment_word_strings.
4945 * config/microblaze/microblaze.h (CONSTANT_ALIGNMENT): Delete.
4946 * config/microblaze/microblaze.c (microblaze_constant_alignment):
4948 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4949 * config/mips/mips.h (CONSTANT_ALIGNMENT): Delete.
4950 * config/mips/mips.c (mips_constant_alignment): New function.
4951 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4952 * config/mmix/mmix.h (CONSTANT_ALIGNMENT): Delete.
4953 * config/mmix/mmix-protos.h (mmix_constant_alignment): Delete.
4954 * config/mmix/mmix.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4955 (mmix_constant_alignment): Make static. Use the same interface
4957 * config/moxie/moxie.h (CONSTANT_ALIGNMENT): Delete.
4958 * config/moxie/moxie.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4959 constant_alignment_word_strings.
4960 * config/nios2/nios2.h (CONSTANT_ALIGNMENT): Delete.
4961 * config/nios2/nios2.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4962 constant_alignment_word_strings.
4963 * config/pa/pa.h (CONSTANT_ALIGNMENT): Delete.
4964 * config/pa/pa.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4965 constant_alignment_word_strings.
4966 * config/powerpcspe/powerpcspe.h (CONSTANT_ALIGNMENT): Delete.
4967 * config/powerpcspe/powerpcspe.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4968 (rs6000_constant_alignment): New function.
4969 * config/riscv/riscv.h (CONSTANT_ALIGNMENT): Delete.
4970 * config/riscv/riscv.c (riscv_constant_alignment): New function.
4971 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4972 * config/rs6000/rs6000.h (CONSTANT_ALIGNMENT): Delete.
4973 * config/rs6000/rs6000.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4974 (rs6000_constant_alignment): New function.
4975 * config/s390/s390.h (CONSTANT_ALIGNMENT): Delete.
4976 * config/s390/s390.c (s390_constant_alignment): New function.
4977 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4978 * config/sh/sh.h (CONSTANT_ALIGNMENT): Delete.
4979 * config/sh/sh.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4980 constant_alignment_word_strings.
4981 * config/sparc/sparc.h (CONSTANT_ALIGNMENT): Delete.
4982 * config/sparc/sparc.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4983 (sparc_constant_alignment): New function.
4984 * config/spu/spu.h (CONSTANT_ALIGNMENT): Delete.
4985 * config/spu/spu.c (spu_constant_alignment): New function.
4986 (TARGET_CONSTANT_ALIGNMENT): Redefine.
4987 * config/stormy16/stormy16.h (CONSTANT_ALIGNMENT): Delete.
4988 * config/stormy16/stormy16.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4989 constant_alignment_word_strings.
4990 * config/tilegx/tilegx.h (CONSTANT_ALIGNMENT): Delete.
4991 * config/tilegx/tilegx.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4992 constant_alignment_word_strings.
4993 * config/tilepro/tilepro.h (CONSTANT_ALIGNMENT): Delete.
4994 * config/tilepro/tilepro.c (TARGET_CONSTANT_ALIGNMENT): Redefine to
4995 constant_alignment_word_strings.
4996 * config/visium/visium.h (CONSTANT_ALIGNMENT): Delete.
4997 * config/visium/visium.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
4998 (visium_constant_alignment): New function.
4999 * config/xtensa/xtensa.h (CONSTANT_ALIGNMENT): Delete.
5000 * config/xtensa/xtensa.c (TARGET_CONSTANT_ALIGNMENT): Redefine.
5001 (xtensa_constant_alignment): New function.
5002 * system.h (CONSTANT_ALIGNMENT): Poison.
5004 2017-09-25 Will Schmidt <will_schmidt@vnet.ibm.com>
5006 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5007 for early folding of vector stores (ALTIVEC_BUILTIN_ST_*).
5008 (rs6000_builtin_valid_without_lhs): New helper function.
5009 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5010 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_ST.
5012 2017-09-25 Richard Sandiford <richard.sandiford@linaro.org>
5014 * target.h (vec_perm_indices): Use unsigned short rather than
5016 (auto_vec_perm_indices): Likewise.
5017 * config/aarch64/aarch64.c (aarch64_vectorize_vec_perm_const_ok):
5018 Use unsigned int rather than unsigned char.
5019 * config/arm/arm.c (arm_vectorize_vec_perm_const_ok): Likewise.
5021 2017-09-25 Richard Biener <rguenther@suse.de>
5023 * cfgloop.h (sort_sibling_loops): Declare.
5024 * cfgloop.c (sort_sibling_loops_cmp): New helper.
5025 (sort_sibling_loops): New function sorting the sibling loop list
5027 * graphite.c (graphite_transform_loops): Sort sibling loops.
5029 2017-09-25 Richard Sandiford <richard.sandifird@linaro.org>
5031 * target.def (vec_perm_const_ok): Change sel parameter to
5033 * optabs-query.c (can_vec_perm_p): Update accordingly.
5034 * doc/tm.texi: Regenerate.
5035 * config/aarch64/aarch64.c (expand_vec_perm_d): Change perm to
5036 auto_vec_perm_indices and remove separate nelt field.
5037 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
5038 (aarch64_evpc_ext, aarch64_evpc_rev, aarch64_evpc_dup)
5039 (aarch64_evpc_tbl, aarch64_expand_vec_perm_const_1)
5040 (aarch64_expand_vec_perm_const): Update accordingly.
5041 (aarch64_vectorize_vec_perm_const_ok): Likewise. Change sel
5042 to vec_perm_indices.
5043 * config/arm/arm.c (expand_vec_perm_d): Change perm to
5044 auto_vec_perm_indices and remove separate nelt field.
5045 (arm_evpc_neon_vuzp, arm_evpc_neon_vzip, arm_evpc_neon_vrev)
5046 (arm_evpc_neon_vtrn, arm_evpc_neon_vext, arm_evpc_neon_vtbl)
5047 (arm_expand_vec_perm_const_1, arm_expand_vec_perm_const): Update
5049 (arm_vectorize_vec_perm_const_ok): Likewise. Change sel
5050 to vec_perm_indices.
5051 * config/i386/i386.c (ix86_vectorize_vec_perm_const_ok): Change
5052 sel to vec_perm_indices.
5053 * config/ia64/ia64.c (ia64_vectorize_vec_perm_const_ok): Likewise.
5054 * config/mips/mips.c (mips_vectorize_vec_perm_const_ok): Likewise.
5055 * config/powerpcspe/powerpcspe.c (rs6000_vectorize_vec_perm_const_ok):
5057 * config/rs6000/rs6000.c (rs6000_vectorize_vec_perm_const_ok):
5060 2017-09-25 Pierre-Marie de Rodat <derodat@adacore.com>
5063 * dwarf2out.c (dwarf2out_early_global_decl): Call dwarf2out_decl
5064 on the FUNCTION_DECL function context if it has a DIE that is a
5067 2017-09-25 Richard Biener <rguenther@suse.de>
5069 PR tree-optimization/82285
5070 * tree-vect-patterns.c (vect_recog_bool_pattern): Also handle
5073 2017-09-25 Tom de Vries <tom@codesourcery.com>
5077 * config/nvptx/nvptx.c (nvptx_output_call_insn): Add exit after call to
5080 2017-09-25 Richard Biener <rguenther@suse.de>
5082 * graphite-optimize-isl.c (optimize_isl): Fail and dump if
5083 ISL errors other than isl_error_quota happen. Dump if the
5084 schedule is the same.
5085 * graphite-sese-to-poly.c (build_poly_scop): Fail on ISL
5086 errors instead of aborting inside ISL.
5088 2017-09-25 Iain Sandoe <iain@codesourcery.com>
5091 * config/i386/darwin.h (REAL_LIB_SPEC): New; put libSystem ahead
5092 of libgcc_eh for m64.
5093 * config/i386/darwin64.h: Likewise.
5095 2017-09-25 Richard Biener <rguenther@suse.de>
5098 * dwarf2out.c (gen_enumeration_type_die): Do not add alignment
5099 attribute for incomplete types nor twice for complete ones.
5101 2017-09-24 Uros Bizjak <ubizjak@gmail.com>
5104 * config/i386/i386.c (ix86_print_operand_address_as): Encode
5105 %esp as %rsp to avoid 0x67 prefix if there is no index or base
5108 2017-09-23 Uros Bizjak <ubizjak@gmail.com>
5111 * config/i386/i386.opt (mprefer-avx256): Use
5112 ix86_target_flags variable.
5113 * config/i386/i386.c (ix86_target_string): Move
5114 -mprefer-avx256 to flag2_opts.
5116 2017-09-22 Jakub Jelinek <jakub@redhat.com>
5119 * match.pd: Simplify x == -1 & y == -1 into (x & y) == -1
5120 and x != -1 | y != -1 into (x & y) != -1.
5122 2017-09-22 Steve Ellcey <sellcey@cavium.com>
5124 * config.gcc: Add new case statement to set
5125 default_gnu_indirect_function. Remove it from x86_64-*-linux*,
5126 i[34567]86-*, powerpc*-*-linux*spe*, powerpc*-*-linux*, s390-*-linux*,
5127 s390x-*-linux* case statements. Added aarch64 to the list of
5128 supported architectures.
5130 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5132 PR tree-optimization/82289
5133 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Check
5134 STMT_VINFO_RELEVANT_P.
5136 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5137 Alan Hayward <alan.hayward@arm.com>
5138 David Sherwood <david.sherwood@arm.com>
5140 * tree-vrp.c (extract_range_from_multiplicative_op_1): Assert
5141 for VR_RANGE only; don't allow VR_ANTI_RANGE.
5142 (extract_range_from_binary_expr_1): Don't call
5143 extract_range_from_multiplicative_op_1 if !range_int_cst_p.
5145 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5146 Alan Hayward <alan.hayward@arm.com>
5147 David Sherwood <david.sherwood@arm.com>
5149 * target.def (preferred_vector_alignment): New hook.
5150 * doc/tm.texi.in (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): New
5152 * doc/tm.texi: Regenerate.
5153 * targhooks.h (default_preferred_vector_alignment): Declare.
5154 * targhooks.c (default_preferred_vector_alignment): New function.
5155 * tree-vectorizer.h (dataref_aux): Add a target_alignment field.
5157 (DR_TARGET_ALIGNMENT): New macro.
5158 (aligned_access_p): Update commentary.
5159 (vect_known_alignment_in_bytes): New function.
5160 * tree-vect-data-refs.c (vect_calculate_required_alignment): New
5162 (vect_compute_data_ref_alignment): Set DR_TARGET_ALIGNMENT.
5163 Calculate the misalignment based on the target alignment rather than
5165 (vect_update_misalignment_for_peel): Use DR_TARGET_ALIGMENT
5166 rather than TYPE_ALIGN / BITS_PER_UNIT to update the misalignment.
5167 (vect_enhance_data_refs_alignment): Mask the byte misalignment with
5168 the target alignment, rather than masking the element misalignment
5169 with the number of elements in a vector. Also use the target
5170 alignment when calculating the maximum number of peels.
5171 (vect_find_same_alignment_drs): Use vect_calculate_required_alignment
5172 instead of TYPE_ALIGN_UNIT.
5173 (vect_duplicate_ssa_name_ptr_info): Remove stmt_info parameter.
5174 Measure DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT.
5175 (vect_create_addr_base_for_vector_ref): Update call accordingly.
5176 (vect_create_data_ref_ptr): Likewise.
5177 (vect_setup_realignment): Realign by ANDing with
5178 -DR_TARGET_MISALIGNMENT.
5179 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Calculate
5180 the number of peels based on DR_TARGET_ALIGNMENT.
5181 * tree-vect-stmts.c (get_group_load_store_type): Compare the gap
5182 with the guaranteed alignment boundary when deciding whether
5184 (vectorizable_mask_load_store): Interpret DR_MISALIGNMENT
5185 relative to DR_TARGET_ALIGNMENT instead of TYPE_ALIGN_UNIT.
5186 (ensure_base_align): Remove stmt_info parameter. Get the
5187 target base alignment from DR_TARGET_ALIGNMENT.
5188 (vectorizable_store): Update call accordingly. Interpret
5189 DR_MISALIGNMENT relative to DR_TARGET_ALIGNMENT instead of
5191 (vectorizable_load): Likewise.
5193 2017-09-22 Richard Sandiford <richard.sandiford@linaro.org>
5194 Alan Hayward <alan.hayward@arm.com>
5195 David Sherwood <david.sherwood@arm.com>
5197 * tree-vectorizer.h (vect_get_scalar_dr_size): New function.
5198 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Use it.
5199 (vect_enhance_data_refs_alignment): Likewise.
5201 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5203 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
5204 error. Only quit immediately if parsing is complete.
5205 (BEGIN): Initialize fatal_err and parse_done.
5206 (begin fpu, end fpu): Check number of arguments.
5207 (begin arch, end arch): Likewise.
5208 (begin cpu, end cpu): Likewise.
5209 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
5210 (optalias): Likewise.
5212 2017-09-22 Richard Earnshaw <richard.earnshaw@arm.com>
5214 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
5215 * config/arm/arm-isa.h: Delete. Move definitions to ...
5216 * arm-cpus.in: ... here. Use new feature and fgroup values.
5217 * config/arm/arm.c (arm_option_override): Use lower case for feature
5219 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
5220 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
5221 * config/arm/parsecpu.awk (END): Add new command 'isa'.
5223 (print_isa_bits_for): New function.
5224 (gen_isa): New function.
5225 (gen_comm_data): Use print_isa_bits_for.
5226 (define feature): New keyword.
5227 (define fgroup): New keyword.
5228 * config/arm/t-arm (TM_H): Remove.
5229 (GTM_H): Add arm-isa.h.
5230 (arm-isa.h): Add rule to generate file.
5231 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
5232 case for feature bit names.
5234 2017-09-22 Richard Biener <rguenther@suse.de>
5236 * graphite-isl-ast-to-gimple.c (graphite_verify): Inline into
5238 (graphite_regenerate_ast_isl): Do not reset SCEV. Move debug
5239 print of no dependency loops ...
5240 * graphite.c (graphite_transform_loops): ... here.
5241 (canonicalize_loop_closed_ssa_form): Work from inner to outer
5243 (same_close_phi_node, remove_duplicate_close_phi,
5244 make_close_phi_nodes_unique, defined_in_loop_p): Fold into ...
5245 (canonicalize_loop_closed_ssa): ... here and simplify.
5246 * graphite-optimize-isl.c: Include tree-vectorizer.h.
5247 (optimize_isl): Use dump_printf_loc to tell when we stopped
5248 optimizing because of an ISL timeout.
5250 2017-09-22 Richard Biener <rguenther@suse.de>
5252 PR tree-optimization/82291
5253 * tree-if-conv.c (predicate_mem_writes): Make sure to
5254 remove writes in blocks predicated with false.
5256 2017-09-22 Richard Biener <rguenther@suse.de>
5258 * sese.c: Include cfganal.h.
5259 (if_region_set_false_region): Remove.
5260 (create_if_region_on_edge): Likewise.
5261 (move_sese_in_condition): Re-implement without destroying
5264 2017-09-22 Richard Biener <rguenther@suse.de>
5266 * graphite-isl-ast-to-gimple.c (translate_pending_phi_nodes):
5267 Verify both BBs contain loop PHI nodes before dispatching to
5269 (graphite_regenerate_ast_isl): Do not recompute dominators,
5270 do not verify three times. Restructure for clarity.
5271 * graphite-scop-detection.c (same_close_phi_node,
5272 remove_duplicate_close_phi, make_close_phi_nodes_unique,
5273 defined_in_loop_p, canonicalize_loop_closed_ssa,
5274 canonicalize_loop_closed_ssa_form): Simplify, remove excess
5275 checking and SSA rewrite, move to ...
5276 * graphite.c: ... here. Include ssa.h and tree-ssa-loop-manip.h.
5277 (graphite_initialize): Do not pass in ctx, do not reset the
5278 SCEV cache, compute only dominators.
5279 (graphite_transform_loops): Allocate ISL ctx after
5280 graphite_initialize. Call canonicalize_loop_closed_ssa_form.
5281 Maintain post-dominators only around build_scops.
5282 * sese.c (if_region_set_false_region): Make static. Free
5283 and recompute dominators.
5284 (move_sese_in_condition): Assert we don't get called with
5285 post-dominators computed.
5286 * sese.h (if_region_set_false_region): Remove.
5288 2017-09-22 Sergey Shalnov <sergey.shalnov@intel.com>
5290 * config/i386/sse.md ("mov<mode>_internal"): Use <sseinsnmode>
5291 mode attribute for TARGET_AVX512VL.
5293 2017-09-21 Sergey Shalnov <sergey.shalnov@intel.com>
5295 * config/i386/i386.opt (mprefer-avx256): New option.
5296 * config/i386/i386.c (ix86_target_string): Add -mprefer-avx256
5298 (ix86_preferred_simd_mode): Return 256-bit AVX modes
5299 for TARGET_PREFER_AVX256.
5300 * doc/invoke.texi (x86 Options): Document -mprefer-avx256.
5302 2017-09-21 Jeff Law <law@redhat.com>
5304 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash):
5305 Fix dump output if the only stack space is for pushed registers.
5307 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5309 * config/spu/spu.c (spu_sched_adjust_cost): Update after renaming
5312 2017-09-21 Martin Sebor <msebor@redhat.com>
5315 * doc/extend.texi (attribute ifunc): Avoid relying on ill-formed
5316 code (in C++) or code that triggers warnings.
5318 2017-09-21 Eric Botcazou <ebotcazou@adacore.com>
5320 * stor-layout.c (bit_from_pos): Do not distribute the conversion.
5322 2017-09-21 Segher Boessenkool <segher@kernel.crashing.org>
5324 * haifa-sched.c: Rename insn_cost to insn_sched_cost.
5325 * sched-rgn.c: Ditto.
5326 * sel-sched-ir.c: Ditto.
5328 2017-09-21 Alexander Monakov <amonakov@ispras.ru>
5330 * toplev.h (set_random_seed): Adjust return type.
5331 * toplev.c (init_local_tick): Move eager initialization of random_seed
5332 to get_random_seed. Adjust comment.
5333 (init_random_seed): Inline to get_random_seed, delete.
5334 (get_random_seed): Initialize random_seed lazily.
5335 (set_random_seed): Do not return previous value.
5336 (print_switch_value): Do not call get_random_seed.
5338 2017-09-21 Evgeny Kudryashov <kudryashov@ispras.ru>
5340 * cgraph.c (delete_function_version): New, broken out from...
5341 (cgraph_node::delete_function_version): ...here. Rename to
5342 cgraph_node::delete_function_version_by_decl. Update all uses.
5343 (cgraph_node::remove): Call delete_function_version.
5345 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5348 * tree-inline.c (expand_call_inline): Emit clobber stmts for
5349 VAR_DECLs to which addressable non-volatile parameters are mapped
5350 and for id->retvar after the return value assignment. Clear
5351 id->retval and id->retbnd after inlining.
5353 2017-09-21 Richard Biener <rguenther@suse.de>
5355 PR tree-optimization/82276
5356 PR tree-optimization/82244
5357 * tree-vrp.c (build_assert_expr_for): Set
5358 SSA_NAME_OCCURS_IN_ABNORMAL_PHI if the variable we assert on
5360 (remove_range_assertions): Revert earlier change.
5362 2017-09-21 Wilco Dijkstra <wdijkstr@arm.com>
5365 * config/aarch64/aarch64.h (LIBGCC2_UNWIND_ATTRIBUTE): Define.
5367 2017-09-21 Richard Biener <rguenther@suse.de>
5369 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl):
5370 Restore valid IL after code generation errors.
5371 * graphite.c (graphite_transform_loops): Diagnose code
5372 generation issues as MSG_MISSED_OPTIMIZATION and continue
5373 with processing SCOPs.
5375 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5376 Alan Hayward <alan.hayward@arm.com>
5377 David Sherwood <david.sherwood@arm.com>
5379 * calls.c (compute_argument_addresses): Use simplify_gen_binary
5380 rather than choosing between plus_constant and gen_rtx_<CODE>.
5381 * expr.c (emit_push_insn): Likewise.
5382 (expand_expr_real_2): Likewise.
5384 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5385 Alan Hayward <alan.hayward@arm.com>
5386 David Sherwood <david.sherwood@arm.com>
5388 * loop-unroll.c (split_iv): Call copy_rtx on the step.
5390 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5391 Alan Hayward <alan.hayward@arm.com>
5392 David Sherwood <david.sherwood@arm.com>
5394 * tree.c (find_atomic_core_type): Check tree_fits_uhwi_p before
5395 calling tree_to_uhwi.
5397 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5398 Alan Hayward <alan.hayward@arm.com>
5399 David Sherwood <david.sherwood@arm.com>
5401 * tree-ssa-ccp.c (get_value_for_expr): Use a positive test for
5402 INTEGER_CST rather than a negative test for ADDR_EXPR.
5404 2017-09-21 Richard Sandiford <richard.sandiford@linaro.org>
5405 Alan Hayward <alan.hayward@arm.com>
5406 David Sherwood <david.sherwood@arm.com>
5408 * tree-vrp.c (extract_range_from_binary_expr_1): Check
5409 int_cst_rangeN before calling value_range_constant_singleton (&vrN).
5411 2017-09-21 Richard Biener <rguenther@suse.de>
5413 PR tree-optimization/71351
5414 * graphite-isl-ast-to-gimple.c (translate_isl_ast_to_gimple::
5415 graphite_create_new_loop_guard): Remove, fold remaining parts
5417 (translate_isl_ast_node_for): ... here and simplify.
5419 2017-09-21 Jakub Jelinek <jakub@redhat.com>
5422 * config/i386/i386.md (*movqi_internal): Replace (=q,q) alternative
5423 with (=Q,Q), (=R,R) and (=r,r) alternatives, only enable the
5424 latter two for 64-bit, renumber alternatives, for -Os imov (=q,n)
5425 alternative always use QI mode, for -Os imov (=R,R) alternative
5426 always use SI mode, for imov (=Q,Q) or (=r,r) alternatives
5429 2017-09-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5430 Jeff Law <law@redhat.com>
5432 * config/s390/s390.c (MIN_UNROLL_PROBES): Define.
5433 (allocate_stack_space): New function, partially extracted from
5435 (s390_emit_prologue): Track offset to most recent stack probe.
5436 Code to allocate space moved into allocate_stack_space.
5437 Dump actions when no stack is allocated.
5438 (s390_prologue_plus_offset): New function.
5439 (s390_emit_stack_probe): Likewise.
5441 2017-09-20 Alexandre Oliva <aoliva@redhat.com>
5443 * common.opt (Wa, Wl, Wp, g, gz=): Add
5445 (gno-column-info): Remove.
5446 (gcolumn-info): Drop RejectNegative.
5448 (gno-record-gcc-switches): Remove.
5449 (grecord-gcc-switches): Drop RejectNegative.
5450 (gno-split-dwarf): Remove.
5451 (gsplit-dwarf): Drop RejectNegative.
5452 (gno-strict-dwarf): Remove.
5453 (gstrict-dwarf): Drop RejectNegative.
5454 * config/darwin.opt (gfull, gused): Add RejectNegative.
5455 * dwarf2out.c (gen_producer_string): Drop
5456 gno-record-gcc-switches handler.
5457 * optc-gen.awk: Add g to prefixes with negative forms.
5458 * opts-common.c (remapping_prefix_p): New.
5459 (find_opt): Check it.
5460 (generate_canonical_option): Test g prefix.
5461 (option_map): Add -gno- mapping.
5462 (add_misspelling_candidates): Check remapping_prefix_p.
5464 2017-09-20 Jeff Law <law@redhat.com>
5466 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Fix
5467 thinko in stack clash protection support.
5469 * explow.c (compute_stack_clash_protection_loop_data): Use
5470 CONST_INT_P instead of explicit test. Verify object is a
5471 CONST_INT_P before looking at INTVAL.
5472 (anti_adjust_stack_and_probe_stack_clash): Use CONST_INT_P
5473 instead of explicit test.
5475 2017-09-20 Segher Boessenkool <segher@kernel.crashing.org>
5478 * config/rs6000/rs6000.md (stack_restore_tie): Store to a scratch
5479 address instead of to r1 and r11.
5481 2017-09-20 Sebastian Peryt <sebastian.peryt@intel.com>
5483 * config.gcc: Support "knm".
5484 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "knm".
5485 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
5487 * config/i386/i386.c (m_KNM): Define.
5488 (processor_target_table): Add "knm".
5490 (ix86_option_override_internal): Add "knm".
5491 (ix86_issue_rate): Add PROCESSOR_KNM.
5492 (ix86_adjust_cost): Ditto.
5493 (ia32_multipass_dfa_lookahead): Ditto.
5494 (get_builtin_code_for_version): Handle PROCESSOR_KNM.
5495 (fold_builtin_cpu): Add M_INTEL_KNM.
5496 * config/i386/i386.h (processor_costs): Define TARGET_KNM.
5497 (processor_type): Add PROCESSOR_KNM.
5498 * config/i386/x86-tune.def: Add m_KNM.
5499 * doc/invoke.texi: Add knm as x86 -march=/-mtune= CPU type.
5501 2017-09-20 Richard Biener <rguenther@suse.de>
5503 PR tree-optimization/80213
5504 * graphite-scop-detection.c (trivially_empty_bb_p): Labels
5505 are allowed in empty BBs as well.
5506 (canonicalize_loop_closed_ssa): Also look for other complex
5508 (scop_detection::get_sese): Include the loop-closed PHI block
5510 (scop_detection::merge_sese): Remove code adding extra blocks.
5511 (scop_detection::region_has_one_loop): Adjust for get_sese changes.
5512 (build_scops): Assert the final returned scop is invalid.
5514 2017-09-20 Richard Biener <rguenther@suse.de>
5516 PR tree-optimization/82264
5517 * tree-ssa-sccvn.c (vn_phi_eq): Use safe_dyn_cast to check
5519 (vn_phi_lookup): Likewise.
5520 (vn_phi_insert): Likewise.
5522 2017-09-20 Jakub Jelinek <jakub@redhat.com>
5524 * dwarf2out.c (tree_add_const_value_attribute): For INTEGER_CST
5525 that fits into uhwi or shwi, add DW_AT_const_value regardless
5526 of early_dwarf without going through RTL, using add_AT_unsigned
5529 * dwarf2out.c (DEBUG_LTO_DWO_INFO_SECTION): Reorder defines.
5530 (DEBUG_LTO_ABBREV_SECTION): Likewise.
5531 (DEBUG_LTO_MACINFO_SECTION): Likewise.
5532 (DEBUG_MACRO_SECTION): Likewise.
5533 (DEBUG_LTO_MACRO_SECTION): Likewise.
5534 (DEBUG_STR_DWO_SECTION): Likewise.
5535 (DEBUG_LTO_STR_DWO_SECTION): Likewise.
5536 (DEBUG_LTO_LINE_SECTION): Drop .dwo suffix from the name.
5537 (DEBUG_LTO_DWO_LINE_SECTION): Define.
5538 (DEBUG_LTO_LINE_STR_SECTION): Define.
5539 (init_sections_and_labels): Initialize debug_line_str_section
5540 variable. Initialize debug_loc_section for -gdwarf-5 to
5541 DEBUG_LOCLISTS_SECTION. Formatting fixes.
5543 2017-09-20 Richard Biener <rguenther@suse.de>
5545 * graphite-sese-to-poly.c (extract_affine): Properly handle
5546 POINTER_PLUS_EXPR, BIT_NOT_EXPR and conversion to signed.
5548 2017-09-20 Richard Biener <rguenther@suse.de>
5550 PR tree-optimization/81373
5551 * graphite-scop-detection.c (build_cross_bb_scalars_def):
5552 Force SESE live-out defs to be handled even if they are
5555 2017-09-19 Jeff Law <law@redhat.com>
5557 * combine-stack-adj.c (combine_stack_adjustments_for_block): Do
5558 nothing for stack adjustments with REG_STACK_CHECK.
5559 * sched-deps.c (parse_add_or_inc): Reject insns with
5560 REG_STACK_CHECK from dependency breaking.
5561 * config/i386/i386.c (pro_epilogue_adjust_stack): Return insn.
5562 (ix86_adjust_satck_and_probe_stack_clash): Add REG_STACK_NOTEs.
5563 * reg-notes.def (STACK_CHECK): New note.
5565 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): New.
5566 (ix86_expand_prologue): Dump stack clash info as needed.
5567 Call ix86_adjust_stack_and_probe_stack_clash as needed.
5569 * function.c (dump_stack_clash_frame_info): New function.
5570 * function.h (dump_stack_clash_frame_info): Prototype.
5571 (enum stack_clash_probes): New enum.
5573 * config/alpha/alpha.c (alpha_expand_prologue): Also check
5574 flag_stack_clash_protection.
5575 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Likewise.
5576 (arm_expand_prologue, thumb1_expand_prologue): Likewise.
5577 (arm_frame_pointer_required): Likewise.
5578 * config/ia64/ia64.c (ia64_compute_frame_size): Likewise.
5579 (ia64_expand_prologue): Likewise.
5580 * config/mips/mips.c (mips_expand_prologue): Likewise.
5581 * config/powerpcspe/powerpcspe.c (rs6000_expand_prologue): Likewise.
5582 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
5583 (sparc_flat_expand_prologue): Likewise.
5584 * config/spu/spu.c (spu_expand_prologue): Likewise.
5586 * explow.c: Include "params.h".
5587 (anti_adjust_stack_and_probe_stack_clash): New function.
5588 (get_stack_check_protect): Likewise.
5589 (compute_stack_clash_protection_loop_data): Likewise.
5590 (emit_stack_clash_protection_loop_start): Likewise.
5591 (emit_stack_clash_protection_loop_end): Likewise.
5592 (allocate_dynamic_stack_space): Use get_stack_check_protect.
5593 Use anti_adjust_stack_and_probe_stack_clash.
5594 * explow.h (compute_stack_clash_protection_loop_data): Prototype.
5595 (emit_stack_clash_protection_loop_start): Likewise.
5596 (emit_stack_clash_protection_loop_end): Likewise.
5597 * rtl.h (get_stack_check_protect): Prototype.
5598 * target.def (stack_clash_protection_final_dynamic_probe): New hook.
5599 * targhooks.c (default_stack_clash_protection_final_dynamic_probe): New.
5600 * targhooks.h (default_stack_clash_protection_final_dynamic_probe):
5602 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE):
5604 * doc/tm.texi: Rebuilt.
5605 * config/aarch64/aarch64.c (aarch64_expand_prologue): Use
5606 get_stack_check_protect.
5607 * config/alpha/alpha.c (alpha_expand_prologue): Likewise.
5608 * config/arm/arm.c (arm_expand_prologue): Likewise.
5609 (arm_frame_pointer_required): Likewise.
5610 * config/i386/i386.c (ix86_expand_prologue): Likewise.
5611 * config/ia64/ia64.c (ia64_expand_prologue): Likewise.
5612 * config/mips/mips.c (mips_expand_prologue): Likewise.
5613 * config/powerpcspe/powerpcspe.c (rs6000_emit_prologue): Likewise.
5614 * config/rs6000/rs6000.c (rs6000_emit_prologue): Likewise.
5615 * config/sparc/sparc.c (sparc_expand_prologue): Likewise.
5616 (sparc_flat_expand_prologue): Likewise.
5618 * common.opt (-fstack-clash-protection): New option.
5619 * flag-types.h (enum stack_check_type): Note difference between
5620 -fstack-check= and -fstack-clash-protection.
5621 * params.def (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE): New PARAM.
5622 (PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Likewise.
5623 * toplev.c (process_options): Issue warnings/errors for cases
5624 not handled with -fstack-clash-protection.
5625 * doc/invoke.texi (-fstack-clash-protection): Document new option.
5626 (-fstack-check): Note additional problem with -fstack-check=generic.
5627 Note that -fstack-check is primarily for Ada and refer users
5628 to -fstack-clash-protection for stack-clash-protection.
5629 Document new params for stack clash protection.
5631 2017-09-19 Uros Bizjak <ubizjak@gmail.com>
5633 * config/i386/i386.c (ix86_split_long_move): Do not handle
5634 address used for LEA in a special way.
5636 2017-09-19 Segher Boessenkool <segher@kernel.crashing.org>
5638 * simplify-rtx.c (simplify_binary_operation_1): Fix typo in comment.
5640 2017-09-19 Martin Sebor <msebor@redhat.com>
5643 * cgraphunit.c (handle_alias_pairs): Reject aliases between functions
5644 of incompatible types.
5646 2017-09-19 Will Schmidt <will_schmidt@vnet.ibm.com>
5648 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5649 for early folding of vector loads (ALTIVEC_BUILTIN_LVX_*).
5650 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
5651 Remove obsoleted code for handling ALTIVEC_BUILTIN_VEC_LD.
5653 2017-09-19 Richard Biener <rguenther@suse.de>
5655 PR tree-optimization/82244
5656 * tree-vrp.c (remove_range_assertions): Do not propagate
5657 a constant to abnormals but replace the assert with a copy.
5659 2017-09-19 Alexander Monakov <amonakov@ispras.ru>
5661 PR rtl-optimization/57878
5662 PR rtl-optimization/68988
5663 * lra-assigns.c (reload_pseudo_compare_func): Remove fragmentation
5664 avoidance test involving non_reload_pseudos. Move frequency test
5665 below the general fragmentation avoidance test.
5667 2017-09-19 Richard Biener <rguenther@suse.de>
5669 PR tree-optimization/69728
5670 * graphite-sese-to-poly.c (schedule_error): New global.
5671 (add_loop_schedule): Handle empty domain by failing the
5673 (build_original_schedule): Handle schedule_error.
5675 2017-09-19 Richard Biener <rguenther@suse.de>
5677 * graphite-scop-detection.c (scop_detection::can_represent_loop):
5678 Do not iterate to sibling loops but only to siblings of inner
5681 2017-09-18 Andreas Schwab <schwab@linux-m68k.org>
5684 * config/m68k/m68k.md (moveq feeding equality comparison): Check
5685 that the registers are different.
5687 2017-09-18 Uros Bizjak <ubizjak@gmail.com>
5689 * config/i386/i386.c (fold_builtin_cpu): Add M_AMDFAM17H
5690 to processor_model and "amdfam17h" to arch_names_table.
5691 * doc/extend.texi (__builtin_cpu_is): Document amdfam17h CPU name.
5693 2017-09-18 Jakub Jelinek <jakub@redhat.com>
5696 * doc/extend.texi: Add @findex entry for __builtin_shuffle.
5698 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5699 Alan Hayward <alan.hayward@arm.com>
5700 David Sherwood <david.sherwood@arm.com>
5702 * tree-vectorizer.h (vect_slp_analyze_operations): Replace parameters
5704 * tree-vect-loop.c (vect_analyze_loop_operations): Update call
5706 * tree-vect-slp.c (vect_slp_analyze_node_operations): Add vec_info *
5707 parameter. Set SLP_TREE_NUMBER_OF_VEC_STMTS here rather than in
5708 vect_schedule_slp_instance.
5709 (vect_slp_analyze_operations): Replace parameters with a vec_info *.
5710 Update call to vect_slp_analyze_node_operations. Simplify return
5712 (vect_slp_analyze_bb_1): Update call accordingly.
5713 (vect_schedule_slp_instance): Remove vectorization_factor parameter.
5714 Don't calculate SLP_TREE_NUMBER_OF_VEC_STMTS here.
5715 (vect_schedule_slp): Update call accordingly.
5717 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5718 Alan Hayward <alan.hayward@arm.com>
5719 David Sherwood <david.sherwood@arm.com>
5721 * tree-vect-data-refs.c (vect_get_smallest_scalar_type): Cope
5722 with types that aren't in fact scalar.
5724 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5726 * tree-vect-slp.c (vect_record_max_nunits): New function,
5728 (vect_build_slp_tree_1): ...here.
5729 (vect_build_slp_tree_2): Call it for phis too.
5731 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5733 * tree-vect-stmts.c (vectorizable_mask_load_store): Pass mask_vectype
5734 to vect_get_vec_def_for_operand when getting the mask operand.
5736 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5737 Alan Hayward <alan.hayward@arm.com>
5738 David Sherwood <david.sherwood@arm.com>
5740 * tree-vect-loop.c (vectorizable_live_operation): Fix type of
5743 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5744 Alan Hayward <alan.hayward@arm.com>
5745 David Sherwood <david.sherwood@arm.com>
5747 * tree-vect-loop.c (vectorizable_live_operation): Fix element size
5748 calculation for vector booleans.
5750 2017-09-18 Richard Sandiford <richard.sandiford@linaro.org>
5751 Alan Hayward <alan.hayward@arm.com>
5752 David Sherwood <david.sherwood@arm.com>
5754 * tree-vect-stmts.c (can_vectorize_live_stmts): New function,
5756 (vect_transform_stmt): ...here.
5757 (vect_analyze_stmt): Use it instead of calling
5758 vectorizable_live_operation directly.
5760 2017-09-18 Cesar Philippidis <cesar@codesourcery.com>
5762 * omp-offload.c (oacc_xform_loop): Enable SIMD vectorization on
5763 non-SIMT targets in acc vector loops.
5765 2017-09-18 Claudiu Zissulescu <claziss@synopsys.com>
5767 * configure.ac: Add arc and check if assembler supports gdwarf2.
5768 * configure: Regenerate.
5770 2017-09-18 Richard Biener <rguenther@suse.de>
5772 PR tree-optimization/82220
5773 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Exclude
5774 epilogue niters from the min_profitable_iters compute.
5776 2017-09-18 Jakub Jelinek <jakub@redhat.com>
5779 * config/i386/i386.c (ix86_init_large_pic_reg): Revert 2017-09-01
5780 changes. Turn CODE_LABEL into NOTE_INSN_DELETED_LABEL immediately.
5781 (ix86_init_pic_reg): Revert 2017-09-01 changes.
5783 2017-09-18 Eric Botcazou <ebotcazou@adacore.com>
5786 * dwarf2cfi.c (add_cfis_to_fde): Do not generate DW_CFA_set_loc after
5787 switching to a new text section.
5789 2017-09-18 Richard Biener <rguenther@suse.de>
5791 * graphite-scop-detection.c (scop_detection::stmt_has_simple_data_ref):
5793 (build_alias_set): Reject aliases with no access function.
5795 2017-09-18 Richard Biener <rguenther@suse.de>
5797 PR tree-optimization/79622
5798 * graphite-scop-detection.c (build_cross_bb_scalars_def): Properly
5800 (build_cross_bb_scalars_use): Likewise.
5802 2017-09-18 Pierre-Marie de Rodat <derodat@adacore.com>
5804 * cgraph.h (cgraph_thunk_info): Fix a typo in a comment.
5806 2017-09-18 Alan Modra <amodra@gmail.com>
5809 * gcc/config/rs6000/rs6000.c (rs6000_return_addr): Use
5810 stack_pointer_rtx for count 0. Update comments. Break up
5811 large rtl expression.
5813 2017-09-17 Daniel Santos <daniel.santos@pobox.com>
5815 * config/i386/i386.c (xlogue_layout::STUB_NAME_MAX_LEN):
5816 Increase to 20 bytes.
5817 (xlogue_layout::s_stub_names): Add an additional size-2 diminsion.
5818 (xlogue_layout::get_stub_name): Modify to select the appropairate sse
5819 or avx version of the stub.
5821 2017-09-17 H.J. Lu <hongjiu.lu@intel.com>
5824 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Properly
5825 compute the minimum stack alignment. Also update preferred stack
5826 boundary for leaf functions.
5828 2017-09-16 Richard Sandiford <richard.sandiford@linaro.org>
5830 PR tree-optimization/82228
5831 * tree-vect-loop.c (vectorizable_live_operation): Move initialization
5834 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5836 * common/config/nds32/nds32-common.c
5837 (nds32_option_optimization_table): Refine formatting.
5838 (nds32_option_optimization_table): Use -fsched-pressure and
5839 -fomit-frame-pointer for specific optimization level.
5841 2017-09-16 Chung-Ju Wu <jasonwucj@gmail.com>
5843 * config/nds32/nds32.c: Refine formatting and comments.
5844 * config/nds32/nds32.h: Likewise.
5845 * config/nds32/nds32.md: Likewise.
5846 * config/nds32/nds32-cost.c: Likewise.
5847 * config/nds32/nds32-isr.c: Likewise.
5848 * config/nds32/nds32-md-auxiliary.c: Likewise.
5849 * config/nds32/nds32-multiple.md: Likewise.
5850 * config/nds32/nds32-predicates.c: Likewise.
5852 2017-09-15 Andrew Sutton <andrew.n.sutton@gmail.com>
5853 Jakub Jelinek <jakub@redhat.com>
5855 Add support for -std=c++2a.
5856 * doc/cpp.texi (__cplusplus): Document value for -std=c++2a
5858 * doc/invoke.texi: Document -std=c++2a and -std=gnu++2a.
5860 2017-09-15 Steve Ellcey <sellcey@cavium.com>
5863 * doc/extend.texi (Common Function Attributes): Add
5864 references to ARM, AArch64, and S/390 specific attributes.
5865 (Function Specific Option Pragmas): Add AArch64 and S/390
5866 to list of back ends that support the target pragma.
5868 2017-09-15 Nathan Sidwell <nathan@acm.org>
5870 * doc/standards.texi: Fix C++17 description. Update URLs for
5873 2017-09-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
5875 * common.opt (Wcast-align=strict): New warning option.
5876 * doc/invoke.texi: Document -Wcast-align=strict.
5878 2017-09-15 Pierre-Marie de Rodat <derodat@adacore.com>
5880 * cgraph.h (cgraph_thunk_info): Add comments.
5881 * cgraph.c (cgraph_node::create_thunk): Adjust comment, make
5882 assert for VIRTUAL_* arguments stricter.
5884 2017-09-15 Jackson Woodruff <jackson.woodruff@arm.com>
5886 PR tree-optimization/71026
5887 * match.pd: Move RDIV patterns from fold-const.c
5888 * fold-const.c (distribute_real_division): Removed.
5889 (fold_binary_loc): Remove calls to distribute_real_divison.
5891 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5893 * doc/invoke.texi: Document -std=c++17 and -std=gnu++17 and document
5894 c++1z and gnu++1z as deprecated. Change other references to
5895 -std=c++1z to -std=c++17 and -std=gnu++1z to -std=gnu++17.
5896 Change -Wc++1z-compat to -Wc++17-compat.
5897 * doc/cpp.texi: Document -std=c++17 defines __cplusplus 201703L.
5898 * dwarf2out.c (highest_c_language): Handle C++17.
5899 (gen_compile_unit_die): Likewise.
5901 2017-09-15 Jakub Jelinek <jakub@redhat.com>
5903 PR rtl-optimization/82192
5904 * combine.c (make_extraction): Don't look through non-paradoxical
5905 SUBREGs or TRUNCATE if pos + len is or might be bigger than
5908 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5909 Alan Hayward <alan.hayward@arm.com>
5910 David Sherwood <david.sherwood@arm.com>
5912 * target.def (function_arg_offset): New hook.
5913 * targhooks.h (default_function_arg_offset): Declare.
5914 * targhooks.c (default_function_arg_offset): New function.
5915 * function.c (locate_and_pad_parm): Use
5916 targetm.calls.function_arg_offset instead of FUNCTION_ARG_OFFSET.
5917 * doc/tm.texi.in (FUNCTION_ARG_OFFSET): Replace with...
5918 (TARGET_FUNCTION_ARG_OFFSET): ...this.
5919 * doc/tm.texi: Regenerate.
5920 * config/spu/spu.h (FUNCTION_ARG_OFFSET): Delete.
5921 * config/spu/spu.c (spu_function_arg_offset): New function.
5922 (TARGET_FUNCTION_ARG_OFFSET): Redefine.
5923 * system.h (FUNCTION_ARG_OFFSET): Poison.
5925 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
5926 Alan Hayard <alan.hayward@arm.com>
5927 David Sherwood <david.sherwood@arm.com>
5929 * target.def (truly_noop_truncation): New hook.
5930 (mode_rep_extended): Refer to TARGET_TRULY_NOOP_TRUNCATION rather
5931 than TRULY_NOOP_TRUNCATION.
5932 * hooks.h (hook_bool_uint_uint_true): Declare.
5933 * hooks.c (hook_bool_uint_uint_true): New function.
5934 * doc/tm.texi.in (TRULY_NOOP_TRUNCATION): Replace with...
5935 (TARGET_TRULY_NOOP_TRUNCATION): ...this.
5936 * doc/tm.texi: Regenerate.
5937 * combine.c (make_extraction): Refer to TARGET_TRULY_NOOP_TRUNCATION
5938 rather than TRULY_NOOP_TRUNCATION in comments.
5939 (simplify_comparison): Likewise.
5940 (record_truncated_value): Likewise.
5941 * expmed.c (extract_bit_field_1): Likewise.
5942 (extract_split_bit_field): Likewise.
5943 * convert.c (convert_to_integer_1): Use targetm.truly_noop_truncation
5944 instead of TRULY_NOOP_TRUNCATION.
5945 * function.c (assign_parm_setup_block): Likewise.
5946 * machmode.h (TRULY_NOOP_TRUNCATION_MODES_P): Likewise.
5947 * rtlhooks.c: Include target.h.
5948 * config/aarch64/aarch64.h (TRULY_NOOP_TRUNCATION): Delete.
5949 * config/alpha/alpha.h (TRULY_NOOP_TRUNCATION): Delete.
5950 * config/arc/arc.h (TRULY_NOOP_TRUNCATION): Delete.
5951 * config/arm/arm.h (TRULY_NOOP_TRUNCATION): Delete.
5952 * config/avr/avr.h (TRULY_NOOP_TRUNCATION): Delete.
5953 * config/bfin/bfin.h (TRULY_NOOP_TRUNCATION): Delete.
5954 * config/c6x/c6x.h (TRULY_NOOP_TRUNCATION): Delete.
5955 * config/cr16/cr16.h (TRULY_NOOP_TRUNCATION): Delete.
5956 * config/cris/cris.h (TRULY_NOOP_TRUNCATION): Delete.
5957 * config/epiphany/epiphany.h (TRULY_NOOP_TRUNCATION): Delete.
5958 * config/fr30/fr30.h (TRULY_NOOP_TRUNCATION): Delete.
5959 * config/frv/frv.h (TRULY_NOOP_TRUNCATION): Delete.
5960 * config/ft32/ft32.h (TRULY_NOOP_TRUNCATION): Delete.
5961 * config/h8300/h8300.h (TRULY_NOOP_TRUNCATION): Delete.
5962 * config/i386/i386.h (TRULY_NOOP_TRUNCATION): Delete.
5963 * config/ia64/ia64.h (TRULY_NOOP_TRUNCATION): Delete.
5964 * config/iq2000/iq2000.h (TRULY_NOOP_TRUNCATION): Delete.
5965 * config/lm32/lm32.h (TRULY_NOOP_TRUNCATION): Delete.
5966 * config/m32c/m32c.h (TRULY_NOOP_TRUNCATION): Delete.
5967 * config/m32r/m32r.h (TRULY_NOOP_TRUNCATION): Delete.
5968 * config/m68k/m68k.h (TRULY_NOOP_TRUNCATION): Delete.
5969 * config/mcore/mcore.h (TRULY_NOOP_TRUNCATION): Delete.
5970 * config/microblaze/microblaze.h (TRULY_NOOP_TRUNCATION): Delete.
5971 * config/mips/mips.h (TRULY_NOOP_TRUNCATION): Delete.
5972 * config/mips/mips.c (mips_truly_noop_truncation): New function.
5973 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
5974 * config/mips/mips.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5975 rather than TRULY_NOOP_TRUNCATION in comments.
5976 * config/mmix/mmix.h (TRULY_NOOP_TRUNCATION): Delete.
5977 * config/mn10300/mn10300.h (TRULY_NOOP_TRUNCATION): Delete.
5978 * config/moxie/moxie.h (TRULY_NOOP_TRUNCATION): Delete.
5979 * config/msp430/msp430.h (TRULY_NOOP_TRUNCATION): Delete.
5980 * config/nds32/nds32.h (TRULY_NOOP_TRUNCATION): Delete.
5981 * config/nios2/nios2.h (TRULY_NOOP_TRUNCATION): Delete.
5982 * config/nvptx/nvptx.h (TRULY_NOOP_TRUNCATION): Delete.
5983 * config/pa/pa.h (TRULY_NOOP_TRUNCATION): Delete.
5984 * config/pdp11/pdp11.h (TRULY_NOOP_TRUNCATION): Delete.
5985 * config/powerpcspe/powerpcspe.h (TRULY_NOOP_TRUNCATION): Delete.
5986 * config/riscv/riscv.h (TRULY_NOOP_TRUNCATION): Delete.
5987 * config/riscv/riscv.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
5988 rather than TRULY_NOOP_TRUNCATION in comments.
5989 * config/rl78/rl78.h (TRULY_NOOP_TRUNCATION): Delete.
5990 * config/rs6000/rs6000.h (TRULY_NOOP_TRUNCATION): Delete.
5991 * config/rx/rx.h (TRULY_NOOP_TRUNCATION): Delete.
5992 * config/s390/s390.h (TRULY_NOOP_TRUNCATION): Delete.
5993 * config/sh/sh.h (MAYBE_BASE_REGISTER_RTX_P): Remove
5994 TRULY_NOOP_TRUNCATION condition.
5995 (MAYBE_INDEX_REGISTER_RTX_P): Likewise.
5996 (TRULY_NOOP_TRUNCATION): Delete.
5997 * config/sparc/sparc.h (TRULY_NOOP_TRUNCATION): Delete.
5998 * config/spu/spu.h (TRULY_NOOP_TRUNCATION): Delete.
5999 * config/spu/spu.c (spu_truly_noop_truncation): New function.
6000 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6001 * config/stormy16/stormy16.h (TRULY_NOOP_TRUNCATION): Delete.
6002 * config/tilegx/tilegx.h (TRULY_NOOP_TRUNCATION): Delete.
6003 * config/tilegx/tilegx.c (tilegx_truly_noop_truncation): New fuction.
6004 (TARGET_TRULY_NOOP_TRUNCATION): Redefine.
6005 * config/tilegx/tilegx.md: Refer to TARGET_TRULY_NOOP_TRUNCATION
6006 rather than TRULY_NOOP_TRUNCATION in comments.
6007 * config/tilepro/tilepro.h (TRULY_NOOP_TRUNCATION): Delete.
6008 * config/v850/v850.h (TRULY_NOOP_TRUNCATION): Delete.
6009 * config/vax/vax.h (TRULY_NOOP_TRUNCATION): Delete.
6010 * config/visium/visium.h (TRULY_NOOP_TRUNCATION): Delete.
6011 * config/xtensa/xtensa.h (TRULY_NOOP_TRUNCATION): Delete.
6012 * system.h (TRULY_NOOP_TRUNCATION): Poison.
6014 2017-09-15 Christophe Lyon <christophe.lyon@linaro.org>
6017 * config/arm/arm.md (*cmp_and): Add enabled_for_depr_it attribute.
6018 (*cmp_ior): Likewise.
6019 (*ior_scc_scc): Add alternative for enabled_for_depr_it attribute.
6020 (*ior_scc_scc_cmp): Likewise.
6021 (*and_scc_scc): Likewise.
6022 (*and_scc_scc_cmp): Likewise.
6024 2017-09-15 Richard Sandiford <richard.sandiford@linaro.org>
6025 Alan Hayard <alan.hayward@arm.com>
6026 David Sherwood <david.sherwood@arm.com>
6028 * target.def (can_change_mode_class): New hook.
6029 (mode_rep_extended): Refer to it instead of CANNOT_CHANGE_MODE_CLASS.
6030 (hard_regno_nregs): Likewise.
6031 * hooks.h (hook_bool_mode_mode_reg_class_t_true): Declare.
6032 * hooks.c (hook_bool_mode_mode_reg_class_t_true): New function.
6033 * doc/tm.texi.in (CANNOT_CHANGE_MODE_CLASS): Replace with...
6034 (TARGET_CAN_CHANGE_MODE_CLASS): ...this.
6035 (LOAD_EXTEND_OP): Update accordingly.
6036 * doc/tm.texi: Regenerate.
6037 * doc/rtl.texi: Refer to TARGET_CAN_CHANGE_MODE_CLASS instead of
6038 CANNOT_CHANGE_MODE_CLASS.
6039 * hard-reg-set.h (REG_CANNOT_CHANGE_MODE_P): Replace with...
6040 (REG_CAN_CHANGE_MODE_P): ...this new macro.
6041 * combine.c (simplify_set): Update accordingly.
6042 * emit-rtl.c (validate_subreg): Likewise.
6043 * recog.c (general_operand): Likewise.
6044 * regcprop.c (mode_change_ok): Likewise.
6045 * reload1.c (choose_reload_regs): Likewise.
6046 (inherit_piecemeal_p): Likewise.
6047 * rtlanal.c (simplify_subreg_regno): Likewise.
6048 * postreload.c (reload_cse_simplify_set): Use REG_CAN_CHANGE_MODE_P
6049 instead of CANNOT_CHANGE_MODE_CLASS.
6050 (reload_cse_simplify_operands): Likewise.
6051 * reload.c (push_reload): Use targetm.can_change_mode_class
6052 instead of CANNOT_CHANGE_MODE_CLASS.
6053 (push_reload): Likewise. Also use REG_CAN_CHANGE_MODE_P instead of
6054 REG_CANNOT_CHANGE_MODE_P.
6055 * config/alpha/alpha.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6056 * config/alpha/alpha.c (alpha_can_change_mode_class): New function.
6057 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6058 * config/arm/arm.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6059 * config/arm/arm.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6060 (arm_can_change_mode_class): New function.
6061 * config/arm/neon.md: Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6062 than CANNOT_CHANGE_MODE_CLASS in comments.
6063 * config/i386/i386.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6064 * config/i386/i386-protos.h (ix86_cannot_change_mode_class): Delete.
6065 * config/i386/i386.c (ix86_cannot_change_mode_class): Replace with...
6066 (ix86_can_change_mode_class): ...this new function, inverting the
6067 sense of the return value.
6068 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6069 * config/ia64/ia64.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6070 * config/ia64/ia64.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6071 (ia64_can_change_mode_class): New function.
6072 * config/m32c/m32c.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6073 * config/m32c/m32c-protos.h (m32c_cannot_change_mode_class): Delete.
6074 * config/m32c/m32c.c (m32c_cannot_change_mode_class): Replace with...
6075 (m32c_can_change_mode_class): ...this new function, inverting the
6076 sense of the return value.
6077 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6078 * config/mips/mips.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6079 * config/mips/mips-protos.h (mips_cannot_change_mode_class): Delete.
6080 * config/mips/mips.c (mips_cannot_change_mode_class): Replace with...
6081 (mips_can_change_mode_class): ...this new function, inverting the
6082 sense of the return value.
6083 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6084 * config/msp430/msp430.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6085 * config/msp430/msp430.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6086 (msp430_can_change_mode_class): New function.
6087 * config/nvptx/nvptx.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6088 * config/nvptx/nvptx.c (nvptx_can_change_mode_class): New function.
6089 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6090 * config/pa/pa32-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6091 * config/pa/pa64-regs.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6092 * config/pa/pa-protos.h (pa_cannot_change_mode_class): Delete.
6093 * config/pa/pa.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6094 (pa_cannot_change_mode_class): Replace with...
6095 (pa_can_change_mode_class): ...this new function, inverting the
6096 sense of the return value.
6097 (pa_modes_tieable_p): Refer to TARGET_CAN_CHANGE_MODE_CLASS rather
6098 than CANNOT_CHANGE_MODE_CLASS in comments.
6099 * config/pdp11/pdp11.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6100 * config/pdp11/pdp11-protos.h (pdp11_cannot_change_mode_class): Delete.
6101 * config/pdp11/pdp11.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6102 (pdp11_cannot_change_mode_class): Replace with...
6103 (pdp11_can_change_mode_class): ...this new function, inverting the
6104 sense of the return value.
6105 * config/powerpcspe/powerpcspe.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6106 * config/powerpcspe/powerpcspe-protos.h
6107 (rs6000_cannot_change_mode_class_ptr): Delete.
6108 * config/powerpcspe/powerpcspe.c
6109 (rs6000_cannot_change_mode_class_ptr): Delete.
6110 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6111 (rs6000_option_override_internal): Assign to
6112 targetm.can_change_mode_class instead of
6113 rs6000_cannot_change_mode_class_ptr.
6114 (rs6000_cannot_change_mode_class): Replace with...
6115 (rs6000_can_change_mode_class): ...this new function, inverting the
6116 sense of the return value.
6117 (rs6000_debug_cannot_change_mode_class): Replace with...
6118 (rs6000_debug_can_change_mode_class): ...this new function.
6119 * config/riscv/riscv.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6120 * config/riscv/riscv.c (riscv_can_change_mode_class): New function.
6121 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6122 * config/rs6000/rs6000.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6123 * config/rs6000/rs6000-protos.h (rs6000_cannot_change_mode_class_ptr):
6125 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class_ptr): Delete.
6126 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6127 (rs6000_option_override_internal): Assign to
6128 targetm.can_change_mode_class instead of
6129 rs6000_cannot_change_mode_class_ptr.
6130 (rs6000_cannot_change_mode_class): Replace with...
6131 (rs6000_can_change_mode_class): ...this new function, inverting the
6132 sense of the return value.
6133 (rs6000_debug_cannot_change_mode_class): Replace with...
6134 (rs6000_debug_can_change_mode_class): ...this new function.
6135 * config/s390/s390.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6136 * config/s390/s390-protos.h (s390_cannot_change_mode_class): Delete.
6137 * config/s390/s390.c (s390_cannot_change_mode_class): Replace with...
6138 (s390_can_change_mode_class): ...this new function, inverting the
6139 sense of the return value.
6140 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6141 * config/sh/sh.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6142 * config/sh/sh-protos.h (sh_cannot_change_mode_class): Delete.
6143 * config/sh/sh.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6144 (sh_cannot_change_mode_class): Replace with...
6145 (sh_can_change_mode_class): ...this new function, inverting the
6146 sense of the return value.
6147 * config/sparc/sparc.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6148 * config/sparc/sparc.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6149 (sparc_can_change_mode_class): New function.
6150 * config/spu/spu.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6151 * config/spu/spu.c (spu_can_change_mode_class): New function.
6152 (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6153 * config/visium/visium.h (CANNOT_CHANGE_MODE_CLASS): Delete.
6154 * config/visium/visium.c (TARGET_CAN_CHANGE_MODE_CLASS): Redefine.
6155 (visium_can_change_mode_class): New function.
6156 * system.h (CANNOT_CHANGE_MODE_CLASS): Poison.
6158 2017-09-15 Richard Biener <rguenther@suse.de>
6160 PR tree-optimization/82217
6161 * tree-ssa-sccvn.c (visit_phi): Properly handle all VN_TOP
6162 but not undefined case.
6164 2017-09-15 Jakub Jelinek <jakub@redhat.com>
6167 * postreload.c (reload_cse_simplify_operands): Skip
6168 NOTE_INSN_DELETED_LABEL similarly to skipping CODE_LABEL.
6170 2017-09-15 Richard Biener <rguenther@suse.de>
6172 PR tree-optimization/68823
6173 * graphite-scop-detection.c (build_alias_set): If we have a
6174 possible dependence check whether we can handle them by just
6175 looking at the DRs DR_ACCESS_FNs.
6176 (build_scops): If build_alias_set fails, fail the SCOP.
6178 2017-09-14 Michael Meissner <meissner@linux.vnet.ibm.com>
6180 * config/rs6000/rs6000-builtin.def (BU_FLOAT128_1_HW): New macros
6181 to support float128 built-in functions that require the ISA 3.0
6183 (BU_FLOAT128_3_HW): Likewise.
6184 (SQRTF128): Add support for the IEEE 128-bit square root and fma
6186 (FMAF128): Likewise.
6188 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
6189 support for built-in functions that need the ISA 3.0 IEEE 128-bit
6190 floating point instructions.
6191 (rs6000_invalid_builtin): Likewise.
6192 (rs6000_builtin_mask_names): Likewise.
6193 * config/rs6000/rs6000.h (MASK_FLOAT128_HW): Likewise.
6194 (RS6000_BTM_FLOAT128_HW): Likewise.
6195 (RS6000_BTM_COMMON): Likewise.
6196 * config/rs6000/rs6000.md (fma<mode>4_hw): Add a generator
6198 * doc/extend.texi (RS/6000 built-in functions): Document the
6199 IEEE 128-bit floating point square root and fused multiply-add
6202 2017-09-14 Pat Haugen <pthaugen@us.ibm.com>
6204 * config/rs6000/rs6000.c (rs6000_set_up_by_prologue): Make sure the TOC
6205 reg (r2) isn't in the set of registers defined in the prologue.
6207 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6208 Alan Hayward <alan.hayward@arm.com>
6209 David Sherwood <david.sherwood@arm.com>
6211 * tree-vectorizer.h (_loop_vec_info): Add max_vectorization_factor.
6212 (LOOP_VINFO_MAX_VECT_FACTOR): New macro.
6213 (LOOP_VINFO_ORIG_VECT_FACTOR): Replace with...
6214 (LOOP_VINFO_ORIG_MAX_VECT_FACTOR): ...this new macro.
6215 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Update
6217 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
6218 max_vectorization_factor.
6219 (vect_analyze_loop_2): Set LOOP_VINFO_MAX_VECT_FACTOR.
6221 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6222 Alan Hayward <alan.hayward@arm.com>
6223 David Sherwood <david.sherwood@arm.com>
6225 * tree-vectorizer.h (vect_min_worthwhile_factor): Delete.
6226 (vect_worthwhile_without_simd_p): Declare.
6227 * tree-vect-loop.c (vect_worthwhile_without_simd_p): New function.
6228 (vectorizable_reduction): Use it.
6229 * tree-vect-stmts.c (vectorizable_shift): Likewise.
6230 (vectorizable_operation): Likewise.
6232 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6233 Alan Hayward <alan.hayward@arm.com>
6234 David Sherwood <david.sherwood@arm.com>
6236 * tree-vectorizer.h (vect_get_num_copies): New function.
6237 * tree-vect-data-refs.c (vect_get_data_access_cost): Use it.
6238 * tree-vect-loop.c (vectorizable_reduction): Likewise.
6239 (vectorizable_induction): Likewise.
6240 (vectorizable_live_operation): Likewise.
6241 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
6242 (vectorizable_bswap): Likewise.
6243 (vectorizable_call): Likewise.
6244 (vectorizable_conversion): Likewise.
6245 (vectorizable_assignment): Likewise.
6246 (vectorizable_shift): Likewise.
6247 (vectorizable_operation): Likewise.
6248 (vectorizable_store): Likewise.
6249 (vectorizable_load): Likewise.
6250 (vectorizable_condition): Likewise.
6251 (vectorizable_comparison): Likewise.
6252 (vect_analyze_stmt): Pass the slp node to vectorizable_live_operation.
6254 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6255 Alan Hayward <alan.hayward@arm.com>
6256 David Sherwood <david.sherwood@arm.com>
6258 * tree-vect-loop.c (vectorizable_induction): Use gimple_build instead
6259 of vect_init_vector.
6261 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6262 Alan Hayward <alan.hayward@arm.com>
6263 David Sherwood <david.sherwood@arm.com>
6265 * gimple-fold.h (gimple_build_vector_from_val): Declare, and provide
6266 an inline wrapper that provides a location.
6267 (gimple_build_vector): Likewise.
6268 * gimple-fold.c (gimple_build_vector_from_val): New function.
6269 (gimple_build_vector): Likewise.
6270 * tree-vect-loop.c (get_initial_def_for_reduction): Use the new
6271 functions to build the initial value. Always return a gimple value.
6272 (get_initial_defs_for_reduction): Likewise. Only compute
6274 (vect_create_epilog_for_reduction): Don't call force_gimple_operand or
6275 vect_init_vector on the results from get_initial_def(s)_for_reduction.
6276 (vectorizable_induction): Use gimple_build_vector rather than
6279 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6280 Alan Hayward <alan.hayward@arm.com>
6281 David Sherwood <david.sherwood@arm.com>
6283 * target.h (vec_perm_indices): New typedef.
6284 (auto_vec_perm_indices): Likewise.
6285 * optabs-query.h: Include target.h
6286 (can_vec_perm_p): Take a vec_perm_indices *.
6287 * optabs-query.c (can_vec_perm_p): Likewise.
6288 (can_mult_highpart_p): Update accordingly. Use auto_vec_perm_indices.
6289 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6290 * tree-vect-generic.c (lower_vec_perm): Likewise.
6291 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
6292 (vect_grouped_load_supported): Likewise.
6293 (vect_shift_permute_load_chain): Likewise.
6294 (vect_permute_store_chain): Use auto_vec_perm_indices.
6295 (vect_permute_load_chain): Likewise.
6296 * fold-const.c (fold_vec_perm): Take vec_perm_indices.
6297 (fold_ternary_loc): Update accordingly. Use auto_vec_perm_indices.
6298 Update uses of can_vec_perm_p.
6299 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Replace the
6300 mode with a number of elements. Take a vec_perm_indices *.
6301 (vect_create_epilog_for_reduction): Update accordingly.
6302 Use auto_vec_perm_indices.
6303 (have_whole_vector_shift): Likewise. Update call to can_vec_perm_p.
6304 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
6305 (vect_transform_slp_perm_load): Likewise.
6306 (vect_schedule_slp_instance): Use auto_vec_perm_indices.
6307 * tree-vectorizer.h (vect_gen_perm_mask_any): Take a vec_perm_indices.
6308 (vect_gen_perm_mask_checked): Likewise.
6309 * tree-vect-stmts.c (vect_gen_perm_mask_any): Take a vec_perm_indices.
6310 (vect_gen_perm_mask_checked): Likewise.
6311 (vectorizable_mask_load_store): Use auto_vec_perm_indices.
6312 (vectorizable_store): Likewise.
6313 (vectorizable_load): Likewise.
6314 (perm_mask_for_reverse): Likewise. Update call to can_vec_perm_p.
6315 (vectorizable_bswap): Likewise.
6317 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6318 Alan Hayward <alan.hayward@arm.com>
6319 David Sherwood <david.sherwood@arm.com>
6321 * tree.h (build_vector): Take a vec<tree> instead of a tree *.
6322 * tree.c (build_vector): Likewise.
6323 (build_vector_from_ctor): Update accordingly.
6324 (build_vector_from_val): Likewise.
6325 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
6326 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
6327 * tree-vect-generic.c (add_rshift): Likewise.
6328 (expand_vector_divmod): Likewise.
6329 (optimize_vector_constructor): Likewise.
6330 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
6331 (vect_transform_slp_perm_load): Likewise.
6332 (vect_schedule_slp_instance): Likewise.
6333 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
6334 (vectorizable_call): Likewise.
6335 (vect_gen_perm_mask_any): Likewise. Add elements in order.
6336 * expmed.c (make_tree): Likewise.
6337 * fold-const.c (fold_negate_expr_1): Use auto_vec<tree> when building
6338 a vector passed to build_vector.
6339 (fold_convert_const): Likewise.
6340 (exact_inverse): Likewise.
6341 (fold_ternary_loc): Likewise.
6342 (fold_relational_const): Likewise.
6343 (const_binop): Likewise. Use VECTOR_CST_ELT directly when operating
6344 on VECTOR_CSTs, rather than going through vec_cst_ctor_to_array.
6345 (const_unop): Likewise. Store the reduction accumulator in a
6346 variable rather than an array.
6347 (vec_cst_ctor_to_array): Take the number of elements as a parameter.
6348 (fold_vec_perm): Update calls accordingly. Use auto_vec<tree> for
6349 the new vector, rather than constructing it after the input arrays.
6350 (native_interpret_vector): Use auto_vec<tree> when building
6351 a vector passed to build_vector. Add elements in order.
6352 * tree-vect-loop.c (get_initial_defs_for_reduction): Use
6353 auto_vec<tree> when building a vector passed to build_vector.
6354 (vect_create_epilog_for_reduction): Likewise.
6355 (vectorizable_induction): Likewise.
6356 (get_initial_def_for_reduction): Likewise. Fix indentation of
6358 * config/sparc/sparc.c (sparc_handle_vis_mul8x16): Change n_elts
6360 (sparc_fold_builtin): Use auto_vec<tree> when building a vector
6361 passed to build_vector.
6363 2017-09-14 Richard Sandiford <richard.sandiford@linaro.org>
6364 Alan Hayward <alan.hayward@arm.com>
6365 David Sherwood <david.sherwood@arm.com>
6367 * tree-core.h (tree_base::u): Add an "nelts" field.
6368 (tree_vector): Use VECTOR_CST_NELTS as the length.
6369 * tree.c (tree_size): Likewise.
6370 (make_vector): Initialize VECTOR_CST_NELTS.
6371 * tree.h (VECTOR_CST_NELTS): Use the u.nelts field.
6372 * cfgexpand.c (expand_debug_expr): Use VECTOR_CST_NELTS instead of
6373 TYPE_VECTOR_SUBPARTS.
6374 * expr.c (const_vector_mask_from_tree): Consistently use "units"
6375 as the number of units, setting it from VECTOR_CST_NELTS.
6376 (const_vector_from_tree): Likewise.
6377 * fold-const.c (negate_expr_p): Use VECTOR_CST_NELTS instead of
6378 TYPE_VECTOR_SUBPARTS for the number of elements in a VECTOR_CST.
6379 (fold_negate_expr_1): Likewise.
6380 (fold_convert_const): Likewise.
6381 (const_binop): Likewise. Differentiate the number of output and
6383 (const_unop): Likewise.
6384 (fold_ternary_loc): Use VECTOR_CST_NELTS for the number of elements
6385 in a VECTOR_CST, asserting that it is the same as TYPE_VECTOR_SUBPARTS
6386 in cases that did the opposite.
6388 2017-09-14 Richard Biener <rguenther@suse.de>
6390 * tree-ssa-sccvn.c (visit_phi): Merge undefined values similar
6393 2017-09-14 Eric Botcazou <ebotcazou@adacore.com>
6395 * dwarf2out.c (dwarf2out_source_line): Remove superfluous test.
6397 2017-09-14 Jakub Jelinek <jakub@redhat.com>
6400 * cfgbuild.c (find_bb_boundaries): Ignore debug insns in decisions
6401 if and where to split a bb, except for splitting before debug insn
6402 sequences followed by non-label real insn. Delete debug insns
6403 in between basic blocks.
6405 * combine.c (make_compound_operation_int): Formatting fixes.
6407 * config/alpha/elf.h (LINK_EH_SPEC): Add -static-pie support.
6408 * config/alpha/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6409 * config/netbsd.h (LINK_EH_SPEC): Likewise.
6410 * config/sol2.h (LINK_EH_SPEC): Likewise.
6411 * config/arm/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6412 * config/s390/linux.h (LINK_SPEC): Likewise.
6413 * config/freebsd.h (LINK_EH_SPEC): Likewise.
6414 * config/openbsd.h (LINK_EH_SPEC): Likewise.
6415 * config/lm32/uclinux-elf.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6416 * config/aarch64/aarch64-linux.h (LINUX_TARGET_LINK_SPEC): Likewise.
6417 * config/powerpcspe/sysv4.h (LINK_EH_SPEC): Likewise.
6418 * config/bfin/linux.h (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6419 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Fix a typo.
6420 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Formatting fix.
6422 2017-09-13 Jakub Jelinek <jakub@redhat.com>
6424 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Add -static-pie
6426 (ENDFILE_LINUX_SPEC): Likewise.
6427 (LINK_EH_SPEC): Likewise.
6428 * config/rs6000/linux64.h (LINK_SHLIB_SPEC): Likewise.
6429 (LINK_OS_LINUX_SPEC32): Likewise.
6430 (LINK_OS_LINUX_SPEC64): Likewise.
6431 * config/rs6000/linux.h (LINK_SHLIB_SPEC): Likewise.
6432 (LINK_OS_LINUX_SPEC): Likewise.
6434 2017-09-13 Martin Liska <mliska@suse.cz>
6437 * stmt.c (expand_sjlj_dispatch_table): Use CASE_LOW when
6438 CASE_HIGH is NULL_TREE.
6440 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6441 Alan Hayward <alan.hayward@arm.com>
6442 David Sherwood <david.sherwood@arm.com>
6444 * target.def (secondary_memory_needed): New hook.
6445 (secondary_reload): Refer to TARGET_SECONDARY_MEMORY_NEEDED
6446 instead of SECONDARY_MEMORY_NEEDED.
6447 (secondary_memory_needed_mode): Likewise.
6448 * hooks.h (hook_bool_mode_reg_class_t_reg_class_t_false): Declare.
6449 * hooks.c (hook_bool_mode_reg_class_t_reg_class_t_false): New function.
6450 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED): Replace with...
6451 (TARGET_SECONDARY_MEMORY_NEEDED): ...this.
6452 (SECONDARY_MEMORY_NEEDED_RTX): Update reference accordingly.
6453 * doc/tm.texi: Regenerate.
6454 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED): Delete.
6455 * config/alpha/alpha.c (alpha_secondary_memory_needed): New function.
6456 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6457 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED): Delete.
6458 * config/i386/i386-protos.h (ix86_secondary_memory_needed): Delete.
6459 * config/i386/i386.c (inline_secondary_memory_needed): Put the
6460 mode argument first and change the reg_class arguments to reg_class_t.
6461 (ix86_secondary_memory_needed): Likewise. Remove the strict parameter.
6462 Make static. Update the call to inline_secondary_memory_needed.
6463 (ix86_register_move_cost): Update the call to
6464 inline_secondary_memory_needed.
6465 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6466 * config/ia64/ia64.h (SECONDARY_MEMORY_NEEDED): Delete commented-out
6468 * config/ia64/ia64.c (spill_xfmode_rfmode_operand): Refer to
6469 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6471 * config/mips/mips.h (SECONDARY_MEMORY_NEEDED): Delete.
6472 * config/mips/mips-protos.h (mips_secondary_memory_needed): Delete.
6473 * config/mips/mips.c (mips_secondary_memory_needed): Make static
6474 and match hook interface. Add comment from mips.h.
6475 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6476 * config/mmix/mmix.md (truncdfsf2): Refer to
6477 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6479 * config/pa/pa-64.h (SECONDARY_MEMORY_NEEDED): Rename to...
6480 (PA_SECONDARY_MEMORY_NEEDED): ...this, and put the mode argument first.
6481 * config/pa/pa.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6482 (pa_secondary_memory_needed): New function.
6483 * config/pdp11/pdp11.h (SECONDARY_MEMORY_NEEDED): Delete.
6484 * config/pdp11/pdp11-protos.h (pdp11_secondary_memory_needed): Delete.
6485 * config/pdp11/pdp11.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6486 (pdp11_secondary_memory_needed): Make static and match hook interface.
6487 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED): Delete.
6488 * config/powerpcspe/powerpcspe-protos.h
6489 (rs6000_secondary_memory_needed_ptr): Delete.
6490 * config/powerpcspe/powerpcspe.c (rs6000_secondary_memory_needed_ptr):
6492 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6493 (rs6000_option_override_internal): Assign to
6494 targetm.secondary_memory_needed rather than
6495 rs6000_secondary_memory_needed_ptr.
6496 (rs6000_secondary_memory_needed): Match hook interface.
6497 (rs6000_debug_secondary_memory_needed): Likewise.
6498 * config/riscv/riscv.h (SECONDARY_MEMORY_NEEDED): Delete.
6499 * config/riscv/riscv.c (riscv_secondary_memory_needed): New function.
6500 (riscv_register_move_cost): Use it instead of SECONDARY_MEMORY_NEEDED.
6501 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6502 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED): Delete.
6503 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_ptr):
6505 * config/rs6000/rs6000.c (rs6000_secondary_memory_needed_ptr): Delete.
6506 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6507 (rs6000_option_override_internal): Assign to
6508 targetm.secondary_memory_needed rather than
6509 rs6000_secondary_memory_needed_ptr.
6510 (rs6000_secondary_memory_needed): Match hook interface.
6511 (rs6000_debug_secondary_memory_needed): Likewise.
6512 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED): Delete.
6513 * config/s390/s390.c (s390_secondary_memory_needed): New function.
6514 (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6515 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED): Delete.
6516 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED): Redefine.
6517 (sparc_secondary_memory_needed): New function.
6518 * lra-constraints.c (check_and_process_move): Refer to
6519 TARGET_SECONDARY_MEMORY_NEEDED rather than SECONDARY_MEMORY_NEEDED
6521 (curr_insn_transform): Likewise.
6522 (process_alt_operands): Use targetm.secondary_memory_needed
6523 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6524 (check_secondary_memory_needed_p): Likewise.
6525 (choose_split_class): Likewise.
6526 * reload.c: Unconditionally include code that was previously
6527 conditional on SECONDARY_MEMORY_NEEDED.
6528 (push_secondary_reload): Use targetm.secondary_memory_needed
6529 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6530 (push_reload): Likewise.
6531 * reload1.c: Unconditionally include code that was previously
6532 conditional on SECONDARY_MEMORY_NEEDED.
6533 (choose_reload_regs): Use targetm.secondary_memory_needed
6534 instead of TARGET_SECONDARY_MEMORY_NEEDED.
6535 (gen_reload): Likewise.
6536 * system.h (SECONDARY_MEMORY_NEEDED): Poison.
6538 2017-09-13 Richard Sandiford <richard.sandiford@linaro.org>
6539 Alan Hayward <alan.hayward@arm.com>
6540 David Sherwood <david.sherwood@arm.com>
6542 * target.def (secondary_memory_needed_mode): New hook:
6543 * targhooks.c (default_secondary_memory_needed_mode): Declare.
6544 * targhooks.h (default_secondary_memory_needed_mode): New function.
6545 * doc/tm.texi.in (SECONDARY_MEMORY_NEEDED_MODE): Replace with...
6546 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): ...this.
6547 * doc/tm.texi: Regenerate.
6548 * lra-constraints.c (check_and_process_move): Use
6549 targetm.secondary_memory_needed_mode instead of
6550 TARGET_SECONDARY_MEMORY_NEEDED_MODE.
6551 (curr_insn_transform): Likewise.
6552 * reload.c (get_secondary_mem): Likewise.
6553 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6554 * config/alpha/alpha.c (alpha_secondary_memory_needed_mode): New
6556 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6557 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6558 * config/i386/i386.c (ix86_secondary_memory_needed_mode): New function.
6559 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6560 * config/powerpcspe/powerpcspe.h (SECONDARY_MEMORY_NEEDED_MODE):
6562 * config/powerpcspe/powerpcspe-protos.h
6563 (rs6000_secondary_memory_needed_mode): Delete.
6564 * config/powerpcspe/powerpcspe.c
6565 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6566 (rs6000_secondary_memory_needed_mode): Make static.
6567 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6568 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_mode):
6570 * config/rs6000/rs6000.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6572 (rs6000_secondary_memory_needed_mode): Make static.
6573 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6574 * config/s390/s390.c (s390_secondary_memory_needed_mode): New function.
6575 (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Redefine.
6576 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Delete.
6577 * config/sparc/sparc.c (TARGET_SECONDARY_MEMORY_NEEDED_MODE):
6579 (sparc_secondary_memory_needed_mode): New function.
6580 * system.h (TARGET_SECONDARY_MEMORY_NEEDED_MODE): Poison.
6582 2017-09-13 Jackson Woodruff <jackson.woodruff@arm.com>
6584 * config/aarch64/constraints.md (Umq): New constraint.
6585 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>):
6587 (mov<mode>): Update condition.
6589 2017-09-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6591 * gimple-ssa-store-merging.c (sort_by_bitpos): Compare store order
6592 when bitposition is the same.
6594 2017-09-13 Richard Biener <rguenther@suse.de>
6596 * dwarf2out.c (output_die_symbol): Remove.
6597 (output_die): Do not output a DIEs symbol.
6599 2017-09-13 Richard Biener <rguenther@suse.de>
6602 * gimple-fold.c (gimple_fold_call): Update SSA name in-place to
6603 default-def to avoid breaking iterator update with the weird
6604 interaction with cgraph_update_edges_for_call_stmt_node.
6606 2017-09-13 Richard Biener <rguenther@suse.de>
6608 * tree-cfg.c (verify_gimple_assign_binary): Add verification
6609 for WIDEN_SUM_EXPR, VEC_WIDEN_MULT_{HI,LO,EVEN,ODD}_EXPR,
6610 VEC_PACK_{TRUNC,SAT,FIX_TRUNC}_EXPR.
6611 (verify_gimple_assign_ternary): Add verification for DOT_PROD_EXPR.
6613 2017-09-13 Kugan Vivekanandarajah <kuganv@linaro.org>
6615 * config/aarch64/aarch64.c (aarch64_override_options_after_change_1):
6616 Disable pc relative literal load irrespective of
6617 TARGET_FIX_ERR_A53_84341 for default.
6619 2017-09-12 Eric Botcazou <ebotcazou@adacore.com>
6621 * config/sparc/sparc.c (output_return): Output the source location of
6622 the insn in the delay slot, if any.
6623 (output_sibcall): Likewise.
6625 2017-09-12 H.J. Lu <hongjiu.lu@intel.com>
6628 * common.opt (-static-pie): New alias.
6629 (shared): Negate static-pie.
6630 (-no-pie): Update help text.
6632 (static-pie): New option.
6633 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Add
6634 -static-pie support.
6635 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
6636 (LINK_EH_SPEC): Likewise.
6637 (LINK_GCC_C_SEQUENCE_SPEC): Likewise.
6638 * config/i386/gnu-user.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
6639 * config/i386/gnu-user64.h (GNU_USER_TARGET_LINK_SPEC): Likewise.
6640 * gcc.c (LINK_COMMAND_SPEC): Likewise.
6641 (init_gcc_specs): Likewise.
6642 (init_spec): Likewise.
6643 (display_help): Update help message for -pie.
6644 * doc/invoke.texi: Update -pie, -no-pie and -static. Document
6647 2017-09-12 Wilco Dijkstra <wdijkstr@arm.com>
6649 * config/aarch64/aarch64.md (movsi_aarch64): Remove all '*'.
6650 (movdi_aarch64): Likewise.
6651 (movti_aarch64): Likewise.
6653 2017-09-12 Simon Wright <simon@pushface.org>
6656 * config/darwin-driver.c (darwin_find_version_from_kernel): Eliminate
6657 calculation of the minor version, always output as 0.
6659 2017-09-12 Jakub Jelinek <jakub@redhat.com>
6662 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
6663 ALTIVEC_BUILTIN_VEC_LD if arg1 has array type call default_conversion
6664 on it early, rather than manual conversion late. For
6665 ALTIVEC_BUILTIN_VEC_ST if arg2 has array type call default_conversion
6666 instead of performing manual conversion.
6668 2017-09-12 Carl Love <cel@us.ibm.com>
6670 * config/rs6000/altivec.md (vec_widen_umult_even_v4si,
6671 vec_widen_smult_even_v4si): Add define expands for vmuleuw, vmulesw,
6673 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
6674 VMULOSW): Add definitions.
6675 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
6676 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
6677 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
6678 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
6679 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
6681 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
6683 * config/aarch64/aarch64.md (movdi_aarch64): Set load/store
6685 (movti_aarch64): Likewise.
6686 (movdf_aarch64): Likewise.
6687 (movtf_aarch64): Likewise.
6688 (load_pairdi): Likewise.
6689 (store_pairdi): Likewise.
6690 (load_pairdf): Likewise.
6691 (store_pairdf): Likewise.
6692 (loadwb_pair<GPI:mode>_<P:mode>): Likewise.
6693 (storewb_pair<GPI:mode>_<P:mode>): Likewise.
6694 (ldr_got_small_<mode>): Likewise.
6695 (ldr_got_small_28k_<mode>): Likewise.
6696 (ldr_got_tiny): Likewise.
6697 * config/aarch64/iterators.md (ldst_sz): New.
6698 (ldpstp_sz): Likewise.
6699 * config/aarch64/thunderx.md (thunderx_storepair): Split store_8
6701 (thunderx_load): Split load_8 to load_16.
6702 * config/aarch64/thunderx2t99.md (thunderx2t99_loadpair): Split
6704 (thunderx2t99_storepair_basic): Split store_8 to store_16.
6705 * config/arm/xgene1.md (xgene1_load_pair): Split load_8 to load_16.
6706 (xgene1_store_pair): Split store_8 to store_16.
6707 * config/aarch64/falkor.md (falkor_ld_3_ld): Split load_8 to load_16.
6708 (falkor_st_0_st_sd): Split store_8 to store_16.
6710 2017-09-12 James Greenhalgh <james.greenhalgh@arm.com>
6712 * config/arm/types.md (type): Rename load1/2/3/4 to load_4/8/12/16
6713 and store1/2/3/4 to store_4/8/12/16.
6714 * config/aarch64/aarch64.md: Update for rename.
6715 * config/arm/arm.md: Likewise.: Likewise.
6716 * config/arm/arm.c: Likewise.
6717 * config/arm/thumb1.md: Likewise.
6718 * config/arm/thumb2.md: Likewise.
6719 * config/arm/vfp.md: Likewise.
6720 * config/arm/arm-generic.md: Likewise.
6721 * config/arm/arm1020e.md: Likewise.
6722 * config/arm/arm1026ejs.md: Likewise.
6723 * config/arm/arm1136jfs.md: Likewise.
6724 * config/arm/arm926ejs.md: Likewise.
6725 * config/arm/cortex-a15.md: Likewise.
6726 * config/arm/cortex-a17.md: Likewise.
6727 * config/arm/cortex-a5.md: Likewise.
6728 * config/arm/cortex-a53.md: Likewise.
6729 * config/arm/cortex-a57.md: Likewise.
6730 * config/arm/cortex-a7.md: Likewise.
6731 * config/arm/cortex-a8.md: Likewise.
6732 * config/arm/cortex-a9.md: Likewise.
6733 * config/arm/cortex-m4.md: Likewise.
6734 * config/arm/cortex-m7.md: Likewise.
6735 * config/arm/cortex-r4.md: Likewise.
6736 * config/arm/exynos-m1.md: Likewise.
6737 * config/arm/fa526.md: Likewise.
6738 * config/arm/fa606te.md: Likewise.
6739 * config/arm/fa626te.md: Likewise.
6740 * config/arm/fa726te.md: Likewise.
6741 * config/arm/fmp626.md: Likewise.
6742 * config/arm/iwmmxt.md: Likewise.
6743 * config/arm/ldmstm.md: Likewise.
6744 * config/arm/marvell-pj4.md: Likewise.
6745 * config/arm/xgene1.md: Likewise.
6746 * config/aarch64/thunderx.md: Likewise.
6747 * config/aarch64/thunderx2t99.md: Likewise.
6748 * config/aarch64/falkor.md: Likewise.
6750 2017-09-12 Martin Liska <mliska@suse.cz>
6752 * attribs.c (private_lookup_attribute): New function.
6753 * attribs.h (private_lookup_attribute): Declared here.
6754 (lookup_attribute): Called from this place.
6756 2017-09-12 Richard Biener <rguenther@suse.de>
6758 PR tree-optimization/82157
6759 * tree-ssa-pre.c (remove_dead_inserted_code): Do not remove
6760 stmts with side-effects.
6762 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6763 Alan Hayward <alan.hayward@arm.com>
6764 David Sherwood <david.sherwood@arm.com>
6766 * target.def (hard_regno_nregs): New hook.
6767 (class_max_nregs): Refer to it instead of HARD_REGNO_NREGS.
6768 * targhooks.h (default_hard_regno_nregs): Declare.
6769 * targhooks.c (default_hard_regno_nregs): New function.
6770 * doc/tm.texi.in (HARD_REGNO_NREGS): Replace with...
6771 (TARGET_HARD_REGNO_NREGS): ...this hook.
6772 (HARD_REGNO_NREGS_HAS_PADDING): Update accordingly.
6773 (CLASS_MAX_NREGS): Likewise.
6774 * doc/tm.texi: Regenerate.
6775 * reginfo.c (init_reg_modes_target): Use targetm.hard_regno_nregs
6776 instead of HARD_REGNO_NREGS.
6777 * rtl.h (REG_NREGS): Refer to TARGET_HARD_REGNO_NREGS rather than
6778 HARD_REGNO_NREGS in the comment.
6779 * config/aarch64/aarch64.h (HARD_REGNO_NREGS): Delete.
6780 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_nregs): Delete.
6781 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs): Make static.
6782 Return an unsigned int.
6783 (TARGET_HARD_REGNO_NREGS): Redefine.
6784 * config/alpha/alpha.h (HARD_REGNO_NREGS): Delete.
6785 * config/arc/arc.h (HARD_REGNO_NREGS): Delete.
6786 * config/arc/arc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6787 (arc_hard_regno_nregs): New function.
6788 * config/arm/arm.h (HARD_REGNO_NREGS): Delete.
6789 * config/arm/arm.c (TARGET_HARD_REGNO_NREGS): Redefine.
6790 (arm_hard_regno_nregs): New function.
6791 * config/avr/avr.h (HARD_REGNO_NREGS): Delete.
6792 * config/bfin/bfin.h (HARD_REGNO_NREGS): Delete.
6793 * config/bfin/bfin.c (bfin_hard_regno_nregs): New function.
6794 (TARGET_HARD_REGNO_NREGS): Redefine.
6795 * config/c6x/c6x.h (HARD_REGNO_NREGS): Delete.
6796 * config/cr16/cr16.h (LONG_REG_P): Use targetm.hard_regno_nregs.
6797 (HARD_REGNO_NREGS): Delete.
6798 * config/cr16/cr16.c (TARGET_HARD_REGNO_NREGS): Redefine.
6799 (cr16_hard_regno_nregs): New function.
6800 (cr16_memory_move_cost): Use it instead of HARD_REGNO_NREGS.
6801 * config/cris/cris.h (HARD_REGNO_NREGS): Delete.
6802 * config/cris/cris.c (TARGET_HARD_REGNO_NREGS): Redefine.
6803 (cris_hard_regno_nregs): New function.
6804 * config/epiphany/epiphany.h (HARD_REGNO_NREGS): Delete.
6805 * config/fr30/fr30.h (HARD_REGNO_NREGS): Delete.
6806 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6807 * config/frv/frv.h (HARD_REGNO_NREGS): Delete.
6808 (CLASS_MAX_NREGS): Remove outdated copy of documentation.
6809 * config/frv/frv-protos.h (frv_hard_regno_nregs): Delete.
6810 * config/frv/frv.c (TARGET_HARD_REGNO_NREGS): Redefine.
6811 (frv_hard_regno_nregs): Make static. Take and return an
6813 (frv_class_max_nregs): Remove outdated copy of documentation.
6814 * config/ft32/ft32.h (HARD_REGNO_NREGS): Delete.
6815 * config/h8300/h8300.h (HARD_REGNO_NREGS): Delete.
6816 * config/h8300/h8300-protos.h (h8300_hard_regno_nregs): Delete.
6817 * config/h8300/h8300.c (h8300_hard_regno_nregs): Delete.
6818 * config/i386/i386.h (HARD_REGNO_NREGS): Delete.
6819 * config/i386/i386.c (ix86_hard_regno_nregs): New function.
6820 (TARGET_HARD_REGNO_NREGS): Redefine.
6821 * config/ia64/ia64.h (HARD_REGNO_NREGS): Delete.
6822 (CLASS_MAX_NREGS): Update comment.
6823 * config/ia64/ia64.c (TARGET_HARD_REGNO_NREGS): Redefine.
6824 (ia64_hard_regno_nregs): New function.
6825 * config/iq2000/iq2000.h (HARD_REGNO_NREGS): Delete.
6826 * config/lm32/lm32.h (HARD_REGNO_NREGS): Delete.
6827 * config/m32c/m32c.h (HARD_REGNO_NREGS): Delete.
6828 * config/m32c/m32c-protos.h (m32c_hard_regno_nregs): Delete.
6829 * config/m32c/m32c.c (m32c_hard_regno_nregs_1): Take and return
6831 (m32c_hard_regno_nregs): Likewise. Make static.
6832 (TARGET_HARD_REGNO_NREGS): Redefine.
6833 * config/m32r/m32r.h (HARD_REGNO_NREGS): Delete.
6834 * config/m68k/m68k.h (HARD_REGNO_NREGS): Delete.
6835 * config/m68k/m68k.c (TARGET_HARD_REGNO_NREGS): Redefine.
6836 (m68k_hard_regno_nregs): New function.
6837 * config/mcore/mcore.h (HARD_REGNO_NREGS): Delete.
6838 * config/microblaze/microblaze.h (HARD_REGNO_NREGS): Delete.
6839 * config/mips/mips.h (HARD_REGNO_NREGS): Delete.
6840 * config/mips/mips-protos.h (mips_hard_regno_nregs): Delete.
6841 * config/mips/mips.c (mips_hard_regno_nregs): Make static.
6842 Take and return an unsigned int.
6843 (TARGET_HARD_REGNO_NREGS): Redefine.
6844 * config/mmix/mmix.h (HARD_REGNO_NREGS): Delete.
6845 (CLASS_MAX_NREGS): Use targetm.hard_regno_nregs.
6846 * config/mn10300/mn10300.h (HARD_REGNO_NREGS): Delete.
6847 * config/moxie/moxie.h (HARD_REGNO_NREGS): Delete.
6848 * config/msp430/msp430.h (HARD_REGNO_NREGS): Delete.
6849 * config/msp430/msp430-protos.h (msp430_hard_regno_nregs): Delete.
6850 * config/msp430/msp430.c (TARGET_HARD_REGNO_NREGS): Redefine.
6851 (msp430_hard_regno_nregs): Make static. Take and return an
6853 * config/nds32/nds32.h (HARD_REGNO_NREGS): Delete.
6854 * config/nds32/nds32-protos.h (nds32_hard_regno_nregs): Delete.
6855 * config/nds32/nds32.c (nds32_hard_regno_nregs): Delete.
6856 (nds32_hard_regno_mode_ok): Use targetm.hard_regno_nregs.
6857 * config/nios2/nios2.h (HARD_REGNO_NREGS): Delete.
6858 * config/nvptx/nvptx.h (HARD_REGNO_NREGS): Delete.
6859 * config/nvptx/nvptx.c (nvptx_hard_regno_nregs): New function.
6860 (TARGET_HARD_REGNO_NREGS): Redefine.
6861 * config/pa/pa32-regs.h (HARD_REGNO_NREGS): Rename to...
6862 (PA_HARD_REGNO_NREGS): ...this.
6863 * config/pa/pa64-regs.h (HARD_REGNO_NREGS): Rename to...
6864 (PA_HARD_REGNO_NREGS): ...this.
6865 * config/pa/pa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6866 (pa_hard_regno_nregs): New function.
6867 * config/pdp11/pdp11.h (HARD_REGNO_NREGS): Delete.
6868 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_NREGS): Redefine.
6869 (pdp11_hard_regno_nregs): New function.
6870 * config/powerpcspe/powerpcspe.h (HARD_REGNO_NREGS): Delete.
6871 * config/powerpcspe/powerpcspe.c (TARGET_HARD_REGNO_NREGS): Redefine.
6872 (rs6000_hard_regno_nregs_hook): New function.
6873 * config/riscv/riscv.h (HARD_REGNO_NREGS): Delete.
6874 * config/riscv/riscv-protos.h (riscv_hard_regno_nregs): Delete.
6875 * config/riscv/riscv.c (riscv_hard_regno_nregs): Make static.
6876 Take and return an unsigned int. Move earlier in file.
6877 (TARGET_HARD_REGNO_NREGS): Redefine.
6878 * config/rl78/rl78.h (HARD_REGNO_NREGS): Delete.
6879 * config/rl78/rl78-protos.h (rl78_hard_regno_nregs): Delete.
6880 * config/rl78/rl78.c (TARGET_HARD_REGNO_NREGS): Reefine.
6881 (rl78_hard_regno_nregs): Make static. Take and return an
6883 * config/rs6000/rs6000.h (HARD_REGNO_NREGS): Delete.
6884 * config/rs6000/rs6000.c (TARGET_HARD_REGNO_NREGS): Redefine.
6885 (rs6000_hard_regno_nregs_hook): New function.
6886 * config/rx/rx.h (HARD_REGNO_NREGS): Delete.
6887 * config/rx/rx.c (rx_hard_regno_nregs): New function.
6888 (TARGET_HARD_REGNO_NREGS): Redefine.
6889 * config/s390/s390.h (HARD_REGNO_NREGS): Delete.
6890 * config/s390/s390.c (REGNO_PAIR_OK): Use s390_hard_regno_nregs
6891 instead of HARD_REGNO_NREGS.
6892 (s390_hard_regno_nregs): New function.
6893 (s390_hard_regno_mode_ok): Add comment from s390.h.
6894 (TARGET_HARD_REGNO_NREGS): Redefine.
6895 * config/sh/sh.h (HARD_REGNO_NREGS): Delete.
6896 * config/sh/sh.c (TARGET_HARD_REGNO_NREGS): Redefine.
6897 (sh_hard_regno_nregs): New function.
6898 (sh_pass_in_reg_p): Use it.
6899 * config/sparc/sparc.h (HARD_REGNO_NREGS): Delete.
6900 * config/sparc/sparc.c (TARGET_HARD_REGNO_NREGS): Redefine.
6901 (sparc_hard_regno_nregs): New function.
6902 * config/spu/spu.h (HARD_REGNO_NREGS): Delete.
6903 * config/spu/spu.c (spu_hard_regno_nregs): New function.
6904 (spu_function_arg_advance): Use it, supplying a valid register number.
6905 (TARGET_HARD_REGNO_NREGS): Redefine.
6906 * config/stormy16/stormy16.h (HARD_REGNO_NREGS): Delete.
6907 * config/tilegx/tilegx.h (HARD_REGNO_NREGS): Delete.
6908 * config/tilepro/tilepro.h (HARD_REGNO_NREGS): Delete.
6909 * config/v850/v850.h (HARD_REGNO_NREGS): Delete.
6910 * config/vax/vax.h (HARD_REGNO_NREGS): Delete.
6911 * config/visium/visium.h (HARD_REGNO_NREGS): Delete.
6912 (CLASS_MAX_NREGS): Remove copy of old documentation.
6913 * config/visium/visium.c (TARGET_HARD_REGNO_NREGS): Redefine.
6914 (visium_hard_regno_nregs): New function.
6915 (visium_hard_regno_mode_ok): Use it instead of HARD_REGNO_NREGS.
6916 * config/xtensa/xtensa.h (HARD_REGNO_NREGS): Delete.
6917 * config/xtensa/xtensa.c (TARGET_HARD_REGNO_NREGS): Redefine.
6918 xtensa_hard_regno_nregs): New function.
6919 * system.h (HARD_REGNO_NREGS): Poison.
6921 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6923 * config/arm/arm.h (THUMB_SECONDARY_INPUT_RELOAD_CLASS): Use
6924 hard_regno_nregs instead of HARD_REGNO_NREGS.
6925 (THUMB_SECONDARY_OUTPUT_RELOAD_CLASS): Likewise.
6926 * config/c6x/c6x.c (c6x_expand_prologue): Likewise.
6927 (c6x_expand_epilogue): Likewise.
6928 * config/frv/frv.c (frv_alloc_temp_reg): Likewise.
6929 (frv_read_iacc_argument): Likewise.
6930 * config/sh/sh.c: Include regs.h.
6931 (sh_print_operand): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6932 (regs_used): Likewise.
6933 (output_stack_adjust): Likewise.
6934 * config/xtensa/xtensa.c (xtensa_copy_incoming_a7): Likewise.
6935 * expmed.c: Include regs.h.
6936 (store_bit_field_1): Use hard_regno_nregs instead of HARD_REGNO_NREGS.
6937 * ree.c: Include regs.h.
6938 (combine_reaching_defs): Use hard_regno_nregs instead of
6940 (add_removable_extension): Likewise.
6942 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
6944 * regs.h (hard_regno_nregs): Turn into a function.
6945 (end_hard_regno): Update accordingly.
6946 * caller-save.c (setup_save_areas): Likewise.
6947 (save_call_clobbered_regs): Likewise.
6948 (replace_reg_with_saved_mem): Likewise.
6949 (insert_restore): Likewise.
6950 (insert_save): Likewise.
6951 * combine.c (can_change_dest_mode): Likewise.
6952 (move_deaths): Likewise.
6953 (distribute_notes): Likewise.
6954 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): Likewise.
6955 * config/powerpcspe/powerpcspe.c (rs6000_cannot_change_mode_class)
6956 (rs6000_split_multireg_move): Likewise.
6957 (rs6000_register_move_cost): Likewise.
6958 (rs6000_memory_move_cost): Likewise.
6959 * config/rs6000/rs6000.c (rs6000_cannot_change_mode_class): Likewise.
6960 (rs6000_split_multireg_move): Likewise.
6961 (rs6000_register_move_cost): Likewise.
6962 (rs6000_memory_move_cost): Likewise.
6963 * cselib.c (cselib_reset_table): Likewise.
6964 (cselib_lookup_1): Likewise.
6965 * emit-rtl.c (set_mode_and_regno): Likewise.
6966 * function.c (aggregate_value_p): Likewise.
6967 * ira-color.c (setup_profitable_hard_regs): Likewise.
6968 (check_hard_reg_p): Likewise.
6969 (calculate_saved_nregs): Likewise.
6970 (assign_hard_reg): Likewise.
6971 (improve_allocation): Likewise.
6972 (calculate_spill_cost): Likewise.
6973 * ira-emit.c (modify_move_list): Likewise.
6974 * ira-int.h (ira_hard_reg_set_intersection_p): Likewise.
6975 (ira_hard_reg_in_set_p): Likewise.
6976 * ira.c (setup_reg_mode_hard_regset): Likewise.
6977 (clarify_prohibited_class_mode_regs): Likewise.
6978 (check_allocation): Likewise.
6979 * lra-assigns.c (find_hard_regno_for_1): Likewise.
6980 (lra_setup_reg_renumber): Likewise.
6981 (setup_try_hard_regno_pseudos): Likewise.
6982 (spill_for): Likewise.
6983 (assign_hard_regno): Likewise.
6984 (setup_live_pseudos_and_spill_after_risky_transforms): Likewise.
6985 * lra-constraints.c (in_class_p): Likewise.
6986 (lra_constraint_offset): Likewise.
6987 (simplify_operand_subreg): Likewise.
6988 (lra_constraints): Likewise.
6989 (split_reg): Likewise.
6990 (split_if_necessary): Likewise.
6991 (invariant_p): Likewise.
6992 (inherit_in_ebb): Likewise.
6993 * lra-lives.c (process_bb_lives): Likewise.
6994 * lra-remat.c (reg_overlap_for_remat_p): Likewise.
6995 (get_hard_regs): Likewise.
6996 (do_remat): Likewise.
6997 * lra-spills.c (assign_spill_hard_regs): Likewise.
6998 * mode-switching.c (create_pre_exit): Likewise.
6999 * postreload.c (reload_combine_recognize_pattern): Likewise.
7000 * recog.c (peep2_find_free_register): Likewise.
7001 * regcprop.c (kill_value_regno): Likewise.
7002 (set_value_regno): Likewise.
7003 (copy_value): Likewise.
7004 (maybe_mode_change): Likewise.
7005 (find_oldest_value_reg): Likewise.
7006 (copyprop_hardreg_forward_1): Likewise.
7007 * regrename.c (check_new_reg_p): Likewise.
7008 (regrename_do_replace): Likewise.
7009 * reload.c (push_reload): Likewise.
7010 (combine_reloads): Likewise.
7011 (find_dummy_reload): Likewise.
7012 (operands_match_p): Likewise.
7013 (find_reloads): Likewise.
7014 (find_equiv_reg): Likewise.
7015 (reload_adjust_reg_for_mode): Likewise.
7016 * reload1.c (count_pseudo): Likewise.
7017 (count_spilled_pseudo): Likewise.
7018 (find_reg): Likewise.
7019 (clear_reload_reg_in_use): Likewise.
7020 (free_for_value_p): Likewise.
7021 (allocate_reload_reg): Likewise.
7022 (choose_reload_regs): Likewise.
7023 (reload_adjust_reg_for_temp): Likewise.
7024 (emit_reload_insns): Likewise.
7025 (delete_output_reload): Likewise.
7026 * rtlanal.c (subreg_get_info): Likewise.
7027 * sched-deps.c (sched_analyze_reg): Likewise.
7028 * sel-sched.c (init_regs_for_mode): Likewise.
7029 (mark_unavailable_hard_regs): Likewise.
7030 (choose_best_reg_1): Likewise.
7031 (verify_target_availability): Likewise.
7032 * valtrack.c (dead_debug_insert_temp): Likewise.
7033 * var-tracking.c (track_loc_p): Likewise.
7034 (emit_note_insn_var_location): Likewise.
7035 * varasm.c (make_decl_rtl): Likewise.
7036 * reginfo.c (choose_hard_reg_mode): Likewise.
7037 (init_reg_modes_target): Refer directly to
7038 this_target_regs->x_hard_regno_nregs.
7040 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7042 * ira-costs.c (record_operand_costs): Use in_hard_reg_set_p
7043 instead of hard_regno_nregs.
7045 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7047 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Use
7048 end_hard_regno instead of hard_regno_nregs.
7049 * config/s390/s390.c (s390_reg_clobbered_rtx): Likewise.
7050 * config/sparc/sparc.h (ASM_DECLARE_REGISTER_GLOBAL): Likewise.
7051 * config/visium/visium.c (visium_hard_regno_mode_ok): Likewise.
7052 * ira-color.c (improve_allocation): Likewise.
7053 * lra-assigns.c (find_hard_regno_for_1): Likewise.
7054 * lra-lives.c (mark_regno_live): Likewise.
7055 (mark_regno_dead): Likewise.
7056 * lra-remat.c (operand_to_remat): Likewise.
7057 * lra.c (collect_non_operand_hard_regs): Likewise.
7058 * postreload.c (reload_combine_note_store): Likewise.
7059 (move2add_valid_value_p): Likewise.
7060 * reload.c (regno_clobbered_p): Likewise.
7062 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7064 * config/frv/frv.c (FOR_EACH_REGNO): Use END_REGNO instead of
7066 * config/v850/v850.c (v850_reorg): Likewise.
7067 * reload.c (refers_to_regno_for_reload_p): Likewise.
7068 (find_equiv_reg): Likewise.
7069 * reload1.c (reload_reg_reaches_end_p): Likewise.
7071 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7073 * caller-save.c (add_used_regs): Use REG_NREGS instead of
7075 * config/aarch64/aarch64.c (aarch64_split_combinev16qi): Likewise.
7076 * config/arm/arm.c (output_move_neon): Likewise.
7077 (arm_attr_length_move_neon): Likewise.
7078 (neon_split_vcombine): Likewise.
7079 * config/c6x/c6x.c (c6x_mark_reg_read): Likewise.
7080 (c6x_mark_reg_written): Likewise.
7081 (c6x_dwarf_register_span): Likewise.
7082 * config/i386/i386.c (ix86_save_reg): Likewise.
7083 * config/ia64/ia64.c (mark_reg_gr_used_mask): Likewise.
7084 (rws_access_reg): Likewise.
7085 * config/s390/s390.c (s390_call_saved_register_used): Likewise.
7086 * mode-switching.c (create_pre_exit): Likewise.
7087 * ree.c (combine_reaching_defs): Likewise.
7088 (add_removable_extension): Likewise.
7089 * regcprop.c (find_oldest_value_reg): Likewise.
7090 (copyprop_hardreg_forward_1): Likewise.
7091 * reload.c (reload_inner_reg_of_subreg): Likewise.
7092 (push_reload): Likewise.
7093 (combine_reloads): Likewise.
7094 (find_dummy_reload): Likewise.
7095 (reload_adjust_reg_for_mode): Likewise.
7096 * reload1.c (find_reload_regs): Likewise.
7097 (forget_old_reloads_1): Likewise.
7098 (reload_reg_free_for_value_p): Likewise.
7099 (reload_adjust_reg_for_temp): Likewise.
7100 (emit_reload_insns): Likewise.
7101 (delete_output_reload): Likewise.
7102 * sel-sched.c (choose_best_reg_1): Likewise.
7103 (choose_best_pseudo_reg): Likewise.
7105 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7106 Alan Hayward <alan.hayward@arm.com>
7107 David Sherwood <david.sherwood@arm.com>
7109 * defaults.h (SLOW_UNALIGNED_ACCESS): Delete.
7110 * target.def (slow_unaligned_access): New hook.
7111 * targhooks.h (default_slow_unaligned_access): Declare.
7112 * targhooks.c (default_slow_unaligned_access): New function.
7113 * doc/tm.texi.in (SLOW_UNALIGNED_ACCESS): Replace with...
7114 (TARGET_SLOW_UNALIGNED_ACCESS): ...this.
7115 * doc/tm.texi: Regenerate.
7116 * config/alpha/alpha.h (SLOW_UNALIGNED_ACCESS): Delete.
7117 * config/arm/arm.h (SLOW_UNALIGNED_ACCESS): Delete.
7118 * config/i386/i386.h (SLOW_UNALIGNED_ACCESS): Delete commented-out
7120 * config/powerpcspe/powerpcspe.h (SLOW_UNALIGNED_ACCESS): Delete.
7121 * config/powerpcspe/powerpcspe.c (TARGET_SLOW_UNALIGNED_ACCESS):
7123 (rs6000_slow_unaligned_access): New function.
7124 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7125 (expand_block_compare): Likewise.
7126 (expand_strn_compare): Likewise.
7127 (rs6000_rtx_costs): Likewise.
7128 * config/riscv/riscv.h (SLOW_UNALIGNED_ACCESS): Delete.
7129 (riscv_slow_unaligned_access): Likewise.
7130 * config/riscv/riscv.c (riscv_slow_unaligned_access): Rename to...
7131 (riscv_slow_unaligned_access_p): ...this and make static.
7132 (riscv_option_override): Update accordingly.
7133 (riscv_slow_unaligned_access): New function.
7134 (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7135 * config/rs6000/rs6000.h (SLOW_UNALIGNED_ACCESS): Delete.
7136 * config/rs6000/rs6000.c (TARGET_SLOW_UNALIGNED_ACCESS): Redefine.
7137 (rs6000_slow_unaligned_access): New function.
7138 (rs6000_emit_move): Use it instead of SLOW_UNALIGNED_ACCESS.
7139 (rs6000_rtx_costs): Likewise.
7140 * config/rs6000/rs6000-string.c (expand_block_compare)
7141 (expand_strn_compare): Use targetm.slow_unaligned_access instead
7142 of SLOW_UNALIGNED_ACCESS.
7143 * config/tilegx/tilegx.h (SLOW_UNALIGNED_ACCESS): Delete.
7144 * config/tilepro/tilepro.h (SLOW_UNALIGNED_ACCESS): Delete.
7145 * calls.c (expand_call): Use targetm.slow_unaligned_access instead
7146 of SLOW_UNALIGNED_ACCESS.
7147 * expmed.c (simple_mem_bitfield_p): Likewise.
7148 * expr.c (alignment_for_piecewise_move): Likewise.
7149 (emit_group_load_1): Likewise.
7150 (emit_group_store): Likewise.
7151 (copy_blkmode_from_reg): Likewise.
7152 (emit_push_insn): Likewise.
7153 (expand_assignment): Likewise.
7154 (store_field): Likewise.
7155 (expand_expr_real_1): Likewise.
7156 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
7157 * lra-constraints.c (simplify_operand_subreg): Likewise.
7158 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
7159 * gimple-ssa-store-merging.c: Likewise in block comment at start
7161 * tree-ssa-strlen.c: Include target.h.
7162 (handle_builtin_memcmp): Use targetm.slow_unaligned_access instead
7163 of SLOW_UNALIGNED_ACCESS.
7164 * system.h (SLOW_UNALIGNED_ACCESS): Poison.
7166 2017-09-12 Richard Sandiford <richard.sandiford@linaro.org>
7168 PR rtl-optimization/82185
7169 * expmed.c (emit_store_flag_int): Only test tem if it has been
7172 2017-09-12 Richard Biener <rguenther@suse.de>
7175 * match.pd ((FTYPE) N CMP CST): Fix typo.
7177 2017-09-12 Simon Atanasyan <simon.atanasyan@imgtec.com>
7179 * config/mips/mips.c (mips_attribute_table): Add 'short_call'
7181 (mips_near_type_p): Add 'short_call' attribute as a synonym
7183 * doc/extend.texi (short_call): Document new function attribute.
7185 2017-09-12 Jakub Jelinek <jakub@redhat.com>
7188 * c-common.c (sync_resolve_size): Instead of c_dialect_cxx ()
7189 assertion check that in the condition.
7190 (get_atomic_generic_size): Likewise. Before testing if parameter
7191 has pointer type, if it has array type, call for C++
7192 default_conversion to perform array-to-pointer conversion.
7194 2017-09-12 Richard Biener <rguenther@suse.de>
7196 * tree-vect-generic.c (expand_vector_operations_1): Do nothing
7197 for operations we cannot scalarize.
7199 2017-09-12 Aldy Hernandez <aldyh@redhat.com>
7201 * tree-ssa-threadbackward.c (fsm_find_thread_path): Make GC
7202 vectors heap vectors. Clean up comments.
7203 Make visited_bbs a reference.
7204 (profitable_jump_thread_path): Make GC
7205 vectors heap vectors. Clean up comments.
7207 (convert_and_register_jump_thread_path): Make GC vectors heap
7209 (check_subpath_and_update_thread_path): Same. Clean up comments.
7210 Make visited_bbs a reference.
7211 (handle_phi): Abstract common code to to
7212 register_jump_thread_path_if_profitable.
7213 Rename VAR_BB to DEF_BB.
7215 Make GC vectors heap vectors.
7216 Make visited_bbs a reference.
7217 (handle_assignment): Same.
7218 (register_jump_thread_path_if_profitable): New.
7219 (fsm_find_control_statement_thread_paths): Rename VAR_BB to
7221 Make GC vectors heap vectors. Clean up comments.
7222 Make visited_bbs a reference.
7223 (find_jump_threads_backwards): Make visited_bbs live in the stack.
7224 * tree-ssa-threadupdate.c (delete_jump_thread_path): Fix typo in
7227 2017-09-11 Max Filippov <jcmvbkbc@gmail.com>
7230 * config/xtensa/xtensa.c (xtensa_mem_offset): Check that both
7231 words of E_DImode object are reachable by xtensa_uimm8x4 access.
7233 2017-09-11 Vidya Praveen <vidyapraveen@arm.com>
7235 Revert r251800 and r251799.
7237 2017-09-11 Martin Jambor <mjambor@suse.cz>
7240 * hsa-gen.c (gen_hsa_phi_from_gimple_phi): Process ADDR_EXPRs in
7241 arguments in advance.
7242 * hsa-regalloc.c (naive_process_phi): New parameter predecessors,
7243 use it to find predecessor edges.
7244 (naive_outof_ssa): Collect vector of predecessors.
7246 2017-09-08 Jason Merrill <jason@redhat.com>
7248 PR c++/70029 - ICE with ref-qualifier and -flto
7249 * langhooks.h (struct lang_hooks_for_types): Add
7250 copy_lang_qualifiers.
7251 * attribs.c (build_type_attribute_qual_variant): Use it.
7252 * langhooks-def.h (LANG_HOOKS_COPY_LANG_QUALIFIERS): Default to
7254 (LANG_HOOKS_FOR_TYPES_INITIALIZER): Use it.
7255 * tree.c (verify_type): Re-enable TYPE_CANONICAL main variant check.
7257 2017-09-08 Eric Botcazou <ebotcazou@adacore.com>
7260 * config/sparc/sparc.md (mulsi3): Rename into *mulsi3_sp32.
7261 (*mulsi3_sp64): New instruction.
7262 (mulsi3): New expander.
7264 2017-09-08 Uros Bizjak <ubizjak@gmail.com>
7266 * config/alpha/alpha.c (alpha_print_operand) <case 'S'>: Remove.
7268 2017-09-08 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
7270 * sancov.c: Include memmodel.h.
7272 2017-09-07 Eric Botcazou <ebotcazou@adacore.com>
7275 * config/sparc/sparc.c (sparc_emit_set_symbolic_const64): Deal with too
7278 2017-09-07 Carl Love <cel@us.ibm.com>
7280 * config/rs6000/vsx.md (define_insn "*stxvl"): Add missing argument to
7281 the sldi instruction.
7283 2017-09-07 David Edelsohn <dje.gcc@gmail.com>
7285 * sancov.c: Include tm_p.h.
7287 2017-09-07 Jakub Jelinek <jakub@redhat.com>
7290 * output.h (switch_to_other_text_partition): New declaration.
7291 * varasm.c (switch_to_other_text_partition): New function.
7292 * config/rs6000/rs6000.c (uses_TOC): Return 2 if
7293 NOTE_INSN_SWITCH_TEXT_SECTIONS is seen before finding load_toc_* insn.
7294 (rs6000_elf_declare_function_name): If uses_TOC returned 2, switch
7295 to the other text partition before emitting LCL label and switch back
7296 after emitting the word after it.
7298 2017-09-07 Richard Biener <rguenther@suse.de>
7300 * passes.def (pass_split_crit_edges): Remove instance before PRE.
7301 * tree-ssa-pre.c (pass_pre::execute): Instead manually split
7302 critical edges here, after loop init.
7303 (pass_data_pre): Remove PROP_no_crit_edges flags.
7304 * tree-ssa-sccvn.c (vn_reference_lookup_3): Use vn_valueize
7305 for valueization of call args to avoid leaking VN_TOP.
7306 (visit_use): Assert we do not visit default defs.
7307 (init_scc_vn): Use build_decl for VN_TOP to make name nicer.
7308 Use error_mark_node to more easily detect leaking VN_TOP.
7309 All default-defs are varying, not VN_TOP. Mark them visited.
7310 (run_scc_vn): Make code match comment.
7312 2017-09-07 Michael Meissner <meissner@linux.vnet.ibm.com>
7314 * config/rs6000/rs6000-cpus.def (OTHER_VSX_VECTOR_MASKS): Delete
7315 OPTION_MASK_FLOAT128_KEYWORD.
7316 (POWERPC_MASKS): Likewise.
7317 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Delete
7318 support for the -mfloat128-type option, and make -mfloat128
7319 default on PowerPC Linux systems. Define or undefine
7320 __FLOAT128__ and __FLOAT128_HARDWARE__ for the current options.
7321 Define __float128 to be __ieee128 if IEEE 128-bit support is
7322 enabled, or undefine it.
7323 (rs6000_cpu_cpp_builtins): Delete defining __FLOAT128__ here.
7324 Delete defining __FLOAT128_TYPE__.
7325 * config/rs6000/rs6000.opt (x_TARGET_FLOAT128_TYPE): Delete the
7326 -mfloat128-type option and make -mfloat128 default on PowerPC
7328 (TARGET_FLOAT128_TYPE): Likewise.
7329 (-mfloat128-type): Likewise.
7330 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7331 Delete the -mfloat128-type option and make -mfloat128 default on
7332 PowerPC Linux systems. Always use __ieee128 to be the keyword for
7333 the IEEE 128-bit type, and map __float128 to __ieee128 if IEEE
7334 128-bit floating point is enabled. Change tests from using
7335 -mfloat128-type to -mfloat128.
7336 (rs6000_mangle_type): Use the correct mangling for the __float128
7337 type even if normal long double is restricted to 64-bits.
7338 (floatn_mode): Enable the _Float128 type by default on VSX Linux
7340 * config/rs6000/rs6000.h (MASK_FLOAT128_TYPE): Delete.
7341 (MASK_FLOAT128_KEYWORD): Define new shortcut macro.
7342 (RS6000BTM_FLOAT128): Define in terms of -mfloat128, not
7344 * doc/invoke.texi (RS/6000 and PowerPC Options): Update
7345 documentation for -mfloat128.
7347 2017-09-06 Olivier Hainque <hainque@adacore.com>
7349 * config.gcc (powerpc-wrs-vxworksspe): Now match as vxworks*spe.
7351 2017-09-06 Wish Wu <wishwu007@gmail.com>
7352 Jakub Jelinek <jakub@redhat.com>
7354 * asan.c (initialize_sanitizer_builtins): Add
7355 BT_FN_VOID_UINT8_UINT8, BT_FN_VOID_UINT16_UINT16,
7356 BT_FN_VOID_UINT32_UINT32, BT_FN_VOID_UINT64_UINT64,
7357 BT_FN_VOID_FLOAT_FLOAT, BT_FN_VOID_DOUBLE_DOUBLE and
7358 BT_FN_VOID_UINT64_PTR variables.
7359 * builtin-types.def (BT_FN_VOID_UINT8_UINT8): New fn type.
7360 (BT_FN_VOID_UINT16_UINT16): Likewise.
7361 (BT_FN_VOID_UINT32_UINT32): Likewise.
7362 (BT_FN_VOID_FLOAT_FLOAT): Likewise.
7363 (BT_FN_VOID_DOUBLE_DOUBLE): Likewise.
7364 (BT_FN_VOID_UINT64_PTR): Likewise.
7365 * common.opt (flag_sanitize_coverage): New variable.
7366 (fsanitize-coverage=trace-pc): Remove.
7367 (fsanitize-coverage=): Add.
7368 * flag-types.h (enum sanitize_coverage_code): New enum.
7369 * fold-const.c (fold_range_test): Disable non-short-circuit
7370 optimization if flag_sanitize_coverage.
7371 (fold_truth_andor): Likewise.
7372 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
7373 * opts.c (COVERAGE_SANITIZER_OPT): Define.
7374 (coverage_sanitizer_opts): New array.
7375 (get_closest_sanitizer_option): Add OPTS argument, handle also
7376 OPT_fsanitize_coverage_.
7377 (parse_sanitizer_options): Adjusted to also handle
7378 OPT_fsanitize_coverage_.
7379 (common_handle_option): Add OPT_fsanitize_coverage_.
7380 * sancov.c (instrument_comparison, instrument_switch): New function.
7381 (sancov_pass): Add trace-cmp support.
7382 * sanitizer.def (BUILT_IN_SANITIZER_COV_TRACE_CMP1,
7383 BUILT_IN_SANITIZER_COV_TRACE_CMP2, BUILT_IN_SANITIZER_COV_TRACE_CMP4,
7384 BUILT_IN_SANITIZER_COV_TRACE_CMP8,
7385 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP1,
7386 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP2,
7387 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP4,
7388 BUILT_IN_SANITIZER_COV_TRACE_CONST_CMP8,
7389 BUILT_IN_SANITIZER_COV_TRACE_CMPF, BUILT_IN_SANITIZER_COV_TRACE_CMPD,
7390 BUILT_IN_SANITIZER_COV_TRACE_SWITCH): New builtins.
7391 * doc/invoke.texi: Document -fsanitize-coverage=trace-cmp.
7393 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7395 * config/arm/parsecpu.awk (fatal): Note that we've encountered an
7396 error. Only quit immediately if parsing is complete.
7397 (BEGIN): Initialize fatal_err and parse_done.
7398 (begin fpu, end fpu): Check number of arguments.
7399 (begin arch, end arch): Likewise.
7400 (begin cpu, end cpu): Likewise.
7401 (cname, tune for, tune flags, architecture, fpu, option): Likewise.
7402 (optalias): Likewise.
7404 2017-09-06 Richard Earnshaw <rearnsha@arm.com>
7406 * config.gcc (arm*-*-*): Don't add arm-isa.h to tm_p_file.
7407 * config/arm/arm-isa.h: Delete. Move definitions to ...
7408 * arm-cpus.in: ... here. Use new feature and fgroup values.
7409 * config/arm/arm.c (arm_option_override): Use lower case for feature
7411 * config/arm/arm.h (TARGET_HARD_FLOAT): Likewise.
7412 (TARGET_VFP3, TARGET_VFP5, TARGET_FMA): Likewise.
7413 * config/arm/parsecpu.awk (END): Add new command 'isa'.
7415 (print_isa_bits_for): New function.
7416 (gen_isa): New function.
7417 (gen_comm_data): Use print_isa_bits_for.
7418 (define feature): New keyword.
7419 (define fgroup): New keyword.
7420 * config/arm/t-arm (OPTIONS_H_EXTRA): Add arm-isa.h
7421 (arm-isa.h): Add rule to generate file.
7422 * common/config/arm/arm-common.c: (arm_canon_arch_option): Use lower
7423 case for feature bit names.
7425 2017-09-06 Richard Biener <rguenther@suse.de>
7427 * tree-ssa-pre.c (NECESSARY): Remove.
7428 (create_expression_by_pieces): Do not touch pass-local flags.
7429 (insert_into_preds_of_block): Likewise.
7430 (do_pre_regular_insertion): Likewise.
7431 (eliminate_insert): Likewise.
7432 (eliminate_dom_walker::before_dom_children): Likewise.
7433 (fini_eliminate): Do not look at inserted_exprs.
7434 (mark_operand_necessary): Remove.
7435 (remove_dead_inserted_code): Replace with simple work-list
7436 algorithm based on inserted_exprs and SSA uses.
7437 (pass_pre::execute): Re-order fini_eliminate and
7438 remove_dead_inserted_code.
7440 2017-09-06 Olivier Hainque <hainque@adacore.com>
7442 * config/powerpcspe/vxworks.h (VXCPU_FOR_8548): Correct definition
7443 for VxWorks 7. Adjust surrounding comments.
7445 2017-09-06 Richard Biener <rguenther@suse.de>
7447 * gimple-ssa-strength-reduction.c
7448 (find_candidates_dom_walker::before_dom_children): Also allow
7451 2017-09-06 Richard Biener <rguenther@suse.de>
7453 PR tree-optimization/82108
7454 * tree-vect-stmts.c (vectorizable_load): Fix pointer adjustment
7455 for gap in the non-permutation SLP case.
7457 2017-09-06 Martin Jambor <mjambor@suse.cz>
7459 PR tree-optimization/82078
7460 * tree-sra.c (sort_and_splice_var_accesses): Move call to
7461 add_access_to_work_queue...
7462 (build_accesses_from_assign): ...here.
7463 (propagate_all_subaccesses): Make sure racc is the group
7464 representative, if there is one.
7466 2017-09-06 Jakub Jelinek <jakub@redhat.com>
7469 * varasm.c (categorize_decl_for_section): Use SECCAT_TBSS for TLS vars with
7472 2017-09-06 Richard Biener <rguenther@suse.de>
7474 * gimple-ssa-strength-reduction.c
7475 (find_candidates_dom_walker::before_doom_children): Use a
7476 type and not a mode check.
7478 2017-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
7481 * config/arm/predicates.md (arm_general_adddi_operand): Create new
7483 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Use new predicates.
7485 2017-09-05 Jeff Law <law@redhat.com>
7487 PR tree-optimization/64910
7488 * tree-ssa-reassoc.c (reassociate_bb): Restrict last change to
7489 cases where we have 3 or more operands.
7491 2017-09-05 Jakub Jelinek <jakub@redhat.com>
7494 * omp-low.c (lower_omp_for): Recompute tree invariant if
7495 gimple_omp_for_initial/final is ADDR_EXPR.
7498 * omp-expand.c (expand_omp_simd): Force second operands of COND_EXPR
7499 into gimple val before gimplification fo the COND_EXPR.
7501 2017-09-05 Aldy Hernandez <aldyh@redhat.com>
7503 * tree-ssa-threadupdate.c (duplicate_thread_path): Remove unused
7504 REGION_COPY argument.
7505 (thread_through_all_blocks): Remove unused argument to
7506 duplicate_thread_path.
7508 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7509 Alan Hayward <alan.hayward@arm.com>
7510 David Sherwood <david.sherwood@arm.com>
7512 * config/aarch64/aarch64-protos.h (aarch64_gen_adjusted_ldpstp):
7513 Take a scalar_mode rather than a machine_mode.
7514 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7515 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Likewise.
7516 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
7517 (aarch64_gen_adjusted_ldpstp): Likewise.
7518 (aarch64_expand_vector_init): Use scalar_mode instead of machine_mode.
7520 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7521 Alan Hayward <alan.hayward@arm.com>
7522 David Sherwood <david.sherwood@arm.com>
7524 * config/aarch64/aarch64-protos.h (aarch64_is_extend_from_extract):
7525 Take a scalar_int_mode instead of a machine_mode.
7526 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7527 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7528 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7529 (aarch64_simd_attr_length_rglist): Delete.
7530 * config/aarch64/aarch64.c (aarch64_is_extend_from_extract): Take
7531 a scalar_int_mode instead of a machine_mode.
7532 (aarch64_add_offset): Likewise.
7533 (aarch64_internal_mov_immediate): Likewise
7534 (aarch64_add_constant_internal): Likewise.
7535 (aarch64_add_constant): Likewise.
7536 (aarch64_movw_imm): Likewise.
7537 (aarch64_rtx_arith_op_extract_p): Likewise.
7538 (aarch64_mask_and_shift_for_ubfiz_p): Likewise.
7539 (aarch64_simd_scalar_immediate_valid_for_move): Likewise.
7540 Remove assert that the mode isn't a vector.
7541 (aarch64_output_scalar_simd_mov_immediate): Likewise.
7542 (aarch64_expand_mov_immediate): Update calls after above changes.
7543 (aarch64_output_casesi): Use as_a <scalar_int_mode>.
7544 (aarch64_and_bitmask_imm): Check for scalar integer modes.
7545 (aarch64_move_imm): Likewise.
7546 (aarch64_can_const_movi_rtx_p): Likewise.
7547 (aarch64_strip_extend): Likewise.
7548 (aarch64_extr_rtx_p): Likewise.
7549 (aarch64_rtx_costs): Likewise, using wode_mode as the mode of
7550 a CONST_INT when the mode parameter is VOIDmode.
7551 (aarch64_float_const_rtx_p): Use scalar_int_mode for a temporary.
7553 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7555 * machmode.h (bitwise_mode_for_mode): Return opt_mode.
7556 * stor-layout.c (bitwise_mode_for_mode): Likewise.
7557 (bitwise_type_for_mode): Update accordingly.
7559 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7561 * stor-layout.h (mode_for_size_tree): Return an opt_mode.
7562 * stor-layout.c (mode_for_size_tree): Likewise.
7563 (mode_for_array): Update accordingly.
7564 (layout_decl): Likewise.
7565 (compute_record_mode): Likewise. Only set the mode once.
7567 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7569 * target.def (get_mask_mode): Change return type to opt_mode.
7571 * doc/tm.texi: Regenerate.
7572 * targhooks.h (default_get_mask_mode): Return an opt_mode.
7573 * targhooks.c (default_get_mask_mode): Likewise.
7574 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
7575 * optabs-query.c (can_vec_mask_load_store_p): Update use of
7576 targetm.get_mask_mode.
7577 * tree.c (build_truth_vector_type): Likewise.
7579 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7581 * machmode.h (mode_for_vector): Return an opt_mode.
7582 * stor-layout.c (mode_for_vector): Likewise.
7583 (mode_for_int_vector): Update accordingly.
7584 (layout_type): Likewise.
7585 * config/i386/i386.c (emit_memmov): Likewise.
7586 (ix86_expand_set_or_movmem): Likewise.
7587 (ix86_expand_vector_init): Likewise.
7588 (ix86_get_mask_mode): Likewise.
7589 * config/powerpcspe/powerpcspe.c (rs6000_expand_vec_perm_const_1):
7591 * config/rs6000/rs6000.c (rs6000_expand_vec_perm_const_1): Likewise.
7592 * expmed.c (extract_bit_field_1): Likewise.
7593 * expr.c (expand_expr_real_2): Likewise.
7594 * optabs-query.c (can_vec_perm_p): Likewise.
7595 (can_vec_mask_load_store_p): Likewise.
7596 * optabs.c (expand_vec_perm): Likewise.
7597 * targhooks.c (default_get_mask_mode): Likewise.
7598 * tree-vect-stmts.c (vectorizable_store): Likewise.
7599 (vectorizable_load): Likewise.
7600 (get_vectype_for_scalar_type_and_size): Likewise.
7602 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7604 * machmode.h (mode_for_int_vector): New function.
7605 * stor-layout.c (mode_for_int_vector): Likewise.
7606 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Use it.
7607 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Likewise.
7608 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
7609 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
7610 (s390_expand_vcond): Likewise.
7612 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7614 * machmode.h (opt_machine_mode): New type.
7615 (opt_mode<T>): Allow construction from anything that can be
7617 (is_a, as_a, dyn_cast): Add overloads for opt_mode.
7618 (mode_for_size): Return an opt_machine_mode.
7619 * stor-layout.c (mode_for_size): Likewise.
7620 (mode_for_size_tree): Update call accordingly.
7621 (bitwise_mode_for_mode): Likewise.
7622 (make_fract_type): Likewise.
7623 (make_accum_type): Likewise.
7624 * caller-save.c (replace_reg_with_saved_mem): Update call
7626 * config/alpha/alpha.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7627 * config/i386/i386.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7628 * config/s390/s390.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7629 * config/sparc/sparc.h (SECONDARY_MEMORY_NEEDED_MODE): Likewise.
7630 * expmed.c (extract_bit_field_1): Likewise.
7631 * reload.c (get_secondary_mem): Likewise.
7632 * varasm.c (assemble_integer): Likewise.
7633 * lower-subreg.c (simplify_subreg_concatn): Likewise. Move
7636 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7638 * machmode.h (decimal_float_mode_for_size): New function.
7639 * real.h (REAL_VALUE_TO_TARGET_LONG_DOUBLE): Use float_mode_for_size.
7640 (REAL_VALUE_TO_TARGET_DOUBLE): Likewise.
7641 (REAL_VALUE_TO_TARGET_SINGLE): Likewise.
7642 (REAL_VALUE_TO_TARGET_DECIMAL128): Use decimal_float_mode_for_size.
7643 (REAL_VALUE_TO_TARGET_DECIMAL64): Likewise.
7644 (REAL_VALUE_TO_TARGET_DECIMAL32): Likewise.
7646 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7648 * builtins.c (expand_builtin_powi): Use int_mode_for_size.
7649 (get_builtin_sync_mode): Likewise.
7650 (expand_ifn_atomic_compare_exchange): Likewise.
7651 (expand_builtin_atomic_clear): Likewise.
7652 (expand_builtin_atomic_test_and_set): Likewise.
7653 (fold_builtin_atomic_always_lock_free): Likewise.
7654 * calls.c (compute_argument_addresses): Likewise.
7655 (emit_library_call_value_1): Likewise.
7656 (store_one_arg): Likewise.
7657 * combine.c (combine_instructions): Likewise.
7658 * config/aarch64/aarch64.c (aarch64_function_value): Likewise.
7659 * config/arm/arm.c (arm_function_value): Likewise.
7660 (aapcs_allocate_return_reg): Likewise.
7661 * config/c6x/c6x.c (c6x_expand_movmem): Likewise.
7662 * config/i386/i386.c (construct_container): Likewise.
7663 (ix86_gimplify_va_arg): Likewise.
7664 (ix86_expand_sse_cmp): Likewise.
7665 (emit_memmov): Likewise.
7666 (emit_memset): Likewise.
7667 (expand_small_movmem_or_setmem): Likewise.
7668 (ix86_expand_pextr): Likewise.
7669 (ix86_expand_pinsr): Likewise.
7670 * config/lm32/lm32.c (lm32_block_move_inline): Likewise.
7671 * config/microblaze/microblaze.c (microblaze_block_move_straight):
7673 * config/mips/mips.c (mips_function_value_1) Likewise.
7674 (mips_block_move_straight): Likewise.
7675 (mips_expand_ins_as_unaligned_store): Likewise.
7676 * config/powerpcspe/powerpcspe.c
7677 (rs6000_darwin64_record_arg_advance_flush): Likewise.
7678 (rs6000_darwin64_record_arg_flush): Likewise.
7679 * config/rs6000/rs6000.c
7680 (rs6000_darwin64_record_arg_advance_flush): Likewise.
7681 (rs6000_darwin64_record_arg_flush): Likewise.
7682 * config/sparc/sparc.c (sparc_function_arg_1): Likewise.
7683 (sparc_function_value_1): Likewise.
7684 * config/spu/spu.c (adjust_operand): Likewise.
7685 (spu_emit_branch_or_set): Likewise.
7686 (arith_immediate_p): Likewise.
7687 * emit-rtl.c (gen_lowpart_common): Likewise.
7688 * expr.c (expand_expr_real_1): Likewise.
7689 * function.c (assign_parm_setup_block): Likewise.
7690 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): Likewise.
7691 * reload1.c (alter_reg): Likewise.
7692 * stor-layout.c (mode_for_vector): Likewise.
7693 (layout_type): Likewise.
7695 2017-09-05 Richard Sandiford <richard.sandiford@linaro.org>
7697 * config/spu/spu.c (exp2_immediate_p): Use int_mode_for_mode.
7698 (spu_convert_move): Likewise.
7699 * lower-subreg.c (resolve_simple_move): Likewise.
7701 2017-09-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7704 * config/rs6000/altivec.md (altivec_vsum2sws): Convert from a
7705 define_insn to a define_expand.
7706 (altivec_vsum2sws_direct): New define_insn.
7707 (altivec_vsumsws): Convert from a define_insn to a define_expand.
7709 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7711 * config/arm/arm.c (arm_option_params_internal): Improve setting of
7714 2017-09-05 H.J. Lu <hongjiu.lu@intel.com>
7719 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Don't
7720 realign stack if stack alignment needed is less than incoming
7723 2017-09-05 Marek Polacek <polacek@redhat.com>
7726 * convert.c (convert_to_integer_1) <case NEGATE_EXPR>: Move the ubsan
7729 2017-09-05 Wilco Dijkstra <wdijkstr@arm.com>
7731 * explow.c (get_dynamic_stack_size): Improve dynamic alignment.
7733 2017-09-05 Richard Biener <rguenther@suse.de>
7735 PR tree-optimization/82084
7736 * fold-const.c (can_native_encode_string_p): Handle wide characters.
7738 2017-09-05 Richard Biener <rguenther@suse.de>
7740 PR tree-optimization/82102
7741 * tree-ssa-pre.c (fini_eliminate): Check if lhs is NULL.
7743 2017-09-05 Martin Liska <mliska@suse.cz>
7745 PR tree-optimization/82032
7746 * tree-cfg.c (generate_range_test): New function.
7747 * tree-cfg.h (generate_range_test): Declared here.
7748 * tree-cfgcleanup.c (convert_single_case_switch): New function.
7749 (cleanup_control_expr_graph): Use it.
7750 * tree-switch-conversion.c (try_switch_expansion): Remove
7752 (emit_case_nodes): Use generate_range_test.
7754 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7757 * config/i386/i386.md (*<btsc><mode>_mask): Add
7758 TARGET_USE_BT to insn constraint.
7759 (*btr<mode>_mask): Ditto.
7761 2017-09-04 Wilco Dijkstra <wdijkstr@arm.com>
7763 * config/arm/arm.c (arm_legitimate_index_p): Add comment.
7764 (thumb2_legitimate_index_p): Use correct range for DI/DF mode.
7766 2017-09-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
7769 * config/arm/arm.md (*arm_adddi3, *arm_subdi3): Split early except for
7770 TARGET_NEON and TARGET_IWMMXT.
7771 (anddi3, iordi3, xordi3, one_cmpldi2): Split while expanding except for
7772 TARGET_NEON and TARGET_IWMMXT.
7773 (*one_cmpldi2_insn): Moved the body of one_cmpldi2 here.
7775 2017-09-04 Uros Bizjak <ubizjak@gmail.com>
7777 * config/i386/i386-protos.h (ix86_tls_address_pattern_p) New prototype.
7778 (ix86_rewrite_tls_address): Ditto.
7779 * config/i386/i386.c (ix86_tls_address_pattern_p) New function.
7780 (ix86_rewrite_tls_address_1): Ditto.
7781 (ix86_rewrite_tls_address): Ditto.
7782 * config/i386/predicates.md (tls_address_pattern): New predicate.
7783 * config/i386/i386.md (TLS address splitter): New splitter.
7785 2017-09-04 Richard Biener <rguenther@suse.de>
7787 PR tree-optimization/82084
7788 * fold-const.h (can_native_encode_string_p): Declare.
7789 * fold-const.c (can_native_encode_string_p): Factor out from ...
7790 (native_encode_string): ... here.
7791 * tree-vect-stmts.c (vectorizable_store): Call it to avoid
7792 vectorizing stores from constants we later cannot handle.
7794 2017-09-04 Marek Polacek <polacek@redhat.com>
7797 * doc/invoke.texi: Update -Wtautological-compare documentation.
7799 2017-09-04 Jeff Law <law@redhat.com>
7801 PR tree-optimization/64910
7802 * tree-ssa-reassoc.c (reassociate_bb): For bitwise binary ops,
7803 swap the first and last operand if the last is a constant.
7805 2017-09-04 Marek Polacek <polacek@redhat.com>
7808 * convert.c (do_narrow): When sanitizing signed integer overflows,
7809 bail out for signed types.
7810 (convert_to_integer_1) <case NEGATE_EXPR>: Likewise.
7812 2017-09-04 Richard Biener <rguenther@suse.de>
7814 PR tree-optimization/82060
7815 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
7816 Move devirtualization after stmt folding and before EH/AB/noreturn
7817 cleanup to get the stmt refs canonicalized. Use a bool instead
7818 of gimple_modified_p since that doesn't work for NOPs. Schedule
7819 NOPs generated by folding for removal.
7821 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7822 Alan Hayward <alan.hayward@arm.com>
7823 David Sherwood <david.sherwood@arm.com>
7825 * coretypes.h (pad_direction): New enum.
7826 * defaults.h (DEFAULT_FUNCTION_ARG_PADDING): Delete.
7827 (FUNCTION_ARG_PADDING): Likewise.
7828 * target.def (function_arg_padding): New hook.
7829 * targhooks.h (default_function_arg_padding): Declare.
7830 * targhooks.c (default_function_arg_padding): New function.
7831 * doc/tm.texi.in (FUNCTION_ARG_PADDING): Replace with...
7832 (TARGET_FUNCTION_ARG_PADDING): ...this.
7833 * doc/tm.texi: Regenerate.
7834 * calls.c (store_unaligned_arguments_into_pseudos): Use pad_direction
7835 instead of direction.
7836 (compute_argument_addresses): Likewise.
7837 (load_register_parameters): Likewise.
7838 (emit_library_call_value_1): Likewise.
7839 (store_one_arg): Use targetm.calls.function_arg_padding instead
7840 of FUNCTION_ARG_PADDING.
7841 (must_pass_in_stack_var_size_or_pad): Likewise.
7842 * expr.c (emit_group_load_1): Use pad_direction instead of direction.
7843 (emit_group_store): Likewise.
7844 (emit_single_push_insn_1): Use targetm.calls.function_arg_padding
7845 instead of FUNCTION_ARG_PADDING.
7846 (emit_push_insn): Likewise, and propagate enum change throughout
7848 * function.h (direction): Delete.
7849 (locate_and_pad_arg_data::where_pad): Use pad_direction instead
7851 * function.c (assign_parm_find_stack_rtl): Likewise.
7852 (assign_parm_setup_block_p): Likewise.
7853 (assign_parm_setup_block): Likewise.
7854 (gimplify_parameters): Likewise.
7855 (locate_and_pad_parm): Use targetm.calls.function_arg_padding
7856 instead of FUNCTION_ARG_PADDING, and propagate enum change throughout
7858 * config/aarch64/aarch64.h (FUNCTION_ARG_PADDING): Delete.
7859 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7860 * config/aarch64/aarch64-protos.h (aarch64_pad_arg_upward): Delete.
7861 * config/aarch64/aarch64.c (aarch64_pad_arg_upward): Replace with...
7862 (aarch64_function_arg_padding): ...this new function.
7863 (aarch64_gimplify_va_arg_expr): Use pad_direction instead of direction.
7864 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7865 * config/arm/arm.h (FUNCTION_ARG_PADDING): Delete.
7866 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7867 * config/arm/arm-protos.h (arm_pad_arg_upward): Delete.
7868 * config/arm/arm.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7869 (arm_pad_arg_upward): Replace with...
7870 (arm_function_arg_padding): ...this new function.
7871 * config/c6x/c6x.h (BLOCK_REG_PADDING): Use pad_direction instead
7873 * config/ia64/hpux.h (FUNCTION_ARG_PADDING): Delete.
7874 * config/ia64/ia64-protos.h (ia64_hpux_function_arg_padding): Delete.
7875 * config/ia64/ia64.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7876 (ia64_hpux_function_arg_padding): Replace with...
7877 (ia64_function_arg_padding): ...this new function. Use pad_direction
7878 instead of direction. Check for TARGET_HPUX.
7879 * config/iq2000/iq2000.h (FUNCTION_ARG_PADDING): Delete.
7880 * config/iq2000/iq2000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7881 (iq2000_function_arg_padding): New function.
7882 * config/mips/mips-protos.h (mips_pad_arg_upward): Delete.
7883 * config/mips/mips.c (mips_pad_arg_upward): Replace with...
7884 (mips_function_arg_padding): ...this new function.
7885 (mips_pad_reg_upward): Update accordingly.
7886 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7887 * config/mips/mips.h (PAD_VARARGS_DOWN): Use
7888 targetm.calls.function_arg_padding.
7889 (FUNCTION_ARG_PADDING): Delete.
7890 (BLOCK_REG_PADDING): Use pad_direction instead of direction.
7891 * config/nios2/nios2.h (FUNCTION_ARG_PADDING): Delete.
7892 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7893 * config/nios2/nios2-protos.h (nios2_function_arg_padding): Delete.
7894 (nios2_block_reg_padding): Return pad_direction instead of direction.
7895 * config/nios2/nios2.c (nios2_block_reg_padding): Return pad_direction
7896 instead of direction.
7897 (nios2_function_arg_padding): Likewise. Make static.
7898 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7899 * config/pa/pa.h (FUNCTION_ARG_PADDING): Delete.
7900 (BLOCK_REG_PADDING): Use targetm.calls.function_arg_padding.
7901 * config/pa/pa-protos.h (pa_function_arg_padding): Delete.
7902 * config/pa/pa.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7903 (pa_function_arg_padding): Make static. Return pad_direction instead
7905 * config/powerpcspe/powerpcspe.h (FUNCTION_ARG_PADDING): Delete.
7906 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7907 * config/powerpcspe/aix.h (BLOCK_REG_PADDING): Use pad_direction
7908 instead of direction. Use targetm.calls.function_arg_padding.
7909 * config/powerpcspe/darwin.h (BLOCK_REG_PADDING): Likewise.
7910 * config/powerpcspe/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7911 * config/powerpcspe/linux64.h (BLOCK_REG_PADDING): Likewise.
7912 * config/powerpcspe/powerpcspe-protos.h (function_arg_padding): Delete.
7913 * config/powerpcspe/powerpcspe.c (TARGET_FUNCTION_ARG_PADDING):
7915 (function_arg_padding): Rename to...
7916 (rs6000_function_arg_padding): ...this. Make static. Return
7917 pad_direction instead of direction.
7918 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7919 * config/rs6000/rs6000.h (FUNCTION_ARG_PADDING): Delete.
7920 (PAD_VARARGS_DOWN): Use targetm.calls.function_arg_padding.
7921 * config/rs6000/aix.h (BLOCK_REG_PADDING): Use pad_direction
7922 instead of direction. Use targetm.calls.function_arg_padding.
7923 * config/rs6000/darwin.h (BLOCK_REG_PADDING): Likewise.
7924 * config/rs6000/freebsd64.h (BLOCK_REG_PADDING): Likewise.
7925 * config/rs6000/linux64.h (BLOCK_REG_PADDING): Likewise.
7926 * config/rs6000/rs6000-protos.h (function_arg_padding): Delete.
7927 * config/rs6000/rs6000.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7928 (function_arg_padding): Rename to...
7929 (rs6000_function_arg_padding): ...this. Make static. Return
7930 pad_direction instead of direction.
7931 (rs6000_return_in_msb): Use rs6000_function_arg_padding.
7932 * config/s390/s390.h (FUNCTION_ARG_PADDING): Delete.
7933 * config/s390/s390.c (s390_function_arg_padding): New function.
7934 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7935 * config/sparc/sparc.h (FUNCTION_ARG_PADDING): Delete.
7936 * config/sparc/sparc-protos.h (function_arg_padding): Delete.
7937 * config/sparc/sparc.c (TARGET_FUNCTION_ARG_PADDING): Redefine.
7938 (function_arg_padding): Rename to...
7939 (sparc_function_arg_padding): ...this. Make static. Return
7940 pad_direction instead of direction.
7941 * config/spu/spu.h (FUNCTION_ARG_PADDING): Delete.
7942 * config/spu/spu.c (spu_function_arg_padding): New function.
7943 (TARGET_FUNCTION_ARG_PADDING): Redefine.
7944 * system.h (FUNCTION_ARG_PADDING): Poison.
7946 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
7947 Alan Hayward <alan.hayward@arm.com>
7948 David Sherwood <david.sherwood@arm.com>
7950 * target.def (modes_tieable_p): New hook.
7951 * doc/tm.texi (MODES_TIEABLE_P): Replace with...
7952 (TARGET_MODES_TIEABLE_P): ...this.
7953 * doc/tm.texi.in: Regenerate.
7954 * hooks.h (hook_bool_mode_mode_true): Declare.
7955 * hooks.c (hook_bool_mode_mode_true): New function.
7956 * combine.c (subst): Use targetm.modes_tieable_p instead of
7958 * dse.c (find_shift_sequence): Likewise.
7959 * expmed.c (extract_low_bits): Likewise.
7960 * lower-subreg.c: Include target.h.
7961 (find_decomposable_subregs): Use targetm.modes_tieable_p instead of
7963 * rtlanal.c (rtx_cost): Likewise.
7964 * config/aarch64/aarch64.h (MODES_TIEABLE_P): Delete.
7965 * config/aarch64/aarch64-protos.h (aarch64_modes_tieable_p): Delete.
7966 * config/aarch64/aarch64.c (aarch64_modes_tieable_p): Make static.
7967 (TARGET_MODES_TIEABLE_P): Redefine.
7968 * config/alpha/alpha.h (MODES_TIEABLE_P): Delete.
7969 * config/alpha/alpha.c (alpha_modes_tieable_p): New function.
7970 (TARGET_MODES_TIEABLE_P): Redefine.
7971 * config/arc/arc.h (MODES_TIEABLE_P): Delete.
7972 * config/arc/arc.c (TARGET_MODES_TIEABLE_P): Redefine.
7973 (arc_modes_tieable_p): New function.
7974 * config/arm/arm.h (MODES_TIEABLE_P): Delete.
7975 * config/arm/arm-protos.h (arm_modes_tieable_p): Delete.
7976 * config/arm/arm.c (TARGET_MODES_TIEABLE_P): Redefine.
7977 (arm_modes_tieable_p): Make static.
7978 * config/avr/avr.h (MODES_TIEABLE_P): Delete.
7979 * config/bfin/bfin.h (MODES_TIEABLE_P): Delete.
7980 * config/bfin/bfin.c (bfin_modes_tieable_p): New function.
7981 (TARGET_MODES_TIEABLE_P): Redefine.
7982 * config/c6x/c6x.h (MODES_TIEABLE_P): Delete.
7983 * config/c6x/c6x.c (c6x_modes_tieable_p): New function.
7984 (TARGET_MODES_TIEABLE_P): Redefine.
7985 * config/cr16/cr16.h (MODES_TIEABLE_P): Delete.
7986 * config/cr16/cr16.c (TARGET_MODES_TIEABLE_P): Redefine.
7987 (cr16_modes_tieable_p): New function.
7988 * config/cris/cris.h (MODES_TIEABLE_P): Delete.
7989 * config/epiphany/epiphany.h (MODES_TIEABLE_P): Delete.
7990 * config/fr30/fr30.h (MODES_TIEABLE_P): Delete.
7991 (TRULY_NOOP_TRUNCATION): Update comment.
7992 * config/frv/frv.h (MODES_TIEABLE_P): Delete.
7993 (TRULY_NOOP_TRUNCATION): Update comment.
7994 * config/frv/frv.c (TARGET_MODES_TIEABLE_P): Redefine.
7995 (frv_modes_tieable_p): New function.
7996 * config/ft32/ft32.h (MODES_TIEABLE_P): Delete.
7997 * config/h8300/h8300.h (MODES_TIEABLE_P): Delete.
7998 * config/h8300/h8300.c (h8300_modes_tieable_p): New function.
7999 (TARGET_MODES_TIEABLE_P): Redefine.
8000 * config/i386/i386.h (MODES_TIEABLE_P): Delete.
8001 * config/i386/i386-protos.h (ix86_modes_tieable_p): Delete.
8002 * config/i386/i386.c (ix86_modes_tieable_p): Make static.
8003 (TARGET_MODES_TIEABLE_P): Redefine.
8004 * config/ia64/ia64.h (MODES_TIEABLE_P): Delete.
8005 * config/ia64/ia64.c (TARGET_MODES_TIEABLE_P): Redefine.
8006 (ia64_modes_tieable_p): New function.
8007 * config/iq2000/iq2000.h (MODES_TIEABLE_P): Delete.
8008 * config/iq2000/iq2000.c (TARGET_MODES_TIEABLE_P): Redefine.
8009 (iq2000_modes_tieable_p): New function.
8010 * config/lm32/lm32.h (MODES_TIEABLE_P): Delete.
8011 * config/lm32/lm32.c (TARGET_MODES_TIEABLE_P): Redefine.
8012 (lm32_modes_tieable_p): New function.
8013 * config/m32c/m32c.h (MODES_TIEABLE_P): Delete.
8014 * config/m32c/m32c-protos.h (m32c_modes_tieable_p): Delete.
8015 * config/m32c/m32c.c (m32c_modes_tieable_p): Make static.
8016 (TARGET_MODES_TIEABLE_P): Redefine.
8017 * config/m32r/m32r.h (MODES_TIEABLE_P): Delete.
8018 * config/m32r/m32r.c (TARGET_MODES_TIEABLE_P): Redefine.
8019 (m32r_modes_tieable_p): New function.
8020 * config/m68k/m68k.h (MODES_TIEABLE_P): Delete.
8021 * config/m68k/m68k.c (TARGET_MODES_TIEABLE_P): Redefine.
8022 (m68k_modes_tieable_p): New function.
8023 * config/mcore/mcore.h (MODES_TIEABLE_P): Delete.
8024 * config/mcore/mcore.c (TARGET_MODES_TIEABLE_P): Redefine.
8025 (mcore_modes_tieable_p): New function.
8026 * config/microblaze/microblaze.h (MODES_TIEABLE_P): Delete.
8027 * config/microblaze/microblaze.c (microblaze_modes_tieable_p): New
8029 (TARGET_MODES_TIEABLE_P): Redefine.
8030 * config/mips/mips.h (MODES_TIEABLE_P): Delete.
8031 * config/mips/mips-protos.h (mips_modes_tieable_p): Delete.
8032 * config/mips/mips.c (mips_modes_tieable_p): Make static.
8033 (TARGET_MODES_TIEABLE_P): Redefine.
8034 * config/mmix/mmix.h (MODES_TIEABLE_P): Delete.
8035 * config/mn10300/mn10300.h (MODES_TIEABLE_P): Delete.
8036 * config/mn10300/mn10300-protos.h (mn10300_modes_tieable): Delete.
8037 * config/mn10300/mn10300.c (mn10300_modes_tieable): Rename to...
8038 (mn10300_modes_tieable_p): ...this and make static.
8039 (TARGET_MODES_TIEABLE_P): Redefine.
8040 * config/moxie/moxie.h (MODES_TIEABLE_P): Delete.
8041 * config/msp430/msp430.h (MODES_TIEABLE_P): Delete.
8042 * config/msp430/msp430-protos.h (msp430_modes_tieable_p): Delete.
8043 * config/msp430/msp430.c (TARGET_MODES_TIEABLE_P): Redefine.
8044 (msp430_modes_tieable_p): Make static.
8045 * config/nds32/nds32.h (MODES_TIEABLE_P): Delete.
8046 * config/nds32/nds32.c (nds32_modes_tieable_p): New function.
8047 (TARGET_MODES_TIEABLE_P): Redefine.
8048 * config/nios2/nios2.h (MODES_TIEABLE_P): Delete.
8049 * config/nvptx/nvptx.h (MODES_TIEABLE_P): Delete.
8050 * config/nvptx/nvptx.c (nvptx_modes_tieable_p): New function.
8051 (TARGET_MODES_TIEABLE_P): Redefine.
8052 * config/pa/pa.h (MODES_TIEABLE_P): Delete.
8053 * config/pa/pa-protos.h (pa_modes_tieable_p): Delete.
8054 * config/pa/pa.c (pa_modes_tieable_p): Make static.
8055 (TARGET_MODES_TIEABLE_P): Redefine.
8056 * config/pdp11/pdp11.h (MODES_TIEABLE_P): Delete.
8057 * config/pdp11/pdp11.c (TARGET_MODES_TIEABLE_P): Redefine.
8058 (pdp11_modes_tieable_p): New function.
8059 * config/powerpcspe/powerpcspe.h (MODES_TIEABLE_P): Delete.
8060 * config/powerpcspe/powerpcspe.c (TARGET_MODES_TIEABLE_P): Redefine.
8061 (rs6000_modes_tieable_p): New function.
8062 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8063 * config/powerpcspe/powerpcspe.md: Update comment.
8064 * config/riscv/riscv.h (MODES_TIEABLE_P): Delete.
8065 * config/riscv/riscv.c (riscv_modes_tieable_p): New function.
8066 (TARGET_MODES_TIEABLE_P): Redefine.
8067 * config/rl78/rl78.h (MODES_TIEABLE_P): Delete.
8068 * config/rl78/rl78.c (TARGET_MODES_TIEABLE_P): Redefine.
8069 (rl78_modes_tieable_p): New function.
8070 * config/rs6000/rs6000.h (MODES_TIEABLE_P): Delete.
8071 * config/rs6000/rs6000.c (TARGET_MODES_TIEABLE_P): Redefine.
8072 (rs6000_modes_tieable_p): New function.
8073 (rs6000_debug_reg_global): Use it instead of MODES_TIEABLE_P.
8074 * config/rs6000/rs6000.md: Update comment.
8075 * config/rx/rx.h (MODES_TIEABLE_P): Delete.
8076 * config/rx/rx.c (rx_modes_tieable_p): New function.
8077 (TARGET_MODES_TIEABLE_P): Redefine.
8078 * config/s390/s390.h (MODES_TIEABLE_P): Delete.
8079 * config/s390/s390.c (s390_modes_tieable_p): New function.
8080 (TARGET_MODES_TIEABLE_P): Redefine.
8081 * config/sh/sh.h (MODES_TIEABLE_P): Delete.
8082 * config/sh/sh.c (TARGET_MODES_TIEABLE_P): Redefine.
8083 (sh_modes_tieable_p): New function.
8084 * config/sparc/sparc.h (MODES_TIEABLE_P): Delete.
8085 * config/sparc/sparc-protos.h (sparc_modes_tieable_p): Delete.
8086 * config/sparc/sparc.c (TARGET_MODES_TIEABLE_P): Redefine.
8087 (sparc_modes_tieable_p): Make static.
8088 * config/spu/spu.h (MODES_TIEABLE_P): Delete.
8089 * config/spu/spu.c (spu_modes_tieable_p): New function.
8090 (TARGET_MODES_TIEABLE_P): Redefine.
8091 * config/stormy16/stormy16.h (MODES_TIEABLE_P): Delete.
8092 * config/stormy16/stormy16.c (xstormy16_modes_tieable_p): New function.
8093 (TARGET_MODES_TIEABLE_P): Redefine.
8094 * config/tilegx/tilegx.h (MODES_TIEABLE_P): Delete.
8095 * config/tilepro/tilepro.h (MODES_TIEABLE_P): Delete.
8096 * config/v850/v850.h (MODES_TIEABLE_P): Delete.
8097 * config/v850/v850.c (v850_modes_tieable_p): New function.
8098 (TARGET_MODES_TIEABLE_P): Redefine.
8099 * config/vax/vax.h (MODES_TIEABLE_P): Delete.
8100 * config/visium/visium.h (MODES_TIEABLE_P): Delete.
8101 * config/visium/visium.c (TARGET_MODES_TIEABLE_P): Redefine.
8102 (visium_modes_tieable_p): New function.
8103 * config/xtensa/xtensa.h (MODES_TIEABLE_P): Delete.
8104 * config/xtensa/xtensa.c (TARGET_MODES_TIEABLE_P): Redefine.
8105 (xtensa_modes_tieable_p): New function.
8106 * system.h (MODES_TIEABLE_P): Poison.
8108 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8109 Alan Hayward <alan.hayward@arm.com>
8110 David Sherwood <david.sherwood@arm.com>
8112 * target.def (hard_regno_mode_ok): New hook.
8113 * doc/tm.texi (HARD_REGNO_MODE_OK): Replace with...
8114 (TARGET_HARD_REGNO_MODE_OK): ...this.
8115 * doc/tm.texi.in: Regenerate.
8116 * hooks.h (hook_bool_uint_mode_true): Declare.
8117 * hooks.c (hook_bool_uint_mode_true): New function.
8118 * doc/md.texi: Refer to targetm.hard_regno_mode_ok instead of
8120 * genpreds.c (write_insn_preds_c): Add an include of target.h.
8121 * alias.c (init_alias_target): Use targetm.hard_regno_mode_ok
8122 instead of HARD_REGNO_MODE_OK.
8123 * caller-save.c: Include target.h.
8124 (reg_save_code): Use targetm.hard_regno_mode_ok instead of
8126 * combine.c (can_combine_p): Likewise.
8127 (combinable_i3pat): Likewise.
8128 (can_change_dest_mode): Likewise.
8129 * expr.c (init_expr_target): Likewise.
8130 (convert_move): Likewise.
8131 (convert_modes): Likewise.
8132 * ira.c (setup_prohibited_class_mode_regs): Likewise.
8133 (setup_prohibited_mode_move_regs): Likewise.
8134 * ira.h (target_ira): Likewise.
8135 * lra-assigns.c (find_hard_regno_for_1): Likewise.
8136 * lra-constraints.c (process_alt_operands): Likewise.
8137 (split_reg): Likewise.
8138 * recog.c (peep2_find_free_register): Likewise.
8139 * ree.c (combine_reaching_defs): Likewise.
8140 * regcprop.c (maybe_mode_change): Likewise.
8141 * reginfo.c (init_reg_sets_1): Likewise.
8142 (choose_hard_reg_mode): Likewise.
8143 (simplifiable_subregs): Likewise.
8144 * regrename.c (check_new_reg_p): Likewise.
8145 * reload.c (find_valid_class): Likewise.
8146 (find_valid_class_1): Likewise.
8147 (reload_inner_reg_of_subreg): Likewise.
8148 (push_reload): Likewise.
8149 (combine_reloads): Likewise.
8150 (find_dummy_reload): Likewise.
8151 (find_reloads): Likewise.
8152 * reload1.c (find_reg): Likewise.
8153 (set_reload_reg): Likewise.
8154 (allocate_reload_reg): Likewise.
8155 (choose_reload_regs): Likewise.
8156 (reload_adjust_reg_for_temp): Likewise.
8157 * rtlanal.c (subreg_size_offset_from_lsb): Likewise.
8158 (simplify_subreg_regno): Likewise.
8159 * sel-sched.c (init_regs_for_mode): Likewise.
8160 * varasm.c (make_decl_rtl): Likewise.
8161 * config/aarch64/aarch64.h (HARD_REGNO_MODE_OK): Delete.
8162 (MODES_TIEABLE_P): Use targetm.hard_regno_mode_ok instead of
8164 * config/aarch64/aarch64-protos.h (aarch64_hard_regno_mode_ok): Delete.
8165 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Make static.
8166 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8167 * config/alpha/alpha.h (HARD_REGNO_MODE_OK): Delete.
8168 * config/alpha/alpha.c (alpha_hard_regno_mode_ok): New function.
8169 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8170 * config/arc/arc.h (arc_hard_regno_mode_ok): Delete.
8171 (arc_mode_class): Delete.
8172 (HARD_REGNO_MODE_OK): Delete.
8173 * config/arc/arc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8174 (arc_hard_regno_mode_ok): Rename old array to...
8175 (arc_hard_regno_mode_ok_modes): ...this.
8176 (arc_conditional_register_usage): Update accordingly.
8177 (arc_mode_class): Make static.
8178 (arc_hard_regno_mode_ok): New function.
8179 * config/arm/arm.h (HARD_REGNO_MODE_OK): Delete.
8180 * config/arm/arm-protos.h (arm_hard_regno_mode_ok): Delete.
8181 * config/arm/arm.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8182 (arm_hard_regno_mode_ok): Make static.
8183 * config/arm/arm.md (movdi): Use targetm.hard_regno_mode_ok instead of
8185 * config/avr/avr-protos.h (avr_hard_regno_mode_ok): Delete.
8186 * config/avr/avr.h (HARD_REGNO_MODE_OK): Delete.
8187 * config/avr/avr.c (avr_hard_regno_mode_ok): Make static and
8189 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8190 * config/bfin/bfin-protos.h (hard_regno_mode_ok): Delete.
8191 * config/bfin/bfin.h (HARD_REGNO_MODE_OK): Delete.
8192 * config/bfin/bfin.c (hard_regno_mode_ok): Rename to...
8193 (bfin_hard_regno_mode_ok): ...this. Make static and return a bool.
8194 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8195 * config/bfin/predicates.md (valid_reg_operand): Use
8196 targetm.hard_regno_mode_ok instead of HARD_REGNO_MODE_OK.
8197 * config/c6x/c6x.h (HARD_REGNO_MODE_OK): Delete.
8198 * config/c6x/c6x.c (c6x_hard_regno_mode_ok): New function.
8199 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8200 * config/cr16/cr16.h (HARD_REGNO_MODE_OK): Delete.
8201 * config/cr16/cr16-protos.h (cr16_hard_regno_mode_ok): Delete.
8202 * config/cr16/cr16.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8203 (cr16_hard_regno_mode_ok): Make static and return a bool.
8204 * config/cris/cris.h (HARD_REGNO_MODE_OK): Delete.
8205 * config/cris/cris.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8206 (cris_hard_regno_mode_ok): New function.
8207 * config/epiphany/epiphany.h (epiphany_hard_regno_mode_ok): Delete.
8208 (epiphany_mode_class): Delete.
8209 (HARD_REGNO_MODE_OK): Delete.
8210 * config/epiphany/epiphany-protos.h (hard_regno_mode_ok): Delete.
8211 * config/epiphany/epiphany.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8212 (hard_regno_mode_ok): Rename to...
8213 (epiphany_hard_regno_mode_ok): ...this. Make static and return a bool.
8214 * config/fr30/fr30.h (HARD_REGNO_MODE_OK): Delete.
8215 * config/fr30/fr30.md: Refer to targetm.hard_regno_mode_ok instead of
8217 * config/frv/frv.h (HARD_REGNO_MODE_OK): Delete.
8218 * config/frv/frv-protos.h (frv_hard_regno_mode_ok): Delete.
8219 * config/frv/frv.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8220 (frv_hard_regno_mode_ok): Make static and return a bool.
8221 * config/frv/frv.md: Refer to targetm.hard_regno_mode_ok instead of
8223 * config/ft32/ft32.h (HARD_REGNO_MODE_OK): Delete.
8224 * config/h8300/h8300.h (HARD_REGNO_MODE_OK): Delete.
8225 * config/h8300/h8300-protos.h (h8300_hard_regno_mode_ok): Delete.
8226 * config/h8300/h8300.c (h8300_hard_regno_mode_ok): Make static
8228 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8229 * config/i386/i386.h (HARD_REGNO_MODE_OK): Delete.
8230 * config/i386/i386-protos.h (ix86_hard_regno_mode_ok): Delete.
8231 * config/i386/i386.c (ix86_hard_regno_mode_ok): Make static and
8233 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8234 * config/ia64/ia64.h (HARD_REGNO_MODE_OK): Delete.
8235 * config/ia64/ia64.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8236 (ia64_hard_regno_mode_ok): New function.
8237 * config/iq2000/iq2000.h (HARD_REGNO_MODE_OK): Delete.
8238 * config/iq2000/iq2000.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8239 (iq2000_hard_regno_mode_ok): New function.
8240 * config/lm32/lm32.h (HARD_REGNO_MODE_OK): Delete.
8241 * config/lm32/lm32.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8242 (lm32_hard_regno_mode_ok): New function.
8243 * config/m32c/m32c.h (HARD_REGNO_MODE_OK): Delete.
8244 * config/m32c/m32c-protos.h (m32c_hard_regno_ok): Delete.
8245 * config/m32c/m32c.c (class_can_hold_mode): Use m32c_hard_regno_mode_ok
8246 instead of HARD_REGNO_MODE_OK.
8247 (m32c_hard_regno_ok): Rename to...
8248 (m32c_hard_regno_mode_ok): ...this. Make static and return a bool.
8249 (m32c_cannot_change_mode_class): Update accordingly.
8250 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8251 * config/m32r/m32r.h (m32r_hard_regno_mode_ok): Delete.
8252 (m32r_mode_class): Delete.
8253 (HARD_REGNO_MODE_OK): Delete.
8254 * config/m32r/m32r.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8255 (m32r_hard_regno_mode_ok): Rename to...
8256 (m32r_hard_regno_modes): ...this.
8257 (m32r_mode_class): Make static.
8258 (m32r_hard_regno_mode_ok): New function.
8259 * config/m68k/m68k.h (HARD_REGNO_MODE_OK): Delete.
8260 * config/m68k/m68k-protos.h (m68k_regno_mode_ok): Delete.
8261 * config/m68k/m68k.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8262 (m68k_hard_regno_mode_ok): Make static.
8263 * config/mcore/mcore.h (HARD_REGNO_MODE_OK): Delete.
8264 * config/mcore/mcore.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8265 (mcore_hard_regno_mode_ok): New function.
8266 * config/microblaze/microblaze.h (microblaze_hard_regno_mode_ok)
8267 (HARD_REGNO_MODE_OK): Delete.
8268 * config/microblaze/microblaze.c (microblaze_hard_regno_mode_ok):
8270 (microblaze_hard_regno_mode_ok_p): ...this and make static.
8271 (microblaze_hard_regno_mode_ok): New function.
8272 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8273 * config/mips/mips.h (HARD_REGNO_MODE_OK): Delete.
8274 (mips_hard_regno_mode_ok): Delete.
8275 * config/mips/mips.c (mips_hard_regno_mode_ok): Rename to...
8276 (mips_hard_regno_mode_ok_p): ...this and make static.
8277 (mips_hard_regno_mode_ok_p): Rename to...
8278 (mips_hard_regno_mode_ok_uncached): ...this.
8279 (mips_hard_regno_mode_ok): New function.
8280 (mips_class_max_nregs): Use mips_hard_regno_mode_ok instead
8281 of HARD_REGNO_MODE_OK.
8282 (mips_option_override): Update after above name changes.
8283 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8284 * config/mmix/mmix.h (HARD_REGNO_MODE_OK): Delete.
8285 * config/mn10300/mn10300.h (HARD_REGNO_MODE_OK): Delete.
8286 * config/mn10300/mn10300-protos.h (mn10300_hard_regno_mode_ok): Delete.
8287 * config/mn10300/mn10300.c (mn10300_hard_regno_mode_ok): Make static.
8288 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8289 * config/moxie/moxie.h (HARD_REGNO_MODE_OK): Delete.
8290 * config/msp430/msp430.h (HARD_REGNO_MODE_OK): Delete.
8291 * config/msp430/msp430-protos.h (msp430_hard_regno_mode_ok): Delete.
8292 * config/msp430/msp430.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8293 (msp430_hard_regno_mode_ok): Make static and return a bool.
8294 * config/nds32/nds32.h (HARD_REGNO_MODE_OK): Delete.
8295 * config/nds32/nds32-protos.h (nds32_hard_regno_mode_ok): Delete.
8296 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Make static
8298 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8299 * config/nios2/nios2.h (HARD_REGNO_MODE_OK): Delete.
8300 * config/nvptx/nvptx.h (HARD_REGNO_MODE_OK): Delete.
8301 * config/pa/pa.h (MODES_TIEABLE_P): Update commentary.
8302 * config/pa/pa32-regs.h (HARD_REGNO_MODE_OK): Rename to...
8303 (PA_HARD_REGNO_MODE_OK): ...this
8304 * config/pa/pa64-regs.h (HARD_REGNO_MODE_OK): Rename to...
8305 (PA_HARD_REGNO_MODE_OK): ...this.
8306 * config/pa/pa.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8307 (pa_hard_regno_mode_ok): New function.
8308 * config/pdp11/pdp11.h (HARD_REGNO_MODE_OK): Delete.
8309 * config/pdp11/pdp11.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8310 (pdp11_hard_regno_mode_ok): New function.
8311 * config/powerpcspe/powerpcspe.h (HARD_REGNO_MODE_OK): Delete.
8312 * config/powerpcspe/powerpcspe-protos.h (rs6000_hard_regno_mode_ok_p):
8314 * config/powerpcspe/powerpcspe.c (rs6000_hard_regno_mode_ok_p):
8316 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8317 (rs6000_hard_regno_mode_ok): Rename to...
8318 (rs6000_hard_regno_mode_ok_uncached): ...this.
8319 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8320 (rs6000_hard_regno_mode_ok): New function.
8321 * config/riscv/riscv.h (HARD_REGNO_MODE_OK): Delete.
8322 * config/riscv/riscv-protos.h (riscv_hard_regno_mode_ok_p): Delete.
8323 * config/riscv/riscv.c (riscv_hard_regno_mode_ok_p): Rename to...
8324 (riscv_hard_regno_mode_ok): ...this and make static.
8325 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8326 * config/rl78/rl78.h (HARD_REGNO_MODE_OK): Delete.
8327 * config/rl78/rl78-protos.h (rl78_hard_regno_mode_ok): Delete.
8328 * config/rl78/rl78.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8329 (rl78_hard_regno_mode_ok): Make static and return bool.
8330 * config/rs6000/rs6000.h (HARD_REGNO_MODE_OK): Delete.
8331 * config/rs6000/rs6000-protos.h (rs6000_hard_regno_mode_ok_p):
8333 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok_p): Make static.
8334 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8335 (rs6000_hard_regno_mode_ok): Rename to...
8336 (rs6000_hard_regno_mode_ok_uncached): ...this.
8337 (rs6000_init_hard_regno_mode_ok): Update accordingly.
8338 (rs6000_hard_regno_mode_ok): New function.
8339 * config/rx/rx.h (HARD_REGNO_MODE_OK): Delete.
8340 * config/rx/rx.c (rx_hard_regno_mode_ok): New function.
8341 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8342 * config/s390/s390.h (HARD_REGNO_MODE_OK): Delete.
8343 * config/s390/s390-protos.h (s390_hard_regno_mode_ok): Delete.
8344 * config/s390/s390.c (s390_hard_regno_mode_ok): Make static.
8345 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8346 * config/sh/sh.h (HARD_REGNO_MODE_OK): Delete.
8347 * config/sh/sh-protos.h (sh_hard_regno_mode_ok): Delete.
8348 * config/sh/sh.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8349 (sh_hard_regno_mode_ok): Make static.
8350 * config/sparc/constraints.md: Refer to targetm.hard_regno_mode_ok
8351 instead of HARD_REGNO_MODE_OK.
8352 * config/sparc/sparc.h (hard_regno_mode_classes): Delete.
8353 (sparc_mode_class): Delete.
8354 (HARD_REGNO_MODE_OK): Delete.
8355 * config/sparc/sparc.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8356 (hard_regno_mode_classes): Make static.
8357 (sparc_mode_class): Likewise.
8358 (sparc_hard_regno_mode_ok): New function.
8359 * config/spu/spu.h (HARD_REGNO_MODE_OK): Delete.
8360 * config/stormy16/stormy16.h (HARD_REGNO_MODE_OK): Delete.
8361 * config/stormy16/stormy16.c (xstormy16_hard_regno_mode_ok): New
8363 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8364 * config/tilegx/tilegx.h (HARD_REGNO_MODE_OK): Delete.
8365 * config/tilepro/tilepro.h (HARD_REGNO_MODE_OK): Delete.
8366 * config/v850/v850.h (HARD_REGNO_MODE_OK): Delete.
8367 * config/v850/v850.c (v850_hard_regno_mode_ok): New function.
8368 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8369 * config/vax/vax.h (HARD_REGNO_MODE_OK): Delete.
8370 * config/visium/visium.h (HARD_REGNO_MODE_OK): Delete.
8371 * config/visium/visium.c (TARGET_HARD_REGNO_MODE_OK): Redefine.
8372 (visium_hard_regno_mode_ok): New function.
8373 * config/visium/visium.md: Refer to targetm.hard_regno_mode_ok
8374 instead of HARD_REGNO_MODE_OK.
8375 * config/xtensa/xtensa.h (xtensa_hard_regno_mode_ok): Delete.
8376 (HARD_REGNO_MODE_OK): Delete.
8377 * config/xtensa/xtensa.c (xtensa_hard_regno_mode_ok): Rename to...
8378 (xtensa_hard_regno_mode_ok_p): ...this and make static.
8379 (xtensa_option_override): Update accordingly.
8380 (TARGET_HARD_REGNO_MODE_OK): Redefine.
8381 (xtensa_hard_regno_mode_ok): New function.
8382 * system.h (HARD_REGNO_MODE_OK): Poison.
8384 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8385 Alan Hayward <alan.hayward@arm.com>
8386 David Sherwood <david.sherwood@arm.com>
8388 * target.def (hard_regno_call_part_clobbered): New hook.
8389 * doc/tm.texi.in (HARD_REGNO_CALL_PART_CLOBBERED): Replace with...
8390 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): ...this hook.
8391 * doc/tm.texi: Regenerate.
8392 * hooks.h (hook_bool_uint_mode_false): Declare.
8393 * hooks.c (hook_bool_uint_mode_false): New function.
8394 * regs.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8395 * cselib.c (cselib_process_insn): Use
8396 targetm.hard_regno_call_part_clobbered instead of
8397 HARD_REGNO_CALL_PART_CLOBBERED.
8398 * ira-conflicts.c (ira_build_conflicts): Likewise.
8399 * ira-costs.c (ira_tune_allocno_costs): Likewise.
8400 * lra-constraints.c (need_for_call_save_p): Likewise.
8401 * lra-lives.c: Include target.h.
8402 (check_pseudos_live_through_calls): Use
8403 targetm.hard_regno_call_part_clobbered instead of
8404 HARD_REGNO_CALL_PART_CLOBBERED.
8405 * regcprop.c: Include target.h.
8406 (copyprop_hardreg_forward_1): Use
8407 targetm.hard_regno_call_part_clobbered instead of
8408 HARD_REGNO_CALL_PART_CLOBBERED.
8409 * reginfo.c (choose_hard_reg_mode): Likewise.
8410 * regrename.c (check_new_reg_p): Likewise.
8411 * reload.c (find_equiv_reg): Likewise.
8412 * reload1.c (emit_reload_insns): Likewise.
8413 * sched-deps.c (deps_analyze_insn): Likewise.
8414 * sel-sched.c (init_regs_for_mode): Likewise.
8415 (mark_unavailable_hard_regs): Likewise.
8416 * targhooks.c (default_dwarf_frame_reg_mode): Likewise.
8417 * config/aarch64/aarch64.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8418 * config/aarch64/aarch64.c (aarch64_hard_regno_call_part_clobbered):
8420 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8421 * config/avr/avr.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8422 * config/avr/avr-protos.h (avr_hard_regno_call_part_clobbered):
8424 * config/avr/avr.c (avr_hard_regno_call_part_clobbered): Make static
8426 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8427 * config/i386/i386.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8428 * config/i386/i386.c (ix86_hard_regno_call_part_clobbered): New
8430 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8431 * config/mips/mips.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8432 * config/mips/mips.c (mips_hard_regno_call_part_clobbered): New
8434 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8435 * config/powerpcspe/powerpcspe.h (HARD_REGNO_CALL_PART_CLOBBERED):
8437 * config/powerpcspe/powerpcspe.c
8438 (rs6000_hard_regno_call_part_clobbered): New function.
8439 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8440 * config/rs6000/rs6000.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8441 * config/rs6000/rs6000.c (rs6000_hard_regno_call_part_clobbered):
8443 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8444 * config/s390/s390.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8445 * config/s390/s390.c (s390_hard_regno_call_part_clobbered): New
8447 (TARGET_HARD_REGNO_CALL_PART_CLOBBERED): Redefine.
8448 * config/sh/sh.h (HARD_REGNO_CALL_PART_CLOBBERED): Delete.
8449 * system.h (HARD_REGNO_CALL_PART_CLOBBERED): Poison.
8451 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8452 Alan Hayward <alan.hayward@arm.com>
8453 David Sherwood <david.sherwood@arm.com>
8455 * rtl.h (subreg_memory_offset): Declare.
8456 * emit-rtl.c (subreg_memory_offset): New function.
8457 * expmed.c (store_bit_field_1): Use it.
8458 * expr.c (undefined_operand_subword_p): Likewise.
8459 * simplify-rtx.c (simplify_subreg): Likewise.
8461 2017-09-04 Alexander Monakov <amonakov@ispras.ru>
8463 PR rtl-optimization/57448
8466 * optabs.c (expand_atomic_load): Place compiler memory barriers if
8467 using atomic_load pattern.
8468 (expand_atomic_store): Likewise.
8470 2017-09-04 Jakub Jelinek <jakub@redhat.com>
8473 * gimple-fold.c (gimple_fold_call): Optimize away useless UBSAN_PTR
8474 and UBSAN_BOUNDS internal calls. Clean up IFN_UBSAN_OBJECT_SIZE
8475 handling. Use replace_call_with_value with NULL instead of
8476 gsi_replace, unlink_stmt_vdef and release_defs.
8478 * gdbhooks.py (OptMachineModePrinter.to_string): Use 8 spaces
8481 * lra-remat.c (reg_overlap_for_remat_p): Fix a pasto.
8483 2017-09-04 Richard Sandiford <richard.sandiford@linaro.org>
8486 * rtl.h (emit_library_call_value_1): Declare.
8487 (emit_library_call): Replace declaration with a series of overloads.
8488 Remove the parameter count argument.
8489 (emit_library_call_value): Likewise.
8490 * calls.c (emit_library_call_value_1): Make global. Replace varargs
8491 with an "rtx_mode_t *".
8492 (emit_library_call_value): Delete.
8493 (emit_library_call): Likewise.
8494 * asan.c (asan_emit_stack_protection): Update calls accordingly.
8495 (asan_emit_allocas_unpoison): Likewise.
8496 * builtins.c (expand_builtin_powi): Likewise.
8497 (expand_asan_emit_allocas_unpoison): Likewise.
8498 * cfgexpand.c (expand_main_function): Likewise.
8499 * config/aarch64/aarch64.c (aarch64_trampoline_init): Likewise.
8500 * config/aarch64/aarch64.h (PROFILE_HOOK): Likewise.
8501 * config/alpha/alpha.c (alpha_trampoline_init): Likewise.
8502 * config/arm/arm.c (arm_trampoline_init): Likewise.
8503 (arm_call_tls_get_addr): Likewise.
8504 (arm_expand_divmod_libfunc): Likewise.
8505 * config/bfin/bfin.md (umulsi3_highpart): Likewise.
8506 (smulsi3_highpart): Likewise.
8507 * config/c6x/c6x.c (c6x_initialize_trampoline): Likewise.
8508 (c6x_expand_compare): Likewise.
8509 (c6x_expand_movmem): Likewise.
8510 * config/frv/frv.c (frv_trampoline_init): Likewise.
8511 * config/i386/i386.c (ix86_trampoline_init): Likewise.
8512 (ix86_expand_divmod_libfunc): Likewise.
8513 * config/ia64/ia64.c (ia64_expand_tls_address): Likewise.
8514 (ia64_expand_compare): Likewise.
8515 (ia64_profile_hook): Likewise.
8516 * config/ia64/ia64.md (save_stack_nonlocal): Likewise.
8517 (nonlocal_goto): Likewise.
8518 (restore_stack_nonlocal): Likewise.
8519 * config/m32r/m32r.c (block_move_call): Likewise.
8520 (m32r_trampoline_init): Likewise.
8521 * config/m68k/linux.h (FINALIZE_TRAMPOLINE): Likewise.
8522 * config/m68k/m68k.c (m68k_call_tls_get_addr): Likewise.
8523 (m68k_call_m68k_read_tp): Likewise.
8524 * config/microblaze/microblaze.c (microblaze_call_tls_get_addr)
8525 (microblaze_expand_divide): Likewise.
8526 * config/mips/mips.h (mips_args): Likewise.
8527 * config/mips/sdemtk.h (mips_sync_icache): Likewise.
8528 (MIPS_ICACHE_SYNC): Likewise.
8529 * config/nios2/nios2.c (nios2_emit_expensive_div): Likewise.
8530 (nios2_trampoline_init): Likewise.
8531 * config/pa/pa.c (hppa_tls_call): Likewise.
8532 (pa_trampoline_init): Likewise.
8533 * config/pa/pa.md (canonicalize_funcptr_for_compare): Likewise.
8534 * config/powerpcspe/powerpcspe.c (rs6000_legitimize_tls_address)
8535 (expand_strn_compare): Likewise.
8536 (rs6000_generate_compare): Likewise.
8537 (rs6000_expand_float128_convert): Likewise.
8538 (output_profile_hook): Likewise.
8539 (rs6000_trampoline_init): Likewise.
8540 * config/powerpcspe/powerpcspe.md (neg<mode>2): Likewise.
8541 * config/riscv/riscv.h (PROFILE_HOOK): Likewise.
8542 * config/rs6000/rs6000-string.c (expand_strn_compare): Likewise.
8543 * config/rs6000/rs6000.c (rs6000_legitimize_tls_address): Likewise.
8544 (rs6000_generate_compare): Likewise.
8545 (rs6000_expand_float128_convert): Likewise.
8546 (output_profile_hook): Likewise.
8547 (rs6000_trampoline_init): Likewise.
8548 * config/rs6000/rs6000.md (neg<mode>2): Likewise.
8549 * config/sh/sh.c (sh_trampoline_init): Likewise.
8550 * config/sparc/sparc.c (emit_soft_tfmode_libcall): Likewise.
8551 (sparc_emit_float_lib_cmp): Likewise.
8552 (sparc32_initialize_trampoline): Likewise.
8553 (sparc64_initialize_trampoline): Likewise.
8554 (sparc_profile_hook): Likewise.
8555 * config/spu/spu.c (ea_load_store): Likewise.
8556 * config/spu/spu.md (floatunssidf2): Likewise.
8557 * config/tilegx/tilegx.c (tilegx_trampoline_init): Likewise.
8558 * config/tilepro/tilepro.c (tilepro_trampoline_init): Likewise.
8559 * config/visium/visium.c (expand_block_move_4): Likewise.
8560 (expand_block_move_2): Likewise.
8561 (expand_block_move_1): Likewise.
8562 (expand_block_set_4): Likewise.
8563 (expand_block_set_2): Likewise.
8564 (expand_block_set_1): Likewise.
8565 (visium_trampoline_init): Likewise.
8566 (visium_profile_hook): Likewise.
8567 * config/xtensa/xtensa.c (xtensa_expand_nonlocal_goto): Likewise.
8568 (xtensa_setup_frame_addresses): Likewise.
8569 (xtensa_trampoline_init): Likewise.
8570 * except.c (sjlj_emit_function_enter): Likewise.
8571 (sjlj_emit_function_exit): Likewise.
8572 * explow.c (allocate_dynamic_stack_space): Likewise.
8573 (probe_stack_range): Likewise.
8574 * expr.c (convert_mode_scalar): Likewise.
8575 * optabs.c (expand_binop): Likewise.
8576 (expand_twoval_binop_libfunc): Likewise.
8577 (expand_unop): Likewise.
8578 (prepare_cmp_insn): Likewise.
8579 (prepare_float_lib_cmp): Likewise.
8580 (expand_float): Likewise.
8581 (expand_fix): Likewise.
8582 (expand_fixed_convert): Likewise.
8583 (maybe_emit_sync_lock_test_and_set): Likewise.
8584 (expand_atomic_compare_and_swap): Likewise.
8585 (expand_mem_thread_fence): Likewise.
8586 (expand_atomic_fetch_op): Likewise.
8588 2017-09-03 Gerald Pfeifer <gerald@pfeifer.com>
8590 * doc/generic.texi (OpenACC): Adjust URL.
8591 * doc/invoke.texi (C Dialect Options): Ditto.
8593 2017-09-03 Uros Bizjak <ubizjak@gmail.com>
8595 * config/i386/i386.md (*bt<mode>): Use nonimmediate_operand
8596 predicate for operand 1. Add (m,<S>) constraint.
8597 (*jcc_bt<mode>): Use nonimmediate_operand predicate for operand 1.
8598 Prevent memory operand 1 with register operand 2.
8600 2017-09-01 Segher Boessenkool <segher@kernel.crashing.org>
8602 PR rtl-optimization/82024
8603 * combine.c (try_combine): If the combination result is a PARALLEL,
8604 and we only need to retain the SET in there that would be placed
8605 at I2, check that we can place that at I3 instead, before doing so.
8607 2017-09-01 Jakub Jelinek <jakub@redhat.com>
8610 * config/i386/i386.c (ix86_init_large_pic_reg): Return label
8612 (ix86_init_pic_reg): Remember label from ix86_init_large_pic_reg,
8613 if non-NULL and preceded by NOTE_INSN_BASIC_BLOCK, swap the note
8616 2017-09-01 Joerg Sonnenberger <joerg@bec.de>
8617 Jeff Law <law@redhat.com>
8619 * varasm.c (bss_initializer_p): Do not put constants into .bss
8620 (categorize_decl_for_section): Handle bss_initializer_p returning
8621 false when DECL_INITIAL is NULL.
8623 2017-09-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8626 * config/s390/s390.c (s390_can_inline_p): New function.
8628 2017-09-01 Jeff Law <law@redhat.com>
8630 PR tree-optimization/82052
8631 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
8632 Always initialize the returned slot after a hash table miss
8633 when INSERT is true.
8635 2017-09-01 Alexander Monakov <amonakov@ispras.ru>
8637 * config/s390/s390.md (mem_signal_fence): Remove.
8638 * doc/md.texi (mem_signal_fence): Remove.
8639 * optabs.c (expand_mem_signal_fence): Remove uses of mem_signal_fence.
8641 * target-insns.def (mem_signal_fence): Remove.
8643 2017-09-01 Jakub Jelinek <jakub@redhat.com>
8646 * doc/invoke.texi: Document -fsanitize=pointer-overflow.
8649 * asan.c (create_odr_indicator): Strip name encoding from assembler
8650 name before appending it after __odr_asan_.
8652 2017-09-01 Martin Liska <mliska@suse.cz>
8654 PR tree-optimization/82059
8655 * gimple-ssa-isolate-paths.c (isolate_path): Add profile and
8656 frequency only when an edge is redirected.
8658 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8660 * config/arc/arc-c.c (__ARC_LPC_WIDTH__): Add builtin define.
8661 * config/arc/arc.c (ARC_MAX_LOOP_LENGTH): Define.
8662 (arc_conditional_register_usage): Remove ARC600 lp_count
8664 (arc_file_start): Emit Tag_ARC_CPU_variation.
8665 (arc_can_use_doloop_p): New conditions to use ZOLs.
8666 (hwloop_fail): New function.
8667 (hwloop_optimize): Likewise.
8668 (hwloop_pattern_reg): Likewise.
8669 (arc_doloop_hooks): New struct, to be used with reorg_loops.
8670 (arc_reorg_loops): New function, calls reorg_loops.
8671 (arc_reorg): Call arc_reorg_loops. Remove old ZOL handling.
8672 (arc600_corereg_hazard): Remove ZOL checking, case handled by
8674 (arc_loop_hazard): Remove function, functionality moved into
8676 (arc_hazard): Remove arc_loop_hazard call.
8677 (arc_adjust_insn_length): Remove ZOL handling, functionality moved
8678 into hwloop_optimize.
8679 (arc_label_align): Remove ZOL handling.
8680 * config/arc/arc.h (LOOP_ALIGN): Changed to 0.
8681 * config/arc/arc.md (doloop_begin): Remove pattern.
8682 (doloop_begin_i): Likewise.
8683 (doloop_end_i): Likewise.
8684 (doloop_fallback): Likewise.
8685 (doloop_fallback_m): Likewise.
8686 (doloop_end): Reimplement expand.
8687 (arc_lp): New pattern for LP instruction.
8688 (loop_end): New pattern.
8689 (loop_fail): Likewise.
8690 (decrement_and_branch_until_zero): Likewise.
8691 * config/arc/arc.opt (mlpc-width): New option.
8692 * doc/invoke.texi (mlpc-width): Document option.
8694 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8696 * config/arc/arc.c (arc_ifcvt): Remove use of merge_blocks call.
8697 (arc_ccfsm_advance): Fix checking for delay slots.
8698 (arc_reorg): Add rtl dump after each call to arc_ifcvt.
8700 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8702 * config/arc/arc.md (movqi_insn): Add stores to save constant long
8704 (movhi_insn): Update store instruction constraint which are saving
8705 6-bit short immediates.
8706 (movsi_insn): Consider also short scaled load operations.
8707 (zero_extendhisi2_i): Use Usd constraint instead of T.
8708 (extendhisi2_i): Add q constraint.
8709 (arc_clzsi2): Add type and length attributes.
8710 (arc_ctzsi2): Likewise.
8711 * config/arc/constraints.md (Usc): Update constraint, the
8712 assembler can parse two relocations for a single instruction.
8714 2017-09-01 Claudiu Zissulescu <claziss@synopsys.com>
8716 * config/arc/arc.c (arc_use_anchors_for_symbol_p): New function.
8717 (TARGET_USE_ANCHORS_FOR_SYMBOL_P): Define.
8719 2017-08-31 Olivier Hainque <hainque@adacore.com>
8721 * config.gcc (powerpc-wrs-vxworks|vxworksae|vxworksmils): Now
8722 match as powerpc-wrs-vxworks*.
8724 2017-08-31 James Greenhalgh <james.greenhalgh@arm.com>
8726 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): Fix
8727 register constraint for by-element operand.
8728 (aarch64_mls_elt_merge<mode>): Likewise.
8730 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8732 * config/arc/arc.c (arc_can_follow_jump): Check for short
8735 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8737 * config.gcc: Use g.opt for arc.
8738 * config/arc/arc.c (LEGITIMATE_SCALED_ADDRESS_P): Deleted,
8739 functionality moved to ...
8740 (legitimate_scaled_address_p): New function, ...here.
8741 (LEGITIMATE_SMALL_DATA_OFFSET_P): New define.
8742 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Use the above define.
8743 (legitimate_offset_address_p): Delete TARGET_NO_SDATA_SET
8745 (arc_override_options): Handle G option.
8746 (arc_output_pic_addr_const): Correct function definition.
8747 (arc_legitimate_address_p): Use legitimate_scaled_address_p.
8748 (arc_decl_anon_ns_mem_p): Delete.
8749 (arc_in_small_data_p): Overhaul this function to take into
8750 consideration the value given via G option.
8751 (arc_rewrite_small_data_1): Renamed and corrected old
8752 arc_rewrite_small_data function.
8753 (arc_rewrite_small_data): New function.
8754 (small_data_pattern): Don't use pic_offset_table_rtx.
8755 * config/arc/arc.h (CC1_SPEC): Recognize G option.
8756 * config/arc/simdext.md (movmisalignv2hi): Use
8757 prepare_move_operands function.
8759 (movmisalign*): Likewise.
8760 * doc/invoke.texi (ARC options): Document -G option.
8762 2017-08-31 Claudiu Zissulescu <claziss@synopsys.com>
8764 * config/arc/arc-protos.h (compact_sda_memory_operand): Update
8766 * config/arc/arc.c (arc_print_operand): Output scalled address for
8767 sdata whenever is possible.
8768 (arc_in_small_data_p): Allow sdata for 64bit datum when double
8769 load/stores are available.
8770 (compact_sda_memory_operand): Check for the alignment required by
8771 code density instructions.
8772 * config/arc/arc.md (movsi_insn): Use newly introduced Us0
8774 * config/arc/constraints.md (Usd): Update constraint.
8775 (Us0): New constraint.
8776 (Usc): Update constraint.
8778 2017-08-31 Richard Biener <rguenther@suse.de>
8781 * dwarf2out.c (dwarf2out_early_global_decl): Process each
8784 2017-08-31 Tamar Christina <tamar.christina@arm.com>
8786 * config/aarch64/aarch64-builtins.c (aarch64_init_simd_builtins):
8787 Resize type_signature.
8789 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8790 Alan Hayward <alan.hayward@arm.com>
8791 David Sherwood <david.sherwood@arm.com>
8793 * config/aarch64/aarch64.c (aarch64_base_register_rtx_p): Only allow
8794 subregs whose inner modes can be stored in GPRs.
8795 (aarch64_classify_index): Likewise.
8797 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8798 Alan Hayward <alan.hayward@arm.com>
8799 David Sherwood <david.sherwood@arm.com>
8801 * config/aarch64/iterators.md (V_cmp_result): Rename to...
8802 (V_INT_EQUIV): ...this.
8803 (v_cmp_result): Rename to...
8804 (v_int_equiv): ...this.
8805 * config/aarch64/aarch64.md (xorsign<mode>3): Update accordingly.
8806 * config/aarch64/aarch64-simd.md (xorsign<mode>3): Likewise.
8807 (copysign<mode>3): Likewise.
8808 (aarch64_simd_bsl<mode>_internal): Likewise.
8809 (aarch64_simd_bsl<mode>): Likewise.
8810 (vec_cmp<mode><mode>): Likewise.
8811 (vcond<mode><mode>): Likewise.
8812 (vcond<v_cmp_mixed><mode>): Likewise.
8813 (vcondu<mode><v_cmp_mixed>): Likewise.
8814 (aarch64_cm<optab><mode>): Likewise.
8815 (aarch64_cmtst<mode>): Likewise.
8816 (aarch64_fac<optab><mode>): Likewise.
8817 (vec_perm_const<mode>): Likewise.
8818 (vcond_mask_<mode><v_cmp_result>): Rename to...
8819 (vcond_mask_<mode><v_int_equiv>): ...this.
8820 (vec_cmp<mode><v_cmp_result>): Rename to...
8821 (vec_cmp<mode><v_int_equiv>): ...this.
8823 2017-08-31 Richard Sandiford <richard.sandiford@linaro.org>
8824 Alan Hayward <alan.hayward@arm.com>
8825 David Sherwood <david.sherwood@arm.com>
8827 * config/aarch64/aarch64-modes.def: Remove 32-, 48- and 64-byte
8829 * config/aarch64/iterators.md (VRL2, VRL3, VRL4): Delete.
8830 * config/aarch64/aarch64.md (UNSPEC_LD2_DREG, UNSPEC_LD3_DREG)
8831 (UNSPEC_LD4_DREG): New unspecs.
8832 * config/aarch64/aarch64-simd.md (aarch64_ld2<mode>_dreg_le)
8833 (aarch64_ld2<mode>_dreg_be): Replace with...
8834 (aarch64_ld2<mode>_dreg): ...this pattern and use the new DREG
8836 (aarch64_ld3<mode>_dreg_le)
8837 (aarch64_ld3<mode>_dreg_be): Replace with...
8838 (aarch64_ld3<mode>_dreg): ...this pattern and use the new DREG
8840 (aarch64_ld4<mode>_dreg_le)
8841 (aarch64_ld4<mode>_dreg_be): Replace with...
8842 (aarch64_ld4<mode>_dreg): ...this pattern and use the new DREG
8845 2017-08-30 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8847 PR tree-optimization/81987
8848 * gimple-ssa-strength-reduction.c (insert_initializers): Don't
8849 insert an initializer in a location not dominated by the stride
8852 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8854 * tree-eh.c (lower_try_finally_switch): Set the location of the finally
8855 on the entire header of the finally block in the fallthru case.
8857 2017-08-30 Eric Botcazou <ebotcazou@adacore.com>
8859 * varasm.c (decode_addr_const): Deal with INDIRECT_REF <INTEGER_CST>.
8861 2017-08-30 Pat Haugen <pthaugen@us.ibm.com>
8863 * config/rs6000/rs6000.c (rs6000_emit_prologue_move_from_cr): Rename from
8864 rs6000_emit_move_from_cr and call renamed function.
8865 (rs6000_emit_prologue): Call renamed functions.
8866 * config/rs6000/rs6000.md (prologue_movesi_from_cr): Rename from
8867 movesi_from_cr, remove volatile CRs.
8869 2017-08-30 Jon Beniston <jon@beniston.com>
8870 Richard Biener <rguenther@suse.de>
8872 * tree-vect-patterns.c (vect_pattern_recog_1): Use VECTOR_TYPE_P instead
8873 of VECTOR_MODE_P check.
8874 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Allow single
8875 element vector types.
8877 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8879 * df.h (df_read_modify_subreg_p): Remove in favor of...
8880 * rtl.h (read_modify_subreg_p): ...this new function. Take a
8881 const_rtx instead of an rtx.
8882 * cprop.c (local_cprop_find_used_regs): Update accordingly.
8883 * df-problems.c (df_word_lr_mark_ref): Likewise.
8884 * ira-lives.c (mark_pseudo_reg_live): Likewise.
8885 (mark_pseudo_reg_dead): Likewise.
8886 (mark_ref_dead): Likewise.
8887 * reginfo.c (init_subregs_of_mode): Likewise.
8888 * sched-deps.c (sched_analyze_1): Likewise.
8889 * df-scan.c (df_def_record_1): Likewise.
8890 (df_uses_record): Likewise.
8891 (df_read_modify_subreg_p): Remove in favor of...
8892 * rtlanal.c (read_modify_subreg_p): ...this new function. Take a
8893 const_rtx instead of an rtx.
8895 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8896 Alan Hayward <alan.hayward@arm.com>
8897 David Sherwood <david.sherwood@arm.com>
8899 * rtl.h (partial_subreg_p): New function.
8900 * caller-save.c (save_call_clobbered_regs): Use it.
8901 * calls.c (expand_call): Likewise.
8902 * combine.c (combinable_i3pat): Likewise.
8903 (simplify_set): Likewise.
8904 (make_extraction): Likewise.
8905 (make_compound_operation_int): Likewise.
8906 (gen_lowpart_or_truncate): Likewise.
8907 (force_to_mode): Likewise.
8908 (make_field_assignment): Likewise.
8909 (reg_truncated_to_mode): Likewise.
8910 (record_truncated_value): Likewise.
8911 (move_deaths): Likewise.
8912 * cse.c (record_jump_cond): Likewise.
8913 (cse_insn): Likewise.
8914 * cselib.c (cselib_lookup_1): Likewise.
8915 * expmed.c (extract_bit_field_using_extv): Likewise.
8916 * function.c (assign_parm_setup_reg): Likewise.
8917 * ifcvt.c (noce_convert_multiple_sets): Likewise.
8918 * ira-build.c (create_insn_allocnos): Likewise.
8919 * lra-coalesce.c (merge_pseudos): Likewise.
8920 * lra-constraints.c (match_reload): Likewise.
8921 (simplify_operand_subreg): Likewise.
8922 (curr_insn_transform): Likewise.
8923 * lra-lives.c (process_bb_lives): Likewise.
8924 * lra.c (new_insn_reg): Likewise.
8925 (lra_substitute_pseudo): Likewise.
8926 * regcprop.c (mode_change_ok): Likewise.
8927 (maybe_mode_change): Likewise.
8928 (copyprop_hardreg_forward_1): Likewise.
8929 * reload.c (push_reload): Likewise.
8930 (find_reloads): Likewise.
8931 (find_reloads_subreg_address): Likewise.
8932 * reload1.c (alter_reg): Likewise.
8933 (eliminate_regs_1): Likewise.
8934 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
8936 2017-08-30 David Edelsohn <dje.gcc@gmail.com>
8938 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Revert
8939 back to if statements, including unpack.
8941 2017-08-30 Martin Liska <mliska@suse.cz>
8944 * ipa-icf-gimple.c (func_checker::compare_tree_list_operand):
8946 (func_checker::compare_asm_inputs_outputs): ... this function.
8947 (func_checker::compare_gimple_asm): Use the function to compare
8948 also ASM constrains.
8949 * ipa-icf-gimple.h: Rename the function.
8951 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8952 Alan Hayward <alan.hayward@arm.com>
8953 David Sherwood <david.sherwood@arm.com>
8955 * coretypes.h (complex_mode): New type.
8956 * gdbhooks.py (build_pretty_printer): Handle it.
8957 * machmode.h (complex_mode): New class.
8958 (complex_mode::includes_p): New function.
8959 (is_complex_int_mode): Likewise.
8960 (is_complex_float_mode): Likewise.
8961 * genmodes.c (get_mode_class): Handle complex mode classes.
8962 * function.c (expand_function_end): Use is_complex_int_mode.
8964 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8965 Alan Hayward <alan.hayward@arm.com>
8966 David Sherwood <david.sherwood@arm.com>
8968 * coretypes.h (scalar_mode_pod): New typedef.
8969 * gdbhooks.py (build_pretty_printer): Handle it.
8970 * machmode.h (gt_ggc_mx, gt_pch_nx): New functions.
8971 * fixed-value.h (fixed_value::mode): Change type to scalar_mode_pod.
8972 * fold-const.c (fold_convert_const_int_from_fixed): Use scalar_mode.
8973 * tree-streamer-in.c (unpack_ts_fixed_cst_value_fields): Use
8976 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8977 Alan Hayward <alan.hayward@arm.com>
8978 David Sherwood <david.sherwood@arm.com>
8980 * machmode.h (mode_for_vector): Take a scalar_mode instead
8982 * stor-layout.c (mode_for_vector): Likewise.
8983 * explow.c (promote_mode): Use as_a <scalar_mode>.
8984 * sdbout.c (sdbout_parms): Use is_a <scalar_mode>.
8986 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
8987 Alan Hayward <alan.hayward@arm.com>
8988 David Sherwood <david.sherwood@arm.com>
8990 * target.def (preferred_simd_mode): Take a scalar_mode
8991 instead of a machine_mode.
8992 * targhooks.h (default_preferred_simd_mode): Likewise.
8993 * targhooks.c (default_preferred_simd_mode): Likewise.
8994 * config/arc/arc.c (arc_preferred_simd_mode): Likewise.
8995 * config/arm/arm.c (arm_preferred_simd_mode): Likewise.
8996 * config/c6x/c6x.c (c6x_preferred_simd_mode): Likewise.
8997 * config/epiphany/epiphany.c (epiphany_preferred_simd_mode): Likewise.
8998 * config/i386/i386.c (ix86_preferred_simd_mode): Likewise.
8999 * config/mips/mips.c (mips_preferred_simd_mode): Likewise.
9000 * config/nvptx/nvptx.c (nvptx_preferred_simd_mode): Likewise.
9001 * config/powerpcspe/powerpcspe.c (rs6000_preferred_simd_mode):
9003 * config/rs6000/rs6000.c (rs6000_preferred_simd_mode): Likewise.
9004 * config/s390/s390.c (s390_preferred_simd_mode): Likewise.
9005 * config/sparc/sparc.c (sparc_preferred_simd_mode): Likewise.
9006 * config/aarch64/aarch64.c (aarch64_preferred_simd_mode): Likewise.
9007 (aarch64_simd_scalar_immediate_valid_for_move): Update accordingly.
9008 * doc/tm.texi: Regenerate.
9009 * optabs-query.c (can_vec_mask_load_store_p): Return false for
9012 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9013 Alan Hayward <alan.hayward@arm.com>
9014 David Sherwood <david.sherwood@arm.com>
9016 * target.def (scalar_mode_supported_p): Take a scalar_mode
9017 instead of a machine_mode.
9018 * targhooks.h (default_scalar_mode_supported_p): Likewise.
9019 * targhooks.c (default_scalar_mode_supported_p): Likewise.
9020 * config/aarch64/aarch64.c (aarch64_scalar_mode_supported_p): Likewise.
9021 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
9022 * config/arm/arm.c (arm_scalar_mode_supported_p): Likewise.
9023 * config/avr/avr.c (avr_scalar_mode_supported_p): Likewise.
9024 * config/c6x/c6x.c (c6x_scalar_mode_supported_p): Likewise.
9025 * config/i386/i386.c (ix86_scalar_mode_supported_p): Likewise.
9026 * config/ia64/ia64.c (ia64_scalar_mode_supported_p): Likewise.
9027 * config/mips/mips.c (mips_scalar_mode_supported_p): Likewise.
9028 * config/msp430/msp430.c (msp430_scalar_mode_supported_p): Likewise.
9029 * config/pa/pa.c (pa_scalar_mode_supported_p): Likewise.
9030 * config/pdp11/pdp11.c (pdp11_scalar_mode_supported_p): Likewise.
9031 * config/powerpcspe/powerpcspe.c (rs6000_scalar_mode_supported_p):
9033 * config/rs6000/rs6000.c (rs6000_scalar_mode_supported_p): Likewise.
9034 * config/s390/s390.c (s390_scalar_mode_supported_p): Likewise.
9035 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
9036 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
9037 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p):
9039 * doc/tm.texi: Regenerate.
9041 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9042 Alan Hayward <alan.hayward@arm.com>
9043 David Sherwood <david.sherwood@arm.com>
9045 * coretypes.h (opt_scalar_mode): New typedef.
9046 * gdbhooks.py (build_pretty_printers): Handle it.
9047 * machmode.h (mode_iterator::get_2xwider): Add overload for
9049 * emit-rtl.c (init_emit_once): Use opt_scalar_mode when iterating
9051 * expr.c (convert_mode_scalar): Likewise.
9052 * omp-low.c (omp_clause_aligned_alignment): Likewise.
9053 * optabs.c (expand_float): Likewise.
9054 (expand_fix): Likewise.
9055 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9057 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9058 Alan Hayward <alan.hayward@arm.com>
9059 David Sherwood <david.sherwood@arm.com>
9061 * optabs.c (expand_float): Explicitly check for scalars before
9062 using a branching expansion.
9063 (expand_fix): Likewise.
9065 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9066 Alan Hayward <alan.hayward@arm.com>
9067 David Sherwood <david.sherwood@arm.com>
9069 * expr.c (convert_mode): Split scalar handling out into...
9070 (convert_mode_scalar): ...this new function. Treat the modes
9073 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9074 Alan Hayward <alan.hayward@arm.com>
9075 David Sherwood <david.sherwood@arm.com>
9077 * omp-expand.c (expand_omp_atomic): Use is_int_mode, is_float_mode
9079 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Likewise.
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 * fixed-value.h (fixed_from_double_int): Take a scalar_mode
9086 rather than a machine_mode.
9087 (fixed_from_string): Likewise.
9088 (fixed_convert): Likewise.
9089 (fixed_convert_from_int): Likewise.
9090 (fixed_convert_from_real): Likewise.
9091 (real_convert_from_fixed): Likewise.
9092 * fixed-value.c (fixed_from_double_int): Likewise.
9093 (fixed_from_string): Likewise.
9094 (fixed_convert): Likewise.
9095 (fixed_convert_from_int): Likewise.
9096 (fixed_convert_from_real): Likewise.
9097 (real_convert_from_fixed): Likewise.
9098 * config/avr/avr.c (avr_out_round): Use as_a <scalar_mode>.
9100 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9101 Alan Hayward <alan.hayward@arm.com>
9102 David Sherwood <david.sherwood@arm.com>
9104 * emit-rtl.c (immed_double_const): Use is_a <scalar_mode> instead
9105 of separate mode class checks. Do not allow vector modes here.
9106 (immed_wide_int_const): Use as_a <scalar_mode>.
9107 * explow.c (trunc_int_for_mode): Likewise.
9108 * rtl.h (wi::int_traits<rtx_mode_t>::get_precision): Likewise.
9109 (wi::shwi): Likewise.
9110 (wi::min_value): Likewise.
9111 (wi::max_value): Likewise.
9112 * dwarf2out.c (loc_descriptor): Likewise.
9113 * simplify-rtx.c (simplify_immed_subreg): Fix rtx_mode_t argument
9116 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9117 Alan Hayward <alan.hayward@arm.com>
9118 David Sherwood <david.sherwood@arm.com>
9120 * tree.h (SCALAR_TYPE_MODE): New macro.
9121 * expr.c (expand_expr_addr_expr_1): Use it.
9122 (expand_expr_real_2): Likewise.
9123 * fold-const.c (fold_convert_const_fixed_from_fixed): Likeise.
9124 (fold_convert_const_fixed_from_int): Likewise.
9125 (fold_convert_const_fixed_from_real): Likewise.
9126 (native_encode_fixed): Likewise
9127 (native_encode_complex): Likewise
9128 (native_encode_vector): Likewise.
9129 (native_interpret_fixed): Likewise.
9130 (native_interpret_real): Likewise.
9131 (native_interpret_complex): Likewise.
9132 (native_interpret_vector): Likewise.
9133 * omp-simd-clone.c (simd_clone_adjust_return_type): Likewise.
9134 (simd_clone_adjust_argument_types): Likewise.
9135 (simd_clone_init_simd_arrays): Likewise.
9136 (simd_clone_adjust): Likewise.
9137 * stor-layout.c (layout_type): Likewise.
9138 * tree.c (build_minus_one_cst): Likewise.
9139 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9140 * tree-inline.c (estimate_move_cost): Likewise.
9141 * tree-ssa-math-opts.c (convert_plusminus_to_widen): Likewise.
9142 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise.
9143 (vectorizable_reduction): Likewise.
9144 * tree-vect-patterns.c (vect_recog_widen_mult_pattern): Likewise.
9145 (vect_recog_mixed_size_cond_pattern): Likewise.
9146 (check_bool_pattern): Likewise.
9147 (adjust_bool_pattern): Likewise.
9148 (search_type_for_mask_1): Likewise.
9149 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
9150 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
9151 (vectorizable_load): Likewise.
9152 (vectorizable_store): Likewise.
9153 * ubsan.c (ubsan_encode_value): Likewise.
9154 * varasm.c (output_constant): Likewise.
9156 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9157 Alan Hayward <alan.hayward@arm.com>
9158 David Sherwood <david.sherwood@arm.com>
9160 * coretypes.h (scalar_mode): New class.
9161 * machmode.h (scalar_mode): Likewise.
9162 (scalar_mode::includes_p): New function.
9163 (mode_to_inner): Return a scalar_mode rather than a machine_mode.
9164 * gdbhooks.py (build_pretty_printers): Handle scalar_mode.
9165 * genmodes.c (get_mode_class): Handle remaining scalar modes.
9166 * cfgexpand.c (expand_debug_expr): Use scalar_mode.
9167 * expmed.c (store_bit_field_1): Likewise.
9168 (extract_bit_field_1): Likewise.
9169 * expr.c (write_complex_part): Likewise.
9170 (read_complex_part): Likewise.
9171 (emit_move_complex_push): Likewise.
9172 (expand_expr_real_2): Likewise.
9173 * function.c (assign_parm_setup_reg): Likewise.
9174 (assign_parms_unsplit_complex): Likewise.
9175 * optabs.c (expand_binop): Likewise.
9176 * rtlanal.c (subreg_get_info): Likewise.
9177 * simplify-rtx.c (simplify_immed_subreg): Likewise.
9178 * varasm.c (output_constant_pool_2): Likewise.
9180 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9181 Alan Hayward <alan.hayward@arm.com>
9182 David Sherwood <david.sherwood@arm.com>
9184 * expmed.c (extract_high_half): Use scalar_int_mode and remove
9186 (expmed_mult_highpart_optab): Likewise.
9187 (expmed_mult_highpart): Likewise.
9189 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9190 Alan Hayward <alan.hayward@arm.com>
9191 David Sherwood <david.sherwood@arm.com>
9193 * builtins.h (builtin_strncpy_read_str): Take a scalar_int_mode
9194 instead of a machine_mode.
9195 (builtin_memset_read_str): Likewise.
9196 * builtins.c (c_readstr): Likewise.
9197 (builtin_memcpy_read_str): Likewise.
9198 (builtin_strncpy_read_str): Likewise.
9199 (builtin_memset_read_str): Likewise.
9200 (builtin_memset_gen_str): Likewise.
9201 (expand_builtin_signbit): Use scalar_int_mode for local variables.
9202 * cfgexpand.c (convert_debug_memory_address): Take a scalar_int_mode
9203 instead of a machine_mode.
9204 * combine.c (simplify_if_then_else): Use scalar_int_mode for local
9206 (make_extraction): Likewise.
9207 (try_widen_shift_mode): Take and return scalar_int_modes instead
9209 * config/aarch64/aarch64.c (aarch64_libgcc_cmp_return_mode): Return
9210 a scalar_int_mode instead of a machine_mode.
9211 * config/avr/avr.c (avr_addr_space_address_mode): Likewise.
9212 (avr_addr_space_pointer_mode): Likewise.
9213 * config/cr16/cr16.c (cr16_unwind_word_mode): Likewise.
9214 * config/msp430/msp430.c (msp430_addr_space_pointer_mode): Likewise.
9215 (msp430_unwind_word_mode): Likewise.
9216 * config/spu/spu.c (spu_unwind_word_mode): Likewise.
9217 (spu_addr_space_pointer_mode): Likewise.
9218 (spu_addr_space_address_mode): Likewise.
9219 (spu_libgcc_cmp_return_mode): Likewise.
9220 (spu_libgcc_shift_count_mode): Likewise.
9221 * config/rl78/rl78.c (rl78_addr_space_address_mode): Likewise.
9222 (rl78_addr_space_pointer_mode): Likewise.
9223 (fl78_unwind_word_mode): Likewise.
9224 (rl78_valid_pointer_mode): Take a scalar_int_mode instead of a
9226 * config/alpha/alpha.c (vms_valid_pointer_mode): Likewise.
9227 * config/ia64/ia64.c (ia64_vms_valid_pointer_mode): Likewise.
9228 * config/mips/mips.c (mips_mode_rep_extended): Likewise.
9229 (mips_valid_pointer_mode): Likewise.
9230 * config/tilegx/tilegx.c (tilegx_mode_rep_extended): Likewise.
9231 * config/ft32/ft32.c (ft32_valid_pointer_mode): Likewise.
9232 (ft32_addr_space_pointer_mode): Return a scalar_int_mode instead
9234 (ft32_addr_space_address_mode): Likewise.
9235 * config/m32c/m32c.c (m32c_valid_pointer_mode): Take a
9236 scalar_int_mode instead of a machine_mode.
9237 (m32c_addr_space_pointer_mode): Return a scalar_int_mode instead
9239 (m32c_addr_space_address_mode): Likewise.
9240 * config/powerpcspe/powerpcspe.c (rs6000_abi_word_mode): Likewise.
9241 (rs6000_eh_return_filter_mode): Likewise.
9242 * config/rs6000/rs6000.c (rs6000_abi_word_mode): Likewise.
9243 (rs6000_eh_return_filter_mode): Likewise.
9244 * config/s390/s390.c (s390_libgcc_cmp_return_mode): Likewise.
9245 (s390_libgcc_shift_count_mode): Likewise.
9246 (s390_unwind_word_mode): Likewise.
9247 (s390_valid_pointer_mode): Take a scalar_int_mode rather than a
9249 * target.def (mode_rep_extended): Likewise.
9250 (valid_pointer_mode): Likewise.
9251 (addr_space.valid_pointer_mode): Likewise.
9252 (eh_return_filter_mode): Return a scalar_int_mode rather than
9254 (libgcc_cmp_return_mode): Likewise.
9255 (libgcc_shift_count_mode): Likewise.
9256 (unwind_word_mode): Likewise.
9257 (addr_space.pointer_mode): Likewise.
9258 (addr_space.address_mode): Likewise.
9259 * doc/tm.texi: Regenerate.
9260 * dojump.c (prefer_and_bit_test): Take a scalar_int_mode rather than
9262 (do_jump): Use scalar_int_mode for local variables.
9263 * dwarf2cfi.c (init_return_column_size): Take a scalar_int_mode
9264 rather than a machine_mode.
9265 * dwarf2out.c (convert_descriptor_to_mode): Likewise.
9266 (scompare_loc_descriptor_wide): Likewise.
9267 (scompare_loc_descriptor_narrow): Likewise.
9268 * emit-rtl.c (adjust_address_1): Use scalar_int_mode for local
9270 * except.c (sjlj_emit_dispatch_table): Likewise.
9271 (expand_builtin_eh_copy_values): Likewise.
9272 * explow.c (convert_memory_address_addr_space_1): Likewise.
9273 Take a scalar_int_mode rather than a machine_mode.
9274 (convert_memory_address_addr_space): Take a scalar_int_mode rather
9275 than a machine_mode.
9276 (memory_address_addr_space): Use scalar_int_mode for local variables.
9277 * expmed.h (expand_mult_highpart_adjust): Take a scalar_int_mode
9278 rather than a machine_mode.
9279 * expmed.c (mask_rtx): Likewise.
9280 (init_expmed_one_conv): Likewise.
9281 (expand_mult_highpart_adjust): Likewise.
9282 (extract_high_half): Likewise.
9283 (expmed_mult_highpart_optab): Likewise.
9284 (expmed_mult_highpart): Likewise.
9285 (expand_smod_pow2): Likewise.
9286 (expand_sdiv_pow2): Likewise.
9287 (emit_store_flag_int): Likewise.
9288 (adjust_bit_field_mem_for_reg): Use scalar_int_mode for local
9290 (extract_low_bits): Likewise.
9291 * expr.h (by_pieces_constfn): Take a scalar_int_mode rather than
9293 * expr.c (pieces_addr::adjust): Likewise.
9294 (can_store_by_pieces): Likewise.
9295 (store_by_pieces): Likewise.
9296 (clear_by_pieces_1): Likewise.
9297 (expand_expr_addr_expr_1): Likewise.
9298 (expand_expr_addr_expr): Use scalar_int_mode for local variables.
9299 (expand_expr_real_1): Likewise.
9300 (try_casesi): Likewise.
9301 * final.c (shorten_branches): Likewise.
9302 * fold-const.c (fold_convert_const_int_from_fixed): Change the
9303 type of "mode" to machine_mode.
9304 * internal-fn.c (expand_arith_overflow_result_store): Take a
9305 scalar_int_mode rather than a machine_mode.
9306 (expand_mul_overflow): Use scalar_int_mode for local variables.
9307 * loop-doloop.c (doloop_modify): Likewise.
9308 (doloop_optimize): Likewise.
9309 * optabs.c (expand_subword_shift): Take a scalar_int_mode rather
9310 than a machine_mode.
9311 (expand_doubleword_shift_condmove): Likewise.
9312 (expand_doubleword_shift): Likewise.
9313 (expand_doubleword_clz): Likewise.
9314 (expand_doubleword_popcount): Likewise.
9315 (expand_doubleword_parity): Likewise.
9316 (expand_absneg_bit): Use scalar_int_mode for local variables.
9317 (prepare_float_lib_cmp): Likewise.
9318 * rtl.h (convert_memory_address_addr_space_1): Take a scalar_int_mode
9319 rather than a machine_mode.
9320 (convert_memory_address_addr_space): Likewise.
9321 (get_mode_bounds): Likewise.
9322 (get_address_mode): Return a scalar_int_mode rather than a
9324 * rtlanal.c (get_address_mode): Likewise.
9325 * stor-layout.c (get_mode_bounds): Take a scalar_int_mode rather
9326 than a machine_mode.
9327 * targhooks.c (default_mode_rep_extended): Likewise.
9328 (default_valid_pointer_mode): Likewise.
9329 (default_addr_space_valid_pointer_mode): Likewise.
9330 (default_eh_return_filter_mode): Return a scalar_int_mode rather
9331 than a machine_mode.
9332 (default_libgcc_cmp_return_mode): Likewise.
9333 (default_libgcc_shift_count_mode): Likewise.
9334 (default_unwind_word_mode): Likewise.
9335 (default_addr_space_pointer_mode): Likewise.
9336 (default_addr_space_address_mode): Likewise.
9337 * targhooks.h (default_eh_return_filter_mode): Likewise.
9338 (default_libgcc_cmp_return_mode): Likewise.
9339 (default_libgcc_shift_count_mode): Likewise.
9340 (default_unwind_word_mode): Likewise.
9341 (default_addr_space_pointer_mode): Likewise.
9342 (default_addr_space_address_mode): Likewise.
9343 (default_mode_rep_extended): Take a scalar_int_mode rather than
9345 (default_valid_pointer_mode): Likewise.
9346 (default_addr_space_valid_pointer_mode): Likewise.
9347 * tree-ssa-address.c (addr_for_mem_ref): Use scalar_int_mode for
9349 * tree-ssa-loop-ivopts.c (get_shiftadd_cost): Take a scalar_int_mode
9350 rather than a machine_mode.
9351 * tree-switch-conversion.c (array_value_type): Use scalar_int_mode
9352 for local variables.
9353 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9354 * var-tracking.c (use_narrower_mode): Take a scalar_int_mode rather
9355 than a machine_mode.
9357 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9358 Alan Hayward <alan.hayward@arm.com>
9359 David Sherwood <david.sherwood@arm.com>
9361 * dojump.c (do_jump_by_parts_greater_rtx): Change the type of
9362 the mode argument to scalar_int_mode.
9363 (do_jump_by_parts_zero_rtx): Likewise.
9364 (do_jump_by_parts_equality_rtx): Likewise.
9365 (do_jump_by_parts_greater): Take a mode argument.
9366 (do_jump_by_parts_equality): Likewise.
9367 (do_jump_1): Update calls accordingly.
9369 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9370 Alan Hayward <alan.hayward@arm.com>
9371 David Sherwood <david.sherwood@arm.com>
9373 * is-a.h (safe_dyn_cast): New function.
9374 * rtl.h (rtx_jump_table_data::get_data_mode): New function.
9375 (jump_table_for_label): Likewise.
9376 * final.c (final_addr_vec_align): Take an rtx_jump_table_data *
9377 instead of an rtx_insn *.
9378 (shorten_branches): Use dyn_cast instead of LABEL_P and
9379 JUMP_TABLE_DATA_P. Use jump_table_for_label and
9380 rtx_jump_table_data::get_data_mode.
9381 (final_scan_insn): Likewise.
9383 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9384 Alan Hayward <alan.hayward@arm.com>
9385 David Sherwood <david.sherwood@arm.com>
9387 * combine.c (try_combine): Use is_a <scalar_int_mode> when
9388 trying to combine a full-register integer set with a subreg
9391 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9392 Alan Hayward <alan.hayward@arm.com>
9393 David Sherwood <david.sherwood@arm.com>
9395 * expr.c (expand_expr_addr_expr): Add a new_tmode local variable
9396 that is always either address_mode or pointer_mode.
9398 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9399 Alan Hayward <alan.hayward@arm.com>
9400 David Sherwood <david.sherwood@arm.com>
9402 * expr.c (expand_expr_real_2): Use word_mode instead of innermode
9403 when the two are known to be equal.
9405 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9406 Alan Hayward <alan.hayward@arm.com>
9407 David Sherwood <david.sherwood@arm.com>
9409 * simplify-rtx.c (simplify_const_unary_operation): Use
9410 is_a <scalar_int_mode> instead of checking for a nonzero
9411 precision. Forcibly convert op_mode to a scalar_int_mode
9412 in that case. More clearly differentiate the operand and
9413 result modes and use the former when deciding what the value
9414 of a count-bits operation should be. Use is_int_mode instead
9415 of checking for a MODE_INT. Remove redundant check for whether
9416 this mode has a zero precision.
9418 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9419 Alan Hayward <alan.hayward@arm.com>
9420 David Sherwood <david.sherwood@arm.com>
9422 * optabs.c (widen_leading): Change the type of the mode argument
9423 to scalar_int_mode. Use opt_scalar_int_mode for the mode iterator.
9424 (widen_bswap): Likewise.
9425 (expand_parity): Likewise.
9426 (expand_ctz): Change the type of the mode argument to scalar_int_mode.
9427 (expand_ffs): Likewise.
9428 (epand_unop): Check for scalar integer modes before calling the
9431 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9432 Alan Hayward <alan.hayward@arm.com>
9433 David Sherwood <david.sherwood@arm.com>
9435 * expr.c (const_scalar_mask_from_tree): Add a mode argument.
9437 (expand_expr_real_1): Update call accordingly.
9439 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9440 Alan Hayward <alan.hayward@arm.com>
9441 David Sherwood <david.sherwood@arm.com>
9443 * expmed.c (store_bit_field_using_insv): Add op0_mode and
9444 value_mode arguments. Use scalar_int_mode internally.
9445 (store_bit_field_1): Rename the new integer mode from imode
9446 to op0_mode and use it instead of GET_MODE (op0). Update calls
9447 to store_split_bit_field, store_bit_field_using_insv and
9448 store_fixed_bit_field.
9449 (store_fixed_bit_field): Add op0_mode and value_mode arguments.
9450 Use scalar_int_mode internally. Use a bit count rather than a mode
9451 when calculating the largest bit size for get_best_mode.
9452 Update calls to store_split_bit_field and store_fixed_bit_field_1.
9453 (store_fixed_bit_field_1): Add mode and value_mode arguments.
9454 Remove assertion that OP0 has a scalar integer mode.
9455 (store_split_bit_field): Add op0_mode and value_mode arguments.
9456 Update calls to extract_fixed_bit_field.
9457 (extract_bit_field_using_extv): Add an op0_mode argument.
9458 Use scalar_int_mode internally.
9459 (extract_bit_field_1): Rename the new integer mode from imode to
9460 op0_mode and use it instead of GET_MODE (op0). Update calls to
9461 extract_split_bit_field, extract_bit_field_using_extv and
9462 extract_fixed_bit_field.
9463 (extract_fixed_bit_field): Add an op0_mode argument. Update calls
9464 to extract_split_bit_field and extract_fixed_bit_field_1.
9465 (extract_fixed_bit_field_1): Add a mode argument. Remove assertion
9466 that OP0 has a scalar integer mode. Use as_a <scalar_int_mode>
9468 (extract_split_bit_field): Add an op0_mode argument. Update call
9469 to extract_fixed_bit_field.
9471 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9472 Alan Hayward <alan.hayward@arm.com>
9473 David Sherwood <david.sherwood@arm.com>
9475 * cse.c (cse_insn): Use opt_scalar_int_mode for the mode iterator.
9476 * explow.c (hard_function_value): Likewise.
9477 * expmed.c (extract_fixed_bit_field_1): Likewise. Move the
9478 convert_to_mode call outside the loop.
9479 * expr.c (alignment_for_piecewise_move): Use opt_scalar_int_mode
9480 for the mode iterator. Require the mode specified by max_pieces
9482 (emit_block_move_via_movmem): Use opt_scalar_int_mode for the
9484 (copy_blkmode_to_reg): Likewise.
9485 (set_storage_via_setmem): Likewise.
9486 * optabs.c (prepare_cmp_insn): Likewise.
9487 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
9488 * stor-layout.c (finish_bitfield_representative): Likewise.
9490 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9491 Alan Hayward <alan.hayward@arm.com>
9492 David Sherwood <david.sherwood@arm.com>
9494 * rtl.h (subreg_unpromoted_mode, subreg_promoted_mode): New functions.
9495 * expr.c (convert_move): Use them.
9496 (convert_modes): Likewise.
9497 (store_expr_with_bounds): Likewise.
9499 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9500 Alan Hayward <alan.hayward@arm.com>
9501 David Sherwood <david.sherwood@arm.com>
9503 * rtl.h (rtl_hooks::reg_nonzero_bits): Add a scalar_int_mode
9504 parameter for the mode of "x". Remove the "known_x", "known_mode"
9505 and "known_ret" arguments. Change the type of the mode argument
9507 (rtl_hooks:reg_num_sign_bit_copies): Likewise.
9508 * combine.c (reg_nonzero_bits_for_combine): Update accordingly.
9509 (reg_num_sign_bit_copies_for_combine): Likewise.
9510 * rtlanal.c (nonzero_bits1): Likewise.
9511 (num_sign_bit_copies1): Likewise.
9512 * rtlhooks-def.h (reg_nonzero_bits_general): Likewise.
9513 (reg_num_sign_bit_copies_general): Likewise.
9514 * rtlhooks.c (reg_num_sign_bit_copies_general): Likewise.
9515 (reg_nonzero_bits_general): Likewise.
9517 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9518 Alan Hayward <alan.hayward@arm.com>
9519 David Sherwood <david.sherwood@arm.com>
9521 * rtlanal.c (num_sign_bit_copies): Handle VOIDmode here rather
9522 than in subroutines. Return 1 for non-integer modes.
9523 (cached_num_sign_bit_copies): Change the type of the mode parameter
9525 (num_sign_bit_copies1): Likewise. Remove early exit for other mode
9526 classes. Handle CONST_INT_P first and then check whether X also
9527 has a scalar integer mode. Check the same thing for inner registers
9528 of a SUBREG and for values that are being extended or truncated.
9530 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9531 Alan Hayward <alan.hayward@arm.com>
9532 David Sherwood <david.sherwood@arm.com>
9534 * rtlanal.c (nonzero_bits): Handle VOIDmode here rather than
9535 in subroutines. Return the mode mask for non-integer modes.
9536 (cached_nonzero_bits): Change the type of the mode parameter
9538 (nonzero_bits1): Likewise. Remove early exit for other mode
9539 classes. Handle CONST_INT_P first and then check whether X
9540 also has a scalar integer mode.
9542 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9543 Alan Hayward <alan.hayward@arm.com>
9544 David Sherwood <david.sherwood@arm.com>
9546 * expr.c (widest_int_mode_for_size): Make the comment match the code.
9547 Return a scalar_int_mode and assert that the size is greater than
9549 (by_pieces_ninsns): Update accordingly and remove VOIDmode handling.
9550 (op_by_pieces_d::op_by_pieces_d): Likewise.
9551 (op_by_pieces_d::run): Likewise.
9552 (can_store_by_pieces): Likewise.
9554 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9555 Alan Hayward <alan.hayward@arm.com>
9556 David Sherwood <david.sherwood@arm.com>
9558 * combine.c (extract_left_shift): Add a mode argument and update
9560 (make_compound_operation_int): Change the type of the mode parameter
9561 to scalar_int_mode and update the call to extract_left_shift.
9563 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9564 Alan Hayward <alan.hayward@arm.com>
9565 David Sherwood <david.sherwood@arm.com>
9567 * combine.c (simplify_and_const_int): Change the type of the mode
9568 parameter to scalar_int_mode.
9569 (simplify_and_const_int_1): Likewise. Update recursive call.
9571 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9572 Alan Hayward <alan.hayward@arm.com>
9573 David Sherwood <david.sherwood@arm.com>
9575 * combine.c (simplify_compare_const): Check that the mode is a
9576 scalar_int_mode (rather than VOIDmode) before testing its
9578 (simplify_comparison): Move COMPARISON_P handling out of the
9579 loop and restrict the latter part of the loop to scalar_int_modes.
9580 Check is_a <scalar_int_mode> before calling HWI_COMPUTABLE_MODE_P
9581 and when considering SUBREG_REGs. Use is_int_mode instead of
9582 checking GET_MODE_CLASS against MODE_INT.
9584 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9585 Alan Hayward <alan.hayward@arm.com>
9586 David Sherwood <david.sherwood@arm.com>
9588 * combine.c (try_widen_shift_mode): Move check for equal modes to...
9589 (simplify_shift_const_1): ...here. Use scalar_int_mode for
9590 shift_unit_mode and for modes involved in scalar shifts.
9592 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9593 Alan Hayward <alan.hayward@arm.com>
9594 David Sherwood <david.sherwood@arm.com>
9596 * combine.c (force_int_to_mode): New function, split out from...
9597 (force_to_mode): ...here. Keep xmode up-to-date and use it
9598 instead of GET_MODE (x).
9600 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9601 Alan Hayward <alan.hayward@arm.com>
9602 David Sherwood <david.sherwood@arm.com>
9604 * optabs-query.h (extraction_insn::struct_mode): Change type to
9605 opt_scalar_int_mode and update comment.
9606 (extraction_insn::field_mode): Change type to scalar_int_mode.
9607 (extraction_insn::pos_mode): Likewise.
9608 * combine.c (make_extraction): Update accordingly.
9609 * optabs-query.c (get_traditional_extraction_insn): Likewise.
9610 (get_optab_extraction_insn): Likewise.
9611 * recog.c (simplify_while_replacing): Likewise.
9612 * expmed.c (narrow_bit_field_mem): Change the type of the mode
9613 parameter to opt_scalar_int_mode.
9615 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9616 Alan Hayward <alan.hayward@arm.com>
9617 David Sherwood <david.sherwood@arm.com>
9619 * machmode.h (bit_field_mode_iterator::next_mode): Take a pointer
9620 to a scalar_int_mode instead of a machine_mode.
9621 (bit_field_mode_iterator::m_mode): Change type to opt_scalar_int_mode.
9622 (get_best_mode): Return a boolean and use a pointer argument to store
9623 the selected mode. Replace the limit mode parameter with a bit limit.
9624 * expmed.c (adjust_bit_field_mem_for_reg): Use scalar_int_mode
9625 for the values returned by bit_field_mode_iterator::next_mode.
9626 (store_bit_field): Update call to get_best_mode.
9627 (store_fixed_bit_field): Likewise.
9628 (extract_fixed_bit_field): Likewise.
9629 * expr.c (optimize_bitfield_assignment_op): Likewise.
9630 * fold-const.c (optimize_bit_field_compare): Likewise.
9631 (fold_truth_andor_1): Likewise.
9632 * stor-layout.c (bit_field_mode_iterator::next_mode): As above.
9633 Update for new type of m_mode.
9634 (get_best_mode): As above.
9636 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9637 Alan Hayward <alan.hayward@arm.com>
9638 David Sherwood <david.sherwood@arm.com>
9640 * expmed.c (strict_volatile_bitfield_p): Change the type of fieldmode
9641 to scalar_int_mode. Remove check for SCALAR_INT_MODE_P.
9642 (store_bit_field): Check is_a <scalar_int_mode> before calling
9643 strict_volatile_bitfield_p.
9644 (extract_bit_field): Likewise.
9646 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9647 Alan Hayward <alan.hayward@arm.com>
9648 David Sherwood <david.sherwood@arm.com>
9650 * target.def (cstore_mode): Return a scalar_int_mode.
9651 * doc/tm.texi: Regenerate.
9652 * config/sparc/sparc.c (sparc_cstore_mode): Return a scalar_int_mode.
9653 * targhooks.h (default_cstore_mode): Likewise.
9654 * targhooks.c (default_cstore_mode): Likewise, using a forced
9656 * expmed.c (emit_cstore): Expect the target of the cstore to be
9659 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9660 Alan Hayward <alan.hayward@arm.com>
9661 David Sherwood <david.sherwood@arm.com>
9663 * cfgloop.h (rtx_iv): Change type of extend_mode and mode to
9665 (niter_desc): Likewise mode.
9666 (iv_analyze): Add a mode parameter.
9668 (iv_analyze_expr): Pass the mode paraeter before the rtx it describes
9669 and change its type to scalar_int_mode.
9670 * loop-iv.c: Update commentary at head of file.
9671 (iv_constant): Pass the mode paraeter before the rtx it describes
9672 and change its type to scalar_int_mode. Remove VOIDmode handling.
9673 (iv_subreg): Change the type of the mode parameter to scalar_int_mode.
9674 (iv_extend): Likewise.
9675 (shorten_into_mode): Likewise.
9676 (iv_add): Use scalar_int_mode.
9677 (iv_mult): Likewise.
9678 (iv_shift): Likewise.
9679 (canonicalize_iv_subregs): Likewise.
9680 (get_biv_step_1): Pass the outer_mode parameter before the rtx
9681 it describes and change its mode to scalar_int_mode. Also change
9682 the type of the returned inner_mode to scalar_int_mode.
9683 (get_biv_step): Likewise, turning outer_mode from a pointer
9684 into a direct parameter. Update call to get_biv_step_1.
9685 (iv_analyze_biv): Add an outer_mode parameter. Update calls to
9686 iv_constant and get_biv_step.
9687 (iv_analyze_expr): Pass the mode parameter before the rtx it describes
9688 and change its type to scalar_int_mode. Don't initialise iv->mode
9689 to VOIDmode and remove later checks for its still being VOIDmode.
9690 Update calls to iv_analyze_op and iv_analyze_expr. Check
9691 is_a <scalar_int_mode> when changing the mode under consideration.
9692 (iv_analyze_def): Ignore registers that don't have a scalar_int_mode.
9693 Update call to iv_analyze_expr.
9694 (iv_analyze_op): Add a mode parameter. Reject subregs whose
9695 inner register is not also a scalar_int_mode. Update call to
9697 (iv_analyze): Add a mode parameter. Update call to iv_analyze_op.
9698 (biv_p): Add a mode parameter. Update call to iv_analyze_biv.
9699 (iv_number_of_iterations): Use is_a <scalar_int_mode> instead of
9700 separate mode class checks. Update calls to iv_analyze. Remove
9701 fix-up of VOIDmodes after iv_analyze_biv.
9702 * loop-unroll.c (analyze_iv_to_split_insn): Reject registers that
9703 don't have a scalar_int_mode. Update call to biv_p.
9705 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9706 Alan Hayward <alan.hayward@arm.com>
9707 David Sherwood <david.sherwood@arm.com>
9709 * cfgexpand.c (convert_debug_memory_address): Use
9710 as_a <scalar_int_mode>.
9711 * combine.c (expand_compound_operation): Likewise.
9712 (make_extraction): Likewise.
9713 (change_zero_ext): Likewise.
9714 (simplify_comparison): Likewise.
9715 * cse.c (cse_insn): Likewise.
9716 * dwarf2out.c (minmax_loc_descriptor): Likewise.
9717 (mem_loc_descriptor): Likewise.
9718 (loc_descriptor): Likewise.
9719 * expmed.c (init_expmed_one_mode): Likewise.
9720 (synth_mult): Likewise.
9721 (emit_store_flag_1): Likewise.
9722 (expand_divmod): Likewise. Use HWI_COMPUTABLE_MODE_P instead
9723 of a comparison with size.
9724 * expr.c (expand_assignment): Use as_a <scalar_int_mode>.
9725 (reduce_to_bit_field_precision): Likewise.
9726 * function.c (expand_function_end): Likewise.
9727 * internal-fn.c (expand_arith_overflow_result_store): Likewise.
9728 * loop-doloop.c (doloop_modify): Likewise.
9729 * optabs.c (expand_binop): Likewise.
9730 (expand_unop): Likewise.
9731 (expand_copysign_absneg): Likewise.
9732 (prepare_cmp_insn): Likewise.
9733 (maybe_legitimize_operand): Likewise.
9734 * recog.c (const_scalar_int_operand): Likewise.
9735 * rtlanal.c (get_address_mode): Likewise.
9736 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9737 (simplify_cond_clz_ctz): Likewise.
9738 * tree-nested.c (get_nl_goto_field): Likewise.
9739 * tree.c (build_vector_type_for_mode): Likewise.
9740 * var-tracking.c (use_narrower_mode): Likewise.
9742 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9743 Alan Hayward <alan.hayward@arm.com>
9744 David Sherwood <david.sherwood@arm.com>
9746 * tree.h (SCALAR_INT_TYPE_MODE): New macro.
9747 * builtins.c (expand_builtin_signbit): Use it.
9748 * cfgexpand.c (expand_debug_expr): Likewise.
9749 * dojump.c (do_jump): Likewise.
9750 (do_compare_and_jump): Likewise.
9751 * dwarf2cfi.c (expand_builtin_init_dwarf_reg_sizes): Likewise.
9752 * expmed.c (make_tree): Likewise.
9753 * expr.c (expand_expr_real_2): Likewise.
9754 (expand_expr_real_1): Likewise.
9755 (try_casesi): Likewise.
9756 * fold-const-call.c (fold_const_call_ss): Likewise.
9757 * fold-const.c (unextend): Likewise.
9758 (extract_muldiv_1): Likewise.
9759 (fold_single_bit_test): Likewise.
9760 (native_encode_int): Likewise.
9761 (native_encode_string): Likewise.
9762 (native_interpret_int): Likewise.
9763 * gimple-fold.c (gimple_fold_builtin_memset): Likewise.
9764 * internal-fn.c (expand_addsub_overflow): Likewise.
9765 (expand_neg_overflow): Likewise.
9766 (expand_mul_overflow): Likewise.
9767 (expand_arith_overflow): Likewise.
9768 * match.pd: Likewise.
9769 * stor-layout.c (layout_type): Likewise.
9770 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
9771 * tree-ssa-math-opts.c (convert_mult_to_widen): Likewise.
9772 * tree-ssanames.c (get_range_info): Likewise.
9773 * tree-switch-conversion.c (array_value_type) Likewise.
9774 * tree-vect-patterns.c (vect_recog_rotate_pattern): Likewise.
9775 (vect_recog_divmod_pattern): Likewise.
9776 (vect_recog_mixed_size_cond_pattern): Likewise.
9777 * tree-vrp.c (extract_range_basic): Likewise.
9778 (simplify_float_conversion_using_ranges): Likewise.
9779 * tree.c (int_fits_type_p): Likewise.
9780 * ubsan.c (instrument_bool_enum_load): Likewise.
9781 * varasm.c (mergeable_string_section): Likewise.
9782 (narrowing_initializer_constant_valid_p): Likewise.
9783 (output_constant): Likewise.
9785 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9786 Alan Hayward <alan.hayward@arm.com>
9787 David Sherwood <david.sherwood@arm.com>
9789 * machmode.h (NARROWEST_INT_MODE): New macro.
9790 * expr.c (alignment_for_piecewise_move): Use it instead of
9791 GET_CLASS_NARROWEST_MODE (MODE_INT).
9792 (push_block): Likewise.
9793 * stor-layout.c (bit_field_mode_iterator::bit_field_mode_iterator):
9795 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
9797 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9798 Alan Hayward <alan.hayward@arm.com>
9799 David Sherwood <david.sherwood@arm.com>
9801 * postreload.c (move2add_valid_value_p): Change the type of the
9802 mode parameter to scalar_int_mode.
9803 (move2add_use_add2_insn): Add a mode parameter and use it instead
9805 (move2add_use_add3_insn): Likewise.
9806 (reload_cse_move2add): Update accordingly.
9808 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9809 Alan Hayward <alan.hayward@arm.com>
9810 David Sherwood <david.sherwood@arm.com>
9812 * expr.c (expand_expr_real_2): Use scalar_int_mode for the
9814 * lower-subreg.c (resolve_shift_zext): Use is_a <scalar_int_mode>.
9815 * optabs.c (expand_unop): Likewise.
9817 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9818 Alan Hayward <alan.hayward@arm.com>
9819 David Sherwood <david.sherwood@arm.com>
9821 * dwarf2out.c (typed_binop): Change mode parameter to scalar_int_mode.
9822 (clz_loc_descriptor): Likewise. Remove SCALAR_INT_MODE_P check.
9823 (popcount_loc_descriptor): Likewise.
9824 (bswap_loc_descriptor): Likewise.
9825 (rotate_loc_descriptor): Likewise.
9826 (mem_loc_descriptor): Add is_a <scalar_int_mode> checks before
9827 calling the functions above.
9829 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9830 Alan Hayward <alan.hayward@arm.com>
9831 David Sherwood <david.sherwood@arm.com>
9833 * combine.c (sign_extend_short_imm): Add is_a <scalar_int_mode>
9835 (try_combine): Likewise.
9836 (simplify_if_then_else): Likewise.
9837 * cse.c (cse_insn): Likewise.
9838 * dwarf2out.c (mem_loc_descriptor): Likewise.
9839 * emit-rtl.c (gen_lowpart_common): Likewise.
9840 * simplify-rtx.c (simplify_truncation): Likewise.
9841 (simplify_binary_operation_1): Likewise.
9842 (simplify_const_relational_operation): Likewise.
9843 (simplify_ternary_operation): Likewise.
9844 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Likewise.
9846 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9847 Alan Hayward <alan.hayward@arm.com>
9848 David Sherwood <david.sherwood@arm.com>
9850 * cse.c (cse_insn): Add is_a <scalar_int_mode> checks.
9851 * reload.c (push_reload): Likewise.
9852 (find_reloads): Likewise.
9854 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9855 Alan Hayward <alan.hayward@arm.com>
9856 David Sherwood <david.sherwood@arm.com>
9858 * combine.c (find_split_point): Add is_a <scalar_int_mode> checks.
9859 (make_compound_operation_int): Likewise.
9860 (change_zero_ext): Likewise.
9861 * expr.c (convert_move): Likewise.
9862 (convert_modes): Likewise.
9863 * fwprop.c (forward_propagate_subreg): Likewise.
9864 * loop-iv.c (get_biv_step_1): Likewise.
9865 * optabs.c (widen_operand): Likewise.
9866 * postreload.c (move2add_valid_value_p): Likewise.
9867 * recog.c (simplify_while_replacing): Likewise.
9868 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9869 (simplify_binary_operation_1): Likewise. Remove redundant
9870 mode equality check.
9872 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9873 Alan Hayward <alan.hayward@arm.com>
9874 David Sherwood <david.sherwood@arm.com>
9876 * combine.c (combine_simplify_rtx): Add checks for
9877 is_a <scalar_int_mode>.
9878 (simplify_if_then_else): Likewise.
9879 (make_field_assignment): Likewise.
9880 (simplify_comparison): Likewise.
9881 * ifcvt.c (noce_try_bitop): Likewise.
9882 * loop-invariant.c (canonicalize_address_mult): Likewise.
9883 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
9885 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9886 Alan Hayward <alan.hayward@arm.com>
9887 David Sherwood <david.sherwood@arm.com>
9889 * gimple-fold.c (gimple_fold_builtin_memory_op): Use
9890 is_a <scalar_int_mode> instead of != BLKmode.
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 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>
9897 instead of != VOIDmode.
9898 * combine.c (if_then_else_cond): Likewise.
9899 (change_zero_ext): Likewise.
9900 * dwarf2out.c (mem_loc_descriptor): Likewise.
9901 (loc_descriptor): Likewise.
9902 * rtlanal.c (canonicalize_condition): Likewise.
9903 * simplify-rtx.c (simplify_relational_operation_1): Likewise.
9905 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9906 Alan Hayward <alan.hayward@arm.com>
9907 David Sherwood <david.sherwood@arm.com>
9909 * simplify-rtx.c (simplify_binary_operation_1): Use
9910 is_a <scalar_int_mode> instead of !VECTOR_MODE_P.
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 * wide-int.h (int_traits<unsigned char>) New class.
9917 (int_traits<unsigned short>) Likewise.
9918 * cfgexpand.c (expand_debug_expr): Use is_a <scalar_int_mode>.
9919 Use GET_MODE_UNIT_PRECISION and remove redundant test for
9921 * combine.c (set_nonzero_bits_and_sign_copies): Use
9922 is_a <scalar_int_mode>.
9923 (find_split_point): Likewise.
9924 (combine_simplify_rtx): Likewise.
9925 (simplify_logical): Likewise.
9926 (expand_compound_operation): Likewise.
9927 (expand_field_assignment): Likewise.
9928 (make_compound_operation): Likewise.
9929 (extended_count): Likewise.
9930 (change_zero_ext): Likewise.
9931 (simplify_comparison): Likewise.
9932 * dwarf2out.c (scompare_loc_descriptor): Likewise.
9933 (ucompare_loc_descriptor): Likewise.
9934 (minmax_loc_descriptor): Likewise.
9935 (mem_loc_descriptor): Likewise.
9936 (loc_descriptor): Likewise.
9937 * expmed.c (init_expmed_one_mode): Likewise.
9938 * lra-constraints.c (lra_constraint_offset): Likewise.
9939 * optabs.c (prepare_libcall_arg): Likewise.
9940 * postreload.c (move2add_note_store): Likewise.
9941 * reload.c (operands_match_p): Likewise.
9942 * rtl.h (load_extend_op): Likewise.
9943 * rtlhooks.c (gen_lowpart_general): Likewise.
9944 * simplify-rtx.c (simplify_truncation): Likewise.
9945 (simplify_unary_operation_1): Likewise.
9946 (simplify_binary_operation_1): Likewise.
9947 (simplify_const_binary_operation): Likewise.
9948 (simplify_const_relational_operation): Likewise.
9949 (simplify_subreg): Likewise.
9950 * stor-layout.c (bitwise_mode_for_mode): Likewise.
9951 * var-tracking.c (adjust_mems): Likewise.
9952 (prepare_call_arguments): Likewise.
9954 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9955 Alan Hayward <alan.hayward@arm.com>
9956 David Sherwood <david.sherwood@arm.com>
9958 * machmode.h (is_int_mode): New fuction.
9959 * combine.c (find_split_point): Use it.
9960 (combine_simplify_rtx): Likewise.
9961 (simplify_if_then_else): Likewise.
9962 (simplify_set): Likewise.
9963 (simplify_shift_const_1): Likewise.
9964 (simplify_comparison): Likewise.
9965 * config/aarch64/aarch64.c (aarch64_rtx_costs): Likewise.
9966 * cse.c (notreg_cost): Likewise.
9967 (cse_insn): Likewise.
9968 * cselib.c (cselib_lookup_1): Likewise.
9969 * dojump.c (do_jump_1): Likewise.
9970 (do_compare_rtx_and_jump): Likewise.
9971 * dse.c (get_call_args): Likewise.
9972 * dwarf2out.c (rtl_for_decl_init): Likewise.
9973 (native_encode_initializer): Likewise.
9974 * expmed.c (emit_store_flag_1): Likewise.
9975 (emit_store_flag): Likewise.
9976 * expr.c (convert_modes): Likewise.
9977 (store_field): Likewise.
9978 (expand_expr_real_1): Likewise.
9979 * fold-const.c (fold_read_from_constant_string): Likewise.
9980 * gimple-ssa-sprintf.c (get_format_string): Likewise.
9981 * optabs-libfuncs.c (gen_int_libfunc): Likewise.
9982 * optabs.c (expand_binop): Likewise.
9983 (expand_unop): Likewise.
9984 (expand_abs_nojump): Likewise.
9985 (expand_one_cmpl_abs_nojump): Likewise.
9986 * simplify-rtx.c (mode_signbit_p): Likewise.
9987 (val_signbit_p): Likewise.
9988 (val_signbit_known_set_p): Likewise.
9989 (val_signbit_known_clear_p): Likewise.
9990 (simplify_relational_operation_1): Likewise.
9991 * tree.c (vector_type_mode): Likewise.
9993 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
9994 Alan Hayward <alan.hayward@arm.com>
9995 David Sherwood <david.sherwood@arm.com>
9997 * machmode.h (smallest_mode_for_size): Fix formatting.
9998 (smallest_int_mode_for_size): New function.
9999 * cfgexpand.c (expand_debug_expr): Use smallest_int_mode_for_size
10000 instead of smallest_mode_for_size.
10001 * combine.c (make_extraction): Likewise.
10002 * config/arc/arc.c (arc_expand_movmem): Likewise.
10003 * config/arm/arm.c (arm_expand_divmod_libfunc): Likewise.
10004 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
10005 * config/s390/s390.c (s390_expand_insv): Likewise.
10006 * config/sparc/sparc.c (assign_int_registers): Likewise.
10007 * config/spu/spu.c (spu_function_value): Likewise.
10008 (spu_function_arg): Likewise.
10009 * coverage.c (get_gcov_type): Likewise.
10010 (get_gcov_unsigned_t): Likewise.
10011 * dse.c (find_shift_sequence): Likewise.
10012 * expmed.c (store_bit_field_1): Likewise.
10013 * expr.c (convert_move): Likewise.
10014 (store_field): Likewise.
10015 * internal-fn.c (expand_arith_overflow): Likewise.
10016 * optabs-query.c (get_best_extraction_insn): Likewise.
10017 * optabs.c (expand_twoval_binop_libfunc): Likewise.
10018 * stor-layout.c (layout_type): Likewise.
10019 (initialize_sizetypes): Likewise.
10020 * targhooks.c (default_get_mask_mode): Likewise.
10021 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Likewise.
10023 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10024 Alan Hayward <alan.hayward@arm.com>
10025 David Sherwood <david.sherwood@arm.com>
10027 * machmode.h (opt_mode::else_blk): New function.
10028 (int_mode_for_mode): Declare.
10029 * stor-layout.c (int_mode_for_mode): Return an opt_scalar_int_mode.
10030 * builtins.c (expand_builtin_signbit): Adjust for new int_mode_for_mode
10032 * cfgexpand.c (expand_debug_expr): Likewise.
10033 * combine.c (gen_lowpart_or_truncate): Likewise.
10034 (gen_lowpart_for_combine): Likewise.
10035 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Likewise.
10036 * config/avr/avr.c (avr_to_int_mode): Likewise.
10037 (avr_out_plus_1): Likewise.
10038 (avr_out_plus): Likewise.
10039 (avr_out_round): Likewise.
10040 * config/i386/i386.c (ix86_split_to_parts): Likewise.
10041 * config/s390/s390.c (s390_expand_vec_compare_cc): Likewise.
10042 (s390_expand_vcond): Likewise.
10043 * config/spu/spu.c (spu_split_immediate): Likewise.
10044 (spu_expand_mov): Likewise.
10045 * dse.c (get_stored_val): Likewise.
10046 * expmed.c (store_bit_field_1): Likewise.
10047 (convert_extracted_bit_field): Use int_mode_for_mode instead of
10049 (extract_bit_field_1): Adjust for new int_mode_for_mode return type.
10050 (extract_low_bits): Likewise.
10051 * expr.c (emit_group_load_1): Likewise. Separate out the BLKmode
10052 handling rather than repeating the check.
10053 (emit_group_store): Likewise.
10054 (emit_move_via_integer): Adjust for new int_mode_for_mode return type.
10055 * optabs.c (expand_absneg_bit): Likewise.
10056 (expand_copysign_absneg): Likewise.
10057 (expand_copysign_bit): Likewise.
10058 * tree-if-conv.c (ifcvt_can_use_mask_load_store): Likewise.
10059 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
10060 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
10061 * var-tracking.c (prepare_call_arguments): Likewise.
10062 * config/powerpcspe/powerpcspe.c (rs6000_do_expand_vec_perm): Use
10063 int_mode_for_mode instead of mode_for_size.
10064 * config/rs6000/rs6000.c (rs6000_do_expand_vec_perm): Likewise.
10066 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10067 Alan Hayward <alan.hayward@arm.com>
10068 David Sherwood <david.sherwood@arm.com>
10070 * machmode.h (int_mode_for_size): New function.
10071 * builtins.c (set_builtin_user_assembler_name): Use int_mode_for_size
10072 instead of mode_for_size.
10073 * calls.c (save_fixed_argument_area): Likewise. Make use of BLKmode
10075 * combine.c (expand_field_assignment): Use int_mode_for_size
10076 instead of mode_for_size.
10077 (make_extraction): Likewise.
10078 (simplify_shift_const_1): Likewise.
10079 (simplify_comparison): Likewise.
10080 * dojump.c (do_jump): Likewise.
10081 * dwarf2out.c (mem_loc_descriptor): Likewise.
10082 * emit-rtl.c (init_derived_machine_modes): Likewise.
10083 * expmed.c (flip_storage_order): Likewise.
10084 (convert_extracted_bit_field): Likewise.
10085 * expr.c (copy_blkmode_from_reg): Likewise.
10086 * graphite-isl-ast-to-gimple.c (max_mode_int_precision): Likewise.
10087 * internal-fn.c (expand_mul_overflow): Likewise.
10088 * lower-subreg.c (simple_move): Likewise.
10089 * optabs-libfuncs.c (init_optabs): Likewise.
10090 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
10091 * tree.c (vector_type_mode): Likewise.
10092 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
10093 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Likewise.
10094 * tree-vect-generic.c (expand_vector_parallel): Likewise.
10095 * tree-vect-stmts.c (vectorizable_load): Likewise.
10096 (vectorizable_store): Likewise.
10098 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10099 Alan Hayward <alan.hayward@arm.com>
10100 David Sherwood <david.sherwood@arm.com>
10102 * coretypes.h (pod_mode): New type.
10103 (scalar_int_mode_pod): New typedef.
10104 * machmode.h (pod_mode): New class.
10105 (int_n_data_t::m): Change type to scalar_int_mode_pod.
10106 * genmodes.c (emit_mode_int_n): Update accordingly.
10107 * lower-subreg.h (target_lower_subreg): Change type to
10108 scalar_int_mode_pod.
10109 * gdbhooks.py (build_pretty_printer): Handle pod_mode and
10110 scalar_int_mode_pod.
10112 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10113 Alan Hayward <alan.hayward@arm.com>
10114 David Sherwood <david.sherwood@arm.com>
10116 * config/powerpcspe/powerpcspe.h (rs6000_pmode): Change type from
10117 machine_mode to scalar_int_mode.
10118 * config/powerpcspe/powerpcspe.c (rs6000_pmode): Likewise.
10119 (rs6000_option_override_internal): Remove cast to int.
10120 * config/rs6000/rs6000.h (rs6000_pmode): Change type from
10121 machine_mode to scalar_int_mode.
10122 * config/rs6000/rs6000.c (rs6000_pmode): Likewise.
10123 (rs6000_option_override_internal): Remove cast to int.
10124 * config/s390/s390.h (Pmode): Remove cast to machine_mode.
10125 * config/epiphany/epiphany.h (RTX_OK_FOR_OFFSET_P): Add cast
10127 * config/s390/s390.c (s390_expand_builtin): Likewise.
10128 * coretypes.h (scalar_int_mode): New type.
10129 (opt_scalar_int_mode): New typedef.
10130 * machmode.h (scalar_int_mode): New class.
10131 (scalar_int_mode::includes_p): New function.
10132 (byte_mode): Change type to scalar_int_mode.
10133 (word_mode): Likewise.
10134 (ptr_mode): Likewise.
10135 * emit-rtl.c (byte_mode): Likewise.
10136 (word_mode): Likewise.
10137 (ptr_mode): Likewise.
10138 (init_derived_machine_modes): Update accordingly.
10139 * genmodes.c (get_mode_class): Return scalar_int_mode for MODE_INT
10140 and MODE_PARTIAL_INT.
10141 * gdbhooks.py (build_pretty_printer): Handle scalar_int_mode and
10142 opt_scalar_int_mode.
10144 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10145 Alan Hayward <alan.hayward@arm.com>
10146 David Sherwood <david.sherwood@arm.com>
10148 * target.def (libgcc_floating_mode_supported_p): Take a
10150 * doc/tm.texi: Regenerate.
10151 * targhooks.h (default_libgcc_floating_mode_supported_p): Take a
10153 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10154 * config/aarch64/aarch64.c (aarch64_libgcc_floating_mode_supported_p):
10157 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10158 Alan Hayward <alan.hayward@arm.com>
10159 David Sherwood <david.sherwood@arm.com>
10161 * target.def (default_floatn_mode): Return an opt_scalar_float_mode.
10162 * doc/tm.texi: Regenerate.
10163 * config/arm/arm.c (arm_floatn_mode): Return an opt_scalar_float_mode.
10164 * config/powerpcspe/powerpcspe.c (rs6000_floatn_mode): Likewise.
10165 * config/rs6000/rs6000.c (rs6000_floatn_mode): Likewise.
10166 * targhooks.h (default_floatn_mode): Likewise.
10167 * targhooks.c (default_floatn_mode): Likewise.
10168 * tree.c (build_common_tree_nodes): Update accordingly.
10170 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10171 Alan Hayward <alan.hayward@arm.com>
10172 David Sherwood <david.sherwood@arm.com>
10174 * machmode.h (mode_iterator::start): Provide overload for opt_modes.
10175 (mode_iterator::iterate_p): Likewise.
10176 (mode_iterator::get_wider): Likewise.
10177 * expr.c (init_expr_target): Use opt_scalar_float_mode.
10179 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10180 Alan Hayward <alan.hayward@arm.com>
10181 David Sherwood <david.sherwood@arm.com>
10183 * coretypes.h (opt_scalar_float_mode): New typedef.
10184 * machmode.h (float_mode_for_size): New function.
10185 * emit-rtl.c (double_mode): Delete.
10186 (init_emit_once): Use float_mode_for_size.
10187 * stor-layout.c (layout_type): Likewise.
10188 * gdbhooks.py (build_pretty_printer): Handle opt_scalar_float_mode.
10190 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10191 Alan Hayward <alan.hayward@arm.com>
10192 David Sherwood <david.sherwood@arm.com>
10194 * output.h (assemble_real): Take a scalar_float_mode.
10195 * config/arm/arm.c (arm_assemble_integer): Update accordingly.
10196 * config/arm/arm.md (consttable_4): Likewise.
10197 (consttable_8): Likewise.
10198 (consttable_16): Likewise.
10199 * config/mips/mips.md (consttable_float): Likewise.
10200 * config/s390/s390.c (s390_output_pool_entry): Likewise.
10201 * varasm.c (assemble_real): Take a scalar_float_mode.
10202 (output_constant_pool_2): Update accordingly.
10203 (output_constant): Likewise.
10205 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10206 Alan Hayward <alan.hayward@arm.com>
10207 David Sherwood <david.sherwood@arm.com>
10209 * tree.h (SCALAR_FLOAT_TYPE_MODE): New macro.
10210 * builtins.c (expand_builtin_signbit): Use it instead of TYPE_MODE.
10211 * fold-const.c (fold_convert_const_real_from_fixed): Likewise.
10212 (native_encode_real): Likewise.
10213 (native_interpret_real): Likewise.
10214 * hsa-brig.c (emit_immediate_scalar_to_buffer): Likewise.
10215 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10217 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10218 Alan Hayward <alan.hayward@arm.com>
10219 David Sherwood <david.sherwood@arm.com>
10221 * optabs-libfuncs.c (gen_trunc_conv_libfunc): Use is_a
10222 <scalar_float_mode>. Simplify.
10223 (gen_extend_conv_libfunc): Likewise.
10225 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10226 Alan Hayward <alan.hayward@arm.com>
10227 David Sherwood <david.sherwood@arm.com>
10229 * coretypes.h (scalar_float_mode): New type.
10230 * machmode.h (mode_traits::from_int): Use machine_mode if
10231 USE_ENUM_MODES is defined.
10232 (is_a): New function.
10234 (dyn_cast): Likewise.
10235 (scalar_float_mode): New class.
10236 (scalar_float_mode::includes_p): New function.
10237 (is_float_mode): Likewise.
10238 * gdbhooks.py (MachineModePrinter): New class.
10239 (build_pretty_printer): Use it for scalar_float_mode.
10240 * real.h (FLOAT_MODE_FORMAT): Use as_a <scalar_float_mode>.
10241 (format_helper::format_helper): Turn into a template.
10242 * genmodes.c (get_mode_class): New function.
10243 (emit_insn_modes_h): Give modes the class returned by get_mode_class,
10244 or machine_mode if none.
10245 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Use
10246 as_a <scalar_float_mode>.
10247 * dwarf2out.c (mem_loc_descriptor): Likewise.
10248 (insert_float): Likewise.
10249 (add_const_value_attribute): Likewise.
10250 * simplify-rtx.c (simplify_immed_subreg): Likewise.
10251 * optabs.c (expand_absneg_bit): Take a scalar_float_mode.
10252 (expand_unop): Update accordingly.
10253 (expand_abs_nojump): Likewise.
10254 (expand_copysign_absneg): Take a scalar_float_mode.
10255 (expand_copysign_bit): Likewise.
10256 (expand_copysign): Update accordingly.
10258 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10259 Alan Hayward <alan.hayward@arm.com>
10260 David Sherwood <david.sherwood@arm.com>
10262 * coretypes.h (opt_mode): New class.
10263 * machmode.h (opt_mode): Likewise.
10264 (opt_mode::else_void): New function.
10265 (opt_mode::require): Likewise.
10266 (opt_mode::exists): Likewise.
10267 (GET_MODE_WIDER_MODE): Turn into a function and return an opt_mode.
10268 (GET_MODE_2XWIDER_MODE): Likewise.
10269 (mode_iterator::get_wider): Update accordingly.
10270 (mode_iterator::get_2xwider): Likewise.
10271 (mode_iterator::get_known_wider): Likewise, turning into a template.
10272 * combine.c (make_extraction): Update use of GET_MODE_WIDER_MODE,
10273 forcing a wider mode to exist.
10274 * config/cr16/cr16.h (LONG_REG_P): Likewise.
10275 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10276 * config/c6x/c6x.c (c6x_rtx_costs): Update use of
10277 GET_MODE_2XWIDER_MODE, forcing a wider mode to exist.
10278 * lower-subreg.c (init_lower_subreg): Likewise.
10279 * optabs-libfuncs.c (init_sync_libfuncs_1): Likewise, but not
10280 on the final iteration.
10281 * config/i386/i386.c (ix86_expand_set_or_movmem): Check whether
10282 a wider mode exists before asking for a move pattern.
10283 (get_mode_wider_vector): Update use of GET_MODE_WIDER_MODE,
10284 forcing a wider mode to exist.
10285 (expand_vselect_vconcat): Update use of GET_MODE_2XWIDER_MODE,
10286 returning false if no such mode exists.
10287 * config/ia64/ia64.c (expand_vselect_vconcat): Likewise.
10288 * config/mips/mips.c (mips_expand_vselect_vconcat): Likewise.
10289 * expmed.c (init_expmed_one_mode): Update use of GET_MODE_WIDER_MODE.
10290 Avoid checking for a MODE_INT if we already know the mode is not a
10292 (extract_high_half): Update use of GET_MODE_WIDER_MODE,
10293 forcing a wider mode to exist.
10294 (expmed_mult_highpart_optab): Likewise.
10295 (expmed_mult_highpart): Likewise.
10296 * expr.c (expand_expr_real_2): Update use of GET_MODE_WIDER_MODE,
10298 * lto-streamer-in.c (lto_input_mode_table): Likewise.
10299 * optabs-query.c (find_widening_optab_handler_and_mode): Likewise.
10300 * stor-layout.c (bit_field_mode_iterator::next_mode): Likewise.
10301 * internal-fn.c (expand_mul_overflow): Update use of
10302 GET_MODE_2XWIDER_MODE.
10303 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10304 * tree-ssa-math-opts.c (convert_mult_to_widen): Update use of
10305 GET_MODE_WIDER_MODE.
10306 (convert_plusminus_to_widen): Likewise.
10307 * tree-switch-conversion.c (array_value_type): Likewise.
10308 * var-tracking.c (emit_note_insn_var_location): Likewise.
10309 * tree-vrp.c (simplify_float_conversion_using_ranges): Likewise.
10310 Return false inside rather than outside the loop if no wider mode
10312 * optabs.c (expand_binop): Update use of GET_MODE_WIDER_MODE
10313 and GET_MODE_2XWIDER_MODE
10314 (can_compare_p): Use else_void.
10315 * gdbhooks.py (OptMachineModePrinter): New class.
10316 (build_pretty_printer): Use it for opt_mode.
10318 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10319 Alan Hayward <alan.hayward@arm.com>
10320 David Sherwood <david.sherwood@arm.com>
10322 * tree-switch-conversion.c (array_value_type): Only read TYPE_MODE
10323 once. Use get_narrowest_mode instead of GET_CLASS_NARROWEST_MODE.
10325 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10326 Alan Hayward <alan.hayward@arm.com>
10327 David Sherwood <david.sherwood@arm.com>
10329 * machmode.h (mode_traits): New structure.
10330 (get_narrowest_mode): New function.
10331 (mode_iterator::start): Likewise.
10332 (mode_iterator::iterate_p): Likewise.
10333 (mode_iterator::get_wider): Likewise.
10334 (mode_iterator::get_known_wider): Likewise.
10335 (mode_iterator::get_2xwider): Likewise.
10336 (FOR_EACH_MODE_IN_CLASS): New mode iterator.
10337 (FOR_EACH_MODE): Likewise.
10338 (FOR_EACH_MODE_FROM): Likewise.
10339 (FOR_EACH_MODE_UNTIL): Likewise.
10340 (FOR_EACH_WIDER_MODE): Likewise.
10341 (FOR_EACH_2XWIDER_MODE): Likewise.
10342 * builtins.c (expand_builtin_strlen): Use new mode iterators.
10343 * combine.c (simplify_comparison): Likewise
10344 * config/i386/i386.c (type_natural_mode): Likewise.
10345 * cse.c (cse_insn): Likewise.
10346 * dse.c (find_shift_sequence): Likewise.
10347 * emit-rtl.c (init_derived_machine_modes): Likewise.
10348 (init_emit_once): Likewise.
10349 * explow.c (hard_function_value): Likewise.
10350 * expmed.c (extract_fixed_bit_field_1): Likewise.
10351 (extract_bit_field_1): Likewise.
10352 (expand_divmod): Likewise.
10353 (emit_store_flag_1): Likewise.
10354 * expr.c (init_expr_target): Likewise.
10355 (convert_move): Likewise.
10356 (alignment_for_piecewise_move): Likewise.
10357 (widest_int_mode_for_size): Likewise.
10358 (emit_block_move_via_movmem): Likewise.
10359 (copy_blkmode_to_reg): Likewise.
10360 (set_storage_via_setmem): Likewise.
10361 (compress_float_constant): Likewise.
10362 * omp-low.c (omp_clause_aligned_alignment): Likewise.
10363 * optabs-query.c (get_best_extraction_insn): Likewise.
10364 * optabs.c (expand_binop): Likewise.
10365 (expand_twoval_unop): Likewise.
10366 (expand_twoval_binop): Likewise.
10367 (widen_leading): Likewise.
10368 (widen_bswap): Likewise.
10369 (expand_parity): Likewise.
10370 (expand_unop): Likewise.
10371 (prepare_cmp_insn): Likewise.
10372 (prepare_float_lib_cmp): Likewise.
10373 (expand_float): Likewise.
10374 (expand_fix): Likewise.
10375 (expand_sfix_optab): Likewise.
10376 * postreload.c (move2add_use_add2_insn): Likewise.
10377 * reg-stack.c (reg_to_stack): Likewise.
10378 * reginfo.c (choose_hard_reg_mode): Likewise.
10379 * rtlanal.c (init_num_sign_bit_copies_in_rep): Likewise.
10380 * stor-layout.c (mode_for_size): Likewise.
10381 (smallest_mode_for_size): Likewise.
10382 (mode_for_vector): Likewise.
10383 (finish_bitfield_representative): Likewise.
10384 * tree-ssa-math-opts.c (target_supports_divmod_p): Likewise.
10385 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
10386 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
10387 * var-tracking.c (prepare_call_arguments): Likewise.
10389 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10390 Alan Hayward <alan.hayward@arm.com>
10391 David Sherwood <david.sherwood@arm.com>
10393 * genconditions.c (write_header): Add a "#define USE_ENUM_MODES".
10394 * genmodes.c (emit_insn_modes_h): Define FOOmode to E_FOOmode if
10395 USE_ENUM_MODES is defined and to ((void) 0, E_FOOmode) otherwise.
10396 * machmode.h (mode_size): Move earlier in file.
10397 (mode_precision): Likewise.
10398 (mode_inner): Likewise.
10399 (mode_nunits): Likewise.
10400 (mode_unit_size): Likewise.
10401 (unit_unit_precision): Likewise.
10402 (mode_wider): Likewise.
10403 (mode_2xwider): Likewise.
10404 (machine_mode): New class.
10405 (mode_to_bytes): New function.
10406 (mode_to_bits): Likewise.
10407 (mode_to_precision): Likewise.
10408 (mode_to_inner): Likewise.
10409 (mode_to_unit_size): Likewise.
10410 (mode_to_unit_precision): Likewise.
10411 (mode_to_nunits): Likewise.
10412 (GET_MODE_SIZE): Use mode_to_bytes.
10413 (GET_MODE_BITSIZE): Use mode_to_bits.
10414 (GET_MODE_PRECISION): Use mode_to_precision.
10415 (GET_MODE_INNER): Use mode_to_inner.
10416 (GET_MODE_UNIT_SIZE): Use mode_to_unit_size.
10417 (GET_MODE_UNIT_PRECISION): Use mode_to_unit_precision.
10418 (GET_MODE_NUNITS): Use mode_to_nunits.
10419 * system.h (ALWAYS_INLINE): New macro.
10420 * config/powerpcspe/powerpcspe-c.c
10421 (altivec_resolve_overloaded_builtin): Use machine_mode instead of
10422 int for arg1_mode and arg2_mode.
10424 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10425 Alan Hayward <alan.hayward@arm.com>
10426 David Sherwood <david.sherwood@arm.com>
10428 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
10429 Prefix mode names with E_ in case statements.
10430 * config/aarch64/aarch64-elf.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10431 * config/aarch64/aarch64.c (aarch64_split_simd_combine): Likewise.
10432 (aarch64_split_simd_move): Likewise.
10433 (aarch64_gen_storewb_pair): Likewise.
10434 (aarch64_gen_loadwb_pair): Likewise.
10435 (aarch64_gen_store_pair): Likewise.
10436 (aarch64_gen_load_pair): Likewise.
10437 (aarch64_get_condition_code_1): Likewise.
10438 (aarch64_constant_pool_reload_icode): Likewise.
10439 (get_rsqrte_type): Likewise.
10440 (get_rsqrts_type): Likewise.
10441 (get_recpe_type): Likewise.
10442 (get_recps_type): Likewise.
10443 (aarch64_gimplify_va_arg_expr): Likewise.
10444 (aarch64_simd_container_mode): Likewise.
10445 (aarch64_emit_load_exclusive): Likewise.
10446 (aarch64_emit_store_exclusive): Likewise.
10447 (aarch64_expand_compare_and_swap): Likewise.
10448 (aarch64_gen_atomic_cas): Likewise.
10449 (aarch64_emit_bic): Likewise.
10450 (aarch64_emit_atomic_swap): Likewise.
10451 (aarch64_emit_atomic_load_op): Likewise.
10452 (aarch64_evpc_trn): Likewise.
10453 (aarch64_evpc_uzp): Likewise.
10454 (aarch64_evpc_zip): Likewise.
10455 (aarch64_evpc_ext): Likewise.
10456 (aarch64_evpc_rev): Likewise.
10457 (aarch64_evpc_dup): Likewise.
10458 (aarch64_gen_ccmp_first): Likewise.
10459 (aarch64_gen_ccmp_next): Likewise.
10460 * config/alpha/alpha.c (alpha_scalar_mode_supported_p): Likewise.
10461 (alpha_emit_xfloating_libcall): Likewise.
10462 (emit_insxl): Likewise.
10463 (alpha_arg_type): Likewise.
10464 * config/arc/arc.c (arc_vector_mode_supported_p): Likewise.
10465 (arc_preferred_simd_mode): Likewise.
10466 (arc_secondary_reload): Likewise.
10467 (get_arc_condition_code): Likewise.
10468 (arc_print_operand): Likewise.
10469 (arc_legitimate_constant_p): Likewise.
10470 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10471 * config/arc/arc.md (casesi_load): Likewise.
10472 (casesi_compact_jump): Likewise.
10473 * config/arc/predicates.md (proper_comparison_operator): Likewise.
10474 (cc_use_register): Likewise.
10475 * config/arm/aout.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10476 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
10477 (arm_init_iwmmxt_builtins): Likewise.
10478 * config/arm/arm.c (thumb1_size_rtx_costs): Likewise.
10479 (neon_expand_vector_init): Likewise.
10480 (arm_attr_length_move_neon): Likewise.
10481 (maybe_get_arm_condition_code): Likewise.
10482 (arm_emit_vector_const): Likewise.
10483 (arm_preferred_simd_mode): Likewise.
10484 (arm_output_iwmmxt_tinsr): Likewise.
10485 (thumb1_output_casesi): Likewise.
10486 (thumb2_output_casesi): Likewise.
10487 (arm_emit_load_exclusive): Likewise.
10488 (arm_emit_store_exclusive): Likewise.
10489 (arm_expand_compare_and_swap): Likewise.
10490 (arm_evpc_neon_vuzp): Likewise.
10491 (arm_evpc_neon_vzip): Likewise.
10492 (arm_evpc_neon_vrev): Likewise.
10493 (arm_evpc_neon_vtrn): Likewise.
10494 (arm_evpc_neon_vext): Likewise.
10495 (arm_validize_comparison): Likewise.
10496 * config/arm/neon.md (neon_vc<cmp_op><mode>): Likewise.
10497 * config/avr/avr-c.c (avr_resolve_overloaded_builtin): Likewise.
10498 * config/avr/avr.c (avr_rtx_costs_1): Likewise.
10499 * config/c6x/c6x.c (c6x_vector_mode_supported_p): Likewise.
10500 (c6x_preferred_simd_mode): Likewise.
10501 * config/epiphany/epiphany.c (get_epiphany_condition_code): Likewise.
10502 (epiphany_rtx_costs): Likewise.
10503 * config/epiphany/predicates.md (proper_comparison_operator):
10505 * config/frv/frv.c (condexec_memory_operand): Likewise.
10506 (frv_emit_move): Likewise.
10507 (output_move_single): Likewise.
10508 (output_condmove_single): Likewise.
10509 (frv_hard_regno_mode_ok): Likewise.
10510 (frv_matching_accg_mode): Likewise.
10511 * config/h8300/h8300.c (split_adds_subs): Likewise.
10512 (h8300_rtx_costs): Likewise.
10513 (h8300_print_operand): Likewise.
10514 (compute_mov_length): Likewise.
10515 (output_logical_op): Likewise.
10516 (compute_logical_op_length): Likewise.
10517 (compute_logical_op_cc): Likewise.
10518 (h8300_shift_needs_scratch_p): Likewise.
10519 (output_a_shift): Likewise.
10520 (compute_a_shift_length): Likewise.
10521 (compute_a_shift_cc): Likewise.
10522 (expand_a_rotate): Likewise.
10523 (output_a_rotate): Likewise.
10524 * config/i386/i386.c (classify_argument): Likewise.
10525 (function_arg_advance_32): Likewise.
10526 (function_arg_32): Likewise.
10527 (function_arg_64): Likewise.
10528 (function_value_64): Likewise.
10529 (ix86_gimplify_va_arg): Likewise.
10530 (ix86_legitimate_constant_p): Likewise.
10531 (put_condition_code): Likewise.
10532 (split_double_mode): Likewise.
10533 (ix86_avx256_split_vector_move_misalign): Likewise.
10534 (ix86_expand_vector_logical_operator): Likewise.
10535 (ix86_split_idivmod): Likewise.
10536 (ix86_expand_adjust_ufix_to_sfix_si): Likewise.
10537 (ix86_build_const_vector): Likewise.
10538 (ix86_build_signbit_mask): Likewise.
10539 (ix86_match_ccmode): Likewise.
10540 (ix86_cc_modes_compatible): Likewise.
10541 (ix86_expand_branch): Likewise.
10542 (ix86_expand_sse_cmp): Likewise.
10543 (ix86_expand_sse_movcc): Likewise.
10544 (ix86_expand_int_sse_cmp): Likewise.
10545 (ix86_expand_vec_perm_vpermi2): Likewise.
10546 (ix86_expand_vec_perm): Likewise.
10547 (ix86_expand_sse_unpack): Likewise.
10548 (ix86_expand_int_addcc): Likewise.
10549 (ix86_split_to_parts): Likewise.
10550 (ix86_vectorize_builtin_gather): Likewise.
10551 (ix86_vectorize_builtin_scatter): Likewise.
10552 (avx_vpermilp_parallel): Likewise.
10553 (inline_memory_move_cost): Likewise.
10554 (ix86_tieable_integer_mode_p): Likewise.
10555 (x86_maybe_negate_const_int): Likewise.
10556 (ix86_expand_vector_init_duplicate): Likewise.
10557 (ix86_expand_vector_init_one_nonzero): Likewise.
10558 (ix86_expand_vector_init_one_var): Likewise.
10559 (ix86_expand_vector_init_concat): Likewise.
10560 (ix86_expand_vector_init_interleave): Likewise.
10561 (ix86_expand_vector_init_general): Likewise.
10562 (ix86_expand_vector_set): Likewise.
10563 (ix86_expand_vector_extract): Likewise.
10564 (emit_reduc_half): Likewise.
10565 (ix86_emit_i387_round): Likewise.
10566 (ix86_mangle_type): Likewise.
10567 (ix86_expand_round_sse4): Likewise.
10568 (expand_vec_perm_blend): Likewise.
10569 (canonicalize_vector_int_perm): Likewise.
10570 (ix86_expand_vec_one_operand_perm_avx512): Likewise.
10571 (expand_vec_perm_1): Likewise.
10572 (expand_vec_perm_interleave3): Likewise.
10573 (expand_vec_perm_even_odd_pack): Likewise.
10574 (expand_vec_perm_even_odd_1): Likewise.
10575 (expand_vec_perm_broadcast_1): Likewise.
10576 (ix86_vectorize_vec_perm_const_ok): Likewise.
10577 (ix86_expand_vecop_qihi): Likewise.
10578 (ix86_expand_mul_widen_hilo): Likewise.
10579 (ix86_expand_sse2_abs): Likewise.
10580 (ix86_expand_pextr): Likewise.
10581 (ix86_expand_pinsr): Likewise.
10582 (ix86_preferred_simd_mode): Likewise.
10583 (ix86_simd_clone_compute_vecsize_and_simdlen): Likewise.
10584 * config/i386/sse.md (*andnot<mode>3): Likewise.
10585 (<mask_codefor><code><mode>3<mask_name>): Likewise.
10586 (*<code><mode>3): Likewise.
10587 * config/ia64/ia64.c (ia64_expand_vecint_compare): Likewise.
10588 (ia64_expand_atomic_op): Likewise.
10589 (ia64_arg_type): Likewise.
10590 (ia64_mode_to_int): Likewise.
10591 (ia64_scalar_mode_supported_p): Likewise.
10592 (ia64_vector_mode_supported_p): Likewise.
10593 (expand_vec_perm_broadcast): Likewise.
10594 * config/iq2000/iq2000.c (iq2000_move_1word): Likewise.
10595 (iq2000_function_arg_advance): Likewise.
10596 (iq2000_function_arg): Likewise.
10597 * config/m32c/m32c.c (m32c_preferred_reload_class): Likewise.
10598 * config/m68k/m68k.c (output_dbcc_and_branch): Likewise.
10599 (m68k_libcall_value): Likewise.
10600 (m68k_function_value): Likewise.
10601 (sched_attr_op_type): Likewise.
10602 * config/mcore/mcore.c (mcore_output_move): Likewise.
10603 * config/microblaze/microblaze.c (microblaze_function_arg_advance):
10605 (microblaze_function_arg): Likewise.
10606 * config/mips/mips.c (mips16_build_call_stub): Likewise.
10607 (mips_print_operand): Likewise.
10608 (mips_mode_ok_for_mov_fmt_p): Likewise.
10609 (mips_vector_mode_supported_p): Likewise.
10610 (mips_preferred_simd_mode): Likewise.
10611 (mips_expand_vpc_loongson_even_odd): Likewise.
10612 (mips_expand_vec_unpack): Likewise.
10613 (mips_expand_vi_broadcast): Likewise.
10614 (mips_expand_vector_init): Likewise.
10615 (mips_expand_vec_reduc): Likewise.
10616 (mips_expand_msa_cmp): Likewise.
10617 * config/mips/mips.md (casesi_internal_mips16_<mode>): Likewise.
10618 * config/mn10300/mn10300.c (mn10300_print_operand): Likewise.
10619 (cc_flags_for_mode): Likewise.
10620 * config/msp430/msp430.c (msp430_print_operand): Likewise.
10621 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Likewise.
10622 (nds32_output_casesi_pc_relative): Likewise.
10623 * config/nds32/nds32.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10624 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Likewise.
10625 (nvptx_gen_unpack): Likewise.
10626 (nvptx_gen_pack): Likewise.
10627 (nvptx_gen_shuffle): Likewise.
10628 (nvptx_gen_wcast): Likewise.
10629 (nvptx_preferred_simd_mode): Likewise.
10630 * config/pa/pa.c (pa_secondary_reload): Likewise.
10631 * config/pa/predicates.md (base14_operand): Likewise.
10632 * config/powerpcspe/powerpcspe-c.c
10633 (altivec_resolve_overloaded_builtin): Likewise.
10634 * config/powerpcspe/powerpcspe.c (rs6000_setup_reg_addr_masks):
10636 (rs6000_preferred_simd_mode): Likewise.
10637 (output_vec_const_move): Likewise.
10638 (rs6000_expand_vector_extract): Likewise.
10639 (rs6000_split_vec_extract_var): Likewise.
10640 (reg_offset_addressing_ok_p): Likewise.
10641 (rs6000_legitimate_offset_address_p): Likewise.
10642 (rs6000_legitimize_address): Likewise.
10643 (rs6000_emit_set_const): Likewise.
10644 (rs6000_const_vec): Likewise.
10645 (rs6000_emit_move): Likewise.
10646 (spe_build_register_parallel): Likewise.
10647 (rs6000_darwin64_record_arg_recurse): Likewise.
10648 (swap_selector_for_mode): Likewise.
10649 (spe_init_builtins): Likewise.
10650 (paired_init_builtins): Likewise.
10651 (altivec_init_builtins): Likewise.
10652 (do_load_for_compare): Likewise.
10653 (rs6000_generate_compare): Likewise.
10654 (rs6000_expand_float128_convert): Likewise.
10655 (emit_load_locked): Likewise.
10656 (emit_store_conditional): Likewise.
10657 (rs6000_output_function_epilogue): Likewise.
10658 (rs6000_handle_altivec_attribute): Likewise.
10659 (rs6000_function_value): Likewise.
10660 (emit_fusion_gpr_load): Likewise.
10661 (emit_fusion_p9_load): Likewise.
10662 (emit_fusion_p9_store): Likewise.
10663 * config/powerpcspe/predicates.md (easy_fp_constant): Likewise.
10664 (fusion_gpr_mem_load): Likewise.
10665 (fusion_addis_mem_combo_load): Likewise.
10666 (fusion_addis_mem_combo_store): Likewise.
10667 * config/rs6000/predicates.md (easy_fp_constant): Likewise.
10668 (fusion_gpr_mem_load): Likewise.
10669 (fusion_addis_mem_combo_load): Likewise.
10670 (fusion_addis_mem_combo_store): Likewise.
10671 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10673 * config/rs6000/rs6000-string.c (do_load_for_compare): Likewise.
10674 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
10675 (rs6000_preferred_simd_mode): Likewise.
10676 (output_vec_const_move): Likewise.
10677 (rs6000_expand_vector_extract): Likewise.
10678 (rs6000_split_vec_extract_var): Likewise.
10679 (reg_offset_addressing_ok_p): Likewise.
10680 (rs6000_legitimate_offset_address_p): Likewise.
10681 (rs6000_legitimize_address): Likewise.
10682 (rs6000_emit_set_const): Likewise.
10683 (rs6000_const_vec): Likewise.
10684 (rs6000_emit_move): Likewise.
10685 (rs6000_darwin64_record_arg_recurse): Likewise.
10686 (swap_selector_for_mode): Likewise.
10687 (paired_init_builtins): Likewise.
10688 (altivec_init_builtins): Likewise.
10689 (rs6000_expand_float128_convert): Likewise.
10690 (emit_load_locked): Likewise.
10691 (emit_store_conditional): Likewise.
10692 (rs6000_output_function_epilogue): Likewise.
10693 (rs6000_handle_altivec_attribute): Likewise.
10694 (rs6000_function_value): Likewise.
10695 (emit_fusion_gpr_load): Likewise.
10696 (emit_fusion_p9_load): Likewise.
10697 (emit_fusion_p9_store): Likewise.
10698 * config/rx/rx.c (rx_gen_move_template): Likewise.
10699 (flags_from_mode): Likewise.
10700 * config/s390/predicates.md (s390_alc_comparison): Likewise.
10701 (s390_slb_comparison): Likewise.
10702 * config/s390/s390.c (s390_handle_vectorbool_attribute): Likewise.
10703 (s390_vector_mode_supported_p): Likewise.
10704 (s390_cc_modes_compatible): Likewise.
10705 (s390_match_ccmode_set): Likewise.
10706 (s390_canonicalize_comparison): Likewise.
10707 (s390_emit_compare_and_swap): Likewise.
10708 (s390_branch_condition_mask): Likewise.
10709 (s390_rtx_costs): Likewise.
10710 (s390_secondary_reload): Likewise.
10711 (__SECONDARY_RELOAD_CASE): Likewise.
10712 (s390_expand_cs): Likewise.
10713 (s390_preferred_simd_mode): Likewise.
10714 * config/s390/vx-builtins.md (vec_packsu_u<mode>): Likewise.
10715 * config/sh/sh.c (sh_print_operand): Likewise.
10716 (dump_table): Likewise.
10717 (sh_secondary_reload): Likewise.
10718 * config/sh/sh.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10719 * config/sh/sh.md (casesi_worker_1): Likewise.
10720 (casesi_worker_2): Likewise.
10721 * config/sparc/predicates.md (icc_comparison_operator): Likewise.
10722 (fcc_comparison_operator): Likewise.
10723 * config/sparc/sparc.c (sparc_expand_move): Likewise.
10724 (emit_soft_tfmode_cvt): Likewise.
10725 (sparc_preferred_simd_mode): Likewise.
10726 (output_cbranch): Likewise.
10727 (sparc_print_operand): Likewise.
10728 (sparc_expand_vec_perm_bmask): Likewise.
10729 (vector_init_bshuffle): Likewise.
10730 * config/spu/spu.c (spu_scalar_mode_supported_p): Likewise.
10731 (spu_vector_mode_supported_p): Likewise.
10732 (spu_expand_insv): Likewise.
10733 (spu_emit_branch_or_set): Likewise.
10734 (spu_handle_vector_attribute): Likewise.
10735 (spu_builtin_splats): Likewise.
10736 (spu_builtin_extract): Likewise.
10737 (spu_builtin_promote): Likewise.
10738 (spu_expand_sign_extend): Likewise.
10739 * config/tilegx/tilegx.c (tilegx_scalar_mode_supported_p): Likewise.
10740 (tilegx_simd_int): Likewise.
10741 * config/tilepro/tilepro.c (tilepro_scalar_mode_supported_p): Likewise.
10742 (tilepro_simd_int): Likewise.
10743 * config/v850/v850.c (const_double_split): Likewise.
10744 (v850_print_operand): Likewise.
10745 (ep_memory_offset): Likewise.
10746 * config/vax/vax.c (vax_rtx_costs): Likewise.
10747 (vax_output_int_move): Likewise.
10748 (vax_output_int_add): Likewise.
10749 (vax_output_int_subtract): Likewise.
10750 * config/visium/predicates.md (visium_branch_operator): Likewise.
10751 * config/visium/visium.c (rtx_ok_for_offset_p): Likewise.
10752 (visium_print_operand_address): Likewise.
10753 * config/visium/visium.h (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
10754 * config/xtensa/xtensa.c (xtensa_mem_offset): Likewise.
10755 (xtensa_expand_conditional_branch): Likewise.
10756 (xtensa_copy_incoming_a7): Likewise.
10757 (xtensa_output_literal): Likewise.
10758 * dfp.c (decimal_real_maxval): Likewise.
10759 * targhooks.c (default_libgcc_floating_mode_supported_p): Likewise.
10761 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10762 Alan Hayward <alan.hayward@arm.com>
10763 David Sherwood <david.sherwood@arm.com>
10765 * genmodes.c (mode_size_inline): Add an E_ prefix to mode names.
10766 (mode_nunits_inline): Likewise.
10767 (mode_inner_inline): Likewise.
10768 (mode_unit_size_inline): Likewise.
10769 (mode_unit_precision_inline): Likewise.
10770 (emit_insn_modes_h): Likewise. Also emit a #define of the
10772 (emit_mode_wider): Add an E_ prefix to mode names.
10773 (emit_mode_complex): Likewise.
10774 (emit_mode_inner): Likewise.
10775 (emit_mode_adjustments): Likewise.
10776 (emit_mode_int_n): Likewise.
10777 * config/aarch64/aarch64-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP)
10778 (v2si_UP, v2sf_UP, v1df_UP, di_UP, df_UP, v16qi_UP, v8hi_UP, v8hf_UP)
10779 (v4si_UP, v4sf_UP, v2di_UP, v2df_UP, ti_UP, oi_UP, ci_UP, xi_UP)
10780 (si_UP, sf_UP, hi_UP, hf_UP, qi_UP): Likewise.
10781 (CRC32_BUILTIN, ENTRY): Likewise.
10782 * config/aarch64/aarch64.c (aarch64_push_regs): Likewise.
10783 (aarch64_pop_regs): Likewise.
10784 (aarch64_process_components): Likewise.
10785 * config/alpha/alpha.c (alpha_emit_conditional_move): Likewise.
10786 * config/arm/arm-builtins.c (v8qi_UP, v4hi_UP, v4hf_UP, v2si_UP)
10787 (v2sf_UP, di_UP, v16qi_UP, v8hi_UP, v8hf_UP, v4si_UP, v4sf_UP)
10788 (v2di_UP, ti_UP, ei_UP, oi_UP, hf_UP, si_UP, void_UP): Likewise.
10789 * config/arm/arm.c (arm_init_libfuncs): Likewise.
10790 * config/i386/i386-builtin-types.awk (ix86_builtin_type_vect_mode):
10792 * config/i386/i386-builtin.def (pcmpestr): Likewise.
10793 (pcmpistr): Likewise.
10794 * config/microblaze/microblaze.c (double_memory_operand): Likewise.
10795 * config/mmix/mmix.c (mmix_output_condition): Likewise.
10796 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
10798 * config/rl78/rl78.c (mduc_regs): Likewise.
10799 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
10800 (htm_expand_builtin): Likewise.
10801 * config/sh/sh.h (REGISTER_NATURAL_MODE): Likewise.
10802 * config/sparc/sparc.c (emit_save_or_restore_regs): Likewise.
10803 * config/xtensa/xtensa.c (print_operand): Likewise.
10804 * expmed.h (NUM_MODE_PARTIAL_INT): Likewise.
10805 (NUM_MODE_VECTOR_INT): Likewise.
10806 * genoutput.c (null_operand): Likewise.
10807 (output_operand_data): Likewise.
10808 * genrecog.c (print_parameter_value): Likewise.
10809 * lra.c (debug_operand_data): Likewise.
10811 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10813 * dwarf2out.c (scompare_loc_descriptor_wide)
10814 (scompare_loc_descriptor_narrow): New functions, split out from...
10815 (scompare_loc_descriptor): ...here.
10816 * expmed.c (emit_store_flag_int): New function, split out from...
10817 (emit_store_flag): ...here.
10819 2017-08-30 Richard Biener <rguenther@suse.de>
10821 * dwarf2out.c (dwarf2out_finish): Remove setting AT_pubnames.
10822 (dwarf2out_early_finish): Move setting of AT_pubnames from
10823 early debug output to early finish.
10825 2017-08-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10827 * gcc/config/msp430/msp430.h: Pass -mcode/data-region to the linker
10828 and -mdata-region to the assembler.
10830 2017-08-30 Richard Biener <rguenther@suse.de>
10832 * dwarf2out.c (add_dwarf_attr): Check we don't add duplicate
10834 (gen_subprogram_die): Add DW_AT_object_pointer only early.
10835 (dwarf2out_early_global_decl): Only generate a DIE for the
10836 abstract origin if it doesn't already exist or is a declaration DIE.
10837 (resolve_addr): Do not add the linkage name twice when
10838 generating a stub DIE for the DW_TAG_GNU_call_site target.
10840 2017-08-30 Richard Sandiford <richard.sandiford@linaro.org>
10842 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
10843 Use machine_mode rather than int for arg1_mode.
10845 2017-08-29 Michael Meissner <meissner@linux.vnet.ibm.com>
10848 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Insure
10849 that the second argument of the built-in functions to unpack
10850 128-bit scalar types to 64-bit values is 0 or 1. Change to use a
10851 switch statement instead a lot of if statements.
10852 * config/rs6000/rs6000.md (unpack<mode>, FMOVE128_VSX iterator):
10853 Allow 64-bit values to be in Altivec registers as well as
10854 traditional floating point registers.
10855 (pack<mode>, FMOVE128_VSX iterator): Likewise.
10857 2017-08-29 Alexander Monakov <amonakov@ispras.ru>
10859 * ira-costs.c (record_address_regs): Handle both operands of PLUS for
10860 MAX_REGS_PER_ADDRESS == 1.
10862 2017-08-29 Uros Bizjak <ubizjak@gmail.com>
10864 * config/i386/i386.opt (flag_fentry): Do not init to -1.
10865 * config/i386/i386.c (ix86_option_override_internal): Simplify
10866 setting of opts->x_flag_entry.
10868 2017-08-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10869 Jakub Jelinek <jakub@redhat.com>
10870 Richard Biener <rguenther@suse.de>
10872 PR tree-optimization/81503
10873 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Ensure
10874 folded constant fits in the target type; reorder tests for clarity.
10876 2017-08-29 Martin Liska <mliska@suse.cz>
10878 * passes.def: Include pass_lower_switch.
10879 * stmt.c (dump_case_nodes): Remove and move to
10880 tree-switch-conversion.
10881 (case_values_threshold): Likewise.
10882 (expand_switch_as_decision_tree_p): Likewise.
10883 (emit_case_decision_tree): Likewise.
10884 (expand_case): Likewise.
10885 (balance_case_nodes): Likewise.
10886 (node_has_low_bound): Likewise.
10887 (node_has_high_bound): Likewise.
10888 (node_is_bounded): Likewise.
10889 (emit_case_nodes): Likewise.
10890 (struct simple_case_node): New struct.
10891 (add_case_node): Remove.
10892 (emit_case_dispatch_table): Use vector instead of case_list.
10893 (reset_out_edges_aux): Remove.
10894 (compute_cases_per_edge): Likewise.
10895 (expand_case): Build list of simple_case_node.
10896 (expand_sjlj_dispatch_table): Use it.
10897 * tree-switch-conversion.c (struct case_node): Moved from
10898 stmt.c and adjusted.
10899 (emit_case_nodes): Likewise.
10900 (node_has_low_bound): Likewise.
10901 (node_has_high_bound): Likewise.
10902 (node_is_bounded): Likewise.
10903 (case_values_threshold): Likewise.
10904 (reset_out_edges_aux): Likewise.
10905 (compute_cases_per_edge): Likewise.
10906 (add_case_node): Likewise.
10907 (dump_case_nodes): Likewise.
10908 (balance_case_nodes): Likewise.
10909 (expand_switch_as_decision_tree_p): Likewise.
10910 (emit_jump): Likewise.
10911 (emit_case_decision_tree): Likewise.
10912 (try_switch_expansion): Likewise.
10913 (do_jump_if_equal): Likewise.
10914 (emit_cmp_and_jump_insns): Likewise.
10915 (fix_phi_operands_for_edge): New function.
10916 (record_phi_operand_mapping): Likewise.
10917 (class pass_lower_switch): New pass.
10918 (pass_lower_switch::execute): New function.
10919 (make_pass_lower_switch): Likewise.
10920 (conditional_probability):
10921 * timevar.def: Add TV_TREE_SWITCH_LOWERING.
10922 * tree-pass.h: Add make_pass_lower_switch.
10924 2017-08-29 Jozef Lawrynowicz <jozef.l@somniumtech.com>
10927 * gcc/config/msp430/msp430.c (msp430_attr): Mark interrupt
10930 2017-08-29 Richard Biener <rguenther@suse.de>
10932 * dwarf2out.c (add_dwarf_attr): When checking is enabled verify
10933 we do not add a DW_AT_inline attribute twice.
10934 (gen_subprogram_die): Remove code setting DW_AT_inline on
10935 DECL_ABSTRACT_P nodes.
10937 2017-08-29 Richard Sandiford <richard.sandiford@linaro.org>
10939 * gimplify.c (gimplify_call_expr): Copy the nothrow flag to
10940 calls to internal functions.
10941 (gimplify_modify_expr): Likewise.
10942 * tree-call-cdce.c (use_internal_fn): Likewise.
10943 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
10944 (convert_to_divmod): Set the nothrow flag.
10945 * tree-if-conv.c (predicate_mem_writes): Likewise.
10946 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
10947 (vectorizable_call): Likewise.
10948 (vectorizable_store): Likewise.
10949 (vectorizable_load): Likewise.
10950 * tree-vect-patterns.c (vect_recog_pow_pattern): Likewise.
10951 (vect_recog_mask_conversion_pattern): Likewise.
10953 2017-08-29 Martin Liska <mliska@suse.cz>
10956 * gcc.c (driver_handle_option): Add new argument.
10957 * opts-common.c (handle_option): Pass
10958 target_option_override_hook.
10959 * opts-global.c (lang_handle_option): Add new option.
10960 (set_default_handlers): Add new argument.
10961 (decode_options): Likewise.
10962 * opts.c (target_handle_option): Likewise.
10963 (common_handle_option): Call target_option_override_hook.
10964 * opts.h (struct cl_option_handler_func): Add hook for
10965 target option override.
10966 (struct cl_option_handlers): Likewise.
10967 (set_default_handlers): Add new argument.
10968 (decode_options): Likewise.
10969 (common_handle_option): Likewise.
10970 (target_handle_option): Likewise.
10971 * toplev.c (toplev::main): Pass targetm.target_option.override
10974 2017-08-29 Richard Biener <rguenther@suse.de>
10975 Dominik Infuehr <dominik.infuehr@theobroma-systems.com>
10977 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Properly confine
10978 life to the active subtree.
10980 2017-08-28 Jeff Law <law@redhat.com>
10982 * tree-ssa-dom.c (edge_info::record_simple_equiv): Call
10983 derive_equivalences.
10984 (derive_equivalences_from_bit_ior, record_temporary_equivalences):
10985 Code moved into....
10986 (edge_info::derive_equivalences): New private member function
10988 * tree-ssa-dom.c (class edge_info): Changed from a struct
10989 to a class. Add ctor/dtor, methods and data members.
10990 (edge_info::edge_info): Renamed from allocate_edge_info.
10991 Initialize additional members.
10992 (edge_info::~edge_info): New.
10993 (free_dom_edge_info): Delete the edge info.
10994 (record_edge_info): Use new class & associated member functions.
10995 Tighten forms for testing for edge equivalences.
10996 (record_temporary_equivalences): Iterate over the simple
10997 equivalences rather than assuming there's only one per edge.
10998 (cprop_into_successor_phis): Iterate over the simple
10999 equivalences rather than assuming there's only one per edge.
11000 (optimize_stmt): Use operand_equal_p rather than pointer
11001 equality for mini-DSE code.
11003 2017-08-28 Nathan Sidwell <nathan@acm.org>
11005 * gcc.c (execute): Fold SIGPIPE handling into switch
11006 statement. Adjust internal error message.
11008 2017-08-28 Richard Biener <rguenther@suse.de>
11011 * dwarf2out.c (gen_remaining_tmpl_value_param_die_attributes):
11012 Do nothing for removed DIEs.
11014 2017-08-28 Richard Biener <rguenther@suse.de>
11016 PR tree-optimization/81977
11017 * tree-ssa-sccvn.c (vn_reference_lookup_3): Fix look through
11020 2017-08-28 Alexander Monakov <amonakov@ispras.ru>
11023 * doc/md.texi (mem_thread_fence): Remove mention of mode. Rewrite.
11024 * optabs.c (expand_mem_thread_fence): Emit a compiler barrier when
11025 using targetm.gen_mem_thread_fence.
11027 2017-08-27 Uros Bizjak <ubizjak@gmail.com>
11030 * config/i386/i386.md (*<btsc><mode>): Change operand 2
11031 predicate to register_operand. Reorder operands.
11032 (*btr<mode>): Ditto.
11033 (*<btsc><mode>_mask): Change operand 3 predicate to register_operand.
11034 (*btr<mode>_mask): Ditto.
11036 2017-08-25 Steven Munroe <munroesj@gcc.gnu.org>
11038 * config.gcc (powerpc*-*-*): Add xmmintrin.h and mm_malloc.h.
11039 * config/rs6000/xmmintrin.h: New file.
11040 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include xmmintrin.h.
11042 2017-08-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11045 * config/rs6000/rs6000-p8swap.c (find_alignment_op): Add reference
11046 parameter and_insn and return it.
11047 (recombine_lvx_pattern): Insert a copy to ensure availability of
11048 the base register of the copied masking operation at the point of
11049 the instruction replacement.
11050 (recombine_stvx_pattern): Likewise.
11052 2017-08-25 Michael Meissner <meissner@linux.vnet.ibm.com>
11054 * config/rs6000/rs6000.opt (-mpower9-dform-scalar): Delete
11055 undocumented switches.
11056 (-mpower9-dform-vector): Likewise.
11057 (-mpower9-dform): Likewise.
11058 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
11059 comments to delete references to -mpower9-dform* switches.
11060 * config/rs6000/predicates.md (vsx_quad_dform_memory_operand):
11061 Delete reference to -mpower9-dform* switches, test for
11062 -mpower9-vector instead.
11063 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Likewise.
11064 (OTHER_P9_VECTOR_MASKS): Likewise.
11065 (POWERPC_MASKS): Likewise.
11066 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Change
11067 tests against -mpower9-dform* to -mpower9-vector. Delete code
11068 that checked for -mpower9-dform* consistancy with other options.
11069 Add test for -mpower9-misc to enable other power9 switches.
11070 (rs6000_init_hard_regno_mode_ok): Likewise.
11071 (rs6000_option_override_internal): Likewise.
11072 (rs6000_emit_prologue): Likewise.
11073 (rs6000_emit_epilogue): Likewise.
11074 (rs6000_opt_masks): Delete -mpower9-dform-{scalar,vector}.
11075 (rs6000_disable_incompatiable_switches): Delete -mpower9-dform.
11076 (emit_fusion_p9_load): Change tests for -mpower9-dform-scalar
11078 (emit_fusion_p9_store): Likewise.
11079 * config/rs6000/rs6000.h (TARGET_P9_DFORM_SCALAR): Delete
11080 resetting these macros if the assembler does not support ISA 3.0
11082 (TARGET_P9_DFORM_VECTOR): Likewise.
11083 * config/rs6000/rs6000.md (peepholes to optimize altivec memory):
11084 Change to use -mpower9-vector instead of -mpower9-dform-scalar.
11086 2017-08-25 Alan Modra <amodra@gmail.com>
11088 PR rtl-optimization/81747
11089 * cse.c (cse_extended_basic_block): Don't attempt to record
11090 equivalences for degenerate conditional branches that branch
11091 to their fall-through.
11093 2017-08-24 Martin Sebor <msebor@redhat.com>
11095 PR middle-end/81908
11096 * gimple-fold.c (size_must_be_zero_p): New function.
11097 (gimple_fold_builtin_memory_op): Call it.
11099 2017-08-24 Steven Munroe <munroesj@gcc.gnu.org>
11101 * config/rs6000/mm_malloc.h: New file.
11103 2017-08-24 Bin Cheng <bin.cheng@arm.com>
11105 PR tree-optimization/81913
11106 * tree-ssa-loop-niter.c (number_of_iterations_cond): Skip niter
11107 analysis when either IVs in condition can wrap.
11109 2017-08-24 Uros Bizjak <ubizjak@gmail.com>
11111 * dwarf2out.c (MAX_ARTIFICIAL_LABEL_BYTES): Increase to 40.
11112 * dwarf2cfi.c (MAX_ARTIFICIAL_LABEL_BYTES): Remove.
11114 2017-08-24 Richard Biener <rguenther@suse.de>
11117 * targhooks.c (default_target_can_inline_p): Properly
11118 use target_option_default_node when no DECL_FUNCTION_SPECIFIC_TARGET
11119 is present and always compare.
11120 * config/i386/i386.c (ix86_valid_target_attribute_tree): Do not
11121 infer -mfpmath=sse from TARGET_SSE_P.
11122 (ix86_can_inline_p): Properly use target_option_default_node when
11123 no DECL_FUNCTION_SPECIFIC_TARGET is present and always compare.
11125 2017-08-24 Richard Biener <rguenther@suse.de>
11128 * dwarf2out.c (output_die): Handle flag_generate_offload like
11130 (output_comp_unit): Likewise.
11131 (gen_array_type_die): Likewise.
11132 (dwarf2out_early_finish): Likewise.
11133 (note_variable_value_in_expr): Likewise.
11134 (dwarf2out_finish): Likewise. Adjust assert.
11135 * cgraphunit.c (symbol_table::compile): Move setting of
11136 flag_generate_offload earlier ...
11137 (symbol_table::finalize_compilation_unit): ... here, before
11138 early debug finalization.
11140 2017-08-24 Richard Biener <rguenther@suse.de>
11142 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
11143 and ipa-fnsummary.h.
11144 (ix86_can_inline_p): When ix86_fpmath flags do not match
11145 check whether the callee uses FP math at all.
11147 2017-08-24 Aldy Hernandez <aldyh@redhat.com>
11149 PR middle-end/81931
11150 * tree-ssanames.c (get_nonzero_bits): Use element_precision
11151 instead of TYPE_PRECISION.
11153 2017-08-24 Richard Sandiford <richard.sandiford@linaro.org>
11154 Alan Hayward <alan.hayward@arm.com>
11155 David Sherwood <david.sherwood@arm.com>
11157 * combine.c (make_extraction): Use subreg_offset_from_lsb.
11159 2017-08-23 Daniel Santos <daniel.santos@pobox.com>
11161 * config/i386/i386.h (ix86_frame::stack_realign_allocate_offset):
11163 (ix86_frame::stack_realign_allocate): New field.
11164 (struct machine_frame_state): Modify comments.
11165 (machine_frame_state::sp_realigned_fp_end): New field.
11166 * config/i386/i386.c (ix86_compute_frame_layout): Rework stack frame
11167 layout calculation.
11168 (sp_valid_at): Add assertion to assure no attempt to access invalid
11169 offset of a realigned stack.
11170 (fp_valid_at): Likewise.
11171 (choose_baseaddr): Modify comments.
11172 (ix86_emit_outlined_ms2sysv_save): Adjust to changes in
11173 ix86_expand_prologue.
11174 (ix86_expand_prologue): Modify stack realignment and allocation.
11175 (ix86_expand_epilogue): Modify comments.
11176 * doc/sourcebuild.texi: Add documentation for target selectors avx2,
11177 avx2_runtime, avx512f, and avx512f_runtime.
11179 2017-08-23 Uros Bizjak <ubizjak@gmail.com>
11181 * config/i386/i386.opt: Remove unneeded Init(0) initializations.
11182 (mstackrealign): Do not init to -1.
11183 * config/i386/i386.c (ix86_option_override_internal):
11184 Check opts_set, not opts when setting default value of
11185 opts->x_ix86_force_align_arg_pointer.
11187 2017-08-23 Richard Biener <rguenther@suse.de>
11189 * function.c (fndecl_name): Use verbosity 1 (no arguments) for
11190 lang_hooks.decl_printable_name.
11191 * print-rtl-function.c (print_rtx_function): Likewise.
11192 * tree-pretty-print.c (dump_function_header): Likewise.
11194 2017-08-23 Richard Biener <rguenther@suse.de>
11197 * dwarf2out.c (dwarf2out_abstract_function): Handle LTO with
11198 -g0 at compile-time.
11200 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11202 PR middle-end/19706
11203 * doc/sourcebuild.texi (Other hardware attributes):
11206 2017-08-23 Tamar Christina <tamar.christina@arm.com>
11208 PR middle-end/19706
11209 * tree-ssa-math-opts.c (convert_expand_mult_copysign):
11210 Fix single-use check.
11212 2017-08-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
11214 * gcc.c (execute): Only test for SIGKILL and SIGQUIT if available.
11216 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11218 * doc/install.texi: Modify to add more details on running selected
11221 2017-08-22 Daniel Santos <daniel.santos@pobox.com>
11223 * config/i386/i386.c (ix86_option_override_internal): Error when -mx32
11224 is combined with -mabi=ms.
11225 (ix86_function_type_abi): Limit errors for mixing -mx32 with attribute
11228 2017-08-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11230 PR tree-optimization/81488
11231 * gimple-ssa-strength-reduction (struct slsr_cand_d): Add visited
11232 and cached_basis fields.
11233 (MAX_SPREAD): New constant.
11234 (alloc_cand_and_find_basis): Initialize new fields.
11235 (clear_visited): New function.
11236 (create_phi_basis_1): Rename from create_phi_basis, set visited
11237 and cached_basis fields.
11238 (create_phi_basis): New wrapper function.
11239 (phi_add_costs_1): Rename from phi_add_costs, add spread
11240 parameter, set visited field, short-circuit when limits reached.
11241 (phi_add_costs): New wrapper function.
11242 (record_phi_increments_1): Rename from record_phi_increments, set
11244 (record_phi_increments): New wrapper function.
11245 (phi_incr_cost_1): Rename from phi_incr_cost, set visited field.
11246 (phi_incr_cost): New wrapper function.
11247 (all_phi_incrs_profitable_1): Rename from
11248 all_phi_incrs_profitable, set visited field.
11249 (all_phi_incrs_profitable): New wrapper function.
11251 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11252 Alan Hayward <alan.hayward@arm.com>
11253 David Sherwood <david.sherwood@arm.com>
11255 * rtl.h (paradoxical_subreg_p): Define inline, and add a version
11256 that takes the outer and inner modes.
11257 * doc/rtl.texi: Use paradoxical_subreg_p instead of a GET_MODE_SIZE
11258 comparison as the canonical test for a paradoxical subreg.
11259 * combine.c (simplify_set): Use paradoxical_subreg_p.
11260 (make_extraction): Likewise.
11261 (force_to_mode): Likewise.
11262 (rtx_equal_for_field_assignment_p): Likewise.
11263 (gen_lowpart_for_combine): Likewise.
11264 (simplify_comparison): Likewise.
11265 * cse.c (equiv_constant): Likewise.
11266 * expmed.c (store_bit_field_1): Likewise.
11267 * final.c (alter_subreg): Likewise.
11268 * fwprop.c (propagate_rtx): Likewise.
11269 (forward_propagate_subreg): Likewise.
11270 * ira-conflicts.c (ira_build_conflicts): Likewise.
11271 * lower-subreg.c (simplify_gen_subreg_concatn): Likewise.
11272 * lra-constraints.c (curr_insn_transform): Likewise.
11273 (split_reg): Likewise.
11274 * lra-eliminations.c (move_plus_up): Likewise.
11275 (lra_eliminate_regs_1): Likewise.
11276 * recog.c (general_operand): Likewise.
11277 * ree.c (combine_reaching_defs): Likewise.
11278 * reload.c (push_reload): Likewise.
11279 (find_reloads): Likewise.
11280 * reload1.c (elimination_effects): Likewise.
11281 (compute_reload_subreg_offset): Likewise.
11282 (choose_reload_regs): Likewise.
11283 * rtlanal.c (subreg_lsb_1): Likewise.
11284 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
11285 (simplify_subreg): Likewise.
11286 * var-tracking.c (track_loc_p): Likewise.
11287 * emit-rtl.c (byte_lowpart_offset): Likewise.
11288 (paradoxical_subreg_p): Delete out-of-line definition.
11290 2017-08-22 Jeff Law <law@redhat.com>
11292 PR tree-optimization/81741
11293 PR tree-optimization/71947
11294 * tree-ssa-dom.c: Include tree-inline.h.
11295 (record_temporary_equivalences): Only record SSA_NAME = SSA_NAME
11296 equivalences if one is more expensive to compute than the other.
11297 * tree-ssa-scopedtables.h (class const_or_copies): Make
11298 record_const_or_copy_raw method private.
11299 (class avail_exprs_stack): New method simplify_binary_operation.
11300 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr): Call
11301 avail_exprs_stack::simplify_binary_operation as needed.
11302 (avail_exprs_stack::simplify_binary_operation): New function.
11304 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
11306 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
11307 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
11308 (DOT_SYMBOLS): Likewise.
11309 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
11310 (RELOCATABLE_NEEDS_FIXUP): Likewise.
11311 (RS6000_ABI_NAME): Likewise.
11312 (TARGET_CMODEL): Likewise.
11313 (TOC_SECTION_ASM_OP): Likewise.
11314 (SET_CMODEL): New macro.
11315 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
11317 2017-08-22 Richard Biener <rguenther@suse.de>
11319 * tree-inline.c (remap_type_1): Change asserts on TYPE_SIZE[_UNIT]
11320 to allow for free-lang-data replacements similar to verify_type_variant.
11322 2017-08-22 Richard Sandiford <richard.sandiford@linaro.org>
11323 Alan Hayward <alan.hayward@arm.com>
11324 David Sherwood <david.sherwood@arm.com>
11326 * config/aarch64/aarch64.md (casesi): Use DImode rather than
11327 VOIDmode for the LABEL_REF.
11329 2017-08-22 Richard Biener <rguenther@suse.de>
11331 * tree-cfg.c (gimple_split_edge): Avoid reallocating target PHI nodes.
11333 2017-08-22 Richard Biener <rguenther@suse.de>
11335 * common.opt (feliminate-dwarf2-dups): Ignore.
11336 * doc/invoke.texi (feliminate-dwarf2-dups): Remove documentation.
11337 * dwarf2out.c (push_new_compile_unit, pop_compile_unit,
11338 same_die_p_wrap, compute_section_prefix,
11339 is_symbol_die, assign_symbol_names, break_out_includes): Remove.
11340 (comdat_symbol_id, comdat_symbol_number): Likewise.
11341 (cu_hash_table_entry, cu_hash_table_entry_hasher, cu_hash_type):
11343 (check_duplicate_cu, record_comdat_symbol_number): Likewise.
11344 (output_die): Mark unreachable path unreachable.
11345 (dwarf2out_start_source_file): Do not create DW_TAG_GNU_BINCL.
11346 (dwarf2out_end_source_file): Do not create DW_TAG_GNU_EINCL.
11347 (dwarf2out_init): Remove code handling flag_eliminate_dwarf2_dups.
11348 (dwarf2out_early_finish): Likewise.
11350 2017-08-22 Aldy Hernandez <aldyh@redhat.com>
11352 * wide-int.h (hwi_with_prec::hwi_with_prec): Sign extend.
11354 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
11357 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
11358 not VAR_P. Filter attribute warnings with OPT_Wattributes.
11359 (avr_attribute_table) <io, io_low, address>: Initialize
11360 .decl_required with true.
11362 2017-08-21 Michael Meissner <meissner@linux.vnet.ibm.com>
11364 * config/rs6000/rs6000-cpus.def (-mvsx-scalar-float): Delete
11365 undocumented debugging options.
11366 (-mvsx-scalar-double): Likewise.
11367 (-mallow-df-permute): Likewise.
11368 (-mvectorize-builtins): Likewise.
11369 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
11370 (rs6000_builtin_vectorized_function): Likewise.
11371 (rs6000_builtin_md_vectorized_function): Likewise.
11372 (rs6000_opt_vars): Likewise.
11374 2017-08-21 Uros Bizjak <ubizjak@gmail.com>
11377 * config/i386/i386.md (*btsq_imm): Rename from *btsq.
11378 (*btrq_imm): Rename from *btrq.
11379 (*btcq_imm): Rename from *btcq.
11380 (btsc): New code attribute.
11381 (*<btsc><mode>): New insn pattern.
11382 (*btr<mode>): Ditto.
11383 (*<btsc><mode>_mask): New insn_and_split pattern.
11384 (*btr<mode>_mask): Ditto.
11386 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11387 Alan Hayward <alan.hayward@arm.com>
11388 David Sherwood <david.sherwood@arm.com>
11390 * function.c (pad_below): Simplify padding calculation.
11392 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11393 Alan Hayward <alan.hayward@arm.com>
11394 David Sherwood <david.sherwood@arm.com>
11396 * target.def (function_prologue): Remove frame size argument.
11397 (function_epilogue): Likewise.
11398 * doc/tm.texi: Regenerate.
11399 * final.c (final_start_function): Update call to function_prologue.
11400 (final_end_function): Update call to function_epilogue.
11401 (default_function_pro_epilogue): Remove frame size argument.
11402 * output.h (default_function_pro_epilogue): Likewise.
11403 * config/arm/arm.c (arm_output_function_epilogue): Likewise.
11404 (arm_output_function_prologue): Likewise.
11405 * config/frv/frv.c (frv_function_prologue): Likewise.
11406 (frv_function_epilogue): Likewise.
11407 * config/i386/i386.c (ix86_output_function_epilogue): Likewise.
11408 * config/ia64/ia64.c (ia64_output_function_prologue): Likewise.
11409 (ia64_output_function_epilogue): Likewise.
11410 * config/m32r/m32r.c (m32r_output_function_prologue): Likewise.
11411 (m32r_output_function_epilogue): Likewise.
11412 * config/microblaze/microblaze.c (microblaze_function_prologue)
11413 (microblaze_function_epilogue): Likewise.
11414 * config/mips/mips.c (mips_output_function_prologue): Likewise.
11415 (mips_output_function_epilogue): Likewise.
11416 * config/mmix/mmix.c (mmix_target_asm_function_prologue): Likewise.
11417 (mmix_target_asm_function_epilogue): Likewise.
11418 * config/msp430/msp430.c (msp430_start_function): Likewise.
11419 * config/nds32/nds32.c (nds32_asm_function_prologue): Likewise.
11420 (nds32_asm_function_epilogue): Likewise.
11421 * config/nios2/nios2.c (nios2_asm_function_prologue): Likewise.
11422 * config/pa/pa.c (pa_output_function_prologue): Likewise.
11423 (pa_output_function_epilogue): Likewise.
11424 * config/powerpcspe/powerpcspe.c (rs6000_output_function_prologue)
11425 (rs6000_output_function_epilogue): Likewise.
11426 * config/rl78/rl78.c (rl78_start_function): Likewise.
11427 * config/rs6000/rs6000.c (rs6000_output_function_prologue): Likewise.
11428 (rs6000_output_function_epilogue): Likewise.
11429 * config/rx/rx.c (rx_output_function_prologue): Likewise.
11430 * config/sh/sh.c (sh_output_function_epilogue): Likewise.
11431 * config/sparc/sparc.c (sparc_asm_function_prologue): Likewise.
11432 (sparc_asm_function_epilogue): Likewise.
11434 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11436 * tree.h (type_has_mode_precision_p): New function.
11437 * convert.c (convert_to_integer_1): Use it.
11438 * expr.c (expand_expr_real_2): Likewise.
11439 (expand_expr_real_1): Likewise.
11440 * fold-const.c (fold_single_bit_test_into_sign_test): Likewise.
11441 * match.pd: Likewise.
11442 * tree-ssa-forwprop.c (simplify_rotate): Likewise.
11443 * tree-ssa-math-opts.c (convert_mult_to_fma): Likewise.
11444 * tree-tailcall.c (process_assignment): Likewise.
11445 * tree-vect-loop.c (vectorizable_reduction): Likewise.
11446 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern)
11447 (vect_recog_mult_pattern, vect_recog_divmod_pattern): Likewise.
11448 * tree-vect-stmts.c (vectorizable_conversion): Likewise.
11449 (vectorizable_assignment): Likewise.
11450 (vectorizable_shift): Likewise.
11451 (vectorizable_operation): Likewise.
11452 * tree-vrp.c (register_edge_assert_for_2): Likewise.
11454 2017-08-21 Wilco Dijkstra <wdijkstr@arm.com>
11456 * match.pd: Add pow (C, x) simplification.
11458 2017-08-21 Richard Biener <rguenther@suse.de>
11460 PR tree-optimization/81900
11461 * tree-ssa-pre.c (compute_antic_aux): Properly compute changed
11462 for blocks with abnormal predecessors.
11463 (compute_antic): Do not set visited flag prematurely.
11465 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
11468 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
11470 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11472 * stor-layout.h (vector_type_mode): Move to...
11473 * tree.h (vector_type_mode): ...here.
11474 * stor-layout.c (vector_type_mode): Move to...
11475 * tree.c (vector_type_mode): ...here. Include rtl.h and regs.h.
11477 2017-08-21 Richard Biener <rguenther@suse.de>
11479 * debug.h (struct gcc_debug_hooks): Add die_ref_for_decl and
11480 register_external_die hooks.
11481 (debug_false_tree_charstarstar_uhwistar): Declare.
11482 (debug_nothing_tree_charstar_uhwi): Likewise.
11483 * debug.c (do_nothing_debug_hooks): Adjust.
11484 (debug_false_tree_charstarstar_uhwistar): New do nothing.
11485 (debug_nothing_tree_charstar_uhwi): Likewise.
11486 * dbxout.c (dbx_debug_hooks): Adjust.
11487 (xcoff_debug_hooks): Likewise.
11488 * sdbout.c (sdb_debug_hooks): Likewise.
11489 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
11490 * dwarf2out.c (macinfo_label_base): New global.
11491 (dwarf2out_register_external_die): New function for the
11492 register_external_die hook.
11493 (dwarf2out_die_ref_for_decl): Likewise for die_ref_for_decl.
11494 (dwarf2_debug_hooks): Use them.
11495 (dwarf2_lineno_debug_hooks): Adjust.
11496 (struct die_struct): Add with_offset flag.
11497 (DEBUG_LTO_DWO_INFO_SECTION, DEBUG_LTO_INFO_SECTION,
11498 DEBUG_LTO_DWO_ABBREV_SECTION, DEBUG_LTO_ABBREV_SECTION,
11499 DEBUG_LTO_DWO_MACINFO_SECTION, DEBUG_LTO_MACINFO_SECTION,
11500 DEBUG_LTO_DWO_MACRO_SECTION, DEBUG_LTO_MACRO_SECTION,
11501 DEBUG_LTO_LINE_SECTION, DEBUG_LTO_DWO_STR_OFFSETS_SECTION,
11502 DEBUG_LTO_STR_DWO_SECTION, DEBUG_STR_LTO_SECTION): New macros
11503 defining section names for the early LTO debug variants.
11504 (reset_indirect_string): New helper.
11505 (add_AT_external_die_ref): Helper for dwarf2out_register_external_die.
11506 (print_dw_val): Add support for offsetted symbol references.
11507 (get_ultimate_context): Split out from is_cxx.
11508 (is_cxx): Use get_ultimate_context.
11509 (is_fortran): Add decl overload.
11510 (compute_comp_unit_symbol): Split out worker from
11511 compute_section_prefix.
11512 (compute_section_prefix): Call compute_comp_unit_symbol and
11513 set comdat_type_p here.
11514 (output_die): Skip DIE symbol output for the LTO added one.
11515 Handle DIE symbol references with offset.
11516 (output_comp_unit): Guard section name mangling properly.
11517 For LTO debug sections emit a symbol at the section beginning
11518 which we use to refer to its DIEs.
11519 (add_abstract_origin_attribute): For DIEs registered via
11520 dwarf2out_register_external_die directly refer to the early
11521 DIE rather than indirectly through the shadow one we created.
11522 Remove obsolete call to dwarf2out_abstract_function for
11523 non-function/block origins.
11524 (gen_array_type_die): When generating early LTO debug do
11525 not emit DW_AT_string_length.
11526 (gen_formal_parameter_die): Do not re-create DIEs for PARM_DECLs
11527 late when in LTO. As suggested place a gcc_unreachable for
11528 the DECL_ABSTRACT_P case.
11529 (gen_subprogram_die): Avoid another specification DIE
11530 for early built declarations/definitions for the late LTO case.
11531 (gen_variable_die): Add type references for late duplicated VLA dies
11533 (gen_inlined_subroutine_die): Do not call dwarf2out_abstract_function,
11534 we have the abstract instance already.
11535 (process_scope_var): Adjust decl DIE contexts in LTO which
11536 first puts them in limbo.
11537 (gen_decl_die): Do not generate type DIEs late apart from
11538 types for VLAs or for decls we do not yet have a DIE. Do not
11539 call dwarf2out_abstract_function late.
11540 (dwarf2out_early_global_decl): Make sure to create DIEs
11541 for abstract instances of a decl first.
11542 (dwarf2out_late_global_decl): Adjust comment.
11543 (output_macinfo_op): With multiple macro sections use
11544 macinfo_label_base to distinguish labels.
11545 (output_macinfo): Likewise. Update macinfo_label_base.
11546 Pass in the line info label.
11547 (note_variable_value_in_expr): When generating LTO resolve
11548 all variable values here by generating DIEs as needed.
11549 (init_sections_and_labels): Add early LTO debug flag parameter
11550 and generate different sections and names if set. Add generation
11551 counter for the labels so we can have multiple of them.
11552 (reset_dies): Helper to allow DIEs to be output multiple times.
11553 (dwarf2out_finish): When outputting DIEs to the fat part of an
11554 LTO object first reset DIEs.
11555 (dwarf2out_early_finish): Output early DIEs when generating LTO.
11556 (modified_type_die): Check for decl_ultimate_origin being self
11558 (gen_type_die_with_usage): Likewise.
11559 (gen_typedef_die): Allow decl_ultimate_origin being self.
11560 (set_decl_abstract_flags): Remove.
11561 (set_block_abstract_flags): Likewise.
11562 (dwarf2out_abstract_function): Treat the early generated DIEs
11563 as the abstract copy and only add DW_AT_inline and
11564 DW_AT_artificial here and call set_decl_origin_self.
11565 If the DIE has an abstract origin don't do anything.
11566 * tree.c (free_lang_data): Build a dummy TRANSLATION_UNIT_DECL
11567 if we have none yet (Go fails to build one, PR78628).
11568 (variably_modified_type_p): Prevent endless recursion for Ada
11569 cyclic pointer types.
11570 * lto-streamer-in.c: Include debug.h.
11571 (dref_queue): New global.
11572 (lto_read_tree_1): Stream in DIE references.
11573 (lto_input_tree): Register DIE references.
11574 (input_function): Stream DECL_DEBUG_ARGS.
11575 * lto-streamer-out.c: Include debug.h.
11576 (lto_write_tree_1): Output DIE references.
11577 (DFS::DFS_write_tree_body): Follow DECL_ABSTRACT_ORIGIN.
11578 Force a TRANSLATION_UNIT_DECL DECL_CONTEXT for file-scope decls.
11579 (output_function): Stream DECL_DEBUG_ARGS.
11580 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
11581 Stream DECL_ABSTRACT_ORIGIN.
11582 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
11583 (write_ts_decl_minimal_tree_pointers): Force a TRANSLATION_UNIT_DECL
11584 DECL_CONTEXT for file-scope decls.
11585 * lto-streamer.h (struct dref_entry): Declare.
11586 (dref_queue): Likewise.
11587 * cfgexpand.c (pass_expand::execute): Do not call the
11588 outlining_inline_function hook here.
11589 * lto-wrapper.c (debug_obj): New global.
11590 (tool_cleanup): Unlink it if required.
11591 (debug_objcopy): New function.
11592 (run_gcc): Handle early debug sections in the IL files by
11593 extracting them to separate files, partially linkin them and
11594 feeding the result back as result to the linker.
11595 * config/darwin.h (DEBUG_LTO_INFO_SECTION, DEBUG_LTO_ABBREV_SECTION,
11596 DEBUG_LTO_MACINFO_SECTION, DEBUG_LTO_LINE_SECTION,
11597 DEBUG_STR_LTO_SECTION, DEBUG_LTO_MACRO_SECTION): Put early debug
11598 sections into a separate segment.
11599 * config/darwin.c (darwin_asm_named_section): Handle __GNU_DWARF_LTO
11601 (darwin_asm_dwarf_section): Likewise.
11602 (darwin_asm_output_dwarf_offset): Likewise.
11603 * config/i386/i386.c (make_resolver_func): Set DECL_IGNORED_P.
11605 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11606 Alan Hayward <alan.hayward@arm.com>
11607 David Sherwood <david.sherwood@arm.com>
11609 * read-md.h (md_reader::record_potential_iterator_use): Replace
11610 pointer argument with an rtx and an index.
11611 * read-rtl.c (iterator_group::apply_iterator): Likewise.
11612 (apply_mode_iterator): Likewise.
11613 (apply_code_iterator): Likewise.
11614 (apply_int_iterator): Likewise.
11615 (apply_subst_iterator): Likewise.
11616 (record_iterator_use): Likewise.
11617 (record_attribute_use): Likewise.
11618 (md_reader::record_potential_iterator_use): Likewise. Update calls
11619 to record_iterator_use and apply_iterator.
11620 (iterator_use): Replace ptr with x and index.
11621 (attribute_use): Likewise.
11622 (apply_attribute_uses): Update calls to apply_iterator.
11623 (apply_iterators): Likewise. Update initialization of iterator_use.
11624 (rtx_reader::read_rtx_code): Update calls to record_iterator_use
11625 and record_potential_iterator_use.
11626 (rtx_reader::read_rtx_operand): Likewise.
11628 2017-08-21 Richard Sandiford <richard.sandiford@linaro.org>
11629 Alan Hayward <alan.hayward@arm.com>
11630 David Sherwood <david.sherwood@arm.com>
11632 * varasm.c (const_rtx_hash_1): Don't hash in the mode of a
11635 2017-08-21 Richard Biener <rguenther@suse.de>
11637 PR middle-end/81884
11638 * tree-ssa-alias.c (stmt_kills_ref_p): Handle array accesses
11639 at struct end conservatively when comparing common bases.
11641 2017-08-21 Richard Biener <rguenther@suse.de>
11643 * tree-ssa-loop-im.c (struct lim_aux_data): Add ref index member.
11644 (mem_ref_in_stmt): Remove.
11645 (determine_max_movement): Use ref index to get at the reference.
11646 (invariantness_dom_walker::before_dom_children): Deal with
11647 lim data already initialized.
11648 (gather_mem_refs_stmt): Initialize lim data and record ref index.
11650 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
11652 * config/i386/i386.h (OPTION_MASK_ISA_ROUND): Remove.
11653 (TARGET_ISA_ROUND): Ditto.
11654 (TARGET_ROUND): Ditto.
11655 * config/i386/i386.c: Substitute TARGET_ROUND with TARGET_SSE4_1.
11656 * config/i386/i386.md: Ditto.
11657 * config/i386/sse.md: Ditto.
11658 * config/i386/i386-builtin.def: Substitute OPTION_MASK_ISA_ROUND
11659 with OPTION_MASK_ISA_SSE4_1.
11661 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
11664 * doc/extend.texi (x86 Built-in Functions): Correct the name of
11665 __builtin_ia32_lzcnt_u16.
11667 2017-08-18 Peter Bergner <bergner@vnet.ibm.com>
11670 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
11671 (rs6000_set_current_function): Rewrite function to use it.
11673 2017-08-18 H.J. Lu <hongjiu.lu@intel.com>
11676 * print-tree.c (print_node): Support DECL_WARN_IF_NOT_ALIGN
11677 and TYPE_WARN_IF_NOT_ALIGN.
11678 * stor-layout.c (do_type_align): Merge DECL_WARN_IF_NOT_ALIGN.
11679 (handle_warn_if_not_align): New.
11680 (place_union_field): Call handle_warn_if_not_align.
11681 (place_field): Call handle_warn_if_not_align.
11682 Copy TYPE_WARN_IF_NOT_ALIGN.
11683 (finish_builtin_struct): Copy TYPE_WARN_IF_NOT_ALIGN.
11684 (layout_type): Likewise.
11685 * tree-core.h (tree_type_common): Add warn_if_not_align. Set
11687 (tree_decl_common): Add warn_if_not_align.
11688 * tree.c (build_range_type_1): Copy TYPE_WARN_IF_NOT_ALIGN.
11689 * tree.h (TYPE_WARN_IF_NOT_ALIGN): New.
11690 (SET_TYPE_WARN_IF_NOT_ALIGN): Likewise.
11691 (DECL_WARN_IF_NOT_ALIGN): Likewise.
11692 (SET_DECL_WARN_IF_NOT_ALIGN): Likewise.
11693 * doc/extend.texi: Document warn_if_not_aligned attribute.
11694 * doc/invoke.texi: Document -Wif-not-aligned and -Wpacked-not-aligned.
11696 2017-08-17 Martin Liska <mliska@suse.cz>
11699 * tree-loop-distribution.c (ddrs_table): Change type to pointer type.
11700 (get_data_dependence): Use it as pointer type.
11701 (distribute_loop): Likewise.
11703 2017-08-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11705 * config/rs6000/altivec.md (UNSPEC_VMRGOW_DIRECT): New constant.
11706 (p8_vmrgew_v4sf_direct): Generalize to p8_vmrgew_<mode>_direct.
11707 (p8_vmrgow_<mode>_direct): New define_insn.
11708 * config/rs6000/rs6000.c (altivec_expand_vec_perm_const): Properly
11709 handle endianness for vmrgew and vmrgow permute patterns.
11711 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11713 * config/rs6000/altivec.md (VParity): Remove TARGET_VSX_TIMODE.
11714 * config/rs6000/rs6000-cpus.def: Remove comment.
11715 (ISA_2_7_MASKS_SERVER): Delete OPTION_MASK_VSX_TIMODE;
11716 (POWERPC_MASKS): Likewise.
11717 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Remove unneeded
11718 use of TARGET_VSX_TIMODE.
11719 (rs6000_setup_reg_addr_masks): Change TARGET_VSX_TIMODE to TARGET_VSX.
11720 (rs6000_init_hard_regno_mode_ok): Remove unneeded uses of
11721 TARGET_VSX_TIMODE. Change use of TARGET_VSX_TIMODE to TARGET_VSX.
11722 (rs6000_option_override_internal): Remove dead code.
11723 (rs6000_legitimize_address): Change TARGET_VSX_TIMODE to TARGET_VSX.
11724 (rs6000_legitimize_reload_address): Likewise.
11725 (rs6000_legitimate_address_p): Likewise.
11726 (rs6000_opt_masks): Delete "vsx-timode".
11727 (rs6000_disable_incompatible_switches): Remove mention of -mvsx-timode
11728 from function comment.
11729 * config/rs6000/rs6000.h (MASK_VSX_TIMODE): Delete.
11730 * config/rs6000/rs6000.md (FMOVE128_GPR): Remove TARGET_VSX_TIMODE.
11731 (V16QI, V8HI, V4SI, V4SF, V2DI, V2DF, V1TI): Remove useless empty
11733 * config/rs6000/rs6000.opt (mvsx-timode): Replace with stub.
11734 * config/rs6000/vector.md (VEC_IP): Remove TARGET_VSX_TIMODE.
11735 * config/rs6000/vsx.md (VSX_LE_128): Likewise.
11736 (VSX_TI): Likewise.
11738 (define_peephole2): Likewise.
11740 2017-08-17 Martin Sebor <msebor@redhat.com>
11743 * pretty-print.c (pp_format): Use strnlen in %.*s to avoid reading
11744 past the end of an array.
11745 (test_pp_format): Add test cases.
11747 2017-08-17 Richard Sandiford <richard.sandiford@linaro.org>
11749 * internal-fn.def (CLRSB, CLZ, CTZ, FFS, PARITY, POPCOUNT): Add
11750 missing ECF_NOTHROW flags.
11752 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
11755 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
11756 operands residing in integer registers.
11757 (*vsx_le_perm_load_<mode>): Likewise.
11758 (*vsx_le_perm_store_<mode>): Likewise.
11759 (define_peephole2): Add peepholes to optimize the above.
11761 2017-08-17 Marek Polacek <polacek@redhat.com>
11763 PR middle-end/81814
11764 * fold-const.c (operand_equal_for_comparison_p): Remove code that used
11765 to mimic what shorten_compare did. Change the return type to bool.
11766 (fold_cond_expr_with_comparison): Update call to
11767 operand_equal_for_comparison_p.
11768 (fold_ternary_loc): Likewise.
11770 2017-08-17 Jackson Woodruff <jackson.woodruff@arm.com>
11772 * aarch64-simd.md (mov<mode>): No longer force zero immediate into
11774 (*aarch64_simd_mov<mode>): Add new case for stp using zero immediate.
11776 2017-08-17 Richard Biener <rguenther@suse.de>
11778 * tree-ssa-structalias.c (solve_graph): When propagating
11779 to successors update the graphs succ edges and avoid duplicate work.
11781 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
11784 * config/i386/i386.c (ix86_option_override_internal): Save target
11785 specific options after ix86_stack_protector_guard_reg was changed.
11787 2017-08-17 Richard Biener <rguenther@suse.de>
11789 PR tree-optimization/81827
11790 * tree-ssa-structalias.c (struct variable_info): Add is_reg_var flag.
11791 (new_var_info): Initialize it conservatively.
11792 (get_call_vi): Mark register vars.
11793 (new_scalar_tmp_constraint_exp): Likewise.
11794 (handle_rhs_call): Likewise.
11795 (handle_const_call): Likewise.
11796 (create_function_info_for): Likewise.
11797 (solve_constraints): Sort varinfos to separate register from
11798 non-register vars to pack points-to solution bitmaps during
11801 2017-08-17 Marek Polacek <polacek@redhat.com>
11803 * gimplify.c (gimplify_adjust_omp_clauses): Compare with 0 instead of 1.
11805 2017-08-17 Richard Biener <rguenther@suse.de>
11807 * tree-vrp.c (vrp_int_const_binop): Do not set *overflow_p
11808 to true when overflow is undefined and we saturated the result.
11810 2017-08-17 Alan Modra <amodra@gmail.com>
11813 * config/rs6000/rs6000.c (rs6000_savres_strategy): Revert 2017-08-09.
11814 Don't use store multiple if only one reg needs saving.
11815 (interesting_frame_related_regno): New function.
11816 (rs6000_frame_related): Don't emit frame info for regs that
11818 (rs6000_emit_epilogue): Likewise.
11820 2017-08-16 Nathan Sidwell <nathan@acm.org>
11822 * tree-core.h (tree_type_non_common): Rename binfo to lang_1.
11823 * tree.h (TYPE_BINFO): Use type_non_common.maxval.
11824 (TYPE_LANG_SLOT_1): Use type_non_common.lang_1, for any type.
11825 * tree.c (free_lang_data_in_type): Use else-if chain. Always
11826 clear TYPE_LANG_1. Remove obsolete member-function stripping.
11827 (find_decls_types_r): Comment about TYPE_MAX_VALUES_RAW.
11828 (verify_type): Adjust for TYPE_BINFO move.
11829 * lto-streamer-out.c (DFS::DFS_write_tree_body): No need to
11830 process TYPE_BINFO directly.
11831 (hash_tree): Likewise.
11832 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
11834 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
11837 2017-08-16 David Malcolm <dmalcolm@redhat.com>
11839 * diagnostic-show-locus.c (colorizer::m_caret): Remove unused field.
11841 2017-08-16 Uros Bizjak <ubizjak@gmail.com>
11844 * config/i386/i386.md (*anddi_1_btr): Change predicates of
11845 operand 0 and operand 1 to nomimmediate_operand. Add "m" constraint.
11846 Add ix86_binary_operator_ok to insn constraint.
11847 (*iordi_1_bts): Ditto.
11848 (*xordi_1_btc): Ditto.
11849 (*btsq): Change predicate of operand 0 to nonimmediate_operand.
11850 Update corresponding peephole2 pattern.
11854 2017-08-16 Bin Cheng <bin.cheng@arm.com>
11856 PR tree-optimization/81832
11857 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): Don't
11858 copy loop header which has IFN_LOOP_DIST_ALIAS call.
11860 2017-08-16 Marek Polacek <polacek@redhat.com>
11863 * fold-const.c (fold_indirect_ref_1): Restore original behavior
11864 regarding size_zero_node.
11866 2017-08-16 Martin Liska <mliska@suse.cz>
11869 * config.gcc: Respect previously set extra_objs in case
11872 2017-08-16 Richard Sandiford <richard.sandiford@linaro.org>
11874 PR tree-optimization/81835
11875 * tree-vect-loop.c (vect_is_simple_reduction): Simply checks for
11876 the phi SSA_NAME. Check that the condition in a COND_EXPR does
11877 not depend on the phi.
11879 2017-08-16 Alan Modra <amodra@gmail.com>
11881 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Delete
11884 2017-08-16 Alan Modra <amodra@gmail.com>
11886 * config/rs6000/rs6000.c (rs6000_reg_live_or_pic_offset_p): Merge..
11887 (save_reg_p): ..into this. Update all callers.
11888 (first_reg_to_save): Simplify.
11890 2017-08-16 Alan Modra <amodra@gmail.com>
11892 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't restore
11895 2017-08-15 Joseph Myers <joseph@codesourcery.com>
11899 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
11900 constant count if that count is less than 32.
11902 2017-08-15 Nathan Sidwell <nathan@acm.org>
11904 * gcc.c (execute): Emit friendlier message if inferior is killed
11905 by an external cause.
11907 2017-08-15 Richard Biener <rguenther@suse.de>
11909 PR tree-optimization/81790
11910 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Handle both
11911 CONSTRUCTORs from simplifying and VN.
11913 2017-08-14 Martin Sebor <msebor@redhat.com>
11915 * builtin-attrs.def: Add comments.
11917 2017-08-14 Martin Sebor <msebor@redhat.com>
11920 * doc/extend.texi (attribute nonstring): Document new attribute.
11922 2017-08-14 Martin Sebor <msebor@redhat.com>
11925 * tree-diagnostic.c (default_tree_printer): Handle %G.
11926 * gimple-pretty-print.h (percent_G_format): Declare new function.
11927 * gimple-pretty-print.c (percent_G_format): Define.
11928 * tree-pretty-print.c (percent_K_format): Add argument.
11930 2017-08-14 Martin Sebor <msebor@redhat.com>
11932 PR translation/79998
11933 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call):
11934 Remove a stray space.
11936 2017-08-14 Uros Bizjak <ubizjak@gmail.com>
11939 * config/i386/i386.md (*anddi_1_btr): New insn_and_split pattern.
11940 (*iordi_1_bts): Ditto.
11941 (*xordi_1_btc): Ditto.
11943 2017-08-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11946 * config/rs6000/linux64.h (INVALID_64BIT): Use quoted strings.
11947 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
11949 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Use
11950 quoted strings, and make more translator-friendly.
11951 (darwin_rs6000_override_options): Likewise.
11952 (rs6000_option_override_internal): Likewise.
11953 (rs6000_return_in_memory): Fix overlong line.
11954 (init_cmulative_args): Use quoted strings, and make more
11955 translator-friendly.
11956 (rs6000_pass_by_reference): Fix overlong line.
11957 (def_builtin): Use quoted strings.
11958 (altivec_expand_predicate_builtin): Use quoted strings, and make
11959 more translator-friendly.
11960 (htm_expand_builtin): Use quoted strings.
11961 (cpu_expand_builtin): Use quoted strings, and make more
11962 translator-friendly.
11963 (altivec_expand_builtin): Likewise.
11964 (paired_expand_predicate_builtin): Likewise.
11965 (rs6000_invalid_builtin): Likewise.
11966 (builtin_function_type): Use quoted strings.
11967 (rs6000_expand_split_stack_prologue): Use quoted strings, and make
11968 more translator-friendly.
11969 (rs6000_trampoline_init): Likewise.
11970 (rs6000_handle_altivec_attribute): Likewise.
11971 (rs6000_inner_target_options): Use quoted strings.
11972 (rs6000_disable_incompatible_switches): Likewise.
11973 * config/rs6000/sysv4.h (SUBTARGET_OVERRIDE_OPTIONS): Use quoted
11974 strings, and make more translator-friendly.
11975 (SUBSUBTARGET_OVERRIDE_OPTIONS): Use quoted strings.
11977 2017-08-14 Bin Cheng <bin.cheng@arm.com>
11979 PR tree-optimization/81799
11980 * tree-loop-distribution.c (version_loop_by_alias_check): Force
11981 cond_expr to simple gimple operand.
11983 2017-08-14 Wilco Dijkstra <wdijkstr@arm.com>
11985 PR middle-end/46932
11986 * doc/sourcebuild.texi (autoincdec): Add autoincdec description.
11988 2017-08-14 Georg-Johann Lay <avr@gjlay.de>
11992 * config/avr/avr.opt (mgas-isr-prologues): New Var avr_gasisr_prologues.
11993 * config/avr/avr.md (gasisr, *gasisr): Use it instead of
11994 TARGET_GASISR_PROLOGUES.
11995 * config/avr/avr.c (avr_option_override): Same.
11996 (avr_pass_pre_proep::execute): Same.
11998 2017-08-13 H.J. Lu <hongjiu.lu@intel.com>
12001 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Replace
12002 frame pointer with stack pointer - UNITS_PER_WORD in debug insns.
12004 2017-08-13 Uros Bizjak <ubizjak@gmail.com>
12006 * config/i386/i386.md (*load_tp_<mode>): Redefine as
12007 define_insn_and_split. Split to a memory load from 0 in
12008 DEFAULT_TLS_SEG_REG address space. Merge with *load_tp_x32
12009 using PTR mode iterator.
12010 (*load_tp_x32_zext"): Redefine as define_insn_and_split.
12011 Split to a memory load from 0 in DEFAULT_TLS_SEG_REG address space.
12012 (*add_tp_<mode>): Redefine as define_insn_and_split.
12013 Split to an add with a memory load from 0 in DEFAULT_TLS_SEG_REG
12014 address space. Merge with *add_tp_x32 using PTR mode iterator.
12015 (*add_tp_x32_zext"): Redefine as define_insn_and_split.
12016 Split to an add with a memory load from 0 in
12017 DEFAULT_TLS_SEG_REG address space.
12019 2017-08-12 Andrew Pinski <apinski@cavium.com>
12021 * config/aarch64/aarch64-option-extensions.def (rdma):
12022 Fix feature string to what Linux prints out in /proc/cpuinfo.
12024 2017-08-12 Pierre-Marie de Rodat <derodat@adacore.com>
12027 * dwarf2out.c (modified_type_die): For C typedef types that have
12028 an ultimate origin, process the ultimate origin instead of the
12030 (gen_typedef_die): Assert that input DECLs have no ultimate
12032 (gen_type_die_with_usage): For typedef variants that have an
12033 ultimate origin, just call gen_decl_die on the original DECL.
12034 (process_scope_var): Avoid creating DIEs for local typedefs and
12035 concrete static variables.
12037 2017-08-12 Alan Modra <amodra@gmail.com>
12041 * config/rs6000/sysv4.h (STARTFILE_LINUX_SPEC): Upgrade to
12042 match gnu-user.h startfile.
12043 (ENDFILE_LINUX_SPEC): Similarly.
12045 2017-08-11 Thomas Schwinge <thomas@codesourcery.com>
12048 * config/nvptx/nvptx.c (nvptx_override_options_after_change):
12050 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Remove definition.
12052 2017-08-11 Tamar Christina <tamar.christina@arm.com>
12053 * config/aarch64/aarch64.md (mov<mode>): Change.
12054 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
12055 aarch64_reg_or_fp_float into aarch64_reg_or_fp_zero.
12056 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): Removed.
12058 2017-08-11 Eric Botcazou <ebotcazou@adacore.com>
12060 * tree-sra.c (build_access_from_expr_1): Use more precise diagnostics
12061 for storage order barriers.
12063 2017-08-11 Martin Liska <mliska@suse.cz>
12066 * tree-chkp.c (chkp_get_bounds_for_decl_addr): Do not instrument
12067 variables of void type.
12069 2017-08-11 Martin Liska <mliska@suse.cz>
12071 * asan.c (asan_protect_global): Replace ASM_OUTPUT_DEF with
12072 TARGET_SUPPORTS_ALIASES.
12073 * cgraph.c (cgraph_node::create_same_body_alias): Likewise.
12074 * ipa-visibility.c (can_replace_by_local_alias): Likewise.
12075 (optimize_weakref): Likewise.
12076 * symtab.c (symtab_node::noninterposable_alias): Likewise.
12077 * varpool.c (varpool_node::create_extra_name_alias): Likewise.
12078 * defaults.h: Introduce TARGET_SUPPORTS_ALIASES.
12080 2017-08-11 Martin Liska <mliska@suse.cz>
12083 * config/i386/i386.c (make_resolver_func): Do complete
12084 refactoring of the function.
12086 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12089 * config/i386/i386.opt (mstack-protector-guard-symbol=): New option
12090 * config/i386/i386.c (ix86_stack_protect_guard): Use
12091 ix86_stack_protect_guard_symbol_str to generate varible declaration.
12092 * doc/invoke.texi (x86 Options): Document
12093 -mstack-protector-guard-symbol= option.
12095 2017-08-10 Uros Bizjak <ubizjak@gmail.com>
12097 * config/i386/i386-protos.h (ix86_split_stack_guard): New prototype.
12098 * config/i386/i386.c (ix86_split_stack_guard): New function.
12099 (ix86_xpand_split_stack_prologue): Call ix86_split_stack_guard.
12100 (ix86_legitimate_address_p) <case UNSPEC_STACK_CHECK>: Remove.
12101 (i386_asm_output_addr_const_extra) <case UNSPEC_STACK_CHECK>: Ditto.
12102 (optput_pic_addr_const): Remove UNSPEC_STACK_CHECK handling.
12103 * config/i386/i386.md (unspec): Remove UNSPEC_STACK_CHECK.
12104 (split_stack_space_check): Call ix86_split_stack_guard.
12106 2017-08-10 Martin Sebor <msebor@redhat.com>
12108 * print-tree.c (print_node): Print location using the established
12110 Replace spaces with colons.
12111 (debug_raw, debug): Ditto.
12113 2017-08-10 Martin Sebor <msebor@redhat.com>
12116 * pretty-print.c (pp_format): Correct the handling of %s precision.
12118 2017-08-10 H.J. Lu <hongjiu.lu@intel.com>
12121 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12123 (ix86_finalize_stack_frame_flags): This. Also clear
12124 frame_pointer_needed if -fno-omit-frame-pointer is used without
12126 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12127 with ix86_finalize_stack_frame_flags.
12128 (ix86_expand_epilogue): Likewise.
12129 (ix86_expand_split_stack_prologue): Likewise.
12130 * doc/invoke.texi: Add a note for -fno-omit-frame-pointer.
12132 2017-08-10 Martin Liska <mliska@suse.cz>
12135 * c-attribs.c (handle_target_attribute):
12136 Report warning for an empty string argument of target attribute.
12138 2017-08-09 Jakub Jelinek <jakub@redhat.com>
12141 * omp-low.c (omp_copy_decl): Don't remap FORCED_LABEL or DECL_NONLOCAL
12143 * tree-cfg.c (move_stmt_op): Don't adjust DECL_CONTEXT of FORCED_LABEL
12144 or DECL_NONLOCAL labels.
12145 (move_stmt_r) <case GIMPLE_LABEL>: Adjust DECL_CONTEXT of FORCED_LABEL
12146 or DECL_NONLOCAL labels here.
12148 2017-08-09 Will Schmidt <will_schmidt@vnet.ibm.com>
12150 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add blurb
12151 to indicate when early gimple folding has been disabled.
12152 (rs6000_gimple_fold_builtin): Add debug content.
12153 (rs6000_invalid_builtin): Fix whitespace.
12154 (rs6000_expand_builtin): Fix whitespace.
12155 * config/rs6000/rs6000.opt: Add option for -mfold-gimple.
12157 2017-08-09 Segher Boessenkool <segher@kernel.crashing.org>
12160 * config/rs6000/rs6000.c (rs6000_savres_strategy): Don't use
12161 SAVE_MULTIPLE if not all the registers that saves, should be saved.
12163 2017-08-09 Jim Wilson <jim.wilson@linaro.org>
12165 * config/aarch64/aarch64-cores.def (falkor): Use falkor pipeline.
12166 (qdf24xx): Likewise.
12167 * config/aarch64/aarch64.md: Include falkor.md.
12168 * config/aarch64/falkor.md: New.
12170 2017-08-09 Marek Polacek <polacek@redhat.com>
12173 * diagnostic-core.h (emit_diagnostic_valist): Add declaration.
12174 * diagnostic.c (emit_diagnostic): Add a comment.
12175 (emit_diagnostic_valist): New function.
12177 2017-08-09 Marek Polacek <polacek@redhat.com>
12180 * input.c (make_location): New overload.
12181 * input.h (make_location): Declare.
12183 2017-08-08 Alan Modra <amodra@gmail.com>
12184 H.J. Lu <hongjiu.lu@intel.com>
12187 * gcc.c (NO_PIE_SPEC): Delete.
12188 (PIE_SPEC): Define as !no-pie/pie. Move static|shared|r
12190 (LINK_PIE_SPEC): ..to here.
12191 (LINK_COMMAND_SPEC): Support -no-pie.
12192 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Correct
12193 chain of crtbegin*.o selection, update for PIE_SPEC changes and
12195 (GNU_USER_TARGET_ENDFILE_SPEC): Similarly.
12196 * config/sol2.h (STARTFILE_CRTBEGIN_SPEC): Similarly.
12197 (ENDFILE_CRTEND_SPEC): Similarly.
12199 2017-08-08 Uros Bizjak <ubizjak@gmail.com>
12202 * config/i386/i386.opt (mstack-protector-guard-reg=): New option
12203 (mstack-protector-guard-offset=): Ditto.
12204 * config/i386/i386.c (ix86_option_override): Handle
12205 -mstack-protector-guard-reg= and -mstack-protector-guard-offset=
12207 (ix86_stack_protect_guard): Use ix86_stack_protect_guard_reg and
12208 ix86_stack_protect_guard_offset variables.
12209 (TARGET_STACK_PROTECT_GUARD): Always define.
12210 * doc/invoke.texi (x86 Options): Document -mstack-protector-guard-reg=
12211 and -mstack-protector-guard-offset= options.
12213 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12215 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): Handle
12216 boundary case for the last candidate.
12218 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12220 * doc/invoke.texi: Document -ftree-loop-distribution for O3.
12221 * opts.c (default_options_table): Add OPT_ftree_loop_distribution.
12223 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12225 PR middle-end/19706
12226 * config/aarch64/aarch64.md (xorsign<mode>3): New optabs.
12227 * config/aarch64/aarch64-builtins.c
12228 (aarch64_builtin_vectorized_function): Added CASE_CFN_XORSIGN.
12229 * config/aarch64/aarch64-simd-builtins.def: Added xorsign BINOP.
12230 * config/aarch64/aarch64-simd.md: Added xorsign<mode>3
12232 2017-08-08 Tamar Christina <tamar.christina@arm.com>
12233 Andrew Pinski <pinskia@gmail.com>
12235 PR middle-end/19706
12236 * internal-fn.def (XORSIGN): New.
12237 * optabs.def (xorsign_optab): New.
12238 * tree-ssa-math-opts.c (is_copysign_call_with_1): New.
12239 (convert_expand_mult_copysign): New.
12240 (pass_optimize_widening_mul::execute): Call
12241 convert_expand_mult_copysign.
12243 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12245 PR tree-optimization/81354
12246 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
12247 Insert on edges rather than explicitly creating landing pads.
12248 (analyze_candidates_and_replace): Commit edge inserts.
12250 2017-08-08 Richard Biener <rguenther@suse.de>
12252 PR middle-end/81719
12253 * tree-ssa-loop-niter.c: Include tree-dfa.h.
12254 (expand_simple_operations): Also look through ADDR_EXPRs with
12255 MEM_REF bases treating them as POINTER_PLUS_EXPR.
12257 2017-08-08 Richard Biener <rguenther@suse.de>
12259 PR tree-optimization/81723
12260 * tree-vect-slp.c (struct bst_traits): New hash traits.
12261 (bst_fail): New global.
12262 (vect_build_slp_tree_2): New worker, split out from ...
12263 (vect_build_slp_tree): ... this now wrapping it with using
12264 bst_fail set to cache SLP tree build fails. Properly handle
12266 (vect_analyze_slp_instance): Allocate and free bst_fail.
12268 2017-08-08 Martin Liska <mliska@suse.cz>
12271 * ipa-icf-gimple.c (func_checker::compare_cst_or_decl): Consider
12272 LABEL_DECLs that can be from a different function.
12274 2017-08-08 Bin Cheng <bin.cheng@arm.com>
12276 PR tree-optimization/81744
12277 * tree-predcom.c (prepare_finalizers_chain): Deep copy expr of
12278 loop's number of iterations.
12280 2017-08-08 Martin Liska <mliska@suse.cz>
12282 * asan.c: Include header files.
12283 * attribs.c (build_decl_attribute_variant): New function moved
12285 (build_type_attribute_qual_variant): Likewise.
12286 (cmp_attrib_identifiers): Likewise.
12287 (simple_cst_list_equal): Likewise.
12288 (omp_declare_simd_clauses_equal): Likewise.
12289 (attribute_value_equal): Likewise.
12290 (comp_type_attributes): Likewise.
12291 (build_type_attribute_variant): Likewise.
12292 (lookup_ident_attribute): Likewise.
12293 (remove_attribute): Likewise.
12294 (merge_attributes): Likewise.
12295 (merge_type_attributes): Likewise.
12296 (merge_decl_attributes): Likewise.
12297 (merge_dllimport_decl_attributes): Likewise.
12298 (handle_dll_attribute): Likewise.
12299 (attribute_list_equal): Likewise.
12300 (attribute_list_contained): Likewise.
12301 * attribs.h (lookup_attribute): New function moved from tree.[ch].
12302 (lookup_attribute_by_prefix): Likewise.
12303 * bb-reorder.c: Include header files.
12304 * builtins.c: Likewise.
12305 * calls.c: Likewise.
12306 * cfgexpand.c: Likewise.
12307 * cgraph.c: Likewise.
12308 * cgraphunit.c: Likewise.
12309 * convert.c: Likewise.
12310 * dwarf2out.c: Likewise.
12311 * final.c: Likewise.
12312 * fold-const.c: Likewise.
12313 * function.c: Likewise.
12314 * gimple-expr.c: Likewise.
12315 * gimple-fold.c: Likewise.
12316 * gimple-pretty-print.c: Likewise.
12317 * gimple.c: Likewise.
12318 * gimplify.c: Likewise.
12319 * hsa-common.c: Likewise.
12320 * hsa-gen.c: Likewise.
12321 * internal-fn.c: Likewise.
12322 * ipa-chkp.c: Likewise.
12323 * ipa-cp.c: Likewise.
12324 * ipa-devirt.c: Likewise.
12325 * ipa-fnsummary.c: Likewise.
12326 * ipa-inline.c: Likewise.
12327 * ipa-visibility.c: Likewise.
12329 * lto-cgraph.c: Likewise.
12330 * omp-expand.c: Likewise.
12331 * omp-general.c: Likewise.
12332 * omp-low.c: Likewise.
12333 * omp-offload.c: Likewise.
12334 * omp-simd-clone.c: Likewise.
12335 * opts-global.c: Likewise.
12336 * passes.c: Likewise.
12337 * predict.c: Likewise.
12338 * sancov.c: Likewise.
12339 * sanopt.c: Likewise.
12340 * symtab.c: Likewise.
12341 * toplev.c: Likewise.
12342 * trans-mem.c: Likewise.
12343 * tree-chkp.c: Likewise.
12344 * tree-eh.c: Likewise.
12345 * tree-into-ssa.c: Likewise.
12346 * tree-object-size.c: Likewise.
12347 * tree-parloops.c: Likewise.
12348 * tree-profile.c: Likewise.
12349 * tree-ssa-ccp.c: Likewise.
12350 * tree-ssa-live.c: Likewise.
12351 * tree-ssa-loop.c: Likewise.
12352 * tree-ssa-sccvn.c: Likewise.
12353 * tree-ssa-structalias.c: Likewise.
12354 * tree-ssa.c: Likewise.
12355 * tree-streamer-in.c: Likewise.
12356 * tree-vectorizer.c: Likewise.
12357 * tree-vrp.c: Likewise.
12358 * tsan.c: Likewise.
12359 * ubsan.c: Likewise.
12360 * varasm.c: Likewise.
12361 * varpool.c: Likewise.
12362 * tree.c: Remove functions moved to attribs.[ch].
12363 * tree.h: Likewise.
12364 * config/aarch64/aarch64.c: Add attrs.h header file.
12365 * config/alpha/alpha.c: Likewise.
12366 * config/arc/arc.c: Likewise.
12367 * config/arm/arm.c: Likewise.
12368 * config/avr/avr.c: Likewise.
12369 * config/bfin/bfin.c: Likewise.
12370 * config/c6x/c6x.c: Likewise.
12371 * config/cr16/cr16.c: Likewise.
12372 * config/cris/cris.c: Likewise.
12373 * config/darwin.c: Likewise.
12374 * config/epiphany/epiphany.c: Likewise.
12375 * config/fr30/fr30.c: Likewise.
12376 * config/frv/frv.c: Likewise.
12377 * config/ft32/ft32.c: Likewise.
12378 * config/h8300/h8300.c: Likewise.
12379 * config/i386/winnt.c: Likewise.
12380 * config/ia64/ia64.c: Likewise.
12381 * config/iq2000/iq2000.c: Likewise.
12382 * config/lm32/lm32.c: Likewise.
12383 * config/m32c/m32c.c: Likewise.
12384 * config/m32r/m32r.c: Likewise.
12385 * config/m68k/m68k.c: Likewise.
12386 * config/mcore/mcore.c: Likewise.
12387 * config/microblaze/microblaze.c: Likewise.
12388 * config/mips/mips.c: Likewise.
12389 * config/mmix/mmix.c: Likewise.
12390 * config/mn10300/mn10300.c: Likewise.
12391 * config/moxie/moxie.c: Likewise.
12392 * config/msp430/msp430.c: Likewise.
12393 * config/nds32/nds32-isr.c: Likewise.
12394 * config/nds32/nds32.c: Likewise.
12395 * config/nios2/nios2.c: Likewise.
12396 * config/nvptx/nvptx.c: Likewise.
12397 * config/pa/pa.c: Likewise.
12398 * config/pdp11/pdp11.c: Likewise.
12399 * config/powerpcspe/powerpcspe.c: Likewise.
12400 * config/riscv/riscv.c: Likewise.
12401 * config/rl78/rl78.c: Likewise.
12402 * config/rx/rx.c: Likewise.
12403 * config/s390/s390.c: Likewise.
12404 * config/sh/sh.c: Likewise.
12405 * config/sol2.c: Likewise.
12406 * config/sparc/sparc.c: Likewise.
12407 * config/spu/spu.c: Likewise.
12408 * config/stormy16/stormy16.c: Likewise.
12409 * config/tilegx/tilegx.c: Likewise.
12410 * config/tilepro/tilepro.c: Likewise.
12411 * config/v850/v850.c: Likewise.
12412 * config/vax/vax.c: Likewise.
12413 * config/visium/visium.c: Likewise.
12414 * config/xtensa/xtensa.c: Likewise.
12416 2017-08-07 Michael Meissner <meissner@linux.vnet.ibm.com>
12419 * config/rs6000/vsx.md (vsx_concat_<mode>, VSX_D): Cleanup
12420 constraints since the -mupper-regs-* switches have been
12422 (vsx_concat_<mode>_1): New combiner insns to recognize inserting
12423 into a vector from a double word element that was extracted from
12424 another vector, and eliminate extra XXPERMDI instructions.
12425 (vsx_concat_<mode>_2): Likewise.
12426 (vsx_concat_<mode>_3): Likewise.
12427 (vsx_set_<mode>, VSX_D): Rewrite vector set in terms of vector
12428 concat to allow optimizing inserts from previous extracts.
12430 2017-08-07 Uros Bizjak <ubizjak@gmail.com>
12432 * config/i386/i386.c (ix86_stack_protect_guard): Generate
12433 memory reference to a SSP offset in TLS address space.
12434 (ix86_print_operand) <case '@'>: Remove.
12435 (ix86_print_operand_punct_valid_p): Remove '@' code.
12436 * config/i386/i386.md (unspec): Remove UNSPEC_SP_TLS_SET and
12437 UNSPEC_SP_TLS_TEST.
12438 (stack_tls_protect_set_<mode>): Remove.
12439 (stack_protect_set): Do not call gen_stack_tls_protect_set_<mode>.
12440 (stack_tls_protect_test_<mode>): Remove.
12441 (stack_protect_test): Do not call gen_stack_tls_protect_test_<mode>.
12443 2017-08-07 Olivier Hainque <hainque@adacore.com>
12446 * config/vxworksae.h (VXWORKS_HAVE_TLS): Define.
12448 2017-08-07 Douglas Rupp <rupp@adacore.com>
12450 * Makefile.in (install-mkheaders): Fix typo, where the multi_dir
12451 variable was referenced as multidir in command.
12453 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12456 * gimplify.c (goa_stabilize_expr): Handle BIT_INSERT_EXPR and
12459 2017-08-07 Martin Liska <mliska@suse.cz>
12461 * config/m32c/m32c.c: Add include of stringpool.h and attribs.h.
12462 * config/rl78/rl78.c: Add include of attribs.h.
12463 * config/sh/sh.c: Likewise.
12464 * config/v850/v850.c: Likewise.
12466 2017-08-07 Tom de Vries <tom@codesourcery.com>
12468 PR middle-end/78266
12469 * omp-expand.c (expand_oacc_for): Ensure diff_type is large enough.
12471 2017-08-07 Martin Liska <mliska@suse.cz>
12473 * config/mips/mips.c: Include attribs.h.
12475 2017-08-07 Thomas Koenig <tkoenig@gcc.gnu.org>
12478 * doc/invoke.texi: Document change in behvaior for -Ofast for
12481 2017-08-07 Wilco Dijkstra <wdijkstr@arm.com>
12483 * config/aarch64/aarch64.c (aarch64_pushwb_single_reg):
12485 (aarch64_pop_regs): Likewise.
12486 (aarch64_gen_load_pair): Likewise.
12487 (aarch64_save_callee_saves): Likewise.
12488 (aarch64_restore_callee_saves): Likewise.
12490 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12492 * config/i386/i386.c: Revert the last change.
12494 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12497 * config/i386/i386.c (ix86_finalize_stack_realign_flags): Renamed
12499 (ix86_finalize_stack_frame_flags): This. Also clear
12500 frame_pointer_needed if -fno-omit-frame-pointer is used without
12502 (ix86_expand_prologue): Replace ix86_finalize_stack_realign_flags
12503 with ix86_finalize_stack_frame_flags.
12504 (ix86_expand_epilogue): Likewise.
12505 (ix86_expand_split_stack_prologue): Likewise.
12507 2017-08-07 H.J. Lu <hongjiu.lu@intel.com>
12510 * config/i386/i386.c (get_builtin_code_for_version): Set priority
12511 to P_AES for Westmere.
12513 2017-08-07 Jonathan Yong <10walls@gmail.com>
12515 * config/i386/mingw.opt (fset-stack-executable): Removed.
12516 * config/i386/cygming.opt (fset-stack-executable): Moved
12518 * config/i386/cygwin.h: Define CHECK_EXECUTE_STACK_ENABLED.
12520 2017-08-07 Segher Boessenkool <segher@kernel.crashing.org>
12522 * print-rtl.c (print_exp): Print NOT as "~" instead of as "!".
12524 2017-08-07 Marek Polacek <polacek@redhat.com>
12526 PR middle-end/81737
12527 * fold-const.c (fold_indirect_ref_1): Check type_domain.
12529 2017-08-07 Martin Liska <mliska@suse.cz>
12531 * attribs.h (canonicalize_attr_name): New function.
12532 (cmp_attribs): Move from c-format.c and adjusted.
12533 (is_attribute_p): Moved from tree.h.
12534 * tree-inline.c: Add new includes.
12535 * tree.c (cmp_attrib_identifiers): Use cmp_attribs.
12536 (private_is_attribute_p): Remove.
12537 (private_lookup_attribute): Likewise.
12538 (private_lookup_attribute_by_prefix): Simplify.
12539 (remove_attribute): Use is_attribute_p.
12540 * tree.h: Remove removed declarations.
12542 2017-08-07 Jakub Jelinek <jakub@redhat.com>
12544 PR middle-end/81698
12545 * stmt.c (emit_case_dispatch_table): Add DEFAULT_EDGE argument,
12546 instead of computing it in the function. Formatting fix.
12547 (expand_case): Don't rely on default_edge being the first edge,
12548 clear it if removing it, pass default_edge to
12549 emit_case_dispatch_table.
12550 (expand_sjlj_dispatch_table): Pass NULL as DEFAULT_EDGE, formatting
12553 2017-08-06 Uros Bizjak <ubizjak@gmail.com>
12555 * config/alpha/alpha.c (alpha_reorg): If trap is the last active
12556 insn in the function, emit NOP after the insn.
12558 2017-08-06 Tom de Vries <tom@codesourcery.com>
12560 * omp-expand.c (expand_oacc_for): Add missing edge probability for tile
12563 2017-08-06 Tom de Vries <tom@codesourcery.com>
12565 * omp-expand.c (expand_oacc_for): Add missing edge probability for chunk
12568 2017-08-04 Yury Gribov <tetra2005@gmail.com>
12570 PR tree-optimization/57371
12571 * match.pd: New pattern.
12573 2017-08-04 Marek Polacek <polacek@redhat.com>
12575 PR middle-end/81695
12576 * fold-const.c (fold_indirect_ref_1): For ((int *)&a + 4 -> a[1],
12577 perform the computation in offset_int.
12579 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12581 PR tree-optimization/81136
12582 * tree-vectorizer.h: Include tree-hash-traits.h.
12583 (vec_base_alignments): New typedef.
12584 (vec_info): Add a base_alignments field.
12585 (vect_record_base_alignments): Declare.
12586 * tree-data-ref.h (data_reference): Add an is_conditional_in_stmt
12588 (DR_IS_CONDITIONAL_IN_STMT): New macro.
12589 (create_data_ref): Add an is_conditional_in_stmt argument.
12590 * tree-data-ref.c (create_data_ref): Likewise. Use it to initialize
12591 the is_conditional_in_stmt field.
12592 (data_ref_loc): Add an is_conditional_in_stmt field.
12593 (get_references_in_stmt): Set the is_conditional_in_stmt field.
12594 (find_data_references_in_stmt): Update call to create_data_ref.
12595 (graphite_find_data_references_in_stmt): Likewise.
12596 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Likewise.
12597 * tree-vect-data-refs.c (vect_analyze_data_refs): Likewise.
12598 (vect_record_base_alignment): New function.
12599 (vect_record_base_alignments): Likewise.
12600 (vect_compute_data_ref_alignment): Adjust base_addr and aligned_to
12601 for nested statements even if we fail to compute a misalignment.
12602 Use pooled base alignments for unconditional references.
12603 (vect_find_same_alignment_drs): Compare base addresses instead
12605 (vect_analyze_data_refs_alignment): Call vect_record_base_alignments.
12606 * tree-vect-slp.c (vect_slp_analyze_bb_1): Likewise.
12608 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12610 * tree-vectorizer.h (vec_info): Add a constructor and destructor.
12611 Add an explicit name for the enum. Use auto_vec for slp_instances
12612 and grouped_stores.
12613 (_loop_vec_info): Add a constructor and destructor. Use auto_vec
12615 (_bb_vec_info): Add a constructor and destructor.
12616 (vinfo_for_stmt): Return NULL for uids of -1 as well.
12617 (destroy_loop_vec_info): Delete.
12618 (vect_destroy_datarefs): Likewise.
12619 * tree-vectorizer.c (vect_destroy_datarefs): Delete.
12620 (vec_info::vec_info): New function.
12621 (vec_info::~vec_info): Likewise.
12622 (vectorize_loops): Use delete instead of destroy_loop_vec_info.
12623 * tree-parloops.c (gather_scalar_reductions): Use delete instead of
12624 destroy_loop_vec_info.
12625 * tree-vect-loop.c (new_loop_vec_info): Replace with...
12626 (_loop_vec_info::_loop_vec_info): ...this.
12627 (destroy_loop_vec_info): Replace with...
12628 (_loop_vec_info::~_loop_vec_info): ...this. Unconditionally delete
12629 the stmt_vec_infos. Leave handling of vec_info information to its
12630 destructor. Remove explicit vector releases.
12631 (vect_analyze_loop_form): Use new instead of new_loop_vec_info.
12632 (vect_analyze_loop): Use delete instead of destroy_loop_vec_info.
12633 * tree-vect-slp.c (new_bb_vec_info): Replace with...
12634 (_bb_vec_info::_bb_vec_info): ...this. Don't reserve space in
12635 BB_VINFO_GROUPED_STORES or BB_VINFO_SLP_INSTANCES.
12636 (destroy_bb_vec_info): Replace with...
12637 (_bb_vec_info::~_bb_vec_info): ...this. Leave handling of vec_info
12638 information to its destructor.
12639 (vect_slp_analyze_bb_1): Use new and delete instead of
12640 new_bb_vec_info and destroy_bb_vec_info.
12641 (vect_slp_bb): Replace 2 calls to destroy_bb_vec_info with a
12644 2017-08-04 Richard Sandiford <richard.sandiford@linaro.org>
12646 * tree-data-ref.h (subscript): Add access_fn field.
12647 (data_dependence_relation): Add could_be_independent_p.
12648 (SUB_ACCESS_FN, DDR_COULD_BE_INDEPENDENT_P): New macros.
12649 (same_access_functions): Move to tree-data-ref.c.
12650 * tree-data-ref.c (ref_contains_union_access_p): New function.
12651 (access_fn_component_p): Likewise.
12652 (access_fn_components_comparable_p): Likewise.
12653 (dr_analyze_indices): Add a reference to access_fn_component_p.
12654 (dump_data_dependence_relation): Use SUB_ACCESS_FN instead of
12656 (constant_access_functions): Likewise.
12657 (add_other_self_distances): Likewise.
12658 (same_access_functions): Likewise. (Moved from tree-data-ref.h.)
12659 (initialize_data_dependence_relation): Use XCNEW and remove
12660 explicit zeroing of DDR_REVERSED_P. Look for a subsequence
12661 of access functions that have the same type. Allow the
12662 subsequence to end with different bases in some circumstances.
12663 Record the chosen access functions in SUB_ACCESS_FN.
12664 (build_classic_dist_vector_1): Replace ddr_a and ddr_b with
12665 a_index and b_index. Use SUB_ACCESS_FN instead of DR_ACCESS_FN.
12666 (subscript_dependence_tester_1): Likewise dra and drb.
12667 (build_classic_dist_vector): Update calls accordingly.
12668 (subscript_dependence_tester): Likewise.
12669 * tree-ssa-loop-prefetch.c (determine_loop_nest_reuse): Check
12670 DDR_COULD_BE_INDEPENDENT_P.
12671 * tree-vectorizer.h (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Test
12672 comp_alias_ddrs instead of may_alias_ddrs.
12673 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
12675 (vect_analyze_data_ref_dependence): Use it if
12676 DDR_COULD_BE_INDEPENDENT_P, but fall back to using the recorded
12677 distance vectors if that fails.
12678 (dependence_distance_ge_vf): New function.
12679 (vect_prune_runtime_alias_test_list): Use it. Don't clear
12680 LOOP_VINFO_MAY_ALIAS_DDRS.
12682 2017-08-04 Richard Biener <rguenther@suse.de>
12684 PR middle-end/81705
12685 * fold-const.c (fold_binary_loc): Properly restrict
12686 minus_var0 && minus_var1 case when associating undefined overflow
12689 2017-08-04 Tom de Vries <tom@codesourcery.com>
12691 * omp-simd-clone.c (simd_clone_adjust): Add missing edge probability.
12693 2017-08-03 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12695 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
12696 Don't start diagnostic messages with a capital letter.
12697 * config/rs6000/rs6000.c (rs6000_option_override_internal):
12699 (rs6000_invalid_builtin): Likewise.
12700 (rs6000_trampoline_init): Likewise.
12702 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12705 * bb-reorder.c (pass_partition_blocks::execute): Return TODO_df_finish
12706 after setting changeable df flags.
12708 2017-08-03 Richard Biener <rguenther@suse.de>
12710 * tree-ssa-reassoc.c (should_break_up_subtract): Also break
12711 up if the use is in USE - X.
12713 2017-08-03 Alexander Monakov <amonakov@ispras.ru>
12715 * toplev.c (dumpfile.h): New include.
12716 (internal_error_reentered): New static function. Use it...
12717 (internal_error_function): ...here to handle reentered internal_error.
12719 2017-08-03 Richard Biener <rguenther@suse.de>
12721 PR middle-end/81148
12722 * fold-const.c (split_tree): Add minus_var and minus_con
12723 arguments, remove unused loc arg. Never generate NEGATE_EXPRs
12724 here but always use minus_*.
12725 (associate_trees): Assert we never associate with MINUS_EXPR
12726 and NULL first operand. Do not recurse for PLUS_EXPR operands
12727 when associating as MINUS_EXPR either.
12728 (fold_binary_loc): Track minus_var and minus_con.
12730 2017-08-03 Tom de Vries <tom@codesourcery.com>
12733 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers): If
12734 ACCEL_COMPILER, apply finish_options on
12735 DECL_FUNCTION_SPECIFIC_OPTIMIZATION.
12737 2017-08-03 Tom de Vries <tom@codesourcery.com>
12740 * config/nvptx/nvptx.c (nvptx_option_override): Emit sorry if
12741 function_entry_patch_area_size > 0.
12743 2017-08-03 Jakub Jelinek <jakub@redhat.com>
12746 * calls.c (alloc_max_size): Use HOST_WIDE_INT_UC (10??)
12747 instead of 10??LU, perform unit multiplication in wide_int,
12748 don't change alloc_object_size_limit if the limit is larger
12751 PR tree-optimization/81655
12752 PR tree-optimization/81588
12753 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Handle also
12754 the case when ranges[i].low and high are 1 for unsigned type with
12757 PR middle-end/81052
12758 * omp-low.c (diagnose_sb_0): Handle flag_openmp_simd like flag_openmp.
12759 (pass_diagnose_omp_blocks::gate): Enable also for flag_openmp_simd.
12761 2017-08-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12763 * tree-vrp.h: Add include guard.
12765 2017-08-02 Uros Bizjak <ubizjak@gmail.com>
12768 * config/i386/i386.md (unspecv): Add UNSPECV_UD2.
12769 (ud2): New insn pattern.
12770 * config/i386/i386.c (ix86_expand_epilogue):
12771 For naked functions, generate ud2 instead of trap insn.
12773 2017-08-02 Marek Polacek <polacek@redhat.com>
12776 * alloc-pool.h (base_pool_allocator): Initialize m_elt_size.
12778 2017-08-02 Tom de Vries <tom@codesourcery.com>
12779 Cesar Philippidis <cesar@codesourcery.com>
12781 * config/nvptx/nvptx.c (nvptx_lockless_update, nvptx_lockfull_update):
12782 Add missing edge probabilities.
12784 2017-08-02 Tamar Christina <tamar.christina@arm.com>
12786 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
12787 Correct endianness.
12789 2017-08-02 Jakub Jelinek <jakub@redhat.com>
12791 PR middle-end/79499
12792 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
12793 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
12794 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
12795 prologue_seq sequences - if any.
12797 2017-08-02 Richard Biener <rguenther@suse.de>
12799 * tree-vect-stmts.c (vectorizable_store): Perform vector extracts
12800 via vectors if supported, integer extracts via punning if supported
12801 or otherwise vector extracts.
12803 2017-08-02 Richard Biener <rguenther@suse.de>
12805 * tree-ssa-pre.c (bitmap_insert_into_set_1): Remove and inline
12807 (bitmap_insert_into_set): ... this.
12809 2017-08-02 Richard Biener <rguenther@suse.de>
12811 PR tree-optimization/81633
12813 2015-08-17 Alan Hayward <alan.hayward@arm.com>
12815 PR tree-optimization/71752
12816 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands.
12818 2017-08-01 Daniel Santos <daniel.santos@pobox.com>
12820 * config/i386/i386.h (ix86_frame::outlined_save_offset): Remove field.
12821 (machine_function::call_ms2sysv_pad_out): Remove field.
12822 * config/i386/i386.c (xlogue_layout::get_stack_space_used): Modify.
12823 (ix86_compute_frame_layout): Likewise.
12825 2017-08-01 H.J. Lu <hongjiu.lu@intel.com>
12828 * config/i386/i386.c (ix86_set_func_type): Disallow naked
12829 attribute with interrupt attribute.
12831 2017-08-01 Andrew Pinski <apinski@cavium.com>
12833 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Check
12834 BIT_INSERT_EXPR's operand 1
12835 to see if the types precision matches.
12837 2017-08-01 Martin Liska <mliska@suse.cz>
12839 PR middle-end/70140
12840 * builtins.c (expand_builtin_memcpy_args): Remove.
12841 (expand_builtin_memcpy): Call newly added function
12842 expand_builtin_memory_copy_args.
12843 (expand_builtin_memcpy_with_bounds): Likewise.
12844 (expand_builtin_mempcpy): Remove last argument.
12845 (expand_builtin_mempcpy_with_bounds): Likewise.
12846 (expand_builtin_memory_copy_args): New function created from
12847 expand_builtin_mempcpy_args with small modifications.
12848 (expand_builtin_mempcpy_args): Remove.
12849 (expand_builtin_stpcpy): Remove unused argument.
12850 (expand_builtin): Likewise.
12851 (expand_builtin_with_bounds): Likewise.
12853 2017-08-01 Martin Liska <mliska@suse.cz>
12856 Make mempcpy more optimal (PR middle-end/70140).
12858 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12861 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
12862 __builtin_vec_cmpne verify both arguments are compatible vectors
12863 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
12864 verify arg1_type is a pointer or array type. For __builtin_vec_st,
12865 move computation of aligned to after checking the argument types.
12869 * config/rs6000/vsx.md (vextract_fp_from_shorth,
12870 vextract_fp_from_shortl): Add element mode after mode in gen_vec_init*
12873 2017-08-01 Jerome Lambourg <lambourg@adacore.com>
12874 Doug Rupp <rupp@adacore.com>
12875 Olivier Hainque <hainque@adacore.com>
12877 * config.gcc (arm-wrs-vxworks*): Rework to handle arm-wrs-vxworks7 as
12878 well as arm-wrs-vxworks. Update target_cpu_name from arm6 (arch v3) to
12880 * config/arm/vxworks.h (MAYBE_TARGET_BPABI_CPP_BUILTINS): New, helper
12881 for TARGET_OS_CPP_BUILTIN.
12882 (TARGET_OS_CPP_BUILTIN): Invoke MAYBE_TARGET_BPABI_CPP_BUILTINS(),
12883 refine CPU definitions for arm_arch5 and add those for arm_arch6 and
12885 (MAYBE_ASM_ABI_SPEC): New, helper for SUBTARGET_EXTRA_ASM_SPEC,
12886 passing required abi options to the assembler for EABI configurations.
12887 (EXTRA_CC1_SPEC): New macro, to help prevent the implicit production
12888 of .text.hot and .text.unlikely sections for kernel modules when
12889 using ARM style exceptions.
12890 (CC1_SPEC): Remove obsolete attempt at mimicking Diab toolchain
12891 options. Add EXTRA_CC1_SPEC.
12892 (VXWORKS_ENDIAN_SPEC): Adjust comment and remove handling of Diab
12894 (DWARF2_UNWIND_INFO): Redefine to handle the pre/post VxWorks 7
12896 (ARM_TARGET2_DWARF_FORMAT): Define.
12897 * config/arm/t-vxworks: Adjust multilib control to removal of the
12898 Diab command line options.
12900 2017-08-01 Martin Liska <mliska@suse.cz>
12902 PR gcov-profile/81561
12903 * gcov.c (unblock): Make unblocking safe as we need to preserve
12904 index correspondence of blocks and block_lists.
12906 2017-08-01 Richard Biener <rguenther@suse.de>
12908 PR tree-optimization/81181
12909 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
12910 (compute_antic): ... end of iteration here.
12912 2017-08-01 James Greenhalgh <james.greenhalgh@arm.com>
12914 * common.opt (ftree-vectorize): No longer set flag_tree_vectorize.
12915 (ftree-loop-vectorize): Set as EnabledBy ftree-vectorize.
12916 (ftree-slp-vectorize): Likewise.
12917 * omp-expand (expand_omp_simd): Remove flag_tree_vectorize, as it
12918 can no longer be set independent of flag_tree_loop_vectorize.
12919 * omp-general.c (emp_max_vf): Likewise.
12920 * opts.c (enable_fdo_optimizations): Remove references to
12921 flag_tree_vectorize, these are now implicit.
12922 (common_handle_option): Remove handling for OPT_ftree_vectorize,
12923 and leave it for the options machinery.
12925 2017-08-01 Martin Liska <mliska@suse.cz>
12927 PR middle-end/70140
12928 * builtins.c (expand_builtin_memcpy_args): Remove.
12929 (expand_builtin_memcpy): Call newly added function
12930 expand_builtin_memory_copy_args.
12931 (expand_builtin_memcpy_with_bounds): Likewise.
12932 (expand_builtin_mempcpy): Remove last argument.
12933 (expand_builtin_mempcpy_with_bounds): Likewise.
12934 (expand_builtin_memory_copy_args): New function created from
12935 expand_builtin_mempcpy_args with small modifications.
12936 (expand_builtin_mempcpy_args): Remove.
12937 (expand_builtin_stpcpy): Remove unused argument.
12938 (expand_builtin): Likewise.
12939 (expand_builtin_with_bounds): Likewise.
12941 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12944 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
12945 print "ds:" only for immediates in generic address space.
12947 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
12950 * config/i386/i386.c (ix86_funciton_naked): New prototype.
12951 (ix86_function_ok_for_sibcall): Return false for naked functions.
12953 2017-08-01 Richard Biener <rguenther@suse.de>
12955 * tree-ssa-pre.c (print_pre_expr): Handle NULL expr.
12956 (compute_antic): Seed worklist with exit block predecessors.
12957 * cfganal.c (dfs_find_deadend): For a cycle return the source
12958 of the edge closing it.
12960 2017-08-01 Tamar Christina <tamar.christina@arm.com>
12962 * config/aarch64/aarch64.c
12963 (aarch64_can_const_movi_rtx_p): Move 0 check.
12965 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12967 * tree.h (POINTER_TYPE_OVERFLOW_UNDEFINED): Delete.
12968 * fold-const.c (fold_comparison, fold_binary_loc): Delete use of
12970 * match.pd: Ditto in address comparison pattern.
12972 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12974 PR tree-optimization/81627
12975 * tree-predcom.c (prepare_finalizers): Always rewrite into loop
12976 closed ssa form for store-store chain.
12978 2017-08-01 Bin Cheng <bin.cheng@arm.com>
12980 PR tree-optimization/81620
12981 * tree-predcom.c (add_ref_to_chain): Don't set has_max_use_after
12982 for store-store chain.
12984 2017-08-01 Jakub Jelinek <jakub@redhat.com>
12986 PR tree-optimization/81588
12987 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
12988 ranges[i].in_p, invert comparison code ccode. For >/>=,
12989 swap rhs1 and rhs2 and comparison code unconditionally,
12990 for </<= don't do that. Don't swap rhs1/rhs2 again if
12991 ranges[i].in_p, instead invert comparison code ccode if
12992 opcode or oe->rank is BIT_IOR_EXPR.
12995 * optabs.def (vec_extract_optab, vec_init_optab): Change from
12996 a direct optab to conversion optab.
12997 * optabs.c (expand_vector_broadcast): Use convert_optab_handler
12998 with GET_MODE_INNER as last argument instead of optab_handler.
12999 * expmed.c (extract_bit_field_1): Likewise. Use vector from
13000 vector extraction if possible and optab is available.
13001 * expr.c (store_constructor): Use convert_optab_handler instead
13002 of optab_handler. Use vector initialization from smaller
13003 vectors if possible and optab is available.
13004 * tree-vect-stmts.c (vectorizable_load): Likewise.
13005 * doc/md.texi (vec_extract, vec_init): Document that the optabs
13006 now have two modes.
13007 * config/i386/i386.c (ix86_expand_vector_init): Handle expansion
13008 of vec_init from half-sized vectors with the same element mode.
13009 * config/i386/sse.md (ssehalfvecmode): Add V4TI case.
13010 (ssehalfvecmodelower, ssescalarmodelower): New mode attributes.
13011 (reduc_plus_scal_v8df, reduc_plus_scal_v4df, reduc_plus_scal_v2df,
13012 reduc_plus_scal_v16sf, reduc_plus_scal_v8sf, reduc_plus_scal_v4sf,
13013 reduc_<code>_scal_<mode>, reduc_umin_scal_v8hi): Add element mode
13014 after mode in gen_vec_extract* calls.
13015 (vec_extract<mode>): Renamed to ...
13016 (vec_extract<mode><ssescalarmodelower>): ... this.
13017 (vec_extract<mode><ssehalfvecmodelower>): New expander.
13018 (rotl<mode>3, rotr<mode>3, <shift_insn><mode>3, ashrv2di3): Add
13019 element mode after mode in gen_vec_init* calls.
13020 (VEC_INIT_HALF_MODE): New mode iterator.
13021 (vec_init<mode>): Renamed to ...
13022 (vec_init<mode><ssescalarmodelower>): ... this.
13023 (vec_init<mode><ssehalfvecmodelower>): New expander.
13024 * config/i386/mmx.md (vec_extractv2sf): Renamed to ...
13025 (vec_extractv2sfsf): ... this.
13026 (vec_initv2sf): Renamed to ...
13027 (vec_initv2sfsf): ... this.
13028 (vec_extractv2si): Renamed to ...
13029 (vec_extractv2sisi): ... this.
13030 (vec_initv2si): Renamed to ...
13031 (vec_initv2sisi): ... this.
13032 (vec_extractv4hi): Renamed to ...
13033 (vec_extractv4hihi): ... this.
13034 (vec_initv4hi): Renamed to ...
13035 (vec_initv4hihi): ... this.
13036 (vec_extractv8qi): Renamed to ...
13037 (vec_extractv8qiqi): ... this.
13038 (vec_initv8qi): Renamed to ...
13039 (vec_initv8qiqi): ... this.
13040 * config/rs6000/vector.md (VEC_base_l): New mode attribute.
13041 (vec_init<mode>): Renamed to ...
13042 (vec_init<mode><VEC_base_l>): ... this.
13043 (vec_extract<mode>): Renamed to ...
13044 (vec_extract<mode><VEC_base_l>): ... this.
13045 * config/rs6000/paired.md (vec_initv2sf): Renamed to ...
13046 (vec_initv2sfsf): ... this.
13047 * config/rs6000/altivec.md (splitter, altivec_copysign_v4sf3,
13048 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13049 vec_unpacku_lo_v8hi, mulv16qi3, altivec_vreve<mode>2): Add
13050 element mode after mode in gen_vec_init* calls.
13051 * config/aarch64/aarch64-simd.md (vec_init<mode>): Renamed to ...
13052 (vec_init<mode><Vel>): ... this.
13053 (vec_extract<mode>): Renamed to ...
13054 (vec_extract<mode><Vel>): ... this.
13055 * config/aarch64/iterators.md (Vel): New mode attribute.
13056 * config/s390/s390.c (s390_expand_vec_strlen, s390_expand_vec_movstr):
13057 Add element mode after mode in gen_vec_extract* calls.
13058 * config/s390/vector.md (non_vec_l): New mode attribute.
13059 (vec_extract<mode>): Renamed to ...
13060 (vec_extract<mode><non_vec_l>): ... this.
13061 (vec_init<mode>): Renamed to ...
13062 (vec_init<mode><non_vec_l>): ... this.
13063 * config/s390/s390-builtins.def (s390_vlgvb, s390_vlgvh, s390_vlgvf,
13064 s390_vlgvf_flt, s390_vlgvg, s390_vlgvg_dbl): Add element mode after
13066 * config/arm/iterators.md (V_elem_l): New mode attribute.
13067 * config/arm/neon.md (vec_extract<mode>): Renamed to ...
13068 (vec_extract<mode><V_elem_l>): ... this.
13069 (vec_extractv2di): Renamed to ...
13070 (vec_extractv2didi): ... this.
13071 (vec_init<mode>): Renamed to ...
13072 (vec_init<mode><V_elem_l>): ... this.
13073 (reduc_plus_scal_<mode>, reduc_plus_scal_v2di, reduc_smin_scal_<mode>,
13074 reduc_smax_scal_<mode>, reduc_umin_scal_<mode>,
13075 reduc_umax_scal_<mode>, neon_vget_lane<mode>, neon_vget_laneu<mode>):
13076 Add element mode after gen_vec_extract* calls.
13077 * config/mips/mips-msa.md (vec_init<mode>): Renamed to ...
13078 (vec_init<mode><unitmode>): ... this.
13079 (vec_extract<mode>): Renamed to ...
13080 (vec_extract<mode><unitmode>): ... this.
13081 * config/mips/loongson.md (vec_init<mode>): Renamed to ...
13082 (vec_init<mode><unitmode>): ... this.
13083 * config/mips/mips-ps-3d.md (vec_initv2sf): Renamed to ...
13084 (vec_initv2sfsf): ... this.
13085 (vec_extractv2sf): Renamed to ...
13086 (vec_extractv2sfsf): ... this.
13087 (reduc_plus_scal_v2sf, reduc_smin_scal_v2sf, reduc_smax_scal_v2sf):
13088 Add element mode after gen_vec_extract* calls.
13089 * config/mips/mips.md (unitmode): New mode iterator.
13090 * config/spu/spu.c (spu_expand_prologue, spu_allocate_stack,
13091 spu_builtin_extract): Add element mode after gen_vec_extract* calls.
13092 * config/spu/spu.md (inner_l): New mode attribute.
13093 (vec_init<mode>): Renamed to ...
13094 (vec_init<mode><inner_l>): ... this.
13095 (vec_extract<mode>): Renamed to ...
13096 (vec_extract<mode><inner_l>): ... this.
13097 * config/sparc/sparc.md (veltmode): New mode iterator.
13098 (vec_init<VMALL:mode>): Renamed to ...
13099 (vec_init<VMALL:mode><VMALL:veltmode>): ... this.
13100 * config/ia64/vect.md (vec_initv2si): Renamed to ...
13101 (vec_initv2sisi): ... this.
13102 (vec_initv2sf): Renamed to ...
13103 (vec_initv2sfsf): ... this.
13104 (vec_extractv2sf): Renamed to ...
13105 (vec_extractv2sfsf): ... this.
13106 * config/powerpcspe/vector.md (VEC_base_l): New mode attribute.
13107 (vec_init<mode>): Renamed to ...
13108 (vec_init<mode><VEC_base_l>): ... this.
13109 (vec_extract<mode>): Renamed to ...
13110 (vec_extract<mode><VEC_base_l>): ... this.
13111 * config/powerpcspe/paired.md (vec_initv2sf): Renamed to ...
13112 (vec_initv2sfsf): ... this.
13113 * config/powerpcspe/altivec.md (splitter, altivec_copysign_v4sf3,
13114 vec_unpacku_hi_v16qi, vec_unpacku_hi_v8hi, vec_unpacku_lo_v16qi,
13115 vec_unpacku_lo_v8hi, mulv16qi3): Add element mode after mode in
13116 gen_vec_init* calls.
13118 2017-08-01 Richard Biener <rguenther@suse.de>
13120 PR tree-optimization/81297
13121 * tree-vrp.c (get_single_symbol): Remove assert, instead drop
13122 TREE_OVERFLOW from INTEGER_CSTs.
13124 2017-07-31 Segher Boessenkool <segher@kernel.crashing.org>
13126 * config/rs6000/rs6000.c (enum rs6000_reg_type): Delete trailing comma.
13128 2017-07-31 Carl Love <cel@us.ibm.com>
13130 * config/rs6000/rs6000-c: Add support for built-in functions
13131 vector signed char vec_xl_be (signed long long, signed char *);
13132 vector unsigned char vec_xl_be (signed long long, unsigned char *);
13133 vector signed int vec_xl_be (signed long long, signed int *);
13134 vector unsigned int vec_xl_be (signed long long, unsigned int *);
13135 vector signed long long vec_xl_be (signed long long, signed long long *);
13136 vector unsigned long long vec_xl_be (signed long long, unsigned long long *);
13137 vector signed short vec_xl_be (signed long long, signed short *);
13138 vector unsigned short vec_xl_be (signed long long, unsigned short *);
13139 vector double vec_xl_be (signed long long, double *);
13140 vector float vec_xl_be (signed long long, float *);
13141 * config/rs6000/altivec.h (vec_xl_be): Add #define.
13142 * config/rs6000/rs6000-builtin.def (XL_BE_V16QI, XL_BE_V8HI, XL_BE_V4SI)
13143 XL_BE_V2DI, XL_BE_V4SF, XL_BE_V2DF, XL_BE): Add definitions
13145 * config/rs6000/rs6000.c (altivec_expand_xl_be_builtin): Add function.
13146 (altivec_expand_builtin): Add switch statement to call
13147 altivec_expand_xl_be for each builtin.
13148 (altivec_init_builtins): Add def_builtin for _builtin_vsx_le_be_v8hi,
13149 __builtin_vsx_le_be_v4si, __builtin_vsx_le_be_v2di,
13150 __builtin_vsx_le_be_v4sf, __builtin_vsx_le_be_v2df,
13151 __builtin_vsx_le_be_v16qi.
13152 * doc/extend.texi: Update the built-in documentation file for the
13153 new built-in functions.
13155 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13158 * config/i386/i386.c (ix86_allocate_stack_slots_for_args):
13160 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
13162 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13164 * config.gcc: Add z14.
13165 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
13166 CPU model numbers for z13s and z14.
13167 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
13169 * config/s390/s390-opts.h (enum processor_type): Rename
13170 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13171 * config/s390/s390.c (processor_table): Add field for CPU name to
13172 be passed to Binutils.
13173 (s390_asm_output_machine_for_arch): Use the new field in
13174 processor_table for Binutils.
13175 (s390_expand_builtin): Replace arch12 with z14.
13176 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
13177 (s390_get_sched_attrmask): Likewise.
13178 (s390_get_unit_mask): Likewise.
13179 * config/s390/s390.opt: Add z14 to processor_type enum.
13181 2017-07-31 Martin Jambor <mjambor@suse.cz>
13184 * ipa-fnsummary.c (ipa_fn_summary_generate): Set versionable
13185 regardless of optimization level.
13187 2017-07-31 Jan Hubicka <hubicka@ucw.cz>
13188 Martin Liska <mliska@suse.cz>
13190 * predict.def: Remove old comment and adjust probability.
13191 * gimplify.c (should_warn_for_implicit_fallthrough): Ignore
13192 PREDICT statements.
13194 2017-07-31 Uros Bizjak <ubizjak@gmail.com>
13197 * config/i386/i386.c (ix86_function_naked): New function.
13198 (ix86_can_use_return_insn_p): Return false for naked functions.
13199 (ix86_expand_prologue): Skip prologue for naked functions.
13200 (ix86_expand_epilogue): Skip epilogue for naked functions
13201 and emit trap instruction.
13202 (ix86_warn_func_return): New function.
13203 (ix86_attribute_table): Add "naked" attribute specification.
13204 (TARGET_WARN_FUNC_RETURN): Define.
13205 * doc/extend.texi (x86 Function Attributes) <naked>: Document it.
13207 2017-07-31 Martin Liska <mliska@suse.cz>
13209 * gimple-pretty-print.c (dump_gimple_label): Never dump BB info.
13210 (dump_gimple_bb_header): Always dump BB info.
13211 (pp_cfg_jump): Do not append info about BB when dumping a jump.
13213 2017-07-31 Martin Liska <mliska@suse.cz>
13216 * convert.c (convert_to_integer_1): Guard condition with flag_sanitize_p
13217 also with current_function_decl non-null equality.
13219 2017-07-31 Jakub Jelinek <jakub@redhat.com>
13222 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
13223 change type to the element type, instead add eltype variable and
13224 use it where we are interested in the element type.
13226 PR tree-optimization/81603
13227 * ipa-polymorphic-call.c
13228 (ipa_polymorphic_call_context::ipa_polymorphic_call_context): Perform
13229 offset arithmetic in offset_int, bail out if the resulting bit offset
13230 doesn't fit into shwi.
13232 2017-07-31 Martin Liska <mliska@suse.cz>
13234 * gimplify.c (mostly_copy_tree_r): Remove Java specific hunk.
13235 (gimplify_save_expr): Fix comment.
13237 2017-07-30 H.J. Lu <hongjiu.lu@intel.com>
13240 * config/i386/i386.c (ix86_function_arg): Update arguments for
13242 (ix86_compute_frame_layout): Set the initial stack offset to
13243 INCOMING_FRAME_SP_OFFSET. Update red-zone offset with
13244 INCOMING_FRAME_SP_OFFSET.
13245 (ix86_expand_epilogue): Don't pop the 'ERROR_CODE' off the
13246 stack before exception handler returns.
13247 * config/i386/i386.h (INCOMING_FRAME_SP_OFFSET): Add the
13248 the 'ERROR_CODE' for exception handler.
13250 2017-07-30 Uros Bizjak <ubizjak@gmail.com>
13252 * config/i386/i386.h (ASM_PRINTF_EXTENSIONS): New macro.
13253 (ASM_OUTPUT_REG_PUSH): Rewrite with new operand modifiers.
13254 (ASM_OUTPUT_REG_POP): Ditto.
13255 * config/i386/i386.c (ix86_asm_output_function_label): Use fputs
13256 instead of asm_fprintf to output pure string.
13258 2017-07-29 Jakub Jelinek <jakub@redhat.com>
13260 * debug.h (struct gcc_debug_hooks): Add IMPLICIT argument
13261 to imported_module_or_decl hook.
13262 (debug_nothing_tree_tree_tree_bool): Remove.
13263 (debug_nothing_tree_tree_tree_bool_bool): New declaration.
13264 * debug.c (do_nothing_debug_hooks): Use
13265 debug_nothing_tree_tree_tree_bool_bool instead of
13266 debug_nothing_tree_tree_tree_bool.
13267 * vmsdbgout.c (vmsdbg_debug_hooks): Likewise.
13268 * dbxout.c (dbx_debug_hooks, xcoff_debug_hooks): Likewise.
13269 * sdbout.c (sdb_debug_hooks): Likewise.
13270 * dwarf2out.c (dwarf2_lineno_debug_hooks): Likewise.
13271 (gen_namespace_die): Add DW_AT_export_symbols attribute if
13273 (dwarf2out_imported_module_or_decl): Add IMPLICIT argument,
13274 if true, -gdwarf-5 and decl will have DW_AT_export_symbols
13275 attribute, don't add anything.
13277 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13279 * fold-const.c (fold_build1_stat_loc): Remove _stat from name.
13280 (fold_build2_stat_loc): Likewise.
13281 (fold_build3_stat_loc): Likewise.
13282 * fold-const.h (fold_build1, fold_build2, fold_build3): Adjust.
13283 (fold_build1_loc): Remove macro.
13284 (fold_build2_loc): Likewise.
13285 (fold_build3_loc): Likewise.
13287 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13289 * gimple.c (gimple_build_debug_bind_stat): Remove _stat from name.
13290 (gimple_build_debug_bind_source_stat): Likewise.
13291 * gimple.h (gimple_build_debug_bind): Remove macro.
13292 (gimple_build_debug_bind_source): Likewise.
13294 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13296 * bitmap.c (bitmap_alloc): Adjust.
13297 (bitmap_gc_alloc): Likewise.
13298 * bitmap.h (bitmap_initialize_stat): Remove _stat from name.
13300 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13302 * bitmap.c (bitmap_obstack_alloc_stat): Rename to bitmap_alloc.
13303 (bitmap_gc_alloc_stat): Rename to bitmap_gc_alloc.
13304 * bitmap.h (bitmap_obstack_alloc_stat): Adjust prototype.
13305 (bitmap_gc_alloc_stat): Likewise.
13306 (BITMAP_ALLOC, BITMAP_GGC_ALLOC): Adjust.
13308 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13310 * rtl.c (shallow_copy_rtx_stat): Remove _stat from name.
13311 * rtl.h (shallow_copy_rtx): Remove macro.
13313 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13315 * emit-rtl.c (gen_raw_REG): Adjust.
13316 * gengenrtl.c (gendef): Likewise.
13317 * rtl.c (rtx_alloc_stat): Remove _stat from name.
13318 * rtl.h (rtx_alloc): Remove macro.
13320 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13322 * tree.c (build_tree_list_vec_stat): Remove _stat from name.
13323 (build_tree_list_stat): Likewise.
13324 * tree.h (build_tree_list): Remove macro.
13325 (build_tree_list_vec): Likewise.
13327 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13329 * tree.c (make_vector_stat): Remove _stat from name.
13330 (build_vector_stat): Likewise.
13331 * tree.h (make_vector_stat): Remove macro.
13332 (build_vector_stat): Likewise.
13334 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13336 * tree.h (build_var_debug_value): Remove prototype.
13338 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13340 * tree.c (tree_cons_stat): Remove _stat from name.
13341 * tree.h (tree_cons): Remove macro.
13343 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13345 * tree.c (build_vl_exp_stat): Remove _stat from name.
13346 * tree.h (build_vl_exp): Remove macro.
13348 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13350 * tree.c (build_decl_stat): Remove _stat from name.
13351 * tree.h (build_decl): Remove macro.
13353 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13355 * gimple.c (gimple_build_with_ops_stat): Adjust.
13356 (gimple_alloc_stat): Remove _stat from name.
13357 * gimple.h (gimple_alloc): Remove macro.
13359 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13361 * tree.c (make_tree_vec_stat): Remove _stat from name.
13362 (grow_tree_vec_stat): Likewise.
13363 * tree.h (make_tree_vec_stat): Adjust prototype.
13364 (grow_tree_vec_stat): Likewise.
13365 (make_tree_vec): Remove macro.
13366 (grow_tree_vec): Likewise.
13368 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13370 * fold-const.c (fold_build1_stat_loc): Adjust.
13371 (fold_build2_stat_loc): Likewise.
13372 (fold_build3_stat_loc): Likewise.
13373 * tree.c (build0_stat): Remove _stat from name.
13374 (build1_stat): Likewise.
13375 (build2_stat): Likewise.
13376 (build3_stat): Likewise.
13377 (build4_stat): Likewise.
13378 (build5_stat): Likewise.
13379 * tree.h (build1_loc): Remove macro, and rename _stat function
13381 (build2_loc): Likewise.
13382 (build3_loc): Likewise.
13383 (build4_loc): Likewise.
13384 (build5_loc): Likewise.
13386 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13388 * tree.c (make_int_cst_stat): Remove _stat from name.
13389 * tree.h (make_int_cst_stat): Adjust prototype.
13390 (make_int_cst): Remove macro.
13392 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13394 * tree.c (make_tre_binfo_stat): Remove _stat from name.
13395 * tree.h (make_tree_binfo_stat): Adjust prototype.
13396 (make_tree_binfo): Remove.
13398 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13400 * tree.c (copy_node_stat): Rename to copy_node.
13401 (build_distinct_type_copy): Adjust.
13402 * tree.h (copy_node_stat): Adjust prototype.
13403 (copy_node): Remove macro.
13405 2017-07-28 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
13407 * tree.c (make_node_stat): rename to make_node.
13408 (build_tree_list_stat): Adjust.
13409 (build0_stat): Likewise.
13410 (build2_stat): Likewise.
13411 (build3_stat): Likewise.
13412 (build4_stat): Likewise.
13413 (build5_stat): Likewise.
13414 (build_decl_stat): Likewise.
13415 * tree.h (make_node_stat): Adjust prototype.
13416 (make_node): remove macro.
13418 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
13420 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
13421 (PPC_FEATURE2_SCV): Likewise.
13422 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
13424 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13426 * config/aarch64/aarch64.c
13427 (aarch64_internal_mov_immediate): Add new special pattern.
13428 * config/aarch64/aarch64.md (*movdi_aarch64):
13429 Add reg/32bit const mov case.
13431 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13432 Richard Sandiford <richard.sandiford@linaro.org>
13434 * config/aarch64/aarch64.md (mov<mode>): Generalize.
13435 (*movhf_aarch64, *movsf_aarch64, *movdf_aarch64):
13436 Add integer and movi cases.
13437 (movi-split-hf-df-sf split, fp16): New.
13438 (enabled): Added TARGET_FP_F16INST.
13439 * config/aarch64/iterators.md (GPF_HF): New.
13440 * config/aarch64/predicates.md (aarch64_reg_or_fp_float): New.
13442 2017-07-28 Tamar Christina <tamar.christina@arm.com>
13444 * config/aarch64/aarch64.c
13445 (aarch64_simd_container_mode): Add prototype.
13446 (aarch64_expand_mov_immediate): Add HI support.
13447 (aarch64_reinterpret_float_as_int, aarch64_float_const_rtx_p: New.
13448 (aarch64_can_const_movi_rtx_p): New.
13449 (aarch64_preferred_reload_class):
13450 Remove restrictions of using FP registers for certain SIMD operations.
13451 (aarch64_rtx_costs): Added new cost for CONST_DOUBLE moves.
13452 (aarch64_valid_floating_const): Add integer move validation.
13453 (aarch64_simd_imm_scalar_p): Remove.
13454 (aarch64_output_scalar_simd_mov_immediate): Generalize function.
13455 (aarch64_legitimate_constant_p): Expand list of supported cases.
13456 * config/aarch64/aarch64-protos.h
13457 (aarch64_float_const_rtx_p, aarch64_can_const_movi_rtx_p): New.
13458 (aarch64_reinterpret_float_as_int): New.
13459 (aarch64_simd_imm_scalar_p): Remove.
13460 * config/aarch64/constraints.md (Uvi): New.
13461 (Dd): Split into Ds and new Dd.
13462 * config/aarch64/aarch64.md (*movsi_aarch64):
13464 (*movdi_aarch64): Add SIMD mov case.
13466 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13468 * tree-predcom.c: (struct chain): Handle store-store chain in which
13469 stores for elimination only store loop invariant values.
13470 (execute_pred_commoning_chain): Ditto.
13471 (prepare_initializers_chain_store_elim): Ditto.
13472 (prepare_finalizers): Ditto.
13473 (is_inv_store_elimination_chain): New function.
13474 (initialize_root_vars_store_elim_1): New function.
13476 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13478 * tree-predcom.c: Revise general description of the pass.
13479 (enum chain_type): New enum type for store elimination.
13480 (struct chain): New field supporting store elimination.
13481 (struct component): Ditto.
13482 (dump_chain): Dump store-stores chain.
13483 (release_chain): Release resources.
13484 (split_data_refs_to_components): Compute and create component
13485 contains only stores for elimination.
13486 (get_chain_last_ref_at): New function.
13487 (make_invariant_chain): Initialization.
13488 (make_rooted_chain): Specify chain type in parameter and record it.
13489 (add_looparound_copies): Skip for store-stores chain.
13490 (determine_roots_comp): Compute type of chain and pass it to
13492 (initialize_root_vars_store_elim_2): New function.
13493 (finalize_eliminated_stores): New function.
13494 (remove_stmt): Handle store for elimination.
13495 (execute_pred_commoning_chain): Execute predictive commoning on
13496 store-store chains.
13497 (determine_unroll_factor): Skip unroll for store-stores chain.
13498 (prepare_initializers_chain_store_elim): New function.
13499 (prepare_initializers_chain): Hanlde store-store chain.
13500 (prepare_finalizers_chain, prepare_finalizers): New function.
13501 (tree_predictive_commoning_loop): Return integer value indicating
13502 if loop is unrolled or lcssa form is corrupted.
13503 (tree_predictive_commoning): Rewrite for lcssa form if necessary.
13505 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13507 * tree-predcom.c (initialize_root): Delete.
13508 (execute_pred_commoning_chain): Initialize root vars and replace
13509 reference of non-combined chain directly, rather than call above
13512 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13514 * tree-predcom.c (ref_at_iteration): Add parameter NITERS. Compute
13515 memory reference to DR at (NITERS + ITERS)-th iteration of loop.
13517 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13519 * tree-predcom.c (struct chain): New field init_seq.
13520 (release_chain): Release init_seq.
13521 (prepare_initializers_chain): Record intialization stmts in above
13523 (insert_init_seqs): New function.
13524 (tree_predictive_commoning_loop): Call insert_init_seqs.
13526 2017-07-28 Bin Cheng <bin.cheng@arm.com>
13528 * tree-predcom.c (determine_roots_comp): Skip trivial components.
13530 2017-07-28 Richard Biener <rguenther@suse.de>
13532 * match.pd: Remove superfluous :c.
13533 * genmatch.c (simplify::id): Add member.
13534 (lower_commutative, lower_opt_convert, lower_cond, lower_for):
13536 (current_id): New global.
13537 (dt_node::parent): Move from ...
13538 (dt_operand::parent): ... here. Add for_id member.
13539 (is_a_helper <dt_operand *>::test): DT_TRUE is also a dt_operand.
13540 (decision_tree::find_node): Relax order requirement when
13541 merging DT_TRUE nodes to ones inbetween the current simplify
13542 and the one we try to merge with. Add diagnostic whenever
13543 we need to enforce pattern order by not merging.
13544 (decision_tree::insert): Set current_id.
13545 (decision_tree::print_node): Dump parent node and for_id.
13546 (parser::last_id): Add member.
13547 (parser::push_simplify): Assign unique id.
13548 (parser::parser): Initialize last_id.
13550 2017-07-28 Martin Liska <mliska@suse.cz>
13553 * sanopt.c (sanitize_rewrite_addressable_params): Set VALUE_EXPR after
13554 gimple_build_debug_bind.
13556 2017-07-28 Richard Biener <rguenther@suse.de>
13558 PR tree-optimization/81502
13559 * match.pd: Add pattern combining BIT_INSERT_EXPR with
13561 * tree-cfg.c (verify_expr): Verify types of BIT_FIELD_REF
13563 (verify_gimple_assign_ternary): Likewise for BIT_INSERT_EXPR pos.
13564 * gimple-fold.c (maybe_canonicalize_mem_ref_addr): Use bitsizetype
13565 for BIT_FIELD_REF args.
13566 * fold-const.c (make_bit_field_ref): Likewise.
13567 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
13569 2017-07-28 Jakub Jelinek <jakub@redhat.com>
13572 * sanopt.c (pass_sanopt::execute): Handle IFN_UBSAN_PTR.
13573 * tree-ssa-alias.c (call_may_clobber_ref_p_1): Likewise.
13574 * flag-types.h (enum sanitize_code): Add SANITIZER_POINTER_OVERFLOW.
13575 Or it into SANITIZER_UNDEFINED.
13576 * ubsan.c: Include gimple-fold.h and varasm.h.
13577 (ubsan_expand_ptr_ifn): New function.
13578 (instrument_pointer_overflow): New function.
13579 (maybe_instrument_pointer_overflow): New function.
13580 (instrument_object_size): Formatting fix.
13581 (pass_ubsan::execute): Call instrument_pointer_overflow
13582 and maybe_instrument_pointer_overflow.
13583 * internal-fn.c (expand_UBSAN_PTR): New function.
13584 * ubsan.h (ubsan_expand_ptr_ifn): Declare.
13585 * sanitizer.def (__ubsan_handle_pointer_overflow,
13586 __ubsan_handle_pointer_overflow_abort): New builtins.
13587 * tree-ssa-tail-merge.c (merge_stmts_p): Handle IFN_UBSAN_PTR.
13588 * internal-fn.def (UBSAN_PTR): New internal function.
13589 * opts.c (sanitizer_opts): Add pointer-overflow.
13590 * lto-streamer-in.c (input_function): Handle IFN_UBSAN_PTR.
13591 * fold-const.c (build_range_check): Compute pointer range check in
13592 integral type if pointer arithmetics would be needed. Formatting
13595 2017-07-28 Martin Liska <mliska@suse.cz>
13598 * sanopt.c (sanitize_rewrite_addressable_params): Do not rewrite
13599 parameters that are of a variable-length.
13601 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
13603 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
13605 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
13606 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
13607 (CRT_CALL_STATIC_FUNCTION): Likewise.
13608 (ASM_DEFAULT_SPEC): New define.
13609 (ASM_SPEC32): Likewise.
13610 (ASM_SPEC64): Likewise.
13611 (ASM_SPEC_COMMON): Likewise.
13612 (ASM_SPEC): Likewise.
13613 (INVALID_64BIT): Likewise.
13614 (LINK_OS_DEFAULT_SPEC): Likewise.
13615 (LINK_OS_SPEC32): Likewise.
13616 (LINK_OS_SPEC64): Likewise.
13617 (POWERPC_LINUX): Likewise.
13618 (PTRDIFF_TYPE): Likewise.
13619 (RESTORE_FP_PREFIX): Likewise.
13620 (RESTORE_FP_SUFFIX): Likewise.
13621 (SAVE_FP_PREFIX): Likewise.
13622 (SAVE_FP_SUFFIX): Likewise.
13623 (SIZE_TYPE): Likewise.
13624 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
13625 (TARGET_64BIT): Likewise.
13626 (TARGET_64BIT): Likewise.
13627 (TARGET_AIX): Likewise.
13628 (WCHAR_TYPE_SIZE): Likewise.
13629 (WCHAR_TYPE): Undefine.
13630 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
13631 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
13632 (CPP_OS_RTEMS_SPEC): Delete.
13633 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
13634 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
13636 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
13638 2017-07-28 Jakub Jelinek <jakub@redhat.com>
13640 PR tree-optimization/81578
13641 * tree-parloops.c (build_new_reduction): Bail out if
13642 reduction_code isn't one of the standard OpenMP reductions.
13643 Move the details printing after that decision.
13645 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13647 * config/rs6000/predicates.md (volatile_mem_operand): Remove code
13648 related to reload_in_progress.
13649 (splat_input_operand): Likewise.
13650 * config/rs6000/rs6000-protos.h (rs6000_secondary_memory_needed_rtx):
13652 * config/rs6000/rs6000.c (machine_function): Remove sdmode_stack_slot
13654 (TARGET_EXPAND_TO_RTL_HOOK): Delete.
13655 (TARGET_INSTANTIATE_DECLS): Likewise.
13656 (legitimate_indexed_address_p): Delete reload_in_progress code.
13657 (rs6000_debug_legitimate_address_p): Likewise.
13658 (rs6000_eliminate_indexed_memrefs): Likewise.
13659 (rs6000_emit_le_vsx_store): Likewise.
13660 (rs6000_emit_move_si_sf_subreg): Likewise.
13661 (rs6000_emit_move): Likewise.
13662 (register_to_reg_type): Likewise.
13663 (rs6000_pre_atomic_barrier): Likewise.
13664 (rs6000_machopic_legitimize_pic_address): Likewise.
13665 (rs6000_allocate_stack_temp): Likewise.
13666 (rs6000_address_for_fpconvert): Likewise.
13667 (rs6000_address_for_altivec): Likewise.
13668 (rs6000_secondary_memory_needed_rtx): Delete function.
13669 (rs6000_check_sdmode): Likewise.
13670 (rs6000_alloc_sdmode_stack_slot): Likewise.
13671 (rs6000_instantiate_decls): Likewise.
13672 * config/rs6000/rs6000.h (SECONDARY_MEMORY_NEEDED_RTX): Delete.
13673 * config/rs6000/rs6000.md (splitter for *movsi_got_internal):
13674 Delete reload_in_progress.
13675 (*vec_reload_and_plus_<mptrsize>): Likewise.
13676 * config/rs6000/vsx.md (vsx_mul_v2di): Likewise.
13677 (vsx_div_v2di): Likewise.
13678 (vsx_udiv_v2di): Likewise.
13680 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13682 * config/rs6000/rs6000.opt (mlra): Replace with stub.
13683 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete OPTION_MASK_LRA.
13684 * config/rs6000/rs6000.c (TARGET_LRA_P): Delete.
13685 (rs6000_debug_reg_global): Delete print of LRA status.
13686 (rs6000_option_override_internal): Delete dead LRA related code.
13687 (rs6000_lra_p): Delete function.
13688 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mlra.
13690 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
13692 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
13693 * config/riscv/rtems.h: New file.
13695 2017-07-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13696 Sudakshina Das <sudi.das@arm.com>
13698 * config/aarch64/aarch64.md
13699 (define_split for and<mode>3nr_compare): Move
13700 non aarch64_logical_operand to a register.
13701 (define_split for and_<SHIFT:optab><mode>3nr_compare0): Move non
13702 register immediate operand to a register.
13703 * config/aarch64/predicates.md (aarch64_mov_imm_operand): New.
13705 2017-07-27 Peter Bergner <bergner@vnet.ibm.com>
13707 PR middle-end/81564
13708 * tree-cfg.c (group_case_labels_stmt): Handle already deleted blocks.
13710 2017-07-27 Richard Biener <rguenther@suse.de>
13712 PR tree-optimization/81573
13713 PR tree-optimization/81494
13714 * tree-vect-loop.c (vect_create_epilog_for_reduction): Handle
13715 multi defuse cycle case.
13717 2017-07-27 Richard Biener <rguenther@suse.de>
13719 PR tree-optimization/81571
13720 * tree-vect-slp.c (vect_build_slp_tree): Properly verify reduction
13723 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
13725 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
13726 earlier and only if MASK_FPU is set. Adjust formatting.
13728 2017-07-27 Martin Liska <mliska@suse.cz>
13730 * opt-functions.awk: Add validation of value of Init.
13731 * optc-gen.awk: Pass new argument.
13733 2017-07-27 Martin Liska <mliska@suse.cz>
13735 * auto-profile.c (autofdo_source_profile::update_inlined_ind_target):
13736 Fix wrong condition.
13738 2017-07-27 Martin Liska <mliska@suse.cz>
13740 * auto-profile.c (afdo_annotate_cfg): Assign zero counts to
13741 BBs and edges seen by autoFDO.
13743 2017-07-27 Richard Biener <rguenther@suse.de>
13745 PR tree-optimization/81502
13746 * tree-ssa.c (non_rewritable_lvalue_p): Handle BIT_INSERT_EXPR
13747 with incompatible but same sized type.
13748 (execute_update_addresses_taken): Likewise.
13750 2017-07-27 James Greenhalgh <james.greenhalgh@arm.com>
13752 * tree-ssa-loop-ch.c (pass_ch::process_loop_p): Guard on
13753 flag_tree_loop_vectorize rather than flag_tree_vectorize.
13755 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13758 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
13759 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
13760 Change s_operand to memory_operand.
13762 2017-07-27 Richard Sandiford <richard.sandiford@linaro.org>
13764 * config/rs6000/rs6000-protos.h (rs6000_emit_le_vsx_permute): Declare.
13765 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Replace with...
13766 (rs6000_emit_le_vsx_permute): ...this. Take the destination as input.
13767 Emit instructions rather than returning an expression. Handle TFmode
13768 and KFmode by casting to TImode.
13769 (rs6000_emit_le_vsx_load): Update to use rs6000_emit_le_vsx_permute.
13770 (rs6000_emit_le_vsx_store): Likewise.
13771 * config/rs6000/vsx.md (VSX_TI): New iterator.
13772 (*vsx_le_permute_<mode>): Use it instead of VSX_LE_128.
13773 (*vsx_le_undo_permute_<mode>): Likewise.
13774 (*vsx_le_perm_load_<mode>): Use rs6000_emit_le_vsx_permute to
13775 emit the split sequence.
13776 (*vsx_le_perm_store_<mode>): Likewise.
13778 2017-07-27 Jakub Jelinek <jakub@redhat.com>
13780 PR tree-optimization/81555
13781 PR tree-optimization/81556
13782 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
13783 if true, force CHANGED for the recursive invocation.
13784 (reassociate_bb): Remember original length of ops array, pass
13785 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
13787 * attribs.c (decl_attributes): Imply noinline, noclone and no_icf
13788 attributes for noipa attribute. For naked attribute use
13789 lookup_attribute first before lookup_attribute_spec.
13790 * final.c (rest_of_handle_final): Disable IPA RA for functions with
13792 * ipa-visibility.c (non_local_p): Fix comment typos. Return true
13793 for functions with noipa attribute.
13794 (cgraph_externally_visible_p): Return true for functions with noipa
13796 * cgraph.c (cgraph_node::get_availability): Return AVAIL_INTERPOSABLE
13797 for functions with noipa attribute.
13798 * doc/extend.texi: Document noipa function attribute.
13799 * tree-ssa-structalias.c (refered_from_nonlocal_fn): Set *nonlocal_p
13800 also for functions with noipa attribute.
13801 (ipa_pta_execute): Set nonlocal_p also for nodes with noipa attribute.
13803 2017-07-26 Andrew Pinski <apinski@cavium.com>
13805 * config/aarch64/aarch64.c (thunderx_vector_cost): Decrease cost of
13806 vec_unalign_load_cost and vec_unalign_store_cost.
13808 2017-07-26 Michael Meissner <meissner@linux.vnet.ibm.com>
13810 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Delete
13811 -mvsx-small-integer option.
13812 (ISA_3_0_MASKS_IEEE): Likewise.
13813 (OTHER_VSX_VECTOR_MASKS): Likewise.
13814 (POWERPC_MASKS): Likewise.
13815 * config/rs6000/rs6000.opt (-mvsx-small-integer): Likewise.
13816 * config/rs6000/rs6000.c (rs6000_hard_regno_mode_ok): Simplify
13817 code, only testing for DImode being allowed in non-VSX floating
13819 (rs6000_init_hard_regno_mode_ok): Change TARGET_VSX_SMALL_INTEGER
13820 to TARGET_P8_VECTOR test. Remove redundant VSX test inside of
13822 (rs6000_option_override_internal): Delete -mvsx-small-integer.
13823 (rs6000_expand_vector_set): Change TARGET_VSX_SMALL_INTEGER to
13824 TARGET_P8_VECTOR test.
13825 (rs6000_secondary_reload_simple_move): Likewise.
13826 (rs6000_preferred_reload_class): Delete TARGET_VSX_SMALL_INTEGER,
13827 since TARGET_P9_VECTOR was already tested.
13828 (rs6000_opt_masks): Remove -mvsx-small-integer.
13829 * config/rs6000/vsx.md (vsx_extract_<mode>): Delete
13830 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13832 (vsx_extract_<mode>_p9): Delete TARGET_VSX_SMALL_INTEGER, since a
13833 test for TARGET_VEXTRACTUB was used, and that uses
13835 (p9 extract splitter): Likewise.
13836 (vsx_extract_<mode>_di_p9): Likewise.
13837 (vsx_extract_<mode>_store_p9): Likewise.
13838 (vsx_extract_si): Delete TARGET_VSX_SMALL_INTEGER, since a test
13839 for TARGET_P9_VECTOR was used. Delete code that is now dead with
13840 the elimination of TARGET_VSX_SMALL_INTEGER.
13841 (vsx_extract_<mode>_p8): Likewise.
13842 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_fl_<FL_CONV:mode>): Likewise.
13843 (vsx_ext_<VSX_EXTRACT_I:VS_scalar>_ufl_<FL_CONV:mode>): Likewise.
13844 (vsx_set_<mode>_p9): Likewise.
13845 (vsx_set_v4sf_p9): Likewise.
13846 (vsx_set_v4sf_p9_zero): Likewise.
13847 (vsx_insert_extract_v4sf_p9): Likewise.
13848 (vsx_insert_extract_v4sf_p9_2): Likewise.
13849 * config/rs6000/rs6000.md (sign extend splitter): Change
13850 TARGET_VSX_SMALL_INTEGER to TARGET_P8_VECTOR test.
13851 (floatsi<mode>2_lfiwax_mem): Likewise.
13852 (floatunssi<mode>2_lfiwzx_mem): Likewise.
13853 (float<QHI:mode><FP_ISA3:mode>2): Delete TARGET_VSX_SMALL_INTEGER,
13854 since a test for TARGET_P9_VECTOR was used.
13855 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13856 (floatuns<QHI:mode><FP_ISA3:mode>2): Likewise.
13857 (floatuns<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13858 (fix_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13859 TARGET_P8_VECTOR test.
13860 (fix_trunc<mode>si2_stfiwx): Likewise.
13861 (fix_trunc<mode>si2_internal): Likewise.
13862 (fix_trunc<SFDF:mode><QHI:mode>2): Delete
13863 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13865 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13866 (fixuns_trunc<mode>si2): Change TARGET_VSX_SMALL_INTEGER to
13867 TARGET_P8_VECTOR test.
13868 (fixuns_trunc<mode>si2_stfiwx): Likewise.
13869 (fixuns_trunc<SFDF:mode><QHI:mode>2): Delete
13870 TARGET_VSX_SMALL_INTEGER, since a test for TARGET_P9_VECTOR was
13872 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
13873 (fctiw<u>z_<mode>_smallint): Delete TARGET_VSX_SMALL_INTEGER,
13874 since a test for TARGET_P9_VECTOR was used.
13875 (splitter for loading small constants): Likewise.
13877 2017-07-26 Andrew Pinski <apinski@cavium.com>
13879 * config/aarch64/aarch64.c (thunderx_vector_cost): Fix
13882 2017-07-26 H.J. Lu <hongjiu.lu@intel.com>
13885 * config/i386/i386.c (sp_valid_at): Properly check CFA offset.
13886 (fp_valid_at): Likewise.
13888 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13890 * config/aarch64/aarch64.c (cortexa57_addrcost_table): Remove.
13891 (qdf24xx_addrcost_table): Likewise.
13892 (cortexa57_tunings): Update to use generic_branch_cost.
13893 (cortexa72_tunings): Likewise.
13894 (cortexa73_tunings): Likewise.
13895 (qdf24xx_tunings): Likewise.
13897 2017-07-26 James Greenhalgh <james.greenhalgh@arm.com>
13899 * config/aarch64/aarch64.c (cortexa57_branch_cost): Remove.
13900 (thunderx2t99_branch_cost): Likewise.
13901 (cortexa35_tunings): Update to use generic_branch_cost.
13902 (cortexa53_tunings): Likewise.
13903 (cortexa57_tunings): Likewise.
13904 (cortexa72_tunings): Likewise.
13905 (cortexa73_tunings): Likewise.
13906 (thunderx2t99_tunings): Likewise.
13908 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13910 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
13911 (sparc_option_override): Honour MASK_FSMULD.
13912 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
13913 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
13914 * config/sparc/sparc.opt (mfsmuld): New option.
13915 * doc/invoke.texi (mfsmuld): Document option.
13917 2017-07-26 Marek Polacek <polacek@redhat.com>
13919 PR middle-end/70992
13920 * tree.c (build2_stat): Don't set TREE_CONSTANT on divisions by zero.
13922 2017-07-26 Richard Biener <rguenther@suse.de>
13924 * gimple-match-head.c (do_valueize): Return OP if valueize
13926 (get_def): New helper to get at the def stmt of a SSA name
13927 if valueize allows.
13928 * genmatch.c (dt_node::gen_kids_1): Use get_def instead of
13929 do_valueize to get at the def stmt.
13930 (dt_operand::gen_gimple_expr): Simplify do_valueize calls.
13932 2017-07-26 Wilco Dijkstra <wdijkstr@arm.com>
13934 PR middle-end/46932
13935 * auto-inc-dec.c (parse_add_or_inc): Block autoinc on sfp.
13937 2017-07-26 Martin Liska <mliska@suse.cz>
13940 * function.c (expand_function_start): Make expansion of
13941 nonlocal_goto_save_area after parm_birth_insn.
13943 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
13945 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
13946 from all CPU target flags enable members.
13948 2017-07-26 Richard Biener <rguenther@suse.de>
13950 * genmatch.c (dt_simplify::gen): Make iterator vars const.
13951 (decision_tree::gen): Make 'type' const.
13952 (write_predicate): Likewise.
13954 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
13956 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok):
13957 Eliminate TARGET_UPPER_REGS_{DF,DI,SF} usage.
13958 (rs6000_option_override_internal): Likewise.
13959 (rs6000_expand_vector_set): Likewise.
13960 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Delete.
13961 (TARGET_UPPER_REGS_SF): Likewise.
13962 (TARGET_UPPER_REGS_DI): Likewise.
13963 (TARGET_VEXTRACTUB): Eliminate TARGET_UPPER_REGS_{DF,DI,SF}.
13964 (TARGET_DIRECT_MOVE_64BIT): Likewise.
13965 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Likewise.
13966 (float<QHI:mode><FP_ISA3:mode>2_internal): Likewise.
13967 (Splitters for DI constants in Altivec registers): Likewise.
13968 * config/rs6000/vsx.md (vsx_set_<mode>_p9): Likewise.
13969 (vsx_set_v4sf_p9): Likewise.
13970 (vsx_set_v4sf_p9_zero): Likewise.
13971 (vsx_insert_extract_v4sf_p9): Likewise.
13972 (vsx_insert_extract_v4sf_p9_2): Likewise.
13974 2017-07-25 Carl Love <cel@us.ibm.com>
13976 * doc/extend.texi: Update the built-in documentation file for the
13977 existing built-in functions
13978 vector signed char vec_cnttz (vector signed char);
13979 vector unsigned char vec_cnttz (vector unsigned char);
13980 vector signed short vec_cnttz (vector signed short);
13981 vector unsigned short vec_cnttz (vector unsigned short);
13982 vector signed int vec_cnttz (vector signed int);
13983 vector unsigned int vec_cnttz (vector unsigned int);
13984 vector signed long long vec_cnttz (vector signed long long);
13985 vector unsigned long long vec_cnttz (vector unsigned long long);
13987 2017-07-25 Andrew Pinski <apinski@cavium.com>
13989 * tree-ssa-uninit.c (warn_uninitialized_vars): Don't warn about memory
13990 accesses where the use is for the first operand of a BIT_INSERT.
13992 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
13995 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
13996 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
13998 2017-07-25 Jim Wilson <jim.wilson@linaro.org>
14000 * config/i386/gstabs.h: Delete.
14001 * config/i386/openbsd.h, config/i386/t-openbsd: Likewise.
14003 2017-07-25 Uros Bizjak <ubizjak@gmail.com>
14005 * config/i386/i386.c (ix86_decompose_address): Do not check for
14006 register RTX when looking at index_reg or base_reg.
14007 * config/i386/i386.h (INCOMING_RETURN_ADDR_RTX): Use stack_pointer_rtx.
14009 2017-07-25 Eric Botcazou <ebotcazou@adacore.com>
14011 * gimple.c (gimple_assign_set_rhs_with_ops): Do not ask gsi_replace
14012 to update EH info here.
14014 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14016 * match.pd ((X * CST1) * CST2): Simplify to X * (CST1 * CST2).
14018 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14020 * match.pd ((X * CST) * Y): Reassociate to (X * Y) * CST.
14022 2017-07-25 Torsten Duwe <duwe@suse.de>
14024 * common.opt: Introduce -fpatchable-function-entry
14025 command line option, and its variables function_entry_patch_area_size
14026 and function_entry_patch_area_start.
14027 * opts.c (common_handle_option): Add -fpatchable_function_entry_ case,
14028 including a two-value parser.
14029 * target.def (print_patchable_function_entry): New target hook.
14030 * targhooks.h (default_print_patchable_function_entry): New function.
14031 * targhooks.c (default_print_patchable_function_entry): Likewise.
14032 * toplev.c (process_options): Switch off IPA-RA if
14033 patchable function entries are being generated.
14034 * varasm.c (assemble_start_function): Look at the
14035 patchable-function-entry command line switch and current
14036 function attributes and maybe generate NOP instructions by
14037 calling the print_patchable_function_entry hook.
14038 * doc/extend.texi: Document patchable_function_entry attribute.
14039 * doc/invoke.texi: Document -fpatchable_function_entry
14040 command line option.
14041 * doc/tm.texi.in (TARGET_ASM_PRINT_PATCHABLE_FUNCTION_ENTRY):
14043 * doc/tm.texi: Re-generate.
14045 2017-07-25 Jakub Jelinek <jakub@redhat.com>
14048 * config/i386/constraints.md (Yd, Ye): Use ALL_SSE_REGS for
14049 TARGET_AVX512DQ rather than TARGET_AVX512BW.
14051 2017-07-25 Tamar Christina <tamar.christina@arm.com>
14053 * config/arm/parsecpu.awk (all_cores): Remove duplicates.
14055 2017-07-25 Richard Biener <rguenther@suse.de>
14057 PR tree-optimization/81455
14058 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
14059 not walk in cycles when looking for guards.
14061 2017-07-25 Richard Biener <rguenther@suse.de>
14063 PR tree-optimization/81529
14064 * tree-vect-stmts.c (process_use): Disregard live induction PHIs
14065 when optimizing backedge uses.
14067 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
14069 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
14071 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
14072 to dl_section_ref. On AIX, append an expression to subtract
14073 the size of the section length to dl_section_ref.
14075 2017-07-25 Segher Boessenkool <segher@kernel.crashing.org>
14077 * configure.ac: If any of the config.* scripts fail, exit 1.
14078 * configure: Regenerate.
14080 2017-07-25 Richard Biener <rguenther@suse.de>
14082 PR middle-end/81546
14083 * tree-ssa-operands.c (verify_imm_links): Remove cap on number
14084 of immediate uses, be more verbose on errors.
14086 2017-07-25 Richard Biener <rguenther@suse.de>
14088 PR tree-optimization/81510
14089 * tree-vect-loop.c (vect_is_simple_reduction): When the
14090 reduction stmt is not inside the loop bail out.
14092 2017-07-25 Richard Biener <rguenther@suse.de>
14094 PR tree-optimization/81303
14095 * tree-vect-loop-manip.c (vect_loop_versioning): Build
14096 profitability check against LOOP_VINFO_NITERSM1.
14098 2017-07-25 Alexander Monakov <amonakov@ispras.ru>
14100 * domwalk.c (cmp_bb_postorder): Simplify.
14101 (sort_bbs_postorder): New function. Use it...
14102 (dom_walker::walk): ...here to optimize common cases.
14104 2017-07-25 Martin Liska <mliska@suse.cz>
14107 * ipa-visibility.c (function_and_variable_visibility): Make the
14108 redirection just on target that supports aliasing.
14109 Fix GNU coding style.
14111 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
14114 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
14116 * config/arm/rtems.h: Likewise.
14117 * config/bfin/rtems.h: Likewise.
14118 * config/i386/rtemself.h: Likewise.
14119 * config/lm32/rtems.h: Likewise.
14120 * config/m32c/rtems.h: Likewise.
14121 * config/m68k/rtemself.h: Likewise.
14122 * config/microblaze/rtems.h: Likewise.
14123 * config/mips/rtems.h: Likewise.
14124 * config/moxie/rtems.h: Likewise.
14125 * config/nios2/rtems.h: Likewise.
14126 * config/powerpcspe/rtems.h: Likewise.
14127 * config/rs6000/rtems.h: Likewise.
14128 * config/rtems.h: Likewise.
14129 * config/sh/rtems.h: Likewise.
14130 * config/sh/rtemself.h: Likewise.
14131 * config/sparc/rtemself.h: Likewise.
14133 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14136 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
14137 * gimple-pretty-print.c (dump_profile, dump_probability): Same.
14138 * tree-ssa-structalias.c (alias_get_name): Same.
14140 2017-07-25 Bin Cheng <bin.cheng@arm.com>
14143 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
14144 instructions if no du chain is found.
14146 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
14148 * config/avr/avr-log.c (avr_log_vadump) ['T']: Print NULL-TREE.
14150 2017-07-25 Richard Biener <rguenther@suse.de>
14152 PR middle-end/81505
14153 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
14156 2017-07-24 Michael Meissner <meissner@linux.vnet.ibm.com>
14158 * config/rs6000/rs6000-cpus.def (ISA_2_6_MASKS_SERVER): Delete
14159 upper-regs options.
14160 (ISA_2_7_MASKS_SERVER): Likewise.
14161 (ISA_3_0_MASKS_IEEE): Likewise.
14162 (OTHER_P8_VECTOR_MASKS): Likewise.
14163 (OTHER_VSX_VECTOR_MASKS): Likewise.
14164 (POWERPC_MASKS): Likewise.
14165 (power7 cpu): Use ISA_2_6_MASKS_SERVER instead of using a
14166 duplicate list of options.
14167 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove
14168 explicit -mupper-regs options.
14169 * config/rs6000/rs6000.opt (-mvsx-scalar-memory): Delete
14170 -mupper-regs* options. Delete -mvsx-scalar-memory, which was an
14171 alias for -mupper-regs-df.
14172 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): Likewise.
14173 (rs6000_init_hard_regno_mode_ok): Likewise.
14174 (rs6000_option_override_internal): Likewise.
14175 (rs6000_opt_masks): Likewise.
14176 * config/rs6000/rs6000.h (TARGET_UPPER_REGS_DF): Define upper regs
14177 options in terms of whether -mvsx or -mpower8-vector was used.
14178 (TARGET_UPPER_REGS_DI): Likewise.
14179 (TARGET_UPPER_REGS_SF): Likewise.
14180 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete the
14181 -mupper-regs-* options.
14183 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14185 * passes.c (emergency_dump_function): Print some empty lines and a
14186 header before the RTL dump.
14188 2017-07-24 Segher Boessenkool <segher@kernel.crashing.org>
14190 * cfgrtl.c (rtl_dump_bb): Don't call NEXT_INSN on NULL.
14192 2017-07-24 Wilco Dijkstra <wdijkstr@arm.com>
14195 * config/aarch64/aarch64.c (aarch64_classify_symbol):
14196 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
14198 2017-07-24 Carl Love <cel@us.ibm.com>
14200 * config/rs6000/rs6000-c.c: Add support for built-in functions
14201 vector float vec_extract_fp32_from_shorth (vector unsigned short);
14202 vector float vec_extract_fp32_from_shortl (vector unsigned short);
14203 * config/rs6000/altivec.h (vec_extract_fp_from_shorth,
14204 vec_extract_fp_from_shortl): Add defines for the two builtins.
14205 * config/rs6000/rs6000-builtin.def (VEXTRACT_FP_FROM_SHORTH,
14206 VEXTRACT_FP_FROM_SHORTL): Add BU_P9V_OVERLOAD_1 and BU_P9V_VSX_1
14208 * config/rs6000/vsx.md vsx_xvcvhpsp): Add define_insn.
14209 (vextract_fp_from_shorth, vextract_fp_from_shortl): Add define_expands.
14210 * doc/extend.texi: Update the built-in documentation file for the
14211 new built-in function.
14213 2017-07-24 Jakub Jelinek <jakub@redhat.com>
14216 * tree.def: Remove TYPE_METHODS documentation, adjust TYPE_FIELDS
14218 * doc/generic.texi: Likewise.
14219 * config/i386/winnt-cxx.c (i386_pe_adjust_class_at_definition): Look
14220 for FUNCTION_DECLs in TYPE_FIELDS rather than TYPE_METHODS.
14222 2017-07-24 Jackson Woodruff <jackson.woodruff@arm.com>
14224 * config/aarch64/aarch64-simd.md (aarch64_mla_elt_merge<mode>): New.
14225 (aarch64_mls_elt_merge<mode>): Likewise.
14227 2017-07-23 Krister Walfridsson <krister.walfridsson@gmail.com>
14229 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
14230 having __cxa_atexit.
14232 2017-07-23 Michael Collison <michael.collison@arm.com>
14234 * config/arm/arm.c (arm_option_override): Deprecate
14235 use of -mstructure-size-boundary.
14236 * config/arm/arm.opt: Deprecate -mstructure-size-boundary.
14237 * doc/invoke.texi: Deprecate -mstructure-size-boundary.
14239 2017-07-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14242 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
14243 Reduce cost estimate for direct moves.
14245 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
14248 * config/i386/i386.c (ix86_option_override_internal): Disable
14249 BMI, BMI2 and TBM instructions for -m16.
14251 2017-07-21 Carl Love <cel@us.ibm.com>
14253 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14254 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14255 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14256 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14257 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14258 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14259 VMULOSW): New enum "unspec" values.
14260 (altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14261 altivec_vmulosw): New patterns.
14262 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14263 VMULOSW): Add definitions.
14265 2017-07-21 Jim Wilson <jim.wilson@linaro.org>
14267 * config/aarch64/aarch64-cores.def (falkor): Add AARCH64_FL_RDMA.
14268 (qdf24xx): Likewise.
14269 * config/aarch64/aarch64-options-extensions.def (rdma); New.
14270 * config/aarch64/aarch64.h (AARCH64_FL_RDMA): New.
14271 (AARCH64_FL_V8_1): Renumber.
14272 (AARCH64_FL_FOR_ARCH8_1): Add AARCH64_FL_RDMA.
14273 (AARCH64_ISA_RDMA): Use AARCH64_FL_RDMA.
14274 * config/aarch64/arm_neon.h: Use +rdma instead of arch=armv8.1-a.
14275 * doc/invoke.texi (AArch64 Options): Mention +rmda in -march docs. Add
14276 rdma to feature modifiers list.
14278 2017-07-21 Yury Gribov <tetra2005@gmail.com>
14280 PR middle-end/56727
14281 * ipa-visibility (function_and_variable_visibility): Convert
14282 recursive PLT call to direct call if appropriate.
14284 2017-07-21 Andrew Pinski <apinski@cavium.com>
14286 * tree-ssa-sccvn.c (vn_nary_op_eq): Check BIT_INSERT_EXPR's
14287 operand 1 to see if the types precision matches.
14288 * fold-const.c (operand_equal_p): Likewise.
14290 2017-07-21 Richard Biener <rguenther@suse.de>
14292 PR tree-optimization/81303
14293 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs): Pass
14294 in datarefs vector. Allow NULL dr0 for no peeling cost estimate.
14295 (vect_peeling_hash_get_lowest_cost): Adjust.
14296 (vect_enhance_data_refs_alignment): Likewise. Use
14297 vect_get_peeling_costs_all_drs to compute the penalty for no
14298 peeling to match up costs.
14300 2017-07-21 Richard Biener <rguenther@suse.de>
14302 PR tree-optimization/81500
14303 * tree-vect-loop.c (vect_is_simple_reduction): Properly fail if
14304 we didn't identify a reduction path.
14306 2017-07-21 Tom de Vries <tom@codesourcery.com>
14307 Cesar Philippidis <cesar@codesourcery.com>
14309 PR gcov-profile/81442
14310 * config/nvptx/nvptx.c (nvptx_goacc_reduction_init): Add missing edge
14313 2017-07-21 Tom de Vries <tom@codesourcery.com>
14316 * config/nvptx/nvptx.c (nvptx_override_options_after_change): New
14318 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define to
14319 nvptx_override_options_after_change.
14321 2017-07-21 Ulrich Drepper <drepper@redhat.com>
14323 * dwarf2out.c (output_file_names): Avoid double testing for
14324 dwarf_version >= 5.
14326 2017-07-21 Georg-Johann Lay <avr@gjlay.de>
14328 * doc/invoke.texi (AVR Built-in Functions): Re-layout section.
14330 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14332 * cfgcleanup.c (flow_find_cross_jump): Do not crossjump across
14334 (try_crossjump_to_edge): Do not punt on partitioned functions.
14336 2016-07-21 Jan Hubicka <hubicka@ucw.cz>
14338 * bb-reorder.c (find_rarely_executed_basic_blocks_and_crossing_edges):
14339 Put all BBs reachable only via paths crossing cold region to cold
14341 * cfgrtl.c (find_bbs_reachable_by_hot_paths): New function.
14343 2016-07-21 Richard Biener <rguenther@suse.de>
14345 PR tree-optimization/81303
14346 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Take
14347 into account prologue and epilogue iterations when raising
14348 min_profitable_iters to sth at least covering one vector iteration.
14350 2017-07-21 Tamar Christina <tamar.christina@arm.com>
14352 * config/arm/arm.c (arm_test_cpu_arch_dat):
14355 2017-07-20 Nathan Sidwell <nathan@acm.org>
14357 Remove TYPE_METHODS.
14358 * tree.h (TYPE_METHODS): Delete.
14359 * dwarf2out.c (gen_member_die): Member fns are on TYPE_FIELDS.
14360 * dbxout.c (dbxout_type_fields): Ignore FUNCTION_DECLs.
14361 (dbxout_type_methods): Scan TYPE_FIELDS.
14362 (dbxout_type): Don't check TYPE_METHODS here.
14363 * function.c (use_register_for_decl): Always ignore register for
14364 class types when not optimizing.
14365 * ipa-devirt.c (odr_types_equivalent_p): Delete TYPE_METHODS scan.
14366 * tree.c (free_lang_data_in_type): Stitch out member functions and
14367 templates from TYPE_FIELDS.
14368 (build_distinct_type_copy, verify_type_variant,
14369 verify_type): Member fns are on TYPE_FIELDS.
14370 * tree-dump.c (dequeue_and_dump): No TYPE_METHODS.
14371 * tree-pretty-print.c (dump_generic_node): Likewise.
14373 2017-07-20 Jakub Jelinek <jakub@redhat.com>
14376 * config/i386/i386.c (ix86_expand_vector_init_general): Handle
14377 V2TImode and V4TImode.
14378 (ix86_expand_vector_extract): Likewise.
14379 * config/i386/sse.md (VMOVE): Enable V4TImode even for just
14380 TARGET_AVX512F, instead of only for TARGET_AVX512BW.
14381 (ssescalarmode): Handle V4TImode and V2TImode.
14382 (VEC_EXTRACT_MODE): Add V4TImode and V2TImode.
14383 (*vec_extractv2ti, *vec_extractv4ti): New insns.
14384 (VEXTRACTI128_MODE): New mode iterator.
14385 (splitter for *vec_extractv?ti first element): New.
14386 (VEC_INIT_MODE): New mode iterator.
14387 (vec_init<mode>): Consolidate 3 expanders into one using
14388 VEC_INIT_MODE mode iterator.
14390 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14392 * lra-assigns.c (pseudo_compare_func): Fix comparison step based on
14393 non_spilled_static_chain_regno_p.
14395 2017-07-20 Alexander Monakov <amonakov@ispras.ru>
14397 * gimple-ssa-store-merging.c (sort_by_bitpos): Return 0 on equal bitpos.
14399 2017-07-20 Jan Hubicka <hubicka@ucw.cz>
14401 * bb-reorder.c (connect_traces): Allow copying of blocks within
14404 2017-07-20 Richard Biener <rguenther@suse.de>
14406 * gimple.h (gimple_phi_result): Add gphi * overload.
14407 (gimple_phi_result_ptr): Likewise.
14408 (gimple_phi_arg): Likewise. Adjust index assert to only
14409 allow actual argument accesses rather than all slots available
14411 (gimple_phi_arg_def): Add gphi * overload.
14412 * tree-phinodes.c (make_phi_node): Initialize only actual
14414 (resize_phi_node): Clear memory not covered by old node,
14415 do not initialize excess argument slots.
14416 (reserve_phi_args_for_new_edge): Initialize new argument slot
14419 2017-07-20 Bin Cheng <bin.cheng@arm.com>
14421 PR tree-optimization/81388
14423 2016-07-21 Bin Cheng <bin.cheng@arm.com>
14425 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
14426 by removing computation of may_be_zero.
14428 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14429 Tom de Vries <tom@codesourcery.com>
14431 PR middle-end/81030
14432 * cfgbuild.c (find_many_sub_basic_blocks): Update REG_BR_PROB note
14433 when gimple level profile disagrees with what RTL expander did.
14435 2017-07-20 Richard Biener <rguenther@suse.de>
14437 PR tree-optimization/61171
14438 * tree-vectorizer.h (slp_instance): Add reduc_phis member.
14439 (vect_analyze_stmt): Add slp instance parameter.
14440 (vectorizable_reduction): Likewise.
14441 * tree-vect-loop.c (vect_analyze_loop_operations): Adjust.
14442 (vect_is_simple_reduction): Deal with chains not detected
14443 as SLP reduction chain, specifically not properly associated
14444 chains containing a mix of plus/minus.
14445 (get_reduction_op): Remove.
14446 (get_initial_defs_for_reduction): Simplify, pass in whether
14447 this is a reduction chain, pass in the SLP node for the PHIs.
14448 (vect_create_epilog_for_reduction): Get the SLP instance as
14450 (vectorizable_reduction): Get the SLP instance as arg.
14451 During analysis remember the SLP node with the PHIs in the
14452 instance. Simplify getting at the vectorized reduction PHIs.
14453 * tree-vect-slp.c (vect_slp_analyze_node_operations): Pass
14454 through SLP instance.
14455 (vect_slp_analyze_operations): Likewise.
14456 * tree-vect-stms.c (vect_analyze_stmt): Likewise.
14457 (vect_transform_stmt): Likewise.
14459 2017-07-20 Tom de Vries <tom@codesourcery.com>
14461 PR tree-optimization/81489
14462 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Move
14463 read of phi arg location to before loop that modifies phi.
14465 2017-07-20 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14467 * match.pd (((m1 >/</>=/<= m2) * d -> (m1 >/</>=/<= m2) ? d : 0):
14470 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14472 PR middle-end/81331
14473 * except.c (execute): Fix ordering issue.
14475 2018-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14477 PR rtl-optimization/81423
14478 * combine.c (make_compound_operation_int): Don't try to optimize
14479 the AND of a SUBREG of an LSHIFTRT if that SUBREG is paradoxical.
14481 2017-07-19 Segher Boessenkool <segher@kernel.crashing.org>
14483 PR rtl-optimization/81423
14484 * simplify-rtx.c (simplify_truncation): Handle truncating an IOR
14485 with a constant that is -1 in the truncated to mode.
14487 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14489 * predict.c (propagate_unlikely_bbs_forward): Break out from ...
14490 (determine_unlikely_bbs): ... here.
14491 * predict.h (propagate_unlikely_bbs_forward): Declare.
14492 * cfgexpand.c (pass_expand::execute): Use it.
14493 * bb-reorder.c (sanitize_hot_paths): Do not consider known to be
14495 (find_rarely_executed_basic_blocks_and_crossing_edges): Use
14496 propagate_unlikely_bbs_forward.
14498 2017-07-19 Jan Hubicka <hubicka@ucw.cz>
14500 PR middle-end/81331
14501 * except.c (maybe_add_nop_after_section_switch): New function.
14504 2017-07-19 Tom de Vries <tom@codesourcery.com>
14506 * gimple.h (gimple_phi_set_arg): Make assert more strict.
14508 2017-07-19 Tom de Vries <tom@codesourcery.com>
14510 * gimple.h (gimple_phi_arg): Make assert more strict.
14512 2017-07-19 Steven Munroe <munroesj@gcc.gnu.org>
14514 * config.gcc (powerpc*-*-*): Add mmintrin.h.
14515 * config/rs6000/mmintrin.h: New file.
14516 * config/rs6000/x86intrin.h [__ALTIVEC__]: Include mmintrin.h.
14518 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14520 PR tree-optimization/81346
14521 * match.pd: Optimize (X - 1U) <= INT_MAX-1U into (int) X > 0.
14523 2017-07-19 Tom de Vries <tom@codesourcery.com>
14525 * config/nvptx/nvptx.md (VECIM): Add V2DI.
14527 2017-07-19 Tom de Vries <tom@codesourcery.com>
14529 * config/nvptx/nvptx-modes.def: Add V2DImode.
14530 * config/nvptx/nvptx-protos.h (nvptx_data_alignment): Declare.
14531 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2DImode.
14532 (nvptx_output_mov_insn): Handle lack of mov.b128.
14533 (nvptx_print_operand): Handle 'H' and 'L' codes.
14534 (nvptx_vector_mode_supported): Allow V2DImode.
14535 (nvptx_preferred_simd_mode): New function.
14536 (nvptx_data_alignment): New function.
14537 (TARGET_VECTORIZE_PREFERRED_SIMD_MODE): Redefine to
14538 nvptx_preferred_simd_mode.
14539 * config/nvptx/nvptx.h (STACK_BOUNDARY, BIGGEST_ALIGNMENT): Change from
14541 (DATA_ALIGNMENT): Define. Set to nvptx_data_alignment.
14543 2017-07-19 Tom de Vries <tom@codesourcery.com>
14545 * config/nvptx/nvptx-modes.def: New file. Add V2SImode.
14546 * config/nvptx/nvptx.c (nvptx_ptx_type_from_mode): Handle V2SImode.
14547 (nvptx_vector_mode_supported): New function. Allow V2SImode.
14548 (TARGET_VECTOR_MODE_SUPPORTED_P): Redefine to nvptx_vector_mode_supported.
14549 * config/nvptx/nvptx.md (VECIM): New mode iterator. Add V2SI.
14550 (mov<VECIM>_insn): New define_insn.
14551 (define_expand "mov<VECIM>): New define_expand.
14553 2017-07-19 Tom de Vries <tom@codesourcery.com>
14555 * config/nvptx/nvptx.c (nvptx_print_operand): Handle v2 vector mode.
14557 2017-07-19 Jakub Jelinek <jakub@redhat.com>
14559 PR tree-optimization/81346
14560 * fold-const.h (fold_div_compare, range_check_type): Declare.
14561 * fold-const.c (range_check_type): New function.
14562 (build_range_check): Use range_check_type.
14563 (fold_div_compare): No longer static, rewritten into
14564 a match.pd helper function.
14565 (fold_comparison): Don't call fold_div_compare here.
14566 * match.pd (X / C1 op C2): New optimization using fold_div_compare
14567 as helper function.
14569 2017-07-19 Nathan Sidwell <nathan@acm.org>
14571 * tree.h (TYPE_MINVAL, TYPE_MAXVAL): Rename to ...
14572 (TYPE_MIN_VALUE_RAW, TYPE_MAX_VALUE_RAW): ... these.
14573 * tree.c (find_decls_types_r, verify_type): Use
14574 TYPE_{MIN,MAX}_VALUE_RAW.
14575 * lto-streamer-out.c (DFS::DFS_write_tree_body): Likewise.
14576 (hash_tree): Likewise.
14577 * tree-streamer-in.c (lto_input_ts_type_non_common_tree_pointers):
14579 * tree-streamer-out.c (write_ts_type_non_common_tree_pointers):
14582 2017-07-18 Tom de Vries <tom@codesourcery.com>
14584 PR middle-end/81464
14585 * omp-expand.c (expand_omp_for_static_chunk): Handle
14586 equal-argument loop exit phi.
14588 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
14591 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
14592 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
14593 operand 2 predicate.
14594 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
14595 operand 2 predicate.
14596 (ror,rol -> rorx splitters): Use const_int_operand as
14597 operand 2 predicate.
14599 2017-06-18 Richard Biener <rguenther@suse.de>
14601 PR tree-optimization/81410
14602 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
14603 the gap in the ! slp_perm SLP case after each group.
14605 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14607 PR middle-end/81463
14608 * cfgloopmanip.c (scale_loop_profile): Watch out for zero frequency
14611 2017-07-18 Jan Hubicka <hubicka@ucw.cz>
14613 PR middle-end/81462
14614 * predict.c (set_even_probabilities): Cleanup; do not affect
14615 probabilities that are already known.
14616 (combine_predictions_for_bb): Call even when count is set.
14618 2017-07-18 Nathan Sidwell <nathan@acm.org>
14620 * tree-parloops.c (try_transform_to_exit_first_loop_alt): Use
14623 2017-07-18 Bin Cheng <bin.cheng@arm.com>
14626 * tree-ssa-loop-niter.c (number_of_iterations_exit): Dump missed
14627 optimization for loop niter analysis.
14629 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
14632 * config/avr/avr.c (avr_optimize_casesi): Don't use
14633 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
14635 2017-07-18 Robin Dapp <rdapp@linux.vnet.ibm.com>
14637 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Remove
14638 body_cost_vec from _vect_peel_extended_info.
14639 (vect_peeling_hash_get_lowest_cost): Do not set body_cost_vec.
14640 (vect_peeling_hash_choose_best_peeling): Remove body_cost_vec and
14643 2017-07-18 Bin Cheng <bin.cheng@arm.com>
14645 * config/arm/arm.c (emit_unlikely_jump): Remove unused var.
14647 2017-07-18 Richard Biener <rguenther@suse.de>
14649 PR tree-optimization/80620
14650 PR tree-optimization/81403
14651 * tree-ssa-pre.c (phi_translate_1): Clear range and points-to
14652 info when re-using a VN table entry.
14654 2017-07-18 Richard Biener <rguenther@suse.de>
14656 PR tree-optimization/81418
14657 * tree-vect-loop.c (vectorizable_reduction): Properly compute
14658 vectype_in. Verify that with lane-reducing reduction operations
14659 we have a single def-use cycle.
14661 2017-07-17 Carl Love <cel@us.ibm.com>
14663 Revert commit r249424 2017-06-20 Carl Love <cel@us.ibm.com>
14665 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14666 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
14667 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
14668 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
14669 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
14670 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
14671 VMULOSW): New enum "unspec" values.
14672 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
14673 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
14674 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
14675 altivec_vmulosw): New patterns.
14676 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
14677 VMULOSW): Add definitions.
14678 2017-07-17 Uros Bizjak <ubizjak@gmail.com>
14680 * config/alpha/alpha.c: Include predict.h.
14682 2017-07-17 Yury Gribov <tetra2005@gmail.com>
14684 * tree-vrp.c (compare_assert_loc): Fix comparison function
14685 to return predictable results.
14687 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14689 * config/arc/arc.md (adddi3): Remove support for mexpand-adddi
14691 (subdi3): Likewise.
14692 * config/arc/arc.opt (mexpand-adddi): Deprecate it.
14693 * doc/invoke.texi (mexpand-adddi): Update text.
14695 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14697 * config/arc/arc.md (clzsi2): Expand to an arc_clzsi2 instruction
14698 that also clobbers the CC register. The old expand code is moved
14700 (*arc_clzsi2): ... here.
14701 (ctzsi2): Expand to an arc_ctzsi2 instruction that also clobbers
14702 the CC register. The old expand code is moved to ...
14703 (arc_ctzsi2): ... here.
14705 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14707 * config/arc/arc.opt (mindexed-loads): Use initial value
14708 TARGET_INDEXED_LOADS_DEFAULT.
14709 (mauto-modify-reg): Use initial value
14710 TARGET_AUTO_MODIFY_REG_DEFAULT.
14711 * config/arc/elf.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14712 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14713 * config/arc/linux.h (TARGET_INDEXED_LOADS_DEFAULT): Define.
14714 (TARGET_AUTO_MODIFY_REG_DEFAULT): Likewise.
14716 2017-07-17 Martin Liska <mliska@suse.cz>
14719 * opts.c (finish_options): Do not allow -fgnu-tm
14720 w/ -fsanitize={kernel-,}address. Say sorry.
14722 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14725 * tree-loop-distribution.c (classify_partition): Only assert on
14726 numer of iterations.
14727 (merge_dep_scc_partitions): Delete prameter. Update function call.
14728 (distribute_loop): Remove code handling loop with unknown niters.
14729 (pass_loop_distribution::execute): Skip loop with unknown niters.
14731 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14734 * tree-loop-distribution.c (merge_dep_scc_partitions): Sink call to
14735 function sort_partitions_by_post_order.
14737 2017-07-17 Bin Cheng <bin.cheng@arm.com>
14739 PR tree-optimization/81374
14740 * tree-loop-distribution.c (pass_loop_distribution::execute): Record
14741 the max index of basic blocks, rather than number of basic blocks.
14743 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14745 * config/arc/arc-protos.h (arc_legitimate_pc_offset_p): Remove
14747 (arc_legitimate_pic_operand_p): Likewise.
14748 * config/arc/arc.c (arc_legitimate_pic_operand_p): Remove
14750 (arc_needs_pcl_p): Likewise.
14751 (arc_legitimate_pc_offset_p): Likewise.
14752 (arc_legitimate_pic_addr_p): Remove LABEL_REF case, as this
14753 function is also used in constrains.md.
14754 (arc_legitimate_constant_p): Use arc_legitimate_pic_addr_p to
14755 validate pic constants. Handle CONST_INT, CONST_DOUBLE, MINUS and
14756 PLUS. Only return true/false in known cases, otherwise assert.
14757 (arc_legitimate_address_p): Remove arc_legitimate_pic_addr_p as it
14758 is already called in arc_legitimate_constant_p.
14759 * config/arc/arc.h (CONSTANT_ADDRESS_P): Consider also LABEL for
14761 (LEGITIMATE_PIC_OPERAND_P): Use
14762 arc_raw_symbolic_reference_mentioned_p function.
14763 * config/arc/constraints.md (Cpc): Use arc_legitimate_pic_addr_p
14768 2017-07-17 Claudiu Zissulescu <claziss@synopsys.com>
14769 Andrew Burgess <andrew.burgess@embecosm.com>
14771 * config/arc/arc-protos.h (arc_compute_function_type): Change prototype.
14772 (arc_return_address_register): New function.
14773 * config/arc/arc.c (arc_handle_fndecl_attribute): New function.
14774 (arc_handle_fndecl_attribute): Add naked attribute.
14775 (TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS): Define.
14776 (TARGET_WARN_FUNC_RETURN): Likewise.
14777 (arc_allocate_stack_slots_for_args): New function.
14778 (arc_warn_func_return): Likewise.
14779 (machine_function): Change type fn_type.
14780 (arc_compute_function_type): Consider new naked function type,
14781 change function return type.
14782 (arc_must_save_register): Adapt to handle new
14783 arc_compute_function_type's return type.
14784 (arc_expand_prologue): Likewise.
14785 (arc_expand_epilogue): Likewise.
14786 (arc_return_address_regs): Delete.
14787 (arc_return_address_register): New function.
14788 (arc_epilogue_uses): Use above function.
14789 * config/arc/arc.h (arc_return_address_regs): Delete prototype.
14790 (arc_function_type): Change encoding, add naked type.
14791 (ARC_INTERRUPT_P): Change to handle the new encoding.
14792 (ARC_FAST_INTERRUPT_P): Likewise.
14793 (ARC_NORMAL_P): Define.
14794 (ARC_NAKED_P): Likewise.
14795 (arc_compute_function_type): Delete prototype.
14796 * config/arc/arc.md (in_ret_delay_slot): Use
14797 arc_return_address_register function.
14798 (simple_return): Likewise.
14799 (p_return_i): Likewise.
14801 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14803 PR tree-optimization/81428
14804 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
14805 can't be built for those types.
14807 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14809 Remove stuff dead since r239246.
14811 * config/avr/avr-arch.h (avr_inform_devices): Remove dead proto.
14812 * config/avr/avr-devices.c (mcu_name, comparator, avr_mcus_str)
14813 (avr_inform_devices): Remove dead stuff.
14815 2017-07-17 Tamar Christina <tamar.christina@arm.com>
14817 * config/arm/arm_neon.h: Fix softp typo.
14819 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14821 PR tree-optimization/81365
14822 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
14823 aggregate moves onto bb predecessor edges, make sure there are no
14824 loads that could alias the lhs in between the start of bb and the
14827 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
14830 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
14831 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
14832 [LSHIFTRT, outer_code = TRUNCATE]: Same.
14834 2017-07-17 Jakub Jelinek <jakub@redhat.com>
14836 PR tree-optimization/81396
14837 * tree-ssa-math-opts.c (struct symbolic_number): Add n_ops field.
14838 (init_symbolic_number): Initialize it to 1.
14839 (perform_symbolic_merge): Add n_ops from both operands into the new
14841 (find_bswap_or_nop): Don't consider n->n == cmpnop computations
14842 without base_addr as useless if they need more than one operation.
14843 (bswap_replace): Handle !bswap case for NULL base_addr.
14845 2017-07-17 Tom de Vries <tom@codesourcery.com>
14848 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
14851 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
14853 * config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
14854 conditional builtin define __FIX_LEON3FT_B2BST.
14856 2017-07-17 Daniel Cederman <cederman@gaisler.com>
14858 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
14859 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
14862 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14864 PR rtl-optimization/81424
14865 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
14866 to remove potential trapping from operands if -fnon-call-exceptions.
14868 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14870 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Use
14871 profile_proability for scalling.
14872 * scale_profile_for_vect_loop.c (scale_profile_for_vect_loop): Likewise.
14874 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14876 * cgraph.c (cgraph_edge::redirect_call_stmt_to_caller): Cleanup.
14878 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14880 * cfgloopmanip.c (scale_loop_profile): Avoid use of REG_BR_PROB_BASE
14881 fixpoint arithmetics.
14883 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14885 * tree-ssa-loop-unswitch.c (hoist_guard): Avoid use of REG_BR_PROB_BASE
14886 fixpoint arithmetics.
14888 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14890 * asan.c (create_cond_insert_point): Avoid use of REG_BR_PROB_BASE
14891 fixpoint arithmetics.
14893 2017-07-16 Jan Hubicka <hubicka@ucw.cz>
14895 * profile-count.h (profile_probability::from_reg_br_prob_note,
14896 profile_probability::to_reg_br_prob_note): New functions.
14897 * doc/rtl.texi (REG_BR_PROB_NOTE): Update documentation.
14898 * reg-notes.h (REG_BR_PROB, REG_BR_PRED): Update docs.
14899 * predict.c (probability_reliable_p): Update.
14900 (edge_probability_reliable_p): Update.
14901 (br_prob_note_reliable_p): Update.
14902 (invert_br_probabilities): Update.
14903 (add_reg_br_prob_note): New function.
14904 (combine_predictions_for_insn): Update.
14905 * asan.c (asan_clear_shadow): Update.
14906 * cfgbuild.c (compute_outgoing_frequencies): Update.
14907 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
14908 (update_br_prob_note): Update.
14909 (rtl_verify_edges): Update.
14910 (purge_dead_edges): Update.
14911 (fixup_reorder_chain): Update.
14912 * emit-rtl.c (try_split): Update.
14913 * ifcvt.c (cond_exec_process_insns): Update.
14914 (cond_exec_process_if_block): Update.
14915 (dead_or_predicable): Update.
14916 * internal-fn.c (expand_addsub_overflow): Update.
14917 (expand_neg_overflow): Update.
14918 (expand_mul_overflow): Update.
14919 * loop-doloop.c (doloop_modify): Update.
14920 * loop-unroll.c (compare_and_jump_seq): Update.
14921 * optabs.c (emit_cmp_and_jump_insn_1): Update.
14922 * predict.h: Update.
14923 * reorg.c (mostly_true_jump): Update.
14925 * config/aarch64/aarch64.c (aarch64_emit_unlikely_jump): Update.
14926 * config/alpha/alpha.c (emit_unlikely_jump): Update.
14927 * config/arc/arc.c: (emit_unlikely_jump): Update.
14928 * config/arm/arm.c: (emit_unlikely_jump): Update.
14929 * config/bfin/bfin.c (cbranch_predicted_taken_p): Update.
14930 * config/frv/frv.c (frv_print_operand_jump_hint): Update.
14931 * config/i386/i386.c (ix86_expand_split_stack_prologue): Update.
14932 (ix86_print_operand): Update.
14933 (ix86_split_fp_branch): Update.
14934 (predict_jump): Update.
14935 * config/ia64/ia64.c (ia64_print_operand): Update.
14936 * config/mmix/mmix.c (mmix_print_operand): Update.
14937 * config/powerpcspe/powerpcspe.c (output_cbranch): Update.
14938 (rs6000_expand_split_stack_prologue): Update.
14939 * config/rs6000/rs6000.c: Update.
14940 * config/s390/s390.c (s390_expand_vec_strlen): Update.
14941 (s390_expand_vec_movstr): Update.
14942 (s390_expand_cs_tdsi): Update.
14943 (s390_expand_split_stack_prologue): Update.
14944 * config/sh/sh.c (sh_print_operand): Update.
14945 (expand_cbranchsi4): Update.
14946 (expand_cbranchdi4): Update.
14947 * config/sparc/sparc.c (output_v9branch): Update.
14948 * config/spu/spu.c (get_branch_target): Update.
14949 (ea_load_store_inline): Update.
14950 * config/tilegx/tilegx.c (cbranch_predicted_p): Update.
14951 * config/tilepro/tilepro.c: Update.
14953 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
14955 * gimplify.c (mostly_copy_tree_r): Revert latest change.
14956 (gimplify_save_expr): Likewise.
14958 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14960 * ipa-visibility.c (function_and_variable_visibility): Fix pasto.
14962 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14964 * ipa-fnsummary.c (pass_data_ipa_fn_summary): Use
14966 * timevar.def (TV_IPA_FNSUMMARY): Define.
14968 2017-07-16 Daniel Cederman <cederman@gaisler.com>
14970 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
14971 to back store errata sensitive sequence from being generated.
14972 (sqrtdf2_fix): Likewise.
14974 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
14976 * tree-ssa-threadupdate.c (compute_path_counts,
14977 update_joiner_offpath_counts): Use profile_probability.
14979 2017-07-15 Thomas Preud'homme <thomas.preudhomme@arm.com>
14982 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
14984 * config/arm/arm-c.c (arm_cpu_builtins): Define
14985 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
14987 2017-07-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
14989 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
14990 array entries to represent __ieee128 versions of the
14991 scalar_test_data_class, scalar_test_neg, scalar_extract_exp,
14992 scalar_extract_sig, and scalar_insert_exp built-in functions.
14993 (altivec_resolve_overloaded_builtin): Add special case handling
14994 for the __builtin_scalar_insert_exp function, as represented by
14995 the P9V_BUILTIN_VEC_VSIEDP constant.
14996 * config/rs6000/rs6000-builtin.def (VSEEQP): Add scalar extract
14997 exponent support for __ieee128 argument.
14998 (VSESQP): Add scalar extract signature support for __ieee128
15000 (VSTDCNQP): Add scalar test negative support for __ieee128
15002 (VSIEQP): Add scalar insert exponent support for __int128 argument
15003 with __ieee128 result.
15004 (VSIEQPF): Add scalar insert exponent support for __ieee128
15005 argument with __ieee128 result.
15006 (VSTDCQP): Add scalar test data class support for __ieee128
15008 (VSTDCNQP): Add overload support for scalar test negative with
15009 __ieee128 argument.
15010 (VSTDCQP): Add overload support for scalar test data class
15011 __ieee128 argument.
15012 * config/rs6000/vsx.md (UNSPEC_VSX_SXSIG) Replace
15013 UNSPEC_VSX_SXSIGDP.
15014 (UNSPEC_VSX_SIEXPQP): New constant.
15015 (xsxexpqp): New insn for VSX scalar extract exponent quad
15017 (xsxsigqp): New insn for VSX scalar extract significand quad
15019 (xsiexpqpf): New insn for VSX scalar insert exponent quad
15020 precision with floating point argument.
15021 (xststdcqp): New expand for VSX scalar test data class quad
15023 (xststdcnegqp): New expand for VSX scalar test negative quad
15025 (xststdcqp): New insn to match expansions for VSX scalar test data
15026 class quad precision and VSX scalar test negative quad precision.
15027 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Add
15028 special case operand checking to enforce that second operand of
15029 VSX scalar test data class with quad precision argument is a 7-bit
15031 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
15032 prototypes and descriptions of __ieee128 versions of
15033 scalar_extract_exp, scalar_extract_sig, scalar_insert_exp,
15034 scalar_test_data_class, and scalar_test_neg built-in functions.
15036 2016-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15038 PR tree-optimization/81162
15039 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
15040 replace a negate with an add.
15042 2017-07-14 James Greenhalgh <james.greenhalgh@arm.com>
15044 * doc/invoke.texi (arm/-mcpu): Document +crypto.
15046 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15048 * config/arm/arm-c.c (arm_cpu_builtins): Define
15049 __ARM_FEATURE_NUMERIC_MAXMIN solely based on TARGET_VFP5.
15051 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15053 * config/arm/arm-cpus.in (cortex-r52): Add new entry.
15054 (armv8-r): Set ARM Cortex-R52 as default CPU.
15055 * config/arm/arm-tables.opt: Regenerate.
15056 * config/arm/arm-tune.md: Regenerate.
15057 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
15059 * doc/invoke.texi: Mention -mtune=cortex-r52 and availability of fp.dp
15060 extension for -mcpu=cortex-r52.
15062 2017-07-14 Thomas Preud'homme <thomas.preudhomme@arm.com>
15064 * config/arm/arm-isa.h (isa_bit_FP_ARMv8): Delete enumerator.
15065 (ISA_FP_ARMv8): Define as ISA_FPv5 and ISA_FP_D32.
15066 * config/arm/arm-cpus.in (armv8-r): Define fp.sp as enabling FPv5.
15067 (fp-armv8): Define it as FP_ARMv8 only.
15068 config/arm/arm.h (TARGET_FPU_ARMV8): Delete.
15069 (TARGET_VFP_FP16INST): Define using TARGET_VFP5 rather than
15071 config/arm/arm.c (arm_rtx_costs_internal): Replace checks against
15072 TARGET_FPU_ARMV8 by checks against TARGET_VFP5.
15073 * config/arm/arm-builtins.c (arm_builtin_vectorized_function): Define
15074 first ARM_CHECK_BUILTIN_MODE definition using TARGET_VFP5 rather
15075 than TARGET_FPU_ARMV8.
15076 * config/arm/arm-c.c (arm_cpu_builtins): Likewise for
15077 __ARM_FEATURE_NUMERIC_MAXMIN macro definition.
15078 * config/arm/arm.md (cmov<mode>): Condition on TARGET_VFP5 rather than
15080 * config/arm/neon.md (neon_vrint): Likewise.
15081 (neon_vcvt): Likewise.
15082 (neon_<fmaxmin_op><mode>): Likewise.
15083 (<fmaxmin><mode>3): Likewise.
15084 * config/arm/vfp.md (l<vrint_pattern><su_optab><mode>si2): Likewise.
15085 * config/arm/predicates.md (arm_cond_move_operator): Check against
15086 TARGET_VFP5 rather than TARGET_FPU_ARMV8 and fix spacing.
15088 2017-07-14 Jackson Woodruff <jackson.woodruff@arm.com>
15090 * config/aarch64/aarch64.c (aarch64_print_operand): Move comments
15091 to top of function.
15093 2017-07-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15095 * gimple-ssa-store-merging.c (clear_bit_region): Replace reference to
15096 loop in comment with memset.
15098 2017-07-14 Martin Liska <mliska@suse.cz>
15100 * cfgexpand.c (expand_gimple_basic_block): Remove dead comment.
15101 * dwarf2out.c (is_java): Remove the function.
15102 (output_pubname): Remove usage of the function.
15103 (lower_bound_default): Remove usage of DW_LANG_Java.
15104 (gen_compile_unit_die): Likewise.
15105 * gcc.c: Remove compiler defaults for .java and .zip files.
15106 * gimple-expr.c (remove_suffix): Change as there's no longer
15107 extension than 4-letter one.
15108 * gimplify.c (mostly_copy_tree_r): Remove Java-special part.
15109 (gimplify_save_expr): Likewise.
15110 * ipa-utils.h (polymorphic_type_binfo_p): Remove the comment
15111 as it's possible even for other languages than Java.
15112 * langhooks.h (struct lang_hooks): Remove Java from a comment.
15113 * lto-opts.c (lto_write_options): Remove reference to Java.
15114 * opts.c (strip_off_ending): Update file extension handling.
15115 * tree-cfg.c (verify_gimple_call): Remove comment with Java.
15116 * tree-eh.c (lower_resx): Likewise.
15117 * tree.c (free_lang_data_in_type): Remove dead code.
15118 (find_decls_types_r): Likewise.
15119 (build_common_builtin_nodes): Remove Java from a comment.
15120 (verify_type): Remove dead code.
15121 * varasm.c (assemble_external): Remove Java from a comment.
15123 2017-07-14 Martin Liska <mliska@suse.cz>
15125 * opts.c (finish_options): Add quotes.
15126 (common_handle_option): Likewise.
15128 2017-07-14 Martin Liska <mliska@suse.cz>
15130 * dbxout.c (get_lang_number): Do not handle GNU Pascal.
15131 * dbxout.h (extern void dbxout_stab_value_internal_label_diff):
15132 Remove N_SO_PASCAL.
15133 * dwarf2out.c (lower_bound_default): Do not handle
15135 (gen_compile_unit_die): Likewise.
15136 * gcc.c: Remove default extension binding for GNU Pascal.
15137 * stmt.c: Remove Pascal language from a comment.
15138 * xcoffout.c: Likewise.
15140 2017-07-13 David Malcolm <dmalcolm@redhat.com>
15143 * diagnostic-show-locus.c (fixit_cmp): New function.
15144 (layout::layout): Sort m_fixit_hints.
15145 (column_range::column_range): Assert that the values are valid.
15146 (struct char_span): New struct.
15147 (correction::overwrite): New method.
15148 (struct source_line): New struct.
15149 (line_corrections::add_hint): Add assertions. Reimplement memcpy
15150 calls in terms of classes source_line and char_span, and
15151 correction::overwrite.
15152 (selftest::test_overlapped_fixit_printing_2): New function.
15153 (selftest::diagnostic_show_locus_c_tests): Call it.
15155 2017-07-13 Will Schmidt <will_schmidt@vnet.ibm.com>
15157 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Return
15158 early if there is no lhs.
15160 2017-07-13 Martin Liska <mliska@suse.cz>
15162 * dwarf2out.c (gen_pointer_type_die): Remove dead code.
15163 (gen_reference_type_die): Likewise.
15164 * stor-layout.c: Remove Pascal-related comment.
15166 2017-07-13 Martin Liska <mliska@suse.cz>
15168 * opts.c (finish_options): Add quotes to error messages.
15169 (parse_sanitizer_options): Likewise.
15171 2017-07-13 Thomas Preud'homme <thomas.preudhomme@arm.com>
15173 * doc/invoke.texi (armv8-r): Document +fp.sp ARMv8-R extension.
15175 2017-07-13 Richard Earnshaw <rearnsha@arm.com>
15177 * config/arm/vxworks.h (TARGET_ENDIAN_DEFAULT): Define.
15179 2017-07-13 Maxim Ostapenko <m.ostapenko@samsung.com>
15181 * asan.c (asan_emit_allocas_unpoison): Use ptr_mode for arguments
15183 * builtins.c (expand_asan_emit_allocas_unpoison): Likewise.
15185 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
15188 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
15189 provides the hardware capability bits, define the macro
15190 __BUILTIN_CPU_SUPPORTS__.
15191 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
15192 if GLIBC does not provide the hardware capability bits. Add a
15193 gcc_unreachable call if the built-in cpu function is neither
15194 __builtin_cpu_is nor __builtin_cpu_supports.
15195 (rs6000_get_function_versions_dispatcher): Change the warning
15196 that an old GLIBC is used which does not export the capability
15197 bits to be an error.
15198 * doc/extend.texi (target_clones attribute): Document the
15199 restriction that GLIBC 2.23 or newer is needed on the PowerPC.
15200 (PowerPC built-in functions): Document that GLIBC 2.23 or newer is
15201 needed by __builtin_cpu_is and __builtin_cpu_supports. Document
15202 the macros defined by GCC if the newer GLIBC is available.
15204 2017-07-12 Jeff Law <law@redhat.com>
15206 * config/riscv/riscv.c: Remove unnecessary includes. Reorder
15207 remaining includes slightly.
15208 * config/riscv/riscv-builtins.c: Include profile-count.h.
15210 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15213 * config/avr/avr.c (avr_set_current_function): In diagnostic
15214 messages: Quote keywords and (parts of) identifiers.
15215 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
15218 2017-07-12 Carl Love <cel@us.ibm.com>
15220 * config/rs6000/rs6000-c.c: Add support for built-in functions
15221 vector bool char vec_revb (vector bool char);
15222 vector bool short vec_revb (vector short char);
15223 vector bool int vec_revb (vector bool int);
15224 vector bool long long vec_revb (vector bool long long);
15225 * doc/extend.texi: Update the built-in documentation file for the
15226 new built-in functions.
15228 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15230 * config/s390/s390.md: Remove movcc splitter.
15232 2017-07-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15234 * config/s390/s390.c (s390_rtx_costs): Return proper costs for
15235 load/store on condition.
15237 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
15240 * config/avr/avr.c (avr_encode_section_info)
15241 [progmem && !TREE_READONLY]: Error if progmem object needs
15244 2017-07-11 Michael Collison <michael.collison@arm.com>
15246 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
15249 2017-07-11 Carl Love <cel@us.ibm.com>
15251 * config/rs6000/rs6000-c.c: Add support for builtins
15252 vector unsigned int vec_parity_lsbb (vector signed int);
15253 vector unsigned int vec_parity_lsbb (vector unsigned int);
15254 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
15255 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
15256 vector unsigned long long vec_parity_lsbb (vector signed long long);
15257 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
15258 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
15259 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
15260 * doc/extend.texi: Update the built-in documentation file for the
15261 new built-in functions.
15263 2017-07-11 David Malcolm <dmalcolm@redhat.com>
15265 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
15266 (layout::m_primary_loc): New field.
15267 (layout::layout): Initialize new field. Move location filtering
15268 logic from here to...
15269 (layout::maybe_add_location_range): ...this new method. Add
15270 support for filtering to just the lines already specified by other
15272 (layout::will_show_line_p): New method.
15273 (gcc_rich_location::add_location_if_nearby): New method.
15274 (selftest::test_add_location_if_nearby): New test function.
15275 (selftest::diagnostic_show_locus_c_tests): Call it.
15276 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
15279 2017-07-11 Tom de Vries <tom@codesourcery.com>
15281 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
15282 (bb_first_real_insn): New function.
15283 (nvptx_single): Add extra initialization of broadcasted condition
15286 2017-07-11 Nathan Sidwell <nathan@acm.org>
15288 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
15290 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
15292 * doc/extend.texi (AVR Function Attributes): Remove weblink to
15293 Binutils doc as TEXI will mess them up.
15294 * doc/invoke.texi (AVR Options): Same here.
15296 2017-07-11 Daniel Cederman <cederman@gaisler.com>
15298 * config/sparc/sparc.opt (mfix-ut700): New option.
15299 (mfix-gr712rc): Likewise.
15300 (sparc_fix_b2bst): New variable.
15301 * doc/invoke.texi (SPARC options): Document them.
15302 (ARM options): Fix warnings.
15303 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
15304 instructions to prevent sequences that can trigger the store-store
15305 errata for certain LEON3FT processors.
15306 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
15307 (sparc_option_override): Set sparc_fix_b2bst appropriately.
15308 * config/sparc/sparc.md (fix_b2bst): New attribute.
15309 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
15311 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
15314 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
15316 (*rsqrtsf2_sse): Ditto.
15318 (div<mode>3): Macroize insn from divdf3 and divsf3
15319 using MODEF mode iterator.
15321 2017-07-10 Martin Sebor <msebor@redhat.com>
15323 PR tree-optimization/80397
15324 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
15325 instead of testing for equality to INTEGER_TYPE.
15327 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
15329 * config.gcc: Remove uclibc from arc target spec.
15331 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
15333 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
15335 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
15338 * lto-wrapper.c (remove_option): New function.
15339 (merge_and_complain): Merge PIC/PIE options more realistically.
15341 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15343 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
15347 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
15348 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
15349 * config.in: Regenerate.
15350 * configure: Regenerate.
15351 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
15352 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
15353 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
15354 (TARGET_GASISR_PROLOGUES): ...target mask.
15355 * common/config/avr/avr-common.c
15356 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
15357 Set -mgas-isr-prologues.
15358 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
15359 INSERT_PASS_BEFORE for it.
15360 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
15361 * config/avr/avr.c (avr_option_override)
15362 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
15363 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
15364 (avr_attribute_table) <no_gccisr>: Add new function attribute.
15365 (avr_set_current_function) <is_no_gccisr>: Init machine field.
15366 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
15368 (make_avr_pass_pre_proep): New function.
15369 (emit_push_sfr) <treg>: Add argument to function and use it
15370 instead of TMP_REG.
15371 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
15372 and set machine->gasisr.yes.
15373 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
15374 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
15375 __gcc_isr.n_pushed to .L__stack_usage.
15376 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
15377 (avr_asm_final_postscan_insn): ...this new static function.
15378 * config/avr/avr.h (machine_function)
15379 <is_no_gccisr, use_L__stack_usage>: New fields.
15380 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
15381 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
15382 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
15383 (gasisr, *gasisr): New expander and insn.
15384 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
15385 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
15386 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
15388 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
15390 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
15393 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
15395 Move jump-tables out of .text again.
15398 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
15399 (ASM_OUTPUT_ADDR_VEC): New function.
15400 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
15401 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
15402 INSN_ADDRESSes as asm comment.
15403 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
15404 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
15405 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
15406 * config/avr/avr.md (*tablejump): Adjust comment.
15407 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
15408 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
15410 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
15411 (avr_output_addr_vec): New proto.
15412 (avr_log_t) <insn_addresses>: New field.
15414 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
15417 * config/i386/i386.c (ix86_function_arg_advance): Set
15418 outgoing_args_on_stack to true if there are outgoing arguments
15420 (ix86_function_arg): Likewise.
15421 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
15422 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
15423 * config/i386/i386.h (machine_function): Add
15424 outgoing_args_on_stack.
15426 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
15428 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
15429 supporting pthreds.
15430 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
15432 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
15434 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
15435 (REAL_H): Remove $(MACHMODE_H).
15436 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
15438 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
15439 $(MACHMODE_H) and double-int.h.
15440 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
15442 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
15445 2017-07-07 Andrew Pinski <apinski@cavium.com>
15447 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
15448 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
15450 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
15452 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
15453 Add warning if GCC was not configured to link against a GLIBC that
15454 exports the hardware capability bits.
15455 (make_resolver_func): Make resolver function private and not a
15456 COMDAT function. Create the name with clone_function_name instead
15457 of make_unique_name.
15460 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
15461 correct operand in doing the split.
15463 2017-07-07 Carl Love <cel@us.ibm.com>
15465 * config/rs6000/rs6000-c: Add support for built-in function
15466 vector unsigned short vec_pack_to_short_fp32 (vector float,
15468 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
15469 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
15470 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
15471 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
15472 (convert_4f32_8i16): Add define_expand.
15473 * doc/extend.texi: Update the built-in documentation file for the
15474 new built-in function.
15476 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15478 * config/sparc/m8.md: New file.
15479 * config/sparc/sparc.md: Include m8.md.
15481 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15483 * config/sparc/sparc.opt: New option -mvis4b.
15484 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
15485 (sparc_option_override): Handle VIS4B.
15486 (enum sparc_builtins): Define
15487 SPARC_BUILTIN_DICTUNPACK{8,16,32},
15488 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
15489 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
15490 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
15491 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
15492 (check_constant_argument): New function.
15493 (sparc_vis_init_builtins): Define builtins
15494 __builtin_vis_dictunpack{8,16,32},
15495 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
15496 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
15497 __builtin_vis_fpcmpde{8,16,32}shl and
15498 __builtin_vis_fpcmpur{8,16,32}shl.
15499 (sparc_expand_builtin): Check that the constant operands to
15500 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
15501 constant and in range.
15502 * config/sparc/sparc-c.c (sparc_target_macros): Handle
15504 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
15505 (SPARC_IMM5_P): Likewise.
15506 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
15507 (enabled): Handle vis4b.
15508 (UNSPEC_DICTUNPACK): New unspec.
15509 (UNSPEC_FPCMPSHL): Likewise.
15510 (UNSPEC_FPUCMPSHL): Likewise.
15511 (UNSPEC_FPCMPDESHL): Likewise.
15512 (UNSPEC_FPCMPURSHL): Likewise.
15513 (cpu_feature): New CPU feature `vis4b'.
15514 (dictunpack{8,16,32}): New insns.
15515 (FPCSMODE): New mode iterator.
15516 (fpcscond): New code iterator.
15517 (fpcsucond): Likewise.
15518 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
15519 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
15520 (fpcmpde{8,16,32}{si,di}shl): Likewise.
15521 (fpcmpur{8,16,32}{si,di}shl): Likewise.
15522 * config/sparc/constraints.md: Define constraints `q' for unsigned
15523 2-bit integer constants and `t' for unsigned 5-bit integer
15525 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
15527 (imm5_operand_dictunpack16): Likewise.
15528 (imm5_operand_dictunpack32): Likewise.
15529 (imm2_operand): Likewise.
15530 * doc/invoke.texi (SPARC Options): Document -mvis4b.
15531 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
15532 ditunpack* and fpcmp*shl builtins.
15534 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15536 * config.gcc: Handle m8 in --with-{cpu,tune} options.
15537 * config.in: Add HAVE_AS_SPARC6 define.
15538 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
15540 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
15542 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
15543 (CPP_CPU_SPEC): Handle m8.
15544 (ASM_CPU_SPEC): Likewise.
15545 * config/sparc/sparc-opts.h (enum processor_type): Add
15547 * config/sparc/sparc.c (m8_costs): New struct.
15548 (sparc_option_override): Handle TARGET_CPU_m8.
15549 (sparc32_initialize_trampoline): Likewise.
15550 (sparc64_initialize_trampoline): Likewise.
15551 (sparc_issue_rate): Likewise.
15552 (sparc_register_move_cost): Likewise.
15553 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
15554 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
15555 (ASM_CPU64_DEFAULT_SPEC): Likewise.
15556 (CPP_CPU_SPEC): Handle M8.
15557 (ASM_CPU_SPEC): Likewise.
15558 (AS_M8_FLAG): Define.
15559 * config/sparc/sparc.md: Add m8 to the cpu attribute.
15560 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
15561 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
15563 * configure: Regenerate.
15564 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
15567 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15569 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
15571 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
15572 ("*movdi_insn_sp32"): Do not set v3pipe.
15573 ("*movsi_insn"): Likewise.
15574 ("*movdi_insn_sp64"): Likewise.
15575 ("*movsf_insn"): Likewise.
15576 ("*movdf_insn_sp32"): Likewise.
15577 ("*movdf_insn_sp64"): Likewise.
15578 ("*zero_extendsidi2_insn_sp64"): Likewise.
15579 ("*sign_extendsidi2_insn"): Likewise.
15580 ("*mov<VM32:mode>_insn"): Likewise.
15581 ("*mov<VM64:mode>_insn_sp64"): Likewise.
15582 ("*mov<VM64:mode>_insn_sp32"): Likewise.
15583 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
15584 ("<vlop:code><VL:mode>3"): Likewise.
15585 ("*not_<vlop:code><VL:mode>3"): Likewise.
15586 ("*nand<VL:mode>_vis"): Likewise.
15587 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
15588 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
15589 ("one_cmpl<VL:mode>2"): Likewise.
15590 ("faligndata<VM64:mode>_vis"): Likewise.
15591 ("alignaddrsi_vis"): Likewise.
15592 ("alignaddrdi_vis"): Likweise.
15593 ("alignaddrlsi_vis"): Likewise.
15594 ("alignaddrldi_vis"): Likewise.
15595 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
15596 ("bmaskdi_vis"): Likewise.
15597 ("bmasksi_vis"): Likewise.
15598 ("bshuffle<VM64:mode>_vis"): Likewise.
15599 ("cmask8<P:mode>_vis"): Likewise.
15600 ("cmask16<P:mode>_vis"): Likewise.
15601 ("cmask32<P:mode>_vis"): Likewise.
15602 ("pdistn<P:mode>_vis"): Likewise.
15603 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
15605 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15607 * config/sparc/sparc.md ("subtype"): New insn attribute.
15608 ("*wrgsr_sp64"): Set insn subtype.
15609 ("*rdgsr_sp64"): Likewise.
15610 ("alignaddrsi_vis"): Likewise.
15611 ("alignaddrdi_vis"): Likewise.
15612 ("alignaddrlsi_vis"): Likewise.
15613 ("alignaddrldi_vis"): Likewise.
15614 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
15615 ("fexpand_vis"): Likewise.
15616 ("fpmerge_vis"): Likewise.
15617 ("faligndata<VM64:mode>_vis"): Likewise.
15618 ("bshuffle<VM64:mode>_vis"): Likewise.
15619 ("cmask8<P:mode>_vis"): Likewise.
15620 ("cmask16<P:mode>_vis"): Likewise.
15621 ("cmask32<P:mode>_vis"): Likewise.
15622 ("fchksm16_vis"): Likewise.
15623 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
15624 ("fmean16_vis"): Likewise.
15625 ("fp<plusminus_insn>64_vis"): Likewise.
15626 ("<plusminus_insn>v8qi3"): Likewise.
15627 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
15628 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
15629 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
15630 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
15631 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
15632 ("*movqi_insn"): Likewise.
15633 ("*movhi_insn"): Likewise.
15634 ("*movsi_insn"): Likewise.
15635 ("movsi_pic_gotdata_op"): Likewise.
15636 ("*movdi_insn_sp32"): Likewise.
15637 ("*movdi_insn_sp64"): Likewise.
15638 ("movdi_pic_gotdata_op"): Likewise.
15639 ("*movsf_insn"): Likewise.
15640 ("*movdf_insn_sp32"): Likewise.
15641 ("*movdf_insn_sp64"): Likewise.
15642 ("*zero_extendhisi2_insn"): Likewise.
15643 ("*zero_extendqihi2_insn"): Likewise.
15644 ("*zero_extendqisi2_insn"): Likewise.
15645 ("*zero_extendqidi2_insn"): Likewise.
15646 ("*zero_extendhidi2_insn"): Likewise.
15647 ("*zero_extendsidi2_insn_sp64"): Likewise.
15648 ("ldfsr"): Likewise.
15649 ("prefetch_64"): Likewise.
15650 ("prefetch_32"): Likewise.
15651 ("tie_ld32"): Likewise.
15652 ("tie_ld64"): Likewise.
15653 ("*tldo_ldub_sp32"): Likewise.
15654 ("*tldo_ldub1_sp32"): Likewise.
15655 ("*tldo_ldub2_sp32"): Likewise.
15656 ("*tldo_ldub_sp64"): Likewise.
15657 ("*tldo_ldub1_sp64"): Likewise.
15658 ("*tldo_ldub2_sp64"): Likewise.
15659 ("*tldo_ldub3_sp64"): Likewise.
15660 ("*tldo_lduh_sp32"): Likewise.
15661 ("*tldo_lduh1_sp32"): Likewise.
15662 ("*tldo_lduh_sp64"): Likewise.
15663 ("*tldo_lduh1_sp64"): Likewise.
15664 ("*tldo_lduh2_sp64"): Likewise.
15665 ("*tldo_lduw_sp32"): Likewise.
15666 ("*tldo_lduw_sp64"): Likewise.
15667 ("*tldo_lduw1_sp64"): Likewise.
15668 ("*tldo_ldx_sp64"): Likewise.
15669 ("*mov<VM32:mode>_insn"): Likewise.
15670 ("*mov<VM64:mode>_insn_sp64"): Likewise.
15671 ("*mov<VM64:mode>_insn_sp32"): Likewise.
15673 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15675 * config/sparc/sparc.md ("type"): New insn type viscmp.
15676 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
15678 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
15679 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
15680 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
15681 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
15683 ("n7_vis_logical_11cycle"): Likewise.
15684 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
15685 * config/sparc/niagara2.md ("niag3_vis": Likewise.
15686 * config/sparc/niagara.md ("niag_vis"): Likewise.
15687 * config/sparc/ultra3.md ("us3_fga"): Likewise.
15688 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
15690 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
15692 * config/sparc/sparc.md: New instruction type `bmask'.
15693 (bmaskdi_vis): Use the `bmask' type.
15694 (bmasksi_vis): Likewise.
15695 * config/sparc/ultra3.md (us3_array): Likewise.
15696 * config/sparc/niagara7.md (n7_array): Likewise.
15697 * config/sparc/niagara4.md (n4_array): Likewise.
15698 * config/sparc/niagara2.md (niag2_vis): Likewise.
15699 (niag3_vis): Likewise.
15700 * config/sparc/niagara.md (niag_vis): Likewise.
15702 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15704 * ipa-comdats.c: Remove optimize check from gate.
15705 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
15706 for functions not optimized.
15707 (ipa_fn_summary_read): Skip optimize check.
15708 (ipa_fn_summary_write): Likewise.
15709 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
15711 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
15713 (can_inline_edge_p): Check flag_pcc_struct_return for match.
15714 (check_callers): Give up on caller which is not optimized.
15715 (inline_small_functions): Likewise.
15716 (ipa_inline): Do not give up when not optimizing.
15717 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
15718 away unoptimizes cdtors.
15719 (whole_program_function_and_variable_visibility): Do
15720 ipa_discover_readonly_nonaddressable_vars in LTO mode.
15721 * ipa.c (process_references): Do not check optimize.
15722 (symbol_table::remove_unreachable_nodes): Update optimize check.
15723 (set_writeonly_bit): Update optimize check.
15724 (pass_ipa_cdtor_merge::gate): Do not check optimize.
15725 (pass_ipa_single_use::gate): Remove.
15727 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15729 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
15730 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
15731 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
15732 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
15733 permute_load, permute_store, adjust_extract, adjust_splat,
15734 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
15735 replace_swap_with_copy, dump_swap_insn_table,
15736 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
15737 recombine_lvx_pattern, recombine_stvx_pattern,
15738 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
15739 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
15740 to file rs6000-p8swap.c.
15741 * config/rs6000/rs6000-p8swap.c: New file.
15742 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
15743 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
15744 and rs6000*-*-* targets.
15746 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15748 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
15750 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15752 * lto-wrapper.c (merge_and_complain): Do not merge
15753 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
15754 fsigned_zeros, ftrapping_math, fwrapv.
15755 (append_compiler_options): Do not track these options.
15756 (append_linker_options): Likewie
15758 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15760 * cgraphunit.c (cgraph_node::finalize_function): When
15761 !flag_toplevel_reorde set no_reorder flag.
15762 (varpool_node::finalize_decl): Likewise.
15763 (symbol_table::compile): Drop no toplevel reorder path.
15765 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15767 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
15768 edges; zero probability is not better than uninitialized.
15770 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15772 * asan.h (asan_sanitize_allocas_p): Declare.
15773 * asan.c (asan_sanitize_allocas_p): New function.
15774 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
15775 (handle_builtin_alloca): Likewise.
15776 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
15777 if !asan_sanitize_allocas_p.
15778 * params.def (asan-instrument-allocas): Add new option.
15779 * params.h (ASAN_PROTECT_ALLOCAS): Define.
15780 * opts.c (common_handle_option): Disable allocas sanitization for
15783 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
15785 * asan.c: Include gimple-fold.h.
15786 (get_last_alloca_addr): New function.
15787 (handle_builtin_stackrestore): Likewise.
15788 (handle_builtin_alloca): Likewise.
15789 (asan_emit_allocas_unpoison): Likewise.
15790 (get_mem_refs_of_builtin_call): Add new parameter, remove const
15791 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
15792 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
15793 (instrument_builtin_call): Pass gimple iterator to
15794 get_mem_refs_of_builtin_call.
15795 (last_alloca_addr): New global.
15796 * asan.h (asan_emit_allocas_unpoison): Declare.
15797 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
15798 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
15799 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
15800 if function calls alloca.
15801 * gimple-fold.c (replace_call_with_value): Remove static keyword.
15802 * gimple-fold.h (replace_call_with_value): Declare.
15803 * internal-fn.c: Include asan.h.
15804 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
15805 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
15807 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15809 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
15810 (C_SELFTEST_FLAGS): New.
15811 (CPP_SELFTEST_FLAGS): New.
15812 (SELFTEST_DEPS): New, from deps of s-selftest.
15813 (C_SELFTEST_DEPS): New, from deps of s-selftest.
15814 (CPP_SELFTEST_DEPS): New.
15815 (selftest): Add dependency on s-selftest-c++.
15816 (s-selftest): Rename to...
15817 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
15818 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
15819 than SELFTEST_FLAGS.
15820 (selftest-gdb): Rename to...
15821 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
15823 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
15824 (selftest-valgrind): Rename to...
15825 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
15827 (selftest-valgrind): Reintroduce as an alias for
15828 selftest-c-valgrind.
15829 (s-selftest-c++): New.
15830 (selftest-c++-gdb): New.
15831 (selftest-c++-valgrind): New.
15833 2017-07-06 Olivier Hainque <hainque@adacore.com>
15835 * gcc.c (process_command): When deciding if undefined variables
15836 should be ignored when processing specs, accept "gcc -v" as well.
15838 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
15840 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
15841 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
15843 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
15845 * config/arm/arm-cpus.in (armv8-r): Add new entry.
15846 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
15847 * config/arm/arm-tables.opt: Regenerate.
15848 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
15850 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
15852 2017-07-06 Carl Love <cel@us.ibm.com>
15854 * ChangeLog: Clean up from mid air collision
15856 2017-07-06 Carl Love <cel@us.ibm.com>
15858 * config/rs6000/rs6000-c.c: Add support for built-in functions
15859 vector signed int vec_subc (vector signed int, vector signed int);
15860 vector signed __int128 vec_subc (vector signed __int128,
15861 vector signed __int128);
15862 vector unsigned __int128 vec_subc (vector unsigned __int128,
15863 vector unsigned __int128);
15864 vector signed int vec_sube (vector signed int, vector signed int,
15865 vector signed int);
15866 vector unsigned int vec_sube (vector unsigned int,
15867 vector unsigned int,
15868 vector unsigned int);
15869 vector signed __int128 vec_sube (vector signed __int128,
15870 vector signed __int128,
15871 vector signed__int128);
15872 vector unsigned __int128 vec_sube (vector unsigned __int128,
15873 vector unsigned __int128,
15874 vector unsigned __int128);
15875 vector signed int vec_subec (vector signed int, vector signed int,
15876 vector signed int);
15877 vector unsigned int vec_subec (vector unsigned int,
15878 vector unsigned int,
15879 vector unsigned int);
15880 vector signed __int128 vec_subec (vector signed __int128,
15881 vector signed __int128,
15882 vector signed__int128);
15883 vector unsigned __int128 vec_subec (vector unsigned __int128,
15884 vector unsigned __int128,
15885 vector unsigned __int128);
15886 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
15887 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
15888 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
15889 BU_ALTIVEC_OVERLOAD_X definitions.
15890 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
15891 * doc/extend.texi: Update the built-in documentation file for the new
15892 built-in functions.
15894 2017-07-06 David Malcolm <dmalcolm@redhat.com>
15897 * diagnostic-show-locus.c (layout::layout): Use start and finish
15898 spelling location for the start and finish of each range.
15899 * genmatch.c (linemap_client_expand_location_to_spelling_point):
15900 Add unused aspect param.
15901 * input.c (expand_location_1): Add "aspect" param, and use it
15902 to access the correct part of the location.
15903 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
15905 (expand_location_to_spelling_point): Likewise.
15906 (linemap_client_expand_location_to_spelling_point): Add "aspect"
15907 param, and pass it to expand_location_1.
15909 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
15911 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
15912 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
15913 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
15914 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
15915 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
15916 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
15917 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
15918 _mm_maskz_getmant_ss): New intrinsics.
15919 (__builtin_ia32_getexpss128_mask): Changed to ...
15920 __builtin_ia32_getexpss128_round ... this.
15921 (__builtin_ia32_getexpsd128_mask): Changed to ...
15922 __builtin_ia32_getexpsd128_round ... this.
15923 * config/i386/i386-builtin-types.def
15924 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
15925 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
15926 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
15927 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
15928 __builtin_ia32_getmantss_mask_round): New builtins.
15929 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
15930 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
15931 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
15932 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
15933 * config/i386/sse.md
15934 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
15935 avx512f_sgetexp<mode><mask_scalar_name>
15936 <round_saeonly_scalar_name> ... this.
15937 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
15938 %0, %1, %2<round_saeonly_op3>}): Changed to ...
15939 vgetexp<ssescalarmodesuffix>
15940 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15941 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
15942 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
15943 avx512f_vgetmant<mode><mask_scalar_name>
15944 <round_saeonly_scalar_name> ... this.
15945 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
15946 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
15947 vgetmant<ssescalarmodesuffix>
15948 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
15949 %0<mask_scalar_operand4>, %1, %2
15950 <round_saeonly_scalar_mask_op4>, %3} ... this.
15951 * config/i386/subst.md (mask_scalar_operand4,
15952 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
15953 round_saeonly_scalar_nimm_predicate): New subst attributes.
15955 2017-07-06 Julia Koval <julia.koval@intel.com>
15957 * config/i386/i386.c (ix86_erase_embedded_rounding):
15958 Remove code for old rounding pattern.
15960 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
15962 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
15964 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
15966 * doc/sourcebuild.texi (Test Directives, Variants of
15967 dg-require-support): Add documentation for dg-require-stack-check.
15969 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
15971 * config/i386/subst.md (mask_scalar, round_scalar,
15972 round_saeonly_scalar): New meta-templates.
15973 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
15974 round_scalar_mask_operand3, round_scalar_mask_op3,
15975 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
15976 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
15977 round_saeonly_scalar_constraint,
15978 round_saeonly_scalar_prefix): New subst attribute.
15979 * config/i386/sse.md
15980 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
15981 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
15982 <round_scalar_name> ... this.
15983 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
15984 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
15985 <round_scalar_name> ... this.
15986 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
15987 <sse>_vm<code><mode>3<mask_scalar_name>
15988 <round_saeonly_scalar_name> ... this.
15989 (v<plusminus_mnemonic><ssescalarmodesuffix>
15990 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15991 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15992 v<plusminus_mnemonic><ssescalarmodesuffix>
15993 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
15994 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
15995 (v<multdiv_mnemonic><ssescalarmodesuffix>
15996 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
15997 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
15998 v<multdiv_mnemonic><ssescalarmodesuffix>
15999 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16000 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
16001 (v<maxmin_float><ssescalarmodesuffix>
16002 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
16003 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
16004 v<maxmin_float><ssescalarmodesuffix>
16005 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
16006 %0<mask_scalar_operand3>, %1, %<iptr>2
16007 <round_saeonly_scalar_mask_op3>} ... this.
16009 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
16011 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
16012 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
16014 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
16015 Alan Hayward <alan.hayward@arm.com>
16016 David Sherwood <david.sherwood@arm.com>
16018 * combine.c (simplify_if_then_else): Remove "enum" before
16020 * compare-elim.c (can_eliminate_compare): Likewise.
16021 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
16023 (aarch64_lookup_simd_builtin_type): Likewise.
16024 (aarch64_simd_builtin_type): Likewise.
16025 (aarch64_init_simd_builtin_types): Likewise.
16026 (aarch64_simd_expand_args): Likewise.
16027 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
16029 (aarch64_reverse_mask): Likewise.
16030 (aarch64_simd_emit_reg_reg_move): Likewise.
16031 (aarch64_gen_adjusted_ldpstp): Likewise.
16032 (aarch64_ccmp_mode_to_code): Likewise.
16033 (aarch64_operands_ok_for_ldpstp): Likewise.
16034 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16035 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
16037 (aarch64_min_divisions_for_recip_mul): Likewise.
16038 (aarch64_reassociation_width): Likewise.
16039 (aarch64_get_condition_code_1): Likewise.
16040 (aarch64_simd_emit_reg_reg_move): Likewise.
16041 (aarch64_simd_attr_length_rglist): Likewise.
16042 (aarch64_reverse_mask): Likewise.
16043 (aarch64_operands_ok_for_ldpstp): Likewise.
16044 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
16045 (aarch64_gen_adjusted_ldpstp): Likewise.
16046 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
16048 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
16049 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
16050 (arm_lookup_simd_builtin_type): Likewise.
16051 (arm_simd_builtin_type): Likewise.
16052 (arm_init_simd_builtin_types): Likewise.
16053 (arm_expand_builtin_args): Likewise.
16054 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
16055 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
16056 (ft32_setup_incoming_varargs): Likewise.
16057 (ft32_function_arg): Likewise.
16058 (ft32_function_arg_advance): Likewise.
16059 (ft32_pass_by_reference): Likewise.
16060 (ft32_arg_partial_bytes): Likewise.
16061 (ft32_valid_pointer_mode): Likewise.
16062 (ft32_addr_space_pointer_mode): Likewise.
16063 (ft32_addr_space_legitimate_address_p): Likewise.
16064 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
16066 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
16067 (ix86_emit_outlined_ms2sysv_restore): Likewise.
16068 (iamcu_alignment): Likewise.
16069 (canonicalize_vector_int_perm): Likewise.
16070 (ix86_noce_conversion_profitable_p): Likewise.
16071 (ix86_mpx_bound_mode): Likewise.
16072 (ix86_operands_ok_for_move_multiple): Likewise.
16073 * config/microblaze/microblaze-protos.h
16074 (microblaze_expand_conditional_branch_reg): Likewise.
16075 * config/microblaze/microblaze.c
16076 (microblaze_expand_conditional_branch_reg): Likewise.
16077 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
16079 (rs6000_reassociation_width): Likewise.
16080 (rs6000_invalid_binary_op): Likewise.
16081 (fusion_p9_p): Likewise.
16082 (emit_fusion_p9_load): Likewise.
16083 (emit_fusion_p9_store): Likewise.
16084 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
16086 (riscv_hard_regno_mode_ok_p): Likewise.
16087 (riscv_address_insns): Likewise.
16088 (riscv_split_symbol): Likewise.
16089 (riscv_legitimize_move): Likewise.
16090 (riscv_function_value): Likewise.
16091 (riscv_hard_regno_nregs): Likewise.
16092 (riscv_expand_builtin): Likewise.
16093 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
16094 (riscv_build_integer): Likewise.
16095 (riscv_split_integer): Likewise.
16096 (riscv_legitimate_constant_p): Likewise.
16097 (riscv_cannot_force_const_mem): Likewise.
16098 (riscv_regno_mode_ok_for_base_p): Likewise.
16099 (riscv_valid_base_register_p): Likewise.
16100 (riscv_valid_offset_p): Likewise.
16101 (riscv_valid_lo_sum_p): Likewise.
16102 (riscv_classify_address): Likewise.
16103 (riscv_legitimate_address_p): Likewise.
16104 (riscv_address_insns): Likewise.
16105 (riscv_load_store_insns): Likewise.
16106 (riscv_force_binary): Likewise.
16107 (riscv_split_symbol): Likewise.
16108 (riscv_force_address): Likewise.
16109 (riscv_legitimize_address): Likewise.
16110 (riscv_move_integer): Likewise.
16111 (riscv_legitimize_const_move): Likewise.
16112 (riscv_legitimize_move): Likewise.
16113 (riscv_address_cost): Likewise.
16114 (riscv_subword): Likewise.
16115 (riscv_output_move): Likewise.
16116 (riscv_canonicalize_int_order_test): Likewise.
16117 (riscv_emit_int_order_test): Likewise.
16118 (riscv_function_arg_boundary): Likewise.
16119 (riscv_pass_mode_in_fpr_p): Likewise.
16120 (riscv_pass_fpr_single): Likewise.
16121 (riscv_pass_fpr_pair): Likewise.
16122 (riscv_get_arg_info): Likewise.
16123 (riscv_function_arg): Likewise.
16124 (riscv_function_arg_advance): Likewise.
16125 (riscv_arg_partial_bytes): Likewise.
16126 (riscv_function_value): Likewise.
16127 (riscv_pass_by_reference): Likewise.
16128 (riscv_setup_incoming_varargs): Likewise.
16129 (riscv_print_operand): Likewise.
16130 (riscv_elf_select_rtx_section): Likewise.
16131 (riscv_save_restore_reg): Likewise.
16132 (riscv_for_each_saved_reg): Likewise.
16133 (riscv_register_move_cost): Likewise.
16134 (riscv_hard_regno_mode_ok_p): Likewise.
16135 (riscv_hard_regno_nregs): Likewise.
16136 (riscv_class_max_nregs): Likewise.
16137 (riscv_memory_move_cost): Likewise.
16138 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
16139 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
16140 (rl78_addr_space_address_mode): Likewise.
16141 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
16143 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
16144 (rs6000_reassociation_width): Likewise.
16145 (rs6000_invalid_binary_op): Likewise.
16146 (fusion_p9_p): Likewise.
16147 (emit_fusion_p9_load): Likewise.
16148 (emit_fusion_p9_store): Likewise.
16149 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
16150 (ok_for_simple_move_operands): Likewise.
16151 (ok_for_simple_move_strict_operands): Likewise.
16152 (ok_for_simple_arith_logic_operands): Likewise.
16153 (visium_legitimize_reload_address): Likewise.
16154 (visium_select_cc_mode): Likewise.
16155 (output_cbranch): Likewise.
16156 (visium_split_double_move): Likewise.
16157 (visium_expand_copysign): Likewise.
16158 (visium_expand_int_cstore): Likewise.
16159 (visium_expand_fp_cstore): Likewise.
16160 * config/visium/visium.c (visium_pass_by_reference): Likewise.
16161 (visium_function_arg): Likewise.
16162 (visium_function_arg_advance): Likewise.
16163 (visium_libcall_value): Likewise.
16164 (visium_setup_incoming_varargs): Likewise.
16165 (visium_legitimate_constant_p): Likewise.
16166 (visium_legitimate_address_p): Likewise.
16167 (visium_legitimize_address): Likewise.
16168 (visium_secondary_reload): Likewise.
16169 (visium_register_move_cost): Likewise.
16170 (visium_memory_move_cost): Likewise.
16171 (prepare_move_operands): Likewise.
16172 (ok_for_simple_move_operands): Likewise.
16173 (ok_for_simple_move_strict_operands): Likewise.
16174 (ok_for_simple_arith_logic_operands): Likewise.
16175 (visium_function_value_1): Likewise.
16176 (rtx_ok_for_offset_p): Likewise.
16177 (visium_legitimize_reload_address): Likewise.
16178 (visium_split_double_move): Likewise.
16179 (visium_expand_copysign): Likewise.
16180 (visium_expand_int_cstore): Likewise.
16181 (visium_expand_fp_cstore): Likewise.
16182 (visium_split_cstore): Likewise.
16183 (visium_select_cc_mode): Likewise.
16184 (visium_split_cbranch): Likewise.
16185 (output_cbranch): Likewise.
16186 (visium_print_operand_address): Likewise.
16187 * expmed.c (flip_storage_order): Likewise.
16188 * expmed.h (emit_cstore): Likewise.
16189 (flip_storage_order): Likewise.
16190 * genrecog.c (validate_pattern): Likewise.
16191 * hsa-gen.c (gen_hsa_addr): Likewise.
16192 * internal-fn.c (expand_arith_overflow): Likewise.
16193 * ira-color.c (allocno_copy_cost_saving): Likewise.
16194 * lra-assigns.c (find_hard_regno_for_1): Likewise.
16195 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
16196 (process_invariant_for_inheritance): Likewise.
16197 * lra-eliminations.c (move_plus_up): Likewise.
16198 * omp-low.c (lower_oacc_reductions): Likewise.
16199 * simplify-rtx.c (simplify_subreg): Likewise.
16200 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
16201 (TARGET_CHKP_BOUND_MODE): Likewise..
16202 * targhooks.c (default_chkp_bound_mode): Likewise.
16203 (default_setup_incoming_vararg_bounds): Likewise.
16204 * targhooks.h (default_chkp_bound_mode): Likewise.
16205 (default_setup_incoming_vararg_bounds): Likewise.
16206 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
16207 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
16208 (have_whole_vector_shift): Likewise.
16209 * tree-vect-stmts.c (vectorizable_load): Likewise.
16210 * doc/tm.texi: Regenerate.
16212 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16214 Graceful degrade if Binutils PR21472 is not available.
16217 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
16218 .rodata in flash test fails.
16219 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
16220 * confgure: Regenerate.
16221 * config.in: Regenerate.
16222 * config/avr/avr.c (avr_asm_named_section)
16223 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
16224 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
16225 (avr_asm_init_sections): Same.
16227 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16229 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
16230 (fma<VH:mode>4_intrinsic): Likewise.
16231 (*fmsub<VCVTF:mode>4): Likewise.
16232 (*fmsub<VH:mode>4_intrinsic): Likewise.
16234 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
16237 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
16238 Don't depend on "optimize > 0".
16239 (out_movhi_r_mr, out_movqi_mr_r): Same.
16240 (out_movhi_mr_r, out_movqi_r_mr): Same.
16241 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
16242 io_address_operand on "optimize > 0".
16244 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16246 * tree-loop-distribution.c: Add general explanantion on the pass.
16247 (generate_loops_for_partition): Mark distributed loop.
16248 (pg_add_dependence_edges): New parameter. Handle alias data
16249 dependence specially and record it in the parameter if asked.
16250 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
16251 (init_partition_graph_vertices, add_partition_graph_edge): New.
16252 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
16253 (free_partition_graph_vdata, build_partition_graph): New.
16254 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
16255 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
16256 (data_ref_segment_size, latch_dominated_by_data_ref): New.
16257 (compute_alias_check_pairs, version_loop_by_alias_check): New.
16258 (version_for_distribution_p, finalize_partitions): New.
16259 (distribute_loop): Handle alias data dependence specially. Factor
16260 out loop fusion code as functions and call these functions.
16262 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16264 * tree-loop-distribution.c (classify_partition): New parameter and
16265 better handle reduction statement.
16266 (rdg_build_partitions): Revise comment.
16267 (distribute_loop): Compute statements in all partitions and pass it
16268 to classify_partition.
16270 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16272 * tree-loop-distribution.c (enum partition_type): New.
16273 (struct partition): New field type.
16274 (partition_merge_into): Add parameter. Update partition type.
16275 (data_dep_in_cycle_p, update_type_for_merge): New functions.
16276 (build_rdg_partition_for_vertex): Compute partition type.
16277 (rdg_build_partitions): Dump partition type.
16278 (distribute_loop): Update calls to partition_merge_into.
16280 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16282 * tree-loop-distribution.c (struct ddr_hasher): New.
16283 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
16285 (classify_partition): Call get_data_dependence.
16286 (pg_add_dependence_edges): Ditto.
16287 (distribute_loop): Release data dependence hash table.
16289 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16291 * tree-loop-distribution.c (ref_base_address): Delete.
16292 (similar_memory_accesses): Rename ...
16293 (share_memory_accesses): ... to this. Check if partitions access
16294 the same memory reference.
16295 (distribute_loop): Call share_memory_accesses.
16297 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16299 * tree-loop-distribution.c (struct partition): New field recording
16300 its data reference.
16301 (partition_alloc, partition_free): Init and release data refs.
16302 (partition_merge_into): Merge data refs.
16303 (build_rdg_partition_for_vertex): Collect data refs for partition.
16304 (pg_add_dependence_edges): Change parameters from vector to bitmap.
16306 (distribute_loop): Remve data refs from vertice data of partition
16309 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16311 * tree-loop-distribution.c (params.h): Include header file.
16312 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
16313 (datarefs_vec): New global var.
16314 (create_rdg_vertices): Use datarefs_vec directly.
16315 (free_rdg): Don't free data references.
16316 (build_rdg): Update use. Don't free data references.
16317 (distribute_loop): Compute global variable for data references.
16318 Bail out if there are too many data references.
16320 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16322 * tree-loop-distribution.c (loop_nest): New global var.
16323 (build_rdg): Use loop directly, rather than loop nest.
16324 (pg_add_dependence_edges): Remove loop nest parameter. Use global
16326 (distribute_loop): Compute global variable loop nest. Update use.
16328 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16330 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
16331 (partition_merge_into): New parameter. Dump reason for fusion.
16332 (distribute_loop): Update use of partition_merge_into.
16334 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16336 * tree-loop-distribution.c (bb_top_order_index): New.
16337 (bb_top_order_index_size, bb_top_order_cmp): New.
16338 (stmts_from_loop): Use topological order.
16339 (pass_loop_distribution::execute): Compute and release topological
16340 order for basic blocks.
16342 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16344 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
16347 2017-07-05 Bin Cheng <bin.cheng@arm.com>
16349 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
16350 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
16351 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
16352 * internal-fn.def (LOOP_DIST_ALIAS): New.
16353 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
16354 (fold_loop_internal_call): ... this.
16355 (vect_loop_dist_alias_call): New function.
16356 (set_uid_loop_bbs): Call fold_loop_internal_call.
16357 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
16360 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16363 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
16364 Require dead FLAGS_REG at the beginning of a peephole.
16366 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
16369 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
16370 arguments in the call to __builtin_ia32_sbb_u32.
16371 (_subborrow_u64): Swap _X and _Y arguments in the call to
16372 __builtin_ia32_sbb_u64.
16374 2017-07-04 Jakub Jelinek <jakub@redhat.com>
16377 * tree-vrp.c (compare_assert_loc): Turn into a function template
16378 with stable template parameter. Only test if a->e is NULL,
16379 !a->e == !b->e has been verified already. Use e == NULL or
16380 e != NULL instead of e or ! e tests. If stable is true, don't use
16381 iterative_hash_expr, on the other side allow a or b or both NULL
16382 and sort the NULLs last.
16383 (process_assert_insertions): Sort using compare_assert_loc<false>
16384 instead of compare_assert_loc, later sort using
16385 compare_assert_loc<true> before calling process_assert_insertions_for
16386 in a loop. Use break instead of continue once seen NULL pointer.
16388 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
16390 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
16391 Cortex-R7 and Cortex-R8 processors.
16393 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16395 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
16396 uninitialized while src is not.
16398 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
16400 * common/config/arm/arm-common.c: Adjust include path for
16402 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
16403 (arm-cpu.h): Create in build directory. Adjust dependency rules.
16404 (arm-cpu-data.h): Likewise.
16405 (arm-cpu-cdata.h): Likewise.
16406 * config/arm/arm-cpu.h: Delete.
16407 * config/arm/arm-cpu-cdata.h: Delete.
16408 * config/arm/arm-cpu-data.h: Delete.
16410 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
16412 * config/arm/arm-cpus.in (cortex-a55): New.
16413 (cortex-a75): Likewise.
16414 (cortex-a75.cortex-a55): Likewise.
16415 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
16417 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
16418 * config/arm/arm-cpu-cdata.h: Regenerate.
16419 * config/arm/arm-cpu-data.h: Regenerate.
16420 * config/arm/arm-cpu.h: Regenerate.
16421 * config/arm/arm-tables.opt: Regenerate.
16422 * config/arm/arm-tune.md: Regenerate.
16424 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16426 * haifa-sched.c (sched_create_recovery_edges): Update profile.
16428 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
16430 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
16433 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
16435 PR tree-optimization/81292
16436 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
16437 full_string_p, also call adjust_related_strinfos if the adjustment
16438 is simple, otherwise invalidate related strinfos.
16440 2017-07-04 Martin Liska <mliska@suse.cz>
16443 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
16444 newly created variable as DECL_IGNORED_P.
16446 2017-07-04 Martin Liska <mliska@suse.cz>
16449 * ipa-inline.c (inline_small_functions):
16450 Use xstrdup_for_dump.
16452 2017-07-04 Tom de Vries <tom@codesourcery.com>
16454 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
16456 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16459 * config/darwin.c (darwin_function_switched_text_sections):
16462 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
16464 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
16466 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16468 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
16470 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16472 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
16473 min_profitable_iters, and th as inclusive lower bounds.
16474 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
16475 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
16476 for min_profitable_iters and min_profitable_estimate.
16477 (vect_transform_loop): Treat th as an inclusive lower bound.
16478 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
16480 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
16483 * config/darwin.c (darwin_function_switched_text_sections):
16484 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
16485 in two pieces, and suppress the use of buf.
16487 2017-07-03 Nathan Sidwell <nathan@acm.org>
16489 * hash-table.h (hash_table_mod1): Fix indentation.
16491 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16493 PR middle-end/81290
16494 * predict.c (force_edge_cold): Be more careful about propagation
16496 * profile-count.h (profile_probability::guessed,
16497 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
16499 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
16501 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
16503 * doc/invoke.texi (rcpc architecture extension): Document it.
16505 2017-07-03 Richard Biener <rguenther@suse.de>
16507 PR tree-optimization/60510
16508 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
16509 the scalar reduction PHI and use it.
16510 (vectorizable_reduction): Properly guard the single_defuse_cycle
16511 path for non-SLP reduction chains where we cannot use it.
16512 Rework reduc_def/index and vector type deduction. Rework
16513 vector operand gathering during reduction op code-gen.
16514 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
16515 chains dissolve the chain and leave it to non-SLP reduction
16518 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16520 * tree-data-ref.h (dr_alignment): Declare.
16521 * tree-data-ref.c (dr_alignment): New function.
16522 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
16523 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
16525 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
16527 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16529 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
16530 and base_misalignment fields.
16531 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
16532 * tree-data-ref.c: Include builtins.h.
16533 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
16534 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
16535 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
16536 * tree-vect-data-refs.c: Include tree-cfg.h.
16537 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
16538 fields instead of calculating an alignment here.
16539 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
16540 innermost_loop_behavior fields.
16542 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16544 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
16546 (DR_STEP_ALIGNMENT): New macro.
16547 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
16548 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
16549 (create_data_ref): Print it.
16550 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
16551 to tell whether the step preserves vector (mis)alignment.
16552 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16553 Move the check for an integer step and generalise to all INTEGER_CST.
16554 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
16555 Print the outer step alignment.
16557 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16559 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
16560 with offset_alignment.
16561 (DR_ALIGNED_TO): Delete.
16562 (DR_OFFSET_ALIGNMENT): New macro.
16563 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
16564 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
16565 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
16566 (create_data_ref): Likewise.
16567 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
16568 (vect_analyze_data_refs): Likewise.
16569 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
16570 creating dummy innermost behavior.
16572 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16574 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
16575 with a "innermost_loop_behavior *" and refeence tree.
16576 * tree-data-ref.c (dr_analyze_innermost): Likewise.
16577 (create_data_ref): Update call accordingly.
16578 * tree-predcom.c (find_looparound_phi): Likewise.
16580 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16582 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
16583 fields with dr_wrt_vec_loop.
16584 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
16585 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
16586 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
16587 (vect_dr_behavior): New function.
16588 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
16589 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
16590 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
16591 track whether the step preserves the misalignment.
16592 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
16593 Use vect_dr_behavior.
16594 (vect_setup_realignment): Update call accordingly.
16595 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
16596 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
16597 call to vect_create_addr_base_for_vector_ref.
16598 (vect_create_cond_for_align_checks): Likewise.
16599 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
16600 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
16601 (vect_recog_mask_conversion_pattern): Likewise.
16602 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
16603 (new_stmt_vec_info): Remove redundant zeroing.
16605 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
16607 * common/config/arm/arm-common.c (arm_be8_option): New function.
16608 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
16609 (ISA_ARMv6): Add isa_bit_be8.
16610 * config/arm/arm.h (arm_be8_option): Add prototype.
16611 (BE8_SPEC_FUNCTION): New define.
16612 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
16613 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
16614 (mlittle-endian): Similarly.
16615 (mbe8, mbe32): New options.
16616 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
16617 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
16619 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16621 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
16623 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16625 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
16626 (cleanup_tree_cfg_bb): Use it.
16627 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
16629 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
16631 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16634 * loop-doloop.c (add_test): Update profile.
16636 2017-07-03 Martin Liska <mliska@suse.cz>
16639 * sanopt.c (rewrite_usage_of_param): New function.
16640 (sanitize_rewrite_addressable_params): Likewise.
16641 (pass_sanopt::execute): Call rewrite_usage_of_param.
16643 2017-07-03 Richard Biener <rguenther@suse.de>
16645 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
16646 back to using VIEW_CONVERT_EXPR.
16648 2017-07-03 Martin Liska <mliska@suse.cz>
16651 * doc/extend.texi: Document when a resolver function is
16652 generated for target_clones.
16654 2017-07-03 Martin Liska <mliska@suse.cz>
16656 * asan.c (asan_emit_stack_protection): Unpoison just red zones
16657 and shadow memory of auto variables which are subject of
16658 use-after-scope sanitization.
16659 (asan_expand_mark_ifn): Add do set only when is_poison.
16661 2016-07-03 Richard Biener <rguenther@suse.de>
16663 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
16665 (vect_force_simple_reduction): Record reduction def -> phi mapping.
16666 (vectorizable_reduction): Perform reduction PHI creation when
16667 visiting a reduction PHI and adjust and simplify code generation
16668 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
16669 (vect_transform_loop): Visit reduction PHIs.
16670 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
16671 defs into the SLP tree.
16672 (vect_build_slp_tree): Reduction defs terminate the recursion.
16673 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
16675 (vect_get_vec_defs_for_stmt_copy): Export.
16676 (vect_get_vec_defs): Likewise.
16677 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
16679 (vect_get_vec_defs_for_stmt_copy): Declare.
16680 (vect_get_vec_defs): Likewise.
16682 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
16684 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
16685 parameter with a "loop" parameter and use it instead of the
16686 loop containing DR_STMT. Don't check simple_iv when doing
16687 BB analysis. Describe the two analysis modes in the comment.
16689 2017-07-03 Tom de Vries <tom@codesourcery.com>
16691 PR tree-optimization/69468
16692 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
16693 (find_same_succ_bb): Handle ignore_edge_flags.
16695 2017-07-03 Tom de Vries <tom@codesourcery.com>
16697 PR tree-optimization/81192
16698 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
16700 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
16702 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
16704 2017-07-03 Tom de Vries <tom@codesourcery.com>
16706 PR tree-optimization/81192
16707 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
16708 BB_SAME_SUCC (bb) == NULL.
16710 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16712 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
16715 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16717 * dumpfile.c: Include profile-count.h
16718 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
16720 (insert_cond_bb): Update profile.
16721 * tree-cfg.h (insert_cond_bb): Update prototype.
16722 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
16723 * tree-dump.c: Do not include tree-cfg.
16725 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16727 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
16729 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16731 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
16734 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
16736 * tree-complex.c (expand_complex_div_wide): update profile.
16738 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16739 Alan Hayward <alan.hayward@arm.com>
16740 David Sherwood <david.sherwood@arm.com>
16742 * Makefile.in (MACHMODE_H): Remove insn-modes.h
16743 (CORETYPES_H): New define.
16744 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
16745 (insn-modes-inline.h, s-modes-inline-h): New rules.
16746 (generated_files): Add insn-modes-inline.h.
16747 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
16748 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
16749 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
16750 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
16751 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
16752 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
16753 (build/gencodes.o, build/genconditions.o): Likewise.
16754 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
16755 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
16756 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
16757 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
16758 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
16759 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
16760 * coretypes.h: Include everything up to real.h for generators.
16761 Include insn-modes.h first. Include wide-int-print.h after
16762 wide-int.h. Include insn-modes-inline.h and then machmode.h.
16763 * machmode.h: Don't include insn-modes.h here.
16764 * function-tests.c: Remove includes of signop.h, machmode.h,
16765 double-int.h and wide-int.h.
16767 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
16769 * optc-save-gen.awk: Likewise.
16770 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
16771 * godump.c: Remove include of wide-int-print.h.
16772 * pretty-print.h: Likewise.
16773 * wide-int-print.cc: Likewise.
16774 * wide-int.cc: Likewise.
16775 * hash-map-tests.c: Remove include of signop.h.
16776 * hash-set-tests.c: Likewise.
16777 * rtl-tests.c: Likewise.
16778 * mkconfig.sh: Remove include of machmode.h.
16779 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
16781 (emit_insn_modes_inline_h): ...this new function. Emit the code
16782 into an insn-modes-inline.h header file, adding appropriate
16783 include guards and end comments.
16784 (emit_insn_modes_c_header): Remove include of machmode.h.
16785 (emit_min_insn_modes_c_header): Include coretypes.h rather than
16787 (main): Handle -i flag and call emit_insn_modes_inline_h when
16790 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16792 * tree-ssa-strlen.c (strinfo): Rename the length field to
16793 nonzero_chars. Add a full_string_p field.
16794 (compare_nonzero_chars, zero_length_string_p): New functions.
16795 (get_addr_stridx): Add an offset_out parameter.
16796 Use compare_nonzero_chars.
16797 (get_stridx): Update accordingly. Use compare_nonzero_chars.
16798 (new_strinfo): Update after above changes to strinfo.
16799 (set_endptr_and_length): Set full_string_p.
16800 (get_string_length): Update after above changes to strinfo.
16801 (unshare_strinfo): Update call to new_strinfo.
16802 (maybe_invalidate): Likewise.
16803 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
16804 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
16805 as a uhwi instead of an shwi. Update after above changes to
16806 strinfo and new_strinfo.
16807 (zero_length_string): Assert that chainsi contains full strings.
16808 Use zero_length_string_p. Update call to new_strinfo.
16809 (adjust_related_strinfos): Update after above changes to strinfo.
16810 Copy full_string_p from origsi.
16811 (adjust_last_stmt): Use zero_length_string_p.
16812 (handle_builtin_strlen): Update after above changes to strinfo and
16813 new_strinfo. Install the lhs as the string length if the previous
16814 entry didn't describe a full string.
16815 (handle_builtin_strchr): Update after above changes to strinfo
16817 (handle_builtin_strcpy): Likewise.
16818 (handle_builtin_strcat): Likewise.
16819 (handle_builtin_malloc): Likewise.
16820 (handle_pointer_plus): Likewise.
16821 (handle_builtin_memcpy): Likewise. Track nonzero characters
16822 that aren't necessarily followed by a nul terminator.
16823 (handle_char_store): Likewise.
16825 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16827 PR tree-optimization/80769
16828 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
16829 for malloc and calloc. Document the new invariant that all related
16830 strinfos have delayed lengths or none do.
16831 (verify_related_strinfos): Move earlier in file.
16832 (set_endptr_and_length): New function, split out from...
16833 (get_string_length): ...here. Also set the lengths of related
16835 (zero_length_string): Assert that chainsi has known (rather than
16837 (adjust_related_strinfos): Likewise.
16839 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
16841 PR tree-optimization/81136
16842 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
16843 assert that two references with the same misalignment have the same
16844 compile-time misalignment if those compile-time misalignments
16847 2017-07-01 Andi Kleen <ak@linux.intel.com>
16849 * print-tree.c (print_node): Print all attributes.
16851 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16853 * cfg.c (scale_bbs_frequencies): New function.
16854 * cfg.h (scale_bbs_frequencies): Declare it.
16855 * cfgloopanal.c (single_likely_exit): Cleanup.
16856 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
16858 (scale_loop_profile): Likewise.
16859 (loop_version): Likewise.
16860 (create_empty_loop_on_edge): Update.
16861 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
16862 scale_loop_frequencies, scale_loop_profile, loopify,
16863 loop_version): Update prototypes.
16864 * modulo-sched.c (sms_schedule): Update.
16865 * predict.c (unlikely_executed_edge_p): Also check probability.
16866 (probably_never_executed_edge_p): Fix typo.
16867 * tree-if-conv.c (version_loop_for_if_conversion): Update.
16868 * tree-parloops.c (gen_parallel_loop): Update.
16869 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
16870 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
16871 * tree-ssa-loop-split.c (split_loop): Update.
16872 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
16873 * tree-vect-loop-manip.c (vect_do_peeling): Update.
16874 (vect_loop_versioning): Update.
16875 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
16877 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16879 * trans-mem.c (split_bb_make_tm_edge): Update profile.
16881 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16883 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
16884 to keep profile consistent.
16886 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16888 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
16889 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
16890 * profile-count.h (max_safe_multiplier): Make unsigned.
16891 (profile_count::guessed_zero): New.
16893 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16895 * bb-reorder.c (fix_up_crossing_landing_pad,
16896 fix_crossing_conditional_branches): Use make_single_succ_edge
16897 to keep profile consistent.
16899 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
16901 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
16904 2017-07-01 Jakub Jelinek <jakub@redhat.com>
16907 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
16908 the right scopes, make sure cond_jump isn't preserved between multiple
16909 iterations. Search for fallthru edge whenever there are 3+ edges and
16910 use find_fallthru_edge for it.
16912 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16914 Patch by Alexander Monakov <amonakov@ispras.ru>
16915 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
16916 probabilities consistently.
16918 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16920 * pa.c (pa_expand_compare_and_swap_loop): Update call of
16921 emit_cmp_and_jump_insns.
16923 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
16926 * tree-inline.c (expand_call_inline): Combine profile statuses.
16928 2017-06-30 Andrew Pinski <apinski@cavium.com>
16930 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
16931 fold_stmt returned true.
16933 2017-06-30 Nathan Sidwell <nathan@acm.org>
16935 * ggc.h (empty_string): Delete.
16936 * cfgexpand.c (expand_asm_stmt): Use plain "".
16937 * optabs.c (expand_asm_memory_barrier): Likewise.
16938 * stringpool.c (empty_string): Delete.
16939 (digit_vector, digit_string): Delete.
16940 (ggc_alloc_string): Use plain "", don't optimize single digit
16941 strings. Use ggc_alloc_atomic.
16943 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
16945 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
16946 comparison set and one other set, use the cost of the non-comparison
16949 2017-06-30 Nathan Sidwell <nathan@acm.org>
16951 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
16954 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
16956 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
16957 loops. Remove now unneeded calls to gimple_switch_set_label() that
16958 just set removed labels to NULL_TREE.
16960 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
16962 * tree-ssanames.c (set_range_info_raw): Abstract from ...
16963 (set_range_info): ...here. Only call set_range_info_raw if domain
16965 (set_nonzero_bits): Call set_range_info_raw.
16966 * tree-ssanames.h (set_range_info_raw): New.
16968 2017-06-30 Jakub Jelinek <jakub@redhat.com>
16971 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
16972 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
16973 of nonimmediate_operand and <store_mask_constraint> instead of m
16974 for the input operand. For V8FI iterator, always split if input
16975 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
16976 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
16977 <store_mask_predicate> instead of register_operand and
16978 <store_mask_constraint> instead of v for the input operand. Make
16979 sure both operands aren't MEMs for if not <mask_applied>.
16981 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
16983 * lto-wrapper.c (copy_file) Close both file descriptors before
16986 2017-06-30 Martin Liska <mliska@suse.cz>
16989 * multiple_target.c (create_dispatcher_calls): Make ifunc
16990 also for function that don't have calls or are not referenced.
16992 2017-06-30 Richard Biener <rguenther@suse.de>
16994 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
16995 analyze the first scalar stmt. Move vector type computation
16996 for the BB case here from ...
16997 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
16998 live operation processing in the SLP case properly.
17000 2017-06-30 Richard Biener <rguenther@suse.de>
17002 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
17004 2017-06-30 Martin Liska <mliska@suse.cz>
17007 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
17008 before BUILT_IN_UNWIND_RESUME when ASAN is used.
17010 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
17012 * doc/invoke.texi (AArch64): Add missing options and remove redundant
17015 2017-06-30 Richard Biener <rguenther@suse.de>
17017 PR tree-optimization/81249
17018 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
17019 condition reduction result to original scalar type.
17021 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17023 * profile-count.h (enum profile_quality): Fix typos and whitespace
17026 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17028 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
17029 type for branch probabilities.
17031 2017-06-29 Julian Brown <julian@codesourcery.com>
17032 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17034 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
17035 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
17036 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
17037 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
17039 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
17041 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
17042 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
17043 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
17044 CC usage from generic code to here.
17045 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
17046 CC usage into the target macros.
17048 2017-06-29 Maya Rashish <coypu@sdf.org>
17050 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
17053 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17055 * arm/arm-builtins.c: Include profile-count.h
17056 * except.c (sjlj_emit_function_enter): Use
17057 profile_probability::unlikely.
17059 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17061 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
17062 and tocrel_offset be pointer args rather than implicitly using
17064 (legitimate_constant_pool_address_p, rs6000_emit_move,
17065 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
17066 tocrel_offset and use in toc_relative_expr_p call.
17067 (print_operand, print_operand_address): Use static tocrel_base_oac
17068 and tocrel_offset_oac.
17069 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
17072 2017-06-29 Maya Rashish <coypu@sdf.org>
17074 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
17076 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
17078 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
17079 objects, take into account only the alignment of 'op0' and 'mode1' if
17082 2017-06-29 Steve Ellcey <sellcey@cavium.com>
17084 * ccmp.c (ccmp_tree_comparison_p): New function.
17085 (ccmp_candidate_p): Update to use above function.
17086 (get_compare_parts): New function.
17087 (expand_ccmp_next): Update to use new functions.
17088 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
17090 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
17091 take mode as argument.
17092 * ccmp.h (expand_ccmp_expr): Add mode as argument.
17093 * expr.c (expand_expr_real_1): Pass mode as argument.
17095 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
17097 * combine.c (combine_instructions): Print insns to dump_file, together
17100 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
17102 * asan.c (asan_emit_stack_protection): Update.
17103 (create_cond_insert_point): Update.
17104 * auto-profile.c (afdo_propagate_circuit): Update.
17105 * basic-block.h (struct edge_def): Turn probability to
17106 profile_probability.
17107 (EDGE_FREQUENCY): Update.
17108 * bb-reorder.c (find_traces_1_round): Update.
17109 (better_edge_p): Update.
17110 (sanitize_hot_paths): Update.
17111 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
17112 (make_single_succ_edge): Update.
17113 (check_bb_profile): Update.
17114 (dump_edge_info): Update.
17115 (update_bb_profile_for_threading): Update.
17116 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
17117 probabilitycount to 0.
17118 * cfgbuild.c (compute_outgoing_frequencies): Update.
17119 * cfgcleanup.c (try_forward_edges): Update.
17120 (outgoing_edges_match): Update.
17121 (try_crossjump_to_edge): Update.
17122 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
17123 (expand_gimple_tailcall): Update.
17124 (construct_init_block): Use make_single_succ_edge.
17125 (construct_exit_block): Use make_single_succ_edge.
17126 * cfghooks.c (verify_flow_info): Update.
17127 (redirect_edge_succ_nodup): Update.
17128 (split_edge): Update.
17129 (account_profile_record): Update.
17130 * cfgloopanal.c (single_likely_exit): Update.
17131 * cfgloopmanip.c (scale_loop_profile): Update.
17132 (set_zero_probability): Remove.
17133 (duplicate_loop_to_header_edge): Update.
17134 * cfgloopmanip.h (loop_version): Update prototype.
17135 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
17136 (force_nonfallthru_and_redirect): Update.
17137 (update_br_prob_note): Update.
17138 (rtl_verify_edges): Update.
17139 (purge_dead_edges): Update.
17140 (rtl_lv_add_condition_to_bb): Update.
17141 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
17142 * cgraphunit.c (init_lowered_empty_function): Update.
17143 (cgraph_node::expand_thunk): Update.
17144 * cilk-common.c: Include profile-count.h
17145 * dojump.c (inv): Remove.
17146 (jumpifnot): Update.
17147 (jumpifnot_1): Update.
17148 (do_jump_1): Update.
17150 (do_jump_by_parts_greater_rtx): Update.
17151 (do_compare_rtx_and_jump): Update.
17152 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
17153 do_jump_1. do_compare_rtx_and_jump): Update prototype.
17154 * dwarf2cfi.c: Include profile-count.h
17155 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
17156 (sjlj_emit_dispatch_table): Likewise.
17157 * explow.c: Include profile-count.h
17158 * expmed.c (emit_store_flag_force): Update.
17159 (do_cmp_and_jump): Update.
17160 * expr.c (compare_by_pieces_d::generate): Update.
17161 (compare_by_pieces_d::finish_mode): Update.
17162 (emit_block_move_via_loop): Update.
17163 (store_expr_with_bounds): Update.
17164 (store_constructor): Update.
17165 (expand_expr_real_2): Update.
17166 (expand_expr_real_1): Update.
17167 * expr.h (try_casesi, try_tablejump): Update prototypes.
17168 * gimple-pretty-print.c (dump_probability): Update.
17169 (dump_profile): New.
17170 (dump_gimple_label): Update.
17171 (dump_gimple_bb_header): Update.
17172 * graph.c (draw_cfg_node_succ_edges): Update.
17173 * hsa-gen.c (convert_switch_statements): Update.
17174 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
17175 (find_if_case_1): Update.
17176 (find_if_case_2): Update.
17177 * internal-fn.c (expand_arith_overflow_result_store): Update.
17178 (expand_addsub_overflow): Update.
17179 (expand_neg_overflow): Update.
17180 (expand_mul_overflow): Update.
17181 (expand_vector_ubsan_overflow): Update.
17182 * ipa-cp.c (good_cloning_opportunity_p): Update.
17183 * ipa-split.c (split_function): Use make_single_succ_edge.
17184 * ipa-utils.c (ipa_merge_profiles): Update.
17185 * loop-doloop.c (add_test): Update.
17186 (doloop_modify): Update.
17187 * loop-unroll.c (compare_and_jump_seq): Update.
17188 (unroll_loop_runtime_iterations): Update.
17189 * lra-constraints.c (lra_inheritance): Update.
17190 * lto-streamer-in.c (input_cfg): Update.
17191 * lto-streamer-out.c (output_cfg): Update.
17192 * mcf.c (adjust_cfg_counts): Update.
17193 * modulo-sched.c (sms_schedule): Update.
17194 * omp-expand.c (expand_omp_for_init_counts): Update.
17195 (extract_omp_for_update_vars): Update.
17196 (expand_omp_ordered_sink): Update.
17197 (expand_omp_for_ordered_loops): Update.
17198 (expand_omp_for_generic): Update.
17199 (expand_omp_for_static_nochunk): Update.
17200 (expand_omp_for_static_chunk): Update.
17201 (expand_cilk_for): Update.
17202 (expand_omp_simd): Update.
17203 (expand_omp_taskloop_for_outer): Update.
17204 (expand_omp_taskloop_for_inner): Update.
17205 * omp-simd-clone.c (simd_clone_adjust): Update.
17206 * optabs.c (expand_doubleword_shift): Update.
17207 (expand_abs): Update.
17208 (emit_cmp_and_jump_insn_1): Update.
17209 (expand_compare_and_swap_loop): Update.
17210 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
17211 * predict.c (predictable_edge_p): Update.
17212 (edge_probability_reliable_p): Update.
17213 (set_even_probabilities): Update.
17214 (combine_predictions_for_insn): Update.
17215 (combine_predictions_for_bb): Update.
17216 (propagate_freq): Update.
17217 (estimate_bb_frequencies): Update.
17218 (force_edge_cold): Update.
17219 * profile-count.c (profile_count::dump): Add missing space into dump.
17220 (profile_count::debug): Add newline.
17221 (profile_count::differs_from_p): Explicitly convert to unsigned.
17222 (profile_count::stream_in): Update.
17223 (profile_probability::dump): New member function.
17224 (profile_probability::debug): New member function.
17225 (profile_probability::differs_from_p): New member function.
17226 (profile_probability::differs_lot_from_p): New member function.
17227 (profile_probability::stream_in): New member function.
17228 (profile_probability::stream_out): New member function.
17229 * profile-count.h (profile_count_quality): Rename to ...
17230 (profile_quality): ... this one.
17231 (profile_probability): New.
17232 (profile_count): Update.
17233 * profile.c (compute_branch_probabilities): Update.
17234 * recog.c (peep2_attempt): Update.
17235 * sched-ebb.c (schedule_ebbs): Update.
17236 * sched-rgn.c (find_single_block_region): Update.
17237 (compute_dom_prob_ps): Update.
17238 (schedule_region): Update.
17239 * sel-sched-ir.c (compute_succs_info): Update.
17240 * stmt.c (struct case_node): Update.
17241 (do_jump_if_equal): Update.
17242 (get_outgoing_edge_probs): Update.
17243 (conditional_probability): Update.
17244 (emit_case_dispatch_table): Update.
17245 (expand_case): Update.
17246 (expand_sjlj_dispatch_table): Update.
17247 (emit_case_nodes): Update.
17248 * targhooks.c: Update.
17249 * tracer.c (better_p): Update.
17250 (find_best_successor): Update.
17251 * trans-mem.c (expand_transaction): Update.
17252 * tree-call-cdce.c: Update.
17253 * tree-cfg.c (gimple_split_edge): Upate.
17254 (move_sese_region_to_fn): Upate.
17255 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
17256 * tree-eh.c (lower_resx): Upate.
17257 (cleanup_empty_eh_move_lp): Upate.
17258 * tree-if-conv.c (version_loop_for_if_conversion): Update.
17259 * tree-inline.c (copy_edges_for_bb): Update.
17260 (copy_cfg_body): Update.
17261 * tree-parloops.c (gen_parallel_loop): Update.
17262 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
17263 (gimple_gen_time_profiler): Update.
17264 * tree-ssa-dce.c (remove_dead_stmt): Update.
17265 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
17266 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
17267 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
17268 (unloop_loops): Update.
17269 (try_peel_loop): Update.
17270 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
17271 * tree-ssa-loop-split.c (connect_loops): Update.
17272 (split_loop): Update.
17273 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
17274 (hoist_guard): Update.
17275 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
17276 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
17277 (value_replacement): Update.
17278 * tree-ssa-reassoc.c (branch_fixup): Update.
17279 * tree-ssa-tail-merge.c (replace_block_by): Update.
17280 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
17281 (create_edge_and_update_destination_phis): Update.
17282 (compute_path_counts): Update.
17283 (recompute_probabilities): Update.
17284 (update_joiner_offpath_counts): Update.
17285 (freqs_to_counts_path): Update.
17286 (duplicate_thread_path): Update.
17287 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
17288 (struct switch_conv_info): Update.
17289 (gen_inbound_check): Update.
17290 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
17291 (vect_do_peeling): Update.
17292 (vect_loop_versioning): Update.
17293 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
17294 (optimize_mask_stores): Update.
17295 * ubsan.c (ubsan_expand_null_ifn): Update.
17296 * value-prof.c (gimple_divmod_fixed_value): Update.
17297 (gimple_divmod_fixed_value_transform): Update.
17298 (gimple_mod_pow2): Update.
17299 (gimple_mod_pow2_value_transform): Update.
17300 (gimple_mod_subtract): Update.
17301 (gimple_mod_subtract_transform): Update.
17302 (gimple_ic): Update.
17303 (gimple_stringop_fixed_value): Update.
17304 (gimple_stringops_transform): Update.
17305 * value-prof.h: Update.
17307 2017-06-29 Carl Love <cel@us.ibm.com>
17309 * config/rs6000/rs6000-c.c: Add support for built-in functions
17310 vector signed int vec_signed (vector float);
17311 vector signed long long vec_signed (vector double);
17312 vector signed int vec_signed2 (vector double, vector double);
17313 vector signed int vec_signede (vector double);
17314 vector signed int vec_signedo (vector double);
17315 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
17316 instruction generator.
17317 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17318 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
17319 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
17321 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
17322 vunsignede_v2df): Add define_expands.
17323 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
17324 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
17325 VEC_UNSIGNEDO): Add definitions.
17326 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
17327 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
17328 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
17329 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
17330 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
17331 * config/rs6000/altivec.h (vec_signed, vec_signed2,
17332 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
17333 vec_unsignede, vec_unsignedo): Add builtin defines.
17334 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
17336 * doc/extend.texi: Update the built-in documentation file for the
17337 new built-in functions.
17339 2017-06-29 Richard Biener <rguenther@suse.de>
17341 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
17342 reduction chains to LOOP_VINFO_REDUCTIONS.
17343 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
17344 SLP reductions after processing reduction chains.
17346 2017-06-29 Nathan Sidwell <nathan@acm.org>
17348 * builtins.c (fold_builtin_FUNCTION): Use
17349 lang_hooks.decl_printable_name.
17351 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
17353 PR middle-end/81194
17354 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
17355 with only one label.
17356 * stmt.c (expand_case): Assert NCASES is greater than one.
17358 2017-06-29 Richard Biener <rguenther@suse.de>
17360 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
17362 (group_case_labels): Likewise.
17363 (find_taken_edge): Push sanity checking on val to workers...
17364 (find_taken_edge_cond_expr): ... here
17365 (find_taken_edge_switch_expr): ... and here, handle cases
17366 with just a default label.
17367 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
17368 (group_case_labels): Likewise.
17369 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
17370 group_case_labels does anything cleanup the CFG again.
17372 2017-06-29 Bin Cheng <bin.cheng@arm.com>
17374 PR tree-optimization/81196
17375 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
17376 exit condition comparing two IVs.
17378 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
17380 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
17381 profile to the dummy entry at the end of the list of architectures.
17382 * config/arm/arm-cpu-cdata.h: Regenerated.
17384 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17385 Michael Collison <michael.collison@arm.com>
17388 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
17390 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
17391 (*aarch64_reg_<mode>3_minus_mask): New pattern.
17392 (*aarch64_<optab>_reg_di3_mask2): New pattern.
17393 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
17394 of shift when the shift amount is masked with constant equal to
17395 the size of the mode.
17396 * config/aarch64/predicates.md (subreg_lowpart_operator): New
17399 2017-06-29 Martin Liska <mliska@suse.cz>
17401 * config/i386/i386.opt: Change range from [1,5] to [0,5].
17403 2017-06-29 Yury Gribov <tetra2005@gmail.com>
17406 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
17409 (edge_growth_cache_entry::edge_growth_cache_entry): New
17411 (reset_edge_growth_cache): Update to use constructor.
17413 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
17415 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
17416 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17417 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
17419 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
17421 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
17422 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
17424 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
17426 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
17427 (*-linux-uclibc*): Add t-uclibc tmake_file.
17428 * config/t-musl: New.
17429 * config/t-uclibc: New.
17431 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
17433 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
17435 (gen_comm_data): Emit architectural setting of arch_prof.
17436 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
17438 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
17439 (armv8-m.base, armv8-m.main): Likewise.
17440 * arm-protos.h (arm_build_target): Add profile field.
17441 (arch_option): Likewise.
17442 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
17444 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
17445 arm_active_target.profile.
17447 2017-06-28 Richard Biener <rguenther@suse.de>
17449 PR middle-end/81227
17450 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
17451 TYPE_OVERFLOW_WRAPS.
17452 * match.pd (negate_expr_p): Likewise.
17453 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
17454 fold_build2, not fold_binary.
17456 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17458 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
17459 Convert memory address to Pmode.
17460 (aarch64_print_operand): Assert MEM operands are always Pmode.
17462 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
17465 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
17466 Remove redundant if.
17467 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
17468 * config/arm/aarch-common-protos.h
17469 (aarch_forward_to_shift_is_not_shifted_re): Remove.
17470 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
17472 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
17475 * multiple_target.c (create_dispatcher_calls): Set the default
17476 clone to be static, not public.
17478 2017-06-28 Richard Biener <rguenther@suse.de>
17480 * tree-vect-loop.c (vectorizable_reduction): Move special
17481 cond reduction IV var creation ...
17482 (vect_create_epilog_for_reduction): ... here. Remove induction_index
17483 parameter. Use STMT_VINFO_VECTYPE.
17484 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
17487 2017-06-28 Martin Liska <mliska@suse.cz>
17490 * ipa-visibility.c (non_local_p): Handle visibility.
17492 2017-06-28 Martin Liska <mliska@suse.cz>
17495 * common.opt: Add IntegerRange to various options.
17496 * opt-functions.awk (integer_range_info): New function.
17497 * optc-gen.awk: Add integer_range_info to cl_options struct.
17498 * opts-common.c (decode_cmdline_option): Handle
17499 CL_ERR_INT_RANGE_ARG.
17500 (cmdline_handle_error): Likewise.
17501 * opts.c (print_filtered_help): Show valid interval in
17502 when --help is provided.
17503 * opts.h (struct cl_option): Add range_min and range_max fields.
17504 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
17506 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
17508 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
17509 (x * C EQ/NE y * C): New transformation.
17511 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
17513 * genmultilib (combination_space): Accept '+' in option names.
17515 2017-06-28 Martin Liska <mliska@suse.cz>
17518 * asan.c (instrument_derefs): Bail out inner references
17519 that are hard register variables.
17521 2017-06-28 Jakub Jelinek <jakub@redhat.com>
17524 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
17525 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
17527 2017-06-28 Richard Biener <rguenther@suse.de>
17529 * tree-vectorizer.h (vect_get_vec_defs): Remove.
17530 (vect_get_slp_defs): Adjust.
17531 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
17533 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
17535 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
17536 get_initial_defs_for_reduction instead of vect_get_vec_defs.
17537 (vectorizable_reduction): Adjust.
17538 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
17540 (vect_get_slp_defs): Likewise.
17541 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
17542 (vectorizable_bswap): Adjust.
17543 (vectorizable_call): Likewise.
17544 (vectorizable_conversion): Likewise.
17545 (vectorizable_assignment): Likewise.
17546 (vectorizable_shift): Likewise.
17547 (vectorizable_operation): Likewise.
17548 (vectorizable_store): Likewise.
17549 (vectorizable_condition): Likewise.
17550 (vectorizable_comparison): Likewise.
17552 2017-06-28 Michael Collison <michael.collison@arm.com>
17555 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
17557 (arm_gen_movmemqi): Removed unused variable 'i'.
17558 Convert 'for' loop into 'while' loop.
17559 (arm_expand_prologue): Remove last unnecessary set of insn.
17560 (thumb_pop): Remove unused variable 'pushed_words'.
17561 (thumb_exit): Remove last unnecessary set of regs_to_pop.
17563 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17565 * config/s390/predicates.md: Use s390_rel_address_ok_p.
17566 * config/s390/s390-protos.h: Add prototype of
17567 s390_rel_address_ok_p.
17568 * config/s390/s390.c (s390_got_symbol): New function.
17569 (s390_rel_address_ok_p): New function.
17570 (legitimize_pic_address): Use s390_rel_address_ok_p.
17571 (s390_load_got): Use s390_got_symbol.
17572 (s390_option_override): Issue error if
17573 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
17574 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
17576 * config/s390/s390.opt: New option mpic-data-is-text-relative.
17578 2017-06-27 Andrew Pinski <apinski@cavium.com>
17580 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
17581 (X * copysign (1.0, X)): New pattern.
17582 (X * copysign (1.0, -X)): New pattern.
17583 (copysign (-1.0, CST)): New pattern.
17585 2017-06-27 Joseph Myers <joseph@codesourcery.com>
17587 * genmultilib (combination_space): Remove variable.
17588 Validate reuse rules against regular expression for any sequence
17589 of multilib options in any order.
17591 2017-06-27 Michael Collison <michael.collison@arm.com>
17593 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
17594 call aarch64_split_simd_combine.
17595 * (aarch64_combine_internal<mode>): Delete pattern.
17596 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
17597 Allow register and subreg operands.
17599 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17601 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
17602 specific need, just fallback on defaults.
17603 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
17605 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17606 Olivier Hainque <hainque@adacore.com>
17608 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
17610 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
17611 targets. Pick a default if no particular attempt applied.
17612 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
17615 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17617 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
17618 (x86_64-wrs-vxworks7): Likewise.
17620 2017-06-27 Marek Polacek <polacek@redhat.com>
17623 * ubsan.c (instrument_null): Check get_base_address's result for null.
17625 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
17627 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
17629 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
17631 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
17632 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
17633 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
17634 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
17635 New function types.
17636 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
17637 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
17638 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
17639 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
17640 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
17641 BUILT_IN_FEUPDATEENV): New builtins.
17642 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
17643 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
17644 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
17645 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
17647 (builtin_structptr_types): Adjust size.
17648 * tree.c (builtin_structptr_types): Add four entries.
17650 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17651 Olivier Hainque <hainque@adacore.com>
17653 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
17654 (TLS_SYM): New local macro, forcing reference to __tls__ on
17655 link command lines for VxWorks 7 RTPs, triggering initialization
17657 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
17658 OS features TLS support, true for RTPs on VxWorks 7.
17659 * config/vxworks.c (vxworks_override_options): Setup emutls
17662 2017-06-27 Jakub Jelinek <jakub@redhat.com>
17664 * predict.c (test_prediction_value_range): Use -1U instead of -1
17665 to avoid narrowing conversion warning.
17666 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
17667 to avoid narrowing conversion warning.
17668 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
17670 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
17672 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
17674 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
17675 64bit configurations.
17676 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
17677 (SIZE_TYPE): Likewise.
17678 * config/vxworks.c (vxworks_emutls_var_fields): Use
17679 long_unsigned_type_node instead of unsigned_type_node as the offset
17680 field type, which is "pointer" mode in emutls.c.
17682 2017-06-27 Jakub Jelinek <jakub@redhat.com>
17685 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
17687 PR middle-end/81207
17688 * gimple-fold.c (replace_call_with_call_and_fold): Handle
17689 gimple_vuse copying separately from gimple_vdef copying.
17691 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17693 * value-prof.c (free_hist): Remove call to memset and the enclosing if
17696 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
17697 Olivier Hainque <hainque@adacore.com>
17699 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
17700 for all vxworks7 targets.
17701 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
17702 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
17703 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
17704 variations for VX6/VX7 and 32/64bits later on in ...
17705 (VXWORKS_LIB_SPEC): Leverage new macros.
17706 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
17707 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
17709 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
17711 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
17712 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
17714 2017-06-26 Carl Love <cel@us.ibm.com>
17716 * config/rs6000/rs6000-c.c: Add support for built-in functions
17717 vector bool char vec_reve (vector bool char);
17718 vector signed char vec_reve (vector signed char);
17719 vector unsigned char vec_reve (vector unsigned char);
17720 vector bool int vec_reve (vector bool int);
17721 vector signed int vec_reve (vector signed int);
17722 vector unsigned int vec_reve (vector unsigned int);
17723 vector bool long long vec_reve (vector bool long long);
17724 vector signed long long vec_reve (vector signed long long);
17725 vector unsigned long long vec_reve (vector unsigned long long);
17726 vector bool short vec_reve (vector bool short);
17727 vector signed short vec_reve (vector signed short);
17728 vector double vec_reve (vector double);
17729 vector float vec_reve (vector float);
17730 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
17731 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
17732 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
17733 (altivec_vreve): New pattern.
17734 * config/rs6000/altivec.h (vec_reve): New define.
17735 * doc/extend.texi (vec_rev): Update the built-in documentation file
17736 for the new built-in functions.
17738 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17740 PR tree-optimization/71815
17741 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
17743 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
17745 (slsr_process_phi): Likewise.
17746 (replace_uncond_cands_and_profitable_phis): Don't replace a
17747 multiply candidate with a stride of 1 (copy or cast).
17748 (phi_incr_cost): Call uses_consumed_by_stmt rather than
17750 (lowest_cost_path): Likewise.
17751 (total_savings): Likewise.
17753 2017-06-26 Richard Biener <rguenther@suse.de>
17756 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
17757 Use def_builtin_pure for all gather builtins.
17759 2017-06-26 Richard Biener <rguenther@suse.de>
17761 PR tree-optimization/81203
17762 * tree-tailcall.c (find_tail_calls): Do not move stmts into
17763 non-dominating BBs.
17765 2017-06-26 Marek Polacek <polacek@redhat.com>
17768 * doc/invoke.texi: Document -Wmultistatement-macros.
17770 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
17772 * doc/sourcebuild.texi (ARM-specific attributes): Document new
17773 arm_neon_ok_no_float_abi effective target.
17775 2017-06-26 Richard Biener <rguenther@suse.de>
17777 PR tree-optimization/80928
17778 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
17779 (copy_bbs): Set BB_DUPLICATED flag early.
17780 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
17782 (execute_on_shrinking_pred): Likewise.
17783 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
17784 BB_DUPLICATED blocks.
17785 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
17786 iterate over all PHIs considering removal of *gsi.
17788 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
17790 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
17793 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
17795 * config/rs6000/rs6000-string.c: (expand_block_clear,
17796 do_load_for_compare, select_block_compare_mode,
17797 compute_current_alignment, expand_block_compare,
17798 expand_strncmp_align_check, expand_strn_compare,
17799 expand_block_move, rs6000_output_load_multiple)
17800 Move functions related to string/block move/compare
17801 to a separate file.
17802 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
17803 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
17804 for this function which is now used in two files.
17805 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
17806 * config.gcc: Add rs6000-string.o to extra_objs for
17807 targets powerpc*-*-* and rs6000*-*-*.
17809 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
17812 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
17813 32-bit, since indexed is not valid for DImode.
17814 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
17815 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
17816 (define_peephole2 for Altivec d-form load): Add 32-bit support.
17817 (define_peephole2 for Altivec d-form store): Likewise.
17820 * multiple_target.c (create_dispatcher_calls): Only create the
17821 dispatcher call if the function is the default clone of a
17822 versioned function.
17824 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
17826 PR middle-end/80902
17827 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
17828 a call, force the call to not be a tail call.
17830 2017-06-23 Jeff Law <law@redhat.com>
17832 * doc/contrib.texi: Add entry for Steven Pemberton's work on
17835 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
17837 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
17838 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
17839 handling for early expansion of vector shifts (sl,sr,sra,rl).
17840 (builtin_function_type): Add vector shift right instructions
17841 to the unsigned argument list.
17843 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
17845 rtl-optimizatoin/79286
17846 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
17847 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
17848 trap. PIC register plus a const unspec without offset can never trap.
17850 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
17852 * tree.h (builtin_structptr_type): New type.
17853 (builtin_structptr_types): Declare new array.
17854 * tree.c (builtin_structptr_types): New array.
17855 (free_lang_data, build_common_tree_nodes): Use it.
17857 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
17860 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
17863 2017-06-22 Matt Turner <mattst88@gmail.com>
17865 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
17866 Lake models to skylake case. Assume skylake for unknown
17867 models with clflushopt.
17869 2017-06-22 Jeff Law <law@redhat.com>
17871 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
17872 frame sizes that do not satisfy aarch64_uimm12_shift.
17874 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
17876 * profile-count.h (apply_probability,
17877 apply_scale, probability_in): Fix checks for zero.
17879 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17881 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
17882 * doc/cppdiropts.texi (-I @var{dir}): Document it.
17884 2016-06-22 Richard Biener <rguenther@suse.de>
17886 * tree-vect-loop.c (vect_model_reduction_cost): Handle
17887 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
17888 REDUC_MAX_EXPR support.
17889 (vectorizable_reduction): Likewise.
17890 (vect_create_epilog_for_reduction): Likewise.
17892 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
17894 * match.pd (A / (1 << B) -> A >> B): New.
17895 * generic-match-head.c: Include optabs-tree.h.
17896 * gimple-match-head.c: Likewise.
17897 * optabs-tree.h (target_supports_op_p): New.
17898 * optabs-tree.c (target_supports_op_p): New.
17900 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
17902 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
17903 $gcc_cv_ld --help output.
17904 (gcc_cv_ld_demangle): Likewise.
17905 (gcc_cv_ld_eh_frame_hdr): Likewise.
17906 (gcc_cv_ld_pie): Likewise.
17907 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
17908 (gcc_cv_ld_buildid): Likewise.
17909 (gcc_cv_ld_sysroot): Likewise.
17910 (ld_bndplt_support): Likewise.
17911 (ld_pushpopstate_support): Likewise.
17912 * configure: Regenerate.
17913 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
17915 2017-06-21 Jakub Jelinek <jakub@redhat.com>
17918 * config/i386/sse.md (round<mode>2): Renumber match_dup and
17919 operands indexes to avoid gap between operands and match_dups.
17921 2017-06-21 Andrew Pinski <apinski@cavium.com>
17923 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
17924 Increment Arith_shift and Arith_shift_reg by 1.
17925 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
17927 * config/aarch64/aarch64.c (thunderx_tunings): Enable
17928 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
17929 (aarch64_strip_extend): Add new argument and test for it.
17930 (aarch64_cheap_mult_shift_p): New function.
17931 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
17932 add a cost if it is true.
17933 Update calls to aarch64_strip_extend.
17934 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
17936 2017-06-21 Andrew Pinski <apinski@cavium.com>
17938 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
17940 (thunderxt88): Likewise.
17941 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
17942 (thunderx_prefetch_tune): New variable.
17943 (thunderx2t99_prefetch_tune): Update for the correct values.
17944 (thunderxt88_tunings): New variable.
17945 (thunderx_tunings): Use thunderx_prefetch_tune instead of
17946 generic_prefetch_tune.
17947 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
17949 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17951 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
17952 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
17953 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
17954 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
17955 (aarch64_atomic_cas<mode>, GPI): Likewise.
17957 2017-06-21 Martin Liska <mliska@suse.cz>
17959 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
17960 statements on cold and hot labels.
17961 * predict.c (tree_estimate_probability_bb): Remove the
17962 prediction from this place.
17964 2017-06-21 Martin Liska <mliska@suse.cz>
17966 PR tree-optimization/79489
17967 * gimplify.c (maybe_add_early_return_predict_stmt): New
17969 (gimplify_return_expr): Call the function.
17970 * predict.c (tree_estimate_probability_bb): Remove handling
17972 * predict.def: Update comment about early return predictor.
17973 * gimple-predict.h (is_gimple_predict): New function.
17974 * predict.def: Change default value of early return to 66.
17975 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
17977 * passes.def: Put pass_strip_predict_hints to the beginning of
17980 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
17982 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
17983 FUNCTION_DECL declarations.
17984 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
17986 (dwaf2out_decl): Likewise.
17987 * godump.c (go_early_global_decl): Skip call to the real debug hook
17988 for FUNCTION_DECL declarations.
17989 * passes.c (rest_of_decl_compilation): Skip call to the
17990 early_global_decl debug hook for FUNCTION_DECL declarations, unless
17991 -fdump-go-spec is passed.
17993 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
17995 * config/i386/i386.c (struct builtin_isa): New field pure_p.
17996 Reorder for compactness.
17997 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
17998 (def_builtin_pure, def_builtin_pure2): New functions.
17999 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
18001 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
18003 * match.pd (nop_convert): New predicate.
18004 ((A +- CST1) +- CST2): Allow some NOP conversions.
18006 2017-06-21 Jakub Jelinek <jakub@redhat.com>
18009 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
18010 with ctors/dtors if GOVD_SHARED is set.
18012 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18014 * config/aarch64/aarch64.md (movti_aarch64):
18015 Emit mov rather than orr.
18016 (movtf_aarch64): Likewise.
18017 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
18018 Emit mov rather than orr.
18020 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18022 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
18023 Swap alternatives, make integer dup more expensive.
18025 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
18027 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
18028 Return true for non-tls symbols.
18030 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
18032 * config/aarch64/aarch64-cores.def (cortex-a55): New.
18033 (cortex-a75): Likewise.
18034 (cortex-a75.cortex-a55): Likewise.
18035 * config/aarch64/aarch64-tune.md: Regenerate.
18036 * doc/invoke.texi (-mtune): Document new values for -mtune.
18038 2017-06-21 Tom de Vries <tom@codesourcery.com>
18040 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
18041 stack_size feature.
18042 (Effective-Target Keywords, Other attributes): Suggest using
18043 dg-add-options stack_size feature to get stack limit in stack_size
18044 effective target documentation.
18046 2017-06-21 Julian Brown <julian@codesourcery.com>
18047 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18049 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
18050 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
18051 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
18053 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
18054 attribute type list for neon_multiply.
18055 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
18056 attribute type list for neon_multiply.
18057 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
18058 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
18059 attribute type list for neon_multiply.
18060 * config/arm/types.md (crypto_pmull): Add.
18061 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
18062 attribute type list.
18064 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
18066 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
18069 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18071 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
18072 to make sure not to dereference a NULL cost_classes_ptr pointer.
18074 2017-06-20 Carl Love <cel@us.ibm.com>
18076 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18077 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
18078 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
18079 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
18080 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
18081 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
18082 VMULOSW): New enum "unspec" values.
18083 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
18084 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
18085 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
18086 altivec_vmulosw): New patterns.
18087 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
18088 VMULOSW): Add definitions.
18090 2017-06-20 Julia Koval <julia.koval@intel.com>
18092 * config/i386/i386.c: Fix rounding expand for new pattern.
18093 * config/i386/subst.md: Fix pattern (parallel -> unspec).
18095 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18097 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
18098 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
18100 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18102 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
18105 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
18107 * config/aarch64/aarch64-cores.def: Rearrange to sort by
18108 architecture, then by implementer ID.
18109 * config/aarch64/aarch64-tune.md: Regenerate.
18111 2017-06-20 Richard Biener <rguenther@suse.de>
18113 PR middle-end/81097
18114 * fold-const.c (split_tree): Fold to type before negating.
18116 2017-06-20 David Malcolm <dmalcolm@redhat.com>
18118 * diagnostic-show-locus.c
18119 (selftest::test_fixit_deletion_affecting_newline): New function.
18120 (selftest::diagnostic_show_locus_c_tests): Call it.
18122 2017-06-20 Andreas Schwab <schwab@suse.de>
18125 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
18128 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
18130 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
18131 __ARM_FEATURE_COPROC according to support.
18133 2017-06-20 Jakub Jelinek <jakub@redhat.com>
18135 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
18136 Rewritten to avoid overflow for > 32-bit pointers.
18139 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
18140 by removing enum keyword.
18141 (ubsan_type_descriptor): Likewise. Formatting fix.
18144 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
18145 splitter): Require TARGET_SSE2 in the condition.
18147 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
18150 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
18151 for doing vector set of SFmode on ISA 3.0.
18152 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
18153 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
18155 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
18156 SFmode value into a V4SF variable that was extracted from another
18157 V4SF variable without converting the element to double precision
18158 and back to single precision vector format.
18159 (vsx_insert_extract_v4sf_p9_2): Likewise.
18161 2017-06-19 Jakub Jelinek <jakub@redhat.com>
18163 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
18164 in UWHI to avoid undefined overflow.
18167 * ubsan.h (enum ubsan_encode_value_phase): New.
18168 (ubsan_encode_value): Change second argument to
18169 enum ubsan_encode_value_phase with default value of
18170 UBSAN_ENCODE_VALUE_GENERIC.
18171 * ubsan.c (ubsan_encode_value): Change second argument to
18172 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
18173 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
18174 create_tmp_var_raw instead of create_tmp_var and use a
18176 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
18177 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
18178 ubsan_encode_value callers.
18181 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
18182 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
18183 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
18185 2017-06-19 Richard Biener <rguenther@suse.de>
18187 PR middle-end/81118
18188 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
18189 estimates if we changed anything.
18191 2017-06-19 Richard Biener <rguenther@suse.de>
18193 PR tree-optimization/80887
18194 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
18195 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
18196 simplified lookups, then reset mprts_hook.
18197 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
18199 (try_to_simplify): Likewise.
18201 2017-06-19 Martin Liska <mliska@suse.cz>
18204 * gimplify.c (gimplify_switch_expr):
18205 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
18207 2017-06-19 Martin Liska <mliska@suse.cz>
18209 * doc/install.texi: Document that PGO runs in 4 stages.
18211 2017-06-19 Martin Liska <mliska@suse.cz>
18214 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
18215 to dispatcher function name.
18216 * multiple_target.c (replace_function_decl): New function.
18217 (create_dispatcher_calls): Redirect both edges and references.
18219 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
18221 * profile-count.c (profile_count::dump): Dump quality.
18222 (profile_count::differs_from_p): Update for unsigned val.
18223 * profile-count.h (profile_count_quality): New enum.
18224 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
18226 2017-06-19 Richard Biener <rguenther@suse.de>
18228 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
18229 struct function as arg.
18230 (estimate_numbers_of_iterations): Export overload with loop arg.
18231 (free_numbers_of_iterations_estimates_loop): Use an overload of
18232 free_numbers_of_iterations_estimates instead.
18233 * tree-cfg.c (remove_bb): Adjust.
18234 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
18235 * tree-parloops.c (gen_parallel_loop): Likewise.
18236 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
18238 (tree_unroll_loops_completely): Likewise.
18239 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
18240 Use an overload instead and export.
18241 (estimated_loop_iterations): Adjust.
18242 (max_loop_iterations): Likewise.
18243 (likely_max_loop_iterations): Likewise.
18244 (estimate_numbers_of_iterations): Take struct function as arg
18246 (loop_exits_before_overflow): Adjust.
18247 (free_numbers_of_iterations_estimates_loop): Use an overload.
18248 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
18249 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
18251 2017-06-19 Richard Biener <rguenther@suse.de>
18254 * ipa-prop.c (find_constructor_constant_at_offset): Handle
18255 RANGE_EXPR conservatively.
18257 2017-06-16 Carl Love <cel@us.ibm.com>
18259 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
18260 definitions for vec_float, vec_float2, vec_floato,
18261 vec_floate built-ins.
18262 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
18263 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
18265 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
18266 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
18267 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
18268 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
18269 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
18270 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
18271 vec_floato): Add builtin defines.
18272 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
18273 Update the built-in documentation file for the new built-in
18276 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18278 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
18279 (mthumb): Mark as the negative of -marm.
18281 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18283 * doc/invoke.texi (ARM Options, -mcpu): Document supported
18285 (ARM Options, -mtune): Document that this accepts the same
18286 extension options as -mcpu.
18287 (ARM Options, -mfpu): Document addition of -mfpu=auto.
18289 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18291 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
18292 permitted extensions.
18294 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18296 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
18297 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
18298 (armv8-m.main): Add option +nodsp.
18299 * config/arm/arm-cpu-cdata.h: Regenerated.
18301 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18303 * config/arm/t-fuchsia: New file.
18304 * config.gcc (arm*-*-fuchsia*): Use it.
18306 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18308 * config/arm/t-symbian: Rewrite for new option infrastructure.
18310 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18312 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
18313 (MULTILIB_REQUIRED): Likewise.
18315 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18317 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
18318 (MULTILIB_RESUE): Likewise.
18319 (MULTILIB_MATCHES): Likewise.
18320 (MULTLIB_REQUIRED): Likewise.
18322 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18324 * config/arm/t-rtems: Rewrite for new option framework.
18326 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18328 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
18329 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
18330 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
18331 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
18332 * config/arm/t-multilib: ... here.
18333 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
18334 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
18335 armv7-a and armv8*-a when A-profile libraries have not been built.
18336 * config/arm/t-rmprofile: Rewrite.
18338 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18340 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
18341 with a backslash. Remove the backslash after substituting unescaped
18343 * doc/fragments.texi (MULTILIB_REUSE): Document it.
18345 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18347 * config.gcc: (arm*-*-*): When building a-profile libraries, force
18348 the driver to pass through the default setting of -mfloat-abi.
18349 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
18351 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
18352 (all_feat_combs): New rule.
18353 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
18355 * config/arm/t-aprofile: Rewrite.
18357 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18359 * config/arm/arm.h (FPUTYPE_AUTO): Define.
18360 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
18361 fpu is not specified by the user/command-line.
18362 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
18363 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
18364 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
18365 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
18366 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
18367 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
18369 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18371 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
18372 * config/arm/t-arm-elf: Rewritten.
18374 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18376 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
18377 have some floating-point instructions.
18378 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
18379 (TARGET_MAYBE_HARD_FLOAT): New macro.
18380 * config/arm/arm-builtins.c (arm_init_builtins): Use
18381 TARGET_MAYBE_HARD_FLOAT.
18382 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
18384 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18386 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
18387 (configargs.h): Include it.
18388 (arm_print_hint_for_fpu_option): New function.
18389 (arm_parse_fpu_option): New function.
18390 (candidate_extension): New class.
18391 (arm_canon_for_multilib): New function.
18392 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
18393 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
18394 (ARCH_CANONICAL_SPECS): New macro.
18395 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
18397 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18399 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
18400 are set after handling multilib fragments. Set target_cpu_default2
18403 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18405 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
18407 (arm*-*-*): Set target_cpu_default2 to a quoted string.
18408 * config/arm/parsecpu.awk (check_cpu): Validate any extension
18410 (check_arch): Likewise.
18411 * config/arm/arm.c (arm_configure_build_target): Handle
18412 TARGET_CPU_DEFAULT being a string constant. Scan any feature
18413 options in the default.
18415 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18417 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
18418 when an option is an alias of another.
18419 * config/arm/parsecpu.awk (optalias): New parser token.
18420 (gen_comm_data): Mark non-alias options as such. Emit entries
18421 for extension aliases.
18422 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
18423 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
18424 (armv6kz, armv6zk, armv6t2): Likewise.
18425 (armv7): Make vfpv3-d16 an alias.
18426 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
18428 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
18429 Sort in canonical order.
18430 (armv8-a): Sort in canonical order.
18431 (armv8.1-a, armv8.2-a): Likewise.
18432 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
18434 (cortex-a9): Sort in canonical order.
18435 * config/arm/arm.c (selftests.h): Include it.
18436 (arm_test_cpu_arch_data): New function.
18437 (arm_run_self_tests): New function.
18438 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
18439 (targetm): Move declaration to the end of the file.
18440 * arm-cpu-cdata.h: Regenerated.
18442 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18444 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
18445 call to target_mode_check describing the type of option passed.
18446 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
18447 (arm_target_thumb_only): Use arm_parse_arch_option_name or
18448 arm_parse_cpu_option_name to match parameters against list of
18450 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
18451 arm_arch_core_flags data structure.
18452 * config/arm/arm-cpu_cdata.h: Regenerated.
18454 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18456 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
18458 (arm_print_hint_for_cpu_option): Likewise.
18459 (arm_print_hint_for_arch_option): Likewise.
18460 (arm_parse_cpu_option_name): Likewise.
18461 (arm_parse_arch_option_name): Likewise.
18462 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
18463 of entries in the all_fpus list.
18464 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
18465 (arm_parse_cpu_option_name): Declare.
18466 (arm_parse_arch_option_name): Declare.
18467 (arm_parse_option_features): Declare.
18468 (arm_intialize_isa): Declare.
18469 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
18471 (gen_comm_data): ... here. Make definitions non-static.
18472 * config/arm/arm-cpu-data.h: Regenerated.
18473 * config/arm/arm-cpu-cdata.h: Regenerated.
18475 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18477 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
18478 (cpu_arch_extension): New structure.
18479 (cpu_arch_option, arch_option, cpu_option): New structures.
18480 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
18481 architecture types.
18482 (gen_data): Generate new format data tables.
18483 * config/arm/arm.c (cpu_tune): New structure.
18484 (cpu_option, processors): Delete.
18485 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
18486 (arm_print_hint_for_cpu_option): ... this and ...
18487 (arm_print_hint_for_arch_option): ... this.
18488 (arm_parse_arch_cpu_name): Delete. Replace with ...
18489 (arm_parse_cpu_option_name): ... this and ...
18490 (arm_parse_arch_option_name): ... this.
18491 (arm_unrecognized_feature): Change type of target parameter to
18493 (arm_parse_arch_cpu_features): Delete. Replace with ...
18494 (arm_parse_option_features): ... this.
18495 (arm_configure_build_target): Rework to use new configuration data
18497 (arm_print_tune_info): Rework for new configuration data tables.
18498 * config/arm/arm-cpu-data.h: Regenerated.
18499 * config/arm/arm-cpu.h: Regenerated.
18501 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18503 * Makefile.in (OBJS): Move sbitmap.o from here ...
18504 (OBJS-libcommon): ... to here.
18506 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18508 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
18509 (ISA_ALL_CRYPTO): New macro.
18510 (ISA_ALL_SIMD): New macro
18511 (ISA_ALL_FP): New macro.
18512 * config/arm/arm.c (fpu_bitlist): Update initializer.
18513 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
18515 (arm9e): Add fpu. Add option for nofp
18516 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
18517 (arm926ej-s, arm1026ej-s): Likewise.
18518 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
18519 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
18520 neon-fp16, neon-vfpv4, nofp and nosimd.
18521 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
18522 (cortex-a8): Add fpu. Add option for nofp.
18523 (cortex-a9): Add fpu. Add options for nosimd and nofp.
18524 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
18525 (cortex-r4f): Add fpu.
18526 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
18527 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
18529 (cortex-r8): Likewise.
18530 (cortex-m4): Add fpu. Add option for nofp.
18531 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
18532 (cortex-a17.cortex-a7): Likewise.
18533 (cortex-a32): Add fpu. Add options for crypto and nofp.
18534 (cortex-a35, cortex-a53): Likewise.
18535 (cortex-a57): Add fpu. Add option for crypto.
18536 (cortex-a72, cortex-a73): Likewise.
18537 (exynos-m1): Likewise.
18538 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
18539 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
18540 (cortex-m33): Add fpu. Add option for nofp.
18541 * config/arm/arm-cpu-cdata.h: Regenerated
18542 * config/arm/arm-cpu-data.h: Regenerated.
18544 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18546 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
18547 (armv5te, armv5tej): Likewise.
18548 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
18549 (armv7): Add options fp and vfpv3-d16.
18550 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
18551 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
18553 (armv7ve): Likewise.
18554 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
18555 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
18556 (armv8-a): Add nocrypto option.
18557 (armv8.1-a, armv8.2-a): Likewise.
18558 (armv8-m.main): add options fp, fp.dp and nofp.
18560 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18562 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
18564 (armv8-a+crc): Delete.
18565 (armv8.1-a): Add options simd, crypto and nofp.
18566 (armv8.2-a): Add options fp16, simd, crypto and nofp.
18567 (armv8.2-a+fp16): Delete.
18568 (armv8-m.main): Add option dsp.
18569 (armv8-m.main+dsp): Delete.
18570 (cortex-a8): Add fpu. Add nofp option.
18571 (cortex-a9): Add fpu. Add nofp and nosimd options.
18572 * config/arm/parsecpu.awk (gen_data): Generate option tables and
18573 link to main cpu and architecture data structures.
18574 (gen_comm_data): Only put isa attributes from the main architecture
18576 (option): New statement for architecture and CPU entries.
18577 * arm.c (struct cpu_option): New structure.
18578 (struct processors): Add entry for options.
18579 (arm_unrecognized_feature): New function.
18580 (arm_parse_arch_cpu_name): Ignore any characters after the first
18582 (arm_parse_arch_cpu_feature): New function.
18583 (arm_configure_build_target): Separate out any CPU and architecture
18584 features and parse separately. Don't error out if -mfpu=auto is
18585 used with only an architecture string.
18586 (arm_print_asm_arch_directives): New function.
18587 (arm_file_start): Call it.
18588 * config/arm/arm-cpu-cdata.h: Regenerated.
18589 * config/arm/arm-cpu-data.h: Likewise.
18590 * config/arm/arm-tables.opt: Likewise.
18592 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18594 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
18595 assembler when it is not -mfpu=auto.
18597 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18599 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
18600 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
18601 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
18602 (ASM_CPU_SPEC): Rewrite.
18603 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
18604 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
18605 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
18606 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
18607 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
18608 copied string is NUL-terminated. Also strip any characters prefixed
18610 (arm_rewrite_selected_arch): New function.
18611 (arm_rewrite_march): New function.
18613 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
18615 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
18616 (x_arm_cpu_string, x_arm_tune_string): Likewise.
18617 (march, mcpu, mtune): Convert to string-based options.
18618 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
18619 (arm_parse_arch_cpu_name): New function.
18620 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
18621 identify selected architecture or CPU.
18622 (arm_option_save): New function.
18623 (TARGET_OPTION_SAVE): Redefine.
18624 (arm_option_restore): Restore string options.
18625 (arm_option_print): Print string options.
18627 2017-06-16 Martin Sebor <msebor@redhat.com>
18629 PR tree-optimization/80933
18630 PR tree-optimization/80934
18631 * builtins.c (fold_builtin_3): Do not handle bcmp here.
18632 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
18633 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
18634 (gimple_fold_builtin): Call them.
18636 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18638 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
18639 as unlikely; update profile.
18641 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18643 * predict.c (force_edge_cold): Handle declaring edges impossible
18646 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18648 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
18650 (try_unroll_loop_completely): Fix reporting.
18652 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18654 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
18656 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
18659 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
18660 if given a non-constant argument for an intrinsic which requires a
18663 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18665 * profile.c (compare_freqs): New function.
18666 (branch_prob): Sort edge list.
18667 (find_spanning_tree): Assume that the list is priority sorted.
18669 2017-06-16 Richard Biener <rguenther@suse.de>
18671 PR tree-optimization/81090
18672 * passes.def (pass_record_bounds): Remove.
18673 * tree-pass.h (make_pass_record_bounds): Likewise.
18674 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
18675 make_pass_record_bounds): Likewise.
18676 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
18677 not free niter estimates at the beginning but at the end.
18678 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
18680 2017-06-16 Richard Biener <rguenther@suse.de>
18682 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
18683 initializer to workaround ICE in host GCC 4.8.
18685 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
18687 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
18689 (clone_inlined_nodes): Update.
18691 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18693 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
18694 prefetch settings, and enable prefetching by default at -O3.
18696 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18698 * config/aarch64/aarch64.c (aarch64_override_options_internal):
18699 Set flag_prefetch_loop_arrays according to tuning data.
18701 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
18703 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
18704 New tune structure.
18705 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
18706 [Unrelated to main purpose of the patch] Place the pointer field last
18707 to enable type checking errors when tune structure are wrongly merged.
18708 * config/aarch64/aarch64.c (generic_prefetch_tune,)
18709 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
18710 (thunderx2t99_prefetch_tune): New tune constants.
18711 (tune_params *_tunings): Update all tunings (no functional change).
18712 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
18713 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
18714 from tunings structures.
18716 2017-06-16 Jakub Jelinek <jakub@redhat.com>
18719 * ubsan.c (instrument_null): Add T argument, use it instead
18720 of computing it based on IS_LHS.
18721 (instrument_object_size): Likewise.
18722 (pass_ubsan::execute): Adjust instrument_null and
18723 instrument_object_size callers to pass gimple_get_lhs or
18724 gimple_assign_rhs1 result to it. Use instrument_null instead of
18725 calling get_base_address and instrument_mem_ref. Handle
18726 aggregate call arguments for object-size sanitization.
18728 2017-06-16 Yury Gribov <tetra2005@gmail.com>
18730 PR tree-optimization/81089
18731 * tree-vrp.c (is_masked_range_test): Validate operands of
18734 2017-06-15 Martin Sebor <msebor@redhat.com>
18737 * dumpfile.c (dump_register): Avoid calling memset to initialize
18738 a class with a default ctor.
18739 * gcc.c (struct compiler): Remove const qualification.
18740 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
18741 * hash-table.h: Ditto.
18742 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
18744 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
18745 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
18747 * params.h (struct param_info): Make struct members non-const.
18748 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
18749 with default initialization.
18750 * vec.h (vec_copy_construct, vec_default_construct): New helper
18752 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
18753 with vec_copy_construct.
18754 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
18755 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
18756 * doc/invoke.texi (-Wclass-memaccess): Document.
18758 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18760 * emit-rtl.h (is_leaf): Update comment about local
18761 register allocator.
18763 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
18766 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
18767 for a variable to have a section before checking if the section has a
18769 Set section to.persistent if persistent attribute is set.
18770 Warn if .persistent attribute is used on an automatic variable.
18772 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
18774 PR rtl-optimization/80474
18775 * reorg.c (update_block): Do not ignore instructions in a delay slot.
18777 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
18779 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
18782 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
18784 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
18785 (casesi): Emit bounds checking as RTL.
18786 (casesi_internal_mips16_<mode>): Remove bounds checking.
18788 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
18790 * config/xtensa/xtensa.c (xtensa_option_override): Append
18791 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
18792 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
18793 xtensa_doloop_hooks): Define unconditionally.
18794 (xtensa_reorg_loops): Only call reorg_loops in the presence of
18796 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
18797 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
18798 for it in xtensa_option_override.
18799 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
18800 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
18802 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
18804 * doc/cppopts.texi: Document '-' special value to -MF.
18806 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
18808 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
18809 (cortex_a53_fconst): Likewise.
18810 (cortex_a53_fpmul): Likewise.
18811 (cortex_a53_f_load_64): Likewise.
18812 (cortex_a53_f_load_many): Likewise.
18813 (cortex_a53_advsimd_alu): Likewise.
18814 (cortex_a53_advsimd_alu_q): Likewise.
18815 (cortex_a53_advsimd_mul): Likewise.
18816 (cortex_a53_advsimd_mul_q): Likewise.
18817 (fpmac bypass): Add new bypass for fpmac-fpmac case.
18818 Add missing fmul, r2f_cvt and fconst cases.
18820 2017-06-14 Richard Biener <rguenther@suse.de>
18822 PR middle-end/81088
18823 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
18825 (fold_binary_loc): When associating do not treat pre-existing
18826 TREE_OVERFLOW on literal constants as a reason to allow
18827 TREE_OVERFLOW on associated literal constants.
18829 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18831 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
18832 (MASK_FEATURES): New macro.
18833 * config/sparc/sparc.c (sparc_option_override): Remove the special
18834 handling of -mfpu and generalize it to all MASK_FEATURES switches.
18836 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18838 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
18839 a division of 0 if non-call exceptions are enabled.
18841 2017-06-14 Andrew Pinski <apinski@cavium.com>
18842 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
18845 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
18846 Improve vector initialization code gen for only variable case.
18848 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
18850 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
18852 2017-06-14 Richard Biener <rguenther@suse.de>
18854 PR tree-optimization/81083
18855 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
18858 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18860 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
18861 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
18862 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
18863 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
18864 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
18865 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
18867 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18869 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
18870 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
18872 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18874 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
18876 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18878 * config/rs6000/t-rtems: Don't handle SPE.
18880 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18882 * config/rs6000/t-linux: Don't handle SPE.
18884 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18886 * config/rs6000/eabispe.h: Delete file.
18888 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18890 * config/rs6000/t-spe: Delete file.
18892 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18894 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
18895 (rs6000_legitimate_offset_address_p): Return false for anything in
18896 V2SImode or V2SFmode.
18898 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
18900 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
18901 except V2SF and V2SI. Rearrange the vector modes, and add comments.
18902 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
18904 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
18905 (rs6000_legitimate_offset_address_p): Ditto.
18906 (rs6000_emit_move): Ditto.
18907 (rs6000_init_builtins): Remove V4HI_type_node.
18909 2017-06-13 Martin Liska <mliska@suse.cz>
18912 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
18913 (gate_asan): Likewise.
18914 * asan.h (asan_no_sanitize_address_p): Remove the function.
18915 (sanitize_flags_p): New function.
18916 * builtins.def: Fix coding style.
18917 * common.opt: Use renamed enum value.
18918 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
18919 * doc/extend.texi: Document no_sanitize attribute.
18920 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
18921 to SANITIZE_UNDEFINED_NONDEFAULT.
18922 * gcc.c (sanitize_spec_function): Use the renamed enum value.
18923 * gimple-fold.c (optimize_atomic_compare_exchange_p):
18924 Use sanitize_flags_p.
18925 * gimplify.c (gimplify_function_tree): Likewise.
18926 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
18927 * opts.c (parse_no_sanitize_attribute): New function.
18928 (common_handle_option): Use renamed enum value.
18929 * opts.h (parse_no_sanitize_attribute): Declare.
18930 * tree.c (sanitize_flags_p): New function.
18931 * tree.h: Declared here.
18932 * tsan.c: Use sanitize_flags_p.
18933 * ubsan.c (ubsan_expand_null_ifn): Likewise.
18934 (instrument_mem_ref): Likewise.
18935 (instrument_bool_enum_load): Likewise.
18936 (do_ubsan_in_current_function): Remove the function.
18937 (pass_ubsan::execute): Use sanitize_flags_p.
18938 * ubsan.h: Remove do_ubsan_in_current_function
18939 * tree-cfg.c (print_no_sanitize_attr_value): New function.
18940 (dump_function_to_file): Use it here.
18942 2017-06-13 Martin Jambor <mjambor@suse.cz>
18944 PR tree-optimization/80803
18945 PR tree-optimization/81063
18946 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
18947 (propagate_subaccesses_across_link): Enqueue subtree whenever
18948 necessary instead of relying on the caller.
18950 2017-06-13 Martin Jambor <mjambor@suse.cz>
18952 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
18953 that have a first_link.
18954 (sort_and_splice_var_accesses): Do not check first_link before
18956 (subtree_mark_written_and_enqueue): Likewise.
18957 (propagate_all_subaccesses): Likewise and do not stop at first
18958 parent with a first_link.
18960 2017-06-13 Martin Jambor <mjambor@suse.cz>
18962 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
18965 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18967 * match.pd: New pattern.
18969 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18971 * tree-vrp.c (is_masked_range_test): New function.
18972 (register_edge_assert_for): Determine ranges for
18975 2017-06-13 Yury Gribov <tetra2005@gmail.com>
18977 PR tree-optimization/67328
18978 * fold-const.c (maskable_range_p): New function.
18979 (build_range_check): Generate bittests if possible.
18981 2017-06-13 Martin Liska <mliska@suse.cz>
18983 * gimple-pretty-print.c (dump_probability): Add new argument.
18984 (dump_edge_probability): Dump both probability and count.
18985 (dump_gimple_label): Likewise.
18986 (dump_gimple_bb_header): Likewise.
18988 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
18991 * config/avr/avr-devices.c: Fix indentation.
18992 * config/avr/gen-avr-mmcu-specs.c: Dito.
18994 2017-06-13 Richard Biener <rguenther@suse.de>
18996 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
18997 instead get vector type from stmt_info.
18998 (vectorizable_reduction): Adjust. Remove dead code.
19000 2017-06-13 Richard Biener <rguenther@suse.de>
19002 PR middle-end/81065
19003 * fold-const.c (extract_muldiv_1): Remove bogus distribution
19004 case of C * (x * C2 + C3).
19005 (fold_addr_of_array_ref_difference): Properly fold index difference.
19007 2017-06-12 David S. Miller <davem@davemloft.net>
19010 * config/sparc/sparc.md (return expander): Emit frame blockage if
19011 function uses alloca.
19013 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
19015 * combine.c (make_field_assignment): Check len rather than the mode
19016 precision when calling force_to_mode.
19018 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
19020 Support multilibs and devices that see flash in RAM address range.
19023 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
19024 (avr_mcu_t) <flash_pm_offset>: New field.
19025 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
19026 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
19027 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
19028 (AVR_TINY_PM_OFFSET): Remove macro.
19029 * config/avr/avr.opt (-mshort-calls): New option.
19030 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
19031 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
19032 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
19033 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
19034 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
19035 instead of avr_arch->have_jmp_call.
19036 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
19037 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
19038 avr_arch->flash_pm_offset to define.
19039 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
19040 new field flash_pm_offset. Add entry for avrxmega3.
19041 (avr_texinfo): Add entry for avrxmega3.
19042 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
19043 attiny212, attiny214,
19044 attiny412, attiny414, attiny416, attiny417,
19045 attiny814, attiny816, attiny817,
19046 attiny1614, attiny1616, attiny1617,
19047 attiny3214, attiny3216, attiny3217.
19048 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
19049 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
19050 (avr_print_operand_address) [AVR_TINY]: Same.
19051 (avr_asm_init_sections) <readonly_data_section>: Only patch
19052 callback if avr_arch->flash_pm_offset = 0.
19053 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
19054 for rodata if avr_arch->flash_pm_offset != 0.
19055 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
19056 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
19057 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
19058 (m_options): Append opt_rcall.
19059 (m_dirnames): Append dir_rcall.
19060 * config/avr/t-multilib: Regenerate.
19062 * configure.ac [target=avr]: Check whether avrxmega3 default
19063 linker description file works as needed.
19064 * configure: Regenerate.
19065 * doc/avr-mmcu.texi: Regenerate.
19066 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
19067 <__AVR_ARCH__>: Document avrxmega3 and 103.
19068 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
19069 <__AVR_SHORT_CALLS__>: Document it.
19070 <__AVR_PM_BASE_ADDRESS__>: Document it.
19071 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
19072 (AVR Variable Attributes) <progmem>: Document this is
19073 not needed for avrxmega3.
19074 (AVR Named Address Spaces) <__flash>: Dito.
19076 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
19078 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
19080 2017-06-12 Doug Rupp <rupp@adacore.com>
19082 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
19083 Append vxworks-stdint.h to the tm_file list.
19084 * config/vxworks-stdint.h: New file.
19086 2017-06-12 Martin Liska <mliska@suse.cz>
19088 PR tree-optimization/81041
19089 * tree-profile.c (gimple_gen_ic_func_profiler):
19090 Create an extra BB in profile-generate
19091 (gimple_gen_time_profiler): Likewise.
19093 2017-06-12 Jakub Jelinek <jakub@redhat.com>
19095 PR tree-optimization/81003
19096 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
19097 (update_range_test): Use it instead of force_gimple_operand_gsi.
19099 2017-06-12 Richard Biener <rguenther@suse.de>
19101 PR tree-optimization/81053
19102 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
19103 with backedge value not defined in loop. Simplify def stmt
19106 2017-06-11 Tom de Vries <tom@codesourcery.com>
19109 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
19111 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
19112 nvptx_cannot_force_const_mem.
19114 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19116 * opts.c (finish_options): Move test for flag_split_stack after
19117 it has been initialized.
19119 2017-06-11 Jason Merrill <jason@redhat.com>
19121 * tree.h (id_equal): New.
19122 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
19123 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
19124 instead of strcmp of IDENTIFIER_POINTER.
19126 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19128 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
19129 (mark_all_inlined_calls_cdtor): Fix formating.
19130 (inline_transform): Rescale profile before inlining.
19132 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19134 * cgraph.h (cgraph_edge::clone): Update prototype.
19135 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
19136 (cgraph_node::create_clone): Update.
19137 (cgraph_node::create_version_clone): Update.
19138 * tree-inline.c (copy_bb): Update.
19139 (expand_call_inline): Update.
19141 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19143 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
19144 factored out from ...
19145 (rs6000_emit_prologue): ... here.
19147 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
19149 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
19150 factored out from ...
19151 (rs6000_emit_prologue): ... here.
19153 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19155 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
19157 (handle_missing_profiles): Fix computation of tp_first_run.
19158 (counts_to_freqs): Do not touch freqs when count is 0.
19160 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
19162 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
19165 2017-06-10 Tom de Vries <tom@codesourcery.com>
19167 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
19168 attributes): Document signal effective target.
19170 2017-06-10 Tom de Vries <tom@codesourcery.com>
19172 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
19173 Document effective target stack_size.
19175 2017-06-09 David Malcolm <dmalcolm@redhat.com>
19177 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
19178 to the edit_context if they can be auto-applied.
19180 2017-06-9 Ian Lance Taylor <iant@golang.org>
19182 * opts.c (finish_options): If -fsplit-stack, disable implicit
19183 -forder-blocks-and-partition.
19184 * doc/invoke.texi (Optimize Options): Document that when using
19185 -fsplit-stack -forder-blocks-and-partition is not implicitly
19188 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19190 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
19191 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
19192 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
19193 * builtins.def (abort, trap, unreachable): Declare cold.
19194 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
19195 * tree-core.h (ECF_COLD): New.
19196 * tree.c (set_call_expr_flags): Handle ECF_COLD.
19197 (build_common_builtin_nodes): Mark unreachable and abort as cold.
19199 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19201 * predict.c (unlikely_executed_stmt_p): Cleanup.
19203 2017-06-09 Richard Biener <rguenther@suse.de>
19205 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
19206 model if the ref is always written to.
19208 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19210 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
19212 2017-06-09 Tamar Christina <tamar.christina@arm.com>
19214 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
19217 2017-06-09 Tom de Vries <tom@codesourcery.com>
19220 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
19221 "target cannot support label values" when encountering LABEL_REF.
19223 2017-06-09 Martin Liska <mliska@suse.cz>
19225 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
19226 (gimple_gen_ic_func_profiler): Emit direct comparison
19227 of __gcov_indirect_call_callee with NULL.
19228 (gimple_gen_time_profiler): Change probability from
19229 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
19231 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19233 * profile.c (edge_gcov_counts): Turn to pointer.
19234 (compute_branch_probabilities, compute_branch_probabilities): Update.
19235 (branch_prob): Do not clear edge_gcov_count.
19236 * profile.h (edge_gcov_counts): Turn to pointer.
19237 (edge_gcov_count): Update.
19239 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
19241 * gimple.h (gimple_check_failed): Mark cold.
19243 2017-06-09 Richard Biener <rguenther@suse.de>
19245 PR tree-optimization/66623
19246 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
19247 refactor check_reduction into two parts, properly computing
19248 whether we have to check reduction validity for outer loop
19251 2017-06-09 Richard Biener <rguenther@suse.de>
19253 PR tree-optimization/79483
19254 * graphite-scop-detection.c (order): New global.
19255 (get_order): Compute bb to order mapping that satisfies code
19256 generation constraints.
19257 (cmp_pbbs): New helper.
19258 (build_scops): Start domwalk at entry block, sort generated
19261 2017-06-09 Richard Biener <rguenther@suse.de>
19263 PR middle-end/81007
19264 * ipa-polymorphic-call.c
19265 (ipa_polymorphic_call_context::restrict_to_inner_class):
19266 Skip FIELD_DECLs with error_mark_node type.
19267 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
19270 2017-06-09 Martin Liska <mliska@suse.cz>
19272 * predict.c (struct branch_predictor): New struct.
19273 (test_prediction_value_range): New test.
19274 (predict_c_tests): New function.
19275 * selftest-run-tests.c (selftest::run_tests): Run the function.
19276 * selftest.h: Declare new tests.
19278 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
19281 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
19282 gen_add3_insn did not fail.
19283 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
19284 r0, construct that number in a temporary reg and add that reg to r0.
19285 If asked to put the result in r0 as well, fail.
19287 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
19289 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
19290 for early expansion of vec_eqv.
19292 2017-06-08 Jakub Jelinek <jakub@redhat.com>
19294 PR middle-end/81005
19295 * ubsan.c (instrument_null): Avoid pointless code temporary.
19296 (pass_ubsan::execute): Instrument aggregate arguments of calls.
19298 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
19302 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
19305 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
19308 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19310 * predict.c (unlikely_executed_edge_p): Move ahead.
19311 (probably_never_executed_edge_p): Use it.
19313 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19315 PR middle-end/79988
19316 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
19317 gimple_call_builtin_p call.
19319 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19321 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
19322 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
19323 rtl_check_failed_type2, rtl_check_failed_code1,
19324 rtl_check_failed_code2, rtl_check_failed_code_mode,
19325 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
19326 rtvec_check_failed_bounds, rtl_check_failed_flag,
19327 _fatal_insn_not_found, _fatal_insn): Likewise.
19328 * tree.h (tree_contains_struct_check_failed,
19329 tree_check_failed, tree_not_check_failed,
19330 tree_class_check_failed, tree_range_check_failed,
19331 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
19332 tree_vec_elt_check_failed, phi_node_elt_check_failed,
19333 tree_operand_check_failed, omp_clause_check_failed,
19334 omp_clause_operand_check_failed, omp_clause_range_check_failed):
19337 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19339 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
19340 flag_branch_probabilities.
19341 * ipa-inline.c (edge_badness): Likewise.
19342 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
19343 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
19344 * predict.c (maybe_hot_frequency_p): Likewise.
19345 (probably_never_executed): Likewise.
19346 * sched-ebb.c (schedule_ebbs): Likewise.
19347 * sched-rgn.c (find_single_block_region): Likewise.
19348 * tracer.c (tail_duplicate): Likewise.
19350 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19352 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
19353 longer requires x_flag_profile_use.
19355 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19357 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
19358 instead of flag_reorder_blocks_and_partition.
19359 * dbxout.c (dbxout_function_end): Likewise.
19360 * dwarf2out.c (gen_subprogram_die): Likewise.
19361 * haifa-sched.c (sched_create_recovery_edges): Likewise.
19362 * hw-doloop.c (reorg_loops): Likewise.
19363 * varasm.c (assemble_start_function,
19364 assemble_end_function): Likewise.
19365 (decide_function_section): Do not check for
19366 flag_reorder_blocks_and_partition.
19368 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
19370 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
19372 (chkp_get_hard_register_fake_addr_expr): Ditto.
19373 (chkp_build_addr_expr): Add check for hard reg case.
19374 (chkp_parse_array_and_component_ref): Ditto.
19375 (chkp_find_bounds_1): Ditto.
19376 (chkp_process_stmt): Don't generate bounds store for
19379 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19381 * predict.c (maybe_hot_bb_p): Do not check profile status.
19382 (maybe_hot_edge_p): Likewise.
19383 (probably_never_executed): Check for zero counts even if profile
19385 (unlikely_executed_edge_p): New function.
19386 (unlikely_executed_stmt_p): New function.
19387 (unlikely_executed_bb_p): New function.
19388 (set_even_probabilities): Use unlikely predicates.
19389 (combine_predictions_for_bb): Likewise.
19390 (predict_paths_for_bb): Likewise.
19391 (predict_paths_leading_to_edge): Likewise.
19392 (determine_unlikely_bbs): New function.
19393 (estimate_bb_frequencies): Use it.
19394 (compute_function_frequency): Use zero counts even if profile is
19396 * profile-count.h: Fix typo.
19398 2017-08-08 Julia Koval <julia.koval@intel.com>
19400 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
19401 _mm512_mask_cvtsepi16_storeu_epi8,
19402 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
19403 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
19404 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
19405 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
19406 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
19407 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
19408 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
19409 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
19410 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
19411 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
19412 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
19413 __builtin_ia32_pmovuswb256mem_mask,
19414 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
19415 __builtin_ia32_pmovwb512mem_mask): New builtins.
19417 2017-08-08 Julia Koval <julia.koval@intel.com>
19419 PR target/73350,80862
19420 * config/i386/subst.md (round): Fix round pattern.
19421 * config/i386/i386.c (ix86_erase_embedded_rounding):
19422 Fix erasing rounding for the fixed pattern.
19424 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
19426 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
19428 2017-06-08 Martin Liska <mliska@suse.cz>
19430 PR gcov-profile/80911
19431 * gcov.c (block_info::block_info): New constructor.
19433 2017-06-07 Carl Love <cel@us.ibm.com>
19435 * config/rs6000/rs6000-c: The return type of the following
19436 built-in functions was implemented as int not long long. Fix sign
19437 of return value for the unsigned version of vec_mulo and vec_mule.
19438 vector unsigned long long vec_bperm (vector unsigned long long,
19439 vector unsigned char)
19440 vector signed long long vec_mule (vector signed int,
19442 vector unsigned long long vec_mule (vector unsigned int,
19443 vector unsigned int)
19444 vector signed long long vec_mulo (vector signed int,
19446 vector unsigned long long vec_mulo (vector unsigned int,
19447 vector unsigned int)
19448 * doc/extend.texi: Fix the documentation for the built-in
19451 2017-06-07 Carl Love <cel@us.ibm.com>
19454 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
19457 2017-06-07 Carl Love <cel@us.ibm.com>
19459 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
19460 support, Generate doublehv for signed int/float for BE case only.
19462 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
19464 * doc/invoke.texi (mcx16): Rewrite.
19466 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19468 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
19469 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
19470 *mov<mode>_softfloat, and an anonymous splitter): Use
19471 nonimmediate_operand instead of rs6000_nonimmediate_operand.
19473 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19475 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
19477 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19478 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
19479 (rs6000_debug_reg_global): Adjust.
19480 (rs6000_init_hard_regno_mode_ok): Adjust.
19481 (rs6000_dbx_register_number): Adjust.
19482 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
19483 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19484 Remove SPE_ACC and SPEFSCR.
19485 (REG_ALLOC_ORDER): Ditto.
19486 (FRAME_POINTER_REGNUM): Change to 111.
19487 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
19488 (REG_CLASS_NAMES): Ditto.
19489 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
19490 (REGISTER_NAMES): Ditto.
19491 (ADDITIONAL_REG_NAMES): Ditto.
19492 (rs6000_reg_names): Ditto.
19493 * config/rs6000/rs6000.md: Renumber some register number
19496 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19498 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
19500 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
19501 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
19503 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
19504 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
19505 Delete the SPE high registers.
19506 (REG_ALLOC_ORDER): Ditto.
19507 (enum reg_class): Remove SPE_HIGH_REGS.
19508 (REG_CLASS_NAMES): Ditto.
19509 (REG_CLASS_CONTENTS): Delete the SPE high registers.
19510 (REGISTER_NAMES): Ditto.
19511 (rs6000_reg_names): Ditto.
19512 * doc/tm.texi.in: Remove SPE as example.
19513 * doc/tm.texi: Regenerate.
19515 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19517 * config/rs6000/8540.md (ppc8540_brinc): Delete.
19518 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
19519 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
19520 * config/rs6000/rs6000.md (type): Remove "brinc".
19522 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19524 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
19525 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
19526 * config/rs6000/linuxspe.h: Delete file.
19527 * config/rs6000/rs6000.md: Don't include spe.md.
19528 * config/rs6000/spe.h: Delete file.
19529 * config/rs6000/spe.md: Delete file.
19530 * config/rs6000/t-rs6000: Remove spe.md.
19532 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19534 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
19535 (reg_or_none500mem_operand): Delete.
19536 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
19537 instead of reg_or_none500mem_operand.
19539 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19541 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
19542 handling of SPE flags.
19543 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
19545 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19547 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
19549 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
19550 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
19551 paired_divv2sf3): Similar.
19552 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
19553 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
19554 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
19556 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
19557 Rename the paired_* instruction patterns.
19558 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
19560 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
19561 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
19562 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
19563 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
19564 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
19565 PAIRED_VECTOR_MODE.
19566 (struct machine_function): Delete field spe_insn_chain_scanned_p.
19567 (spe_func_has_64bit_regs_p): Delete.
19568 (spe_expand_predicate_builtin): Delete.
19569 (spe_expand_evsel_builtin): Delete.
19570 (TARGET_DWARF_REGISTER_SPAN): Do not define.
19571 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
19572 (invalid_e500_subreg): Delete.
19573 (rs6000_legitimize_address): Always force_reg op2 as well, for
19574 paired single memory accesses.
19575 (rs6000_member_type_forces_blk): Delete.
19576 (rs6000_spe_function_arg): Delete.
19577 (rs6000_expand_unop_builtin): Delete SPE handling.
19578 (rs6000_expand_binop_builtin): Ditto.
19579 (spe_expand_stv_builtin): Delete.
19580 (bdesc_2arg_spe): Delete.
19581 (spe_expand_builtin): Delete.
19582 (spe_expand_predicate_builtin): Delete.
19583 (spe_expand_evsel_builtin): Delete.
19584 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
19585 (spe_init_builtins): Delete.
19586 (spe_func_has_64bit_regs_p): Delete.
19587 (savres_routine_name): Delete "info" parameter. Adjust callers.
19588 (rs6000_emit_stack_reset): Ditto.
19589 (rs6000_dwarf_register_span): Delete.
19590 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
19591 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
19592 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
19594 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
19596 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
19597 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
19598 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
19599 mulv2sf3, divv2sf3): Delete expanders.
19601 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19603 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
19605 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19607 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
19608 * config/rs6000/rs6000.c: Ditto.
19610 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19612 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
19613 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
19614 comparison_operator.
19616 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19618 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
19619 * config/rs6000/rs6000.opt: Ditto.
19620 * config/rs6000/t-rtems: Ditto.
19622 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19624 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
19625 TARGET_E500_SINGLE by 0, simplify.
19626 * config/rs6000/rs6000.c: Ditto.
19627 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
19628 (spe_build_register_parallel): Delete.
19629 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
19630 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
19631 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
19632 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
19633 (E500_CONVERT): Delete.
19634 * config/rs6000/spe.md: Remove many patterns and all define_constants.
19636 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
19638 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
19639 * config/rs6000/dfp.md: Ditto.
19640 (negdd2, *negdd2_fpr): Merge.
19641 (absdd2, *absdd2_fpr): Merge.
19642 (negtd2, *negtd2_fpr): Merge.
19643 (abstd2, *abstd2_fpr): Merge.
19644 * config/rs6000/e500.h: Delete file.
19645 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
19646 TARGET_FPRS by 1 and simplify.
19647 * config/rs6000/rs6000-c.c: Ditto.
19648 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
19649 TARGET_DF_SPE by 0.
19650 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
19652 * config/rs6000/rs6000.md: Ditto.
19653 (floatdidf2, *floatdidf2_fpr): Merge.
19654 (move_from_CR_gt_bit): Delete.
19655 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
19656 (E500_CR_IOR_COMPARE): Delete.
19657 (All patterns that require !TARGET_FPRS): Delete.
19658 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
19660 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19662 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
19664 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19666 * graphds.c (add_edge): Intitialize edge's attached data.
19667 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
19668 pointer parameter. Call pointed function on each edge during
19669 graph traversing. Skip traversing the edge when the function
19671 (graphds_dfs, graphds_scc): Ditto.
19672 (for_each_edge): New parameter. Pass the new parameter to callback
19674 * graphds.h (skip_edge_callback): New function pointer type.
19675 (graphds_dfs, graphds_scc): New function pointer parameter.
19676 (graphds_edge_callback, for_each_edge): New parameter.
19678 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19680 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
19681 out code checking if runtime alias check is possible to below ...
19682 Call the new function.
19683 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
19684 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
19686 2017-06-07 Marek Polacek <polacek@redhat.com>
19689 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
19690 TYPE_OVERFLOW_WRAPS check.
19692 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19694 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
19695 if versioning is required.
19696 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
19697 peeling with the check for versioning.
19699 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19701 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
19702 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
19703 Set true to new parameter if new ssa variable is defined.
19704 (vect_gen_vector_loop_niters): Refactor. Set range information
19705 for the new vector loop bound variable.
19706 (vect_do_peeling): Ditto.
19708 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19710 * tree-affine.c (ssa.h): Include header file.
19711 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
19712 has wrapping overflow behavior.
19714 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19716 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
19718 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19720 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
19721 (tree_to_aff_combination): ... here.
19723 2017-06-07 Bin Cheng <bin.cheng@arm.com>
19725 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
19726 reg_pressure model function.
19727 (ivopts_global_cost_for_size): Delete.
19728 (determine_set_costs, iv_ca_recount_cost): Call new model function
19729 ivopts_estimate_reg_pressure.
19731 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19733 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
19734 expensive than udiv. Remove floating point cases from mod.
19736 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19738 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
19739 Increase idiv cost.
19741 2017-06-07 Tamar Christina <tamar.christina@arm.com>
19743 * config/aarch64/aarch64.md
19744 (copysignsf3): Fix mask generation.
19746 2017-06-07 Jakub Jelinek <jakub@redhat.com>
19748 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
19750 (class dump_manager): Add register_dumps method.
19751 * dumpfile.c: Include langhooks.h.
19752 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
19753 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
19754 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
19755 (dump_manager::dump_register): Start with 512 entries instead of 32.
19756 (dump_manager::register_dumps): New method.
19757 * toplev.c (general_init): Instead of invoking register_dumps
19758 langhook, invoke register_dumps method on the dump manager.
19759 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
19762 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
19764 * doc/md.texi: Clarify the restrictions on a define_insn condition.
19765 Say that # requires an associated define_split to exist, and that
19766 the define_split must be suitable for use after register allocation.
19768 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19770 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
19771 (compute_outgoing_frequencies): Also initialize zero counts.
19772 (find_many_sub_basic_blocks): Do not produce uninitialized profile
19773 around loops; preserve more of profile when nothing changes.
19775 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
19777 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
19779 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
19780 * config/arm/arm-cpu-cdata.h: Regenerate.
19781 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
19782 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
19783 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
19784 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
19785 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
19787 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
19789 * config/arm/t-rmprofile: Likewise.
19790 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
19792 2017-06-06 David S. Miller <davem@davemloft.net>
19795 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
19796 blockage if function uses alloca.
19798 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19800 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
19801 New "uid" fields to hold pretty-print IDs of group and ref.
19802 Memory references are now identified as <group_id>:<ref_id>
19803 instead of using [random] addresses.
19804 (dump_mem_details): Simplify, no functional change.
19805 (dump_mem_ref): Simplify and make output more concise.
19806 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
19807 (find_or_create_group): Initialize group uid.
19808 (record_ref): Initialize ref uid. Improve debug output.
19809 (prune_group_by_reuse, should_issue_prefetch_p,)
19810 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
19811 (mark_nontemporal_store, determine_loop_nest_reuse):
19812 Improve debug output.
19814 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
19816 * dbgcnt.def (prefetch): New debug counter.
19817 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
19818 (schedule_prefetches): Stop issueing prefetches if debug counter
19821 2017-06-06 Tom de Vries <tom@codesourcery.com>
19823 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
19824 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
19826 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19828 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
19829 Use aarch64_reg_or_zero predicate for operand 4.
19830 (aarch64_compare_and_swap<mode> define_insn_and_split):
19831 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
19832 (aarch64_store_exclusive<mode>): Likewise for operand 2.
19834 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
19836 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
19837 (arm_compute_save_core_reg_mask): This.
19838 (thumb1_compute_save_reg_mask): Rename into ...
19839 (thumb1_compute_save_core_reg_mask): This.
19840 (arm_compute_save_reg0_reg12_mask): Adapt comment.
19841 (arm_compute_frame_layout): Likewise.
19843 2017-06-06 Richard Biener <rguenther@suse.de>
19845 PR tree-optimization/80974
19846 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
19847 keep or clear leaders SSA info.
19849 2017-06-06 Tom de Vries <tom@codesourcery.com>
19851 * config/nvptx/nvptx.c (split_mode_p): New function.
19852 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
19854 2017-06-06 Tom de Vries <tom@codesourcery.com>
19856 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
19858 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19861 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
19864 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19866 * shrink-wrap.c (handle_simple_exit): Update profile.
19867 (try_shrink_wrapping): Upate profile.
19869 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19871 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
19872 (tree_guess_outgoing_edge_probabilities): New.
19873 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
19874 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
19876 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19878 * ipa-split.c (split_function): Initialize return bb profile.
19880 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
19882 * profile.c (compute_branch_probabilities): Also initialize
19883 EXIT_BLOCK profile.
19885 2017-06-06 Richard Biener <rguenther@suse.de>
19887 PR tree-optimization/80928
19888 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
19889 (vect_analyze_loop_operations): Properly guard analysis for
19891 (vect_transform_loop): Likewise.
19892 (vect_analyze_loop_2): Also reset SLP type on PHIs.
19893 (vect_model_induction_cost): Do not cost for pure SLP.
19894 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
19895 of induction in inner loop vectorization.
19896 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
19897 (vect_get_and_check_slp_defs): Handle vect_induction_def.
19898 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
19900 (vect_analyze_slp_cost_1): Cost induction.
19901 (vect_detect_hybrid_slp_stmts): Handle PHIs.
19902 (vect_get_slp_vect_defs): Likewise.
19903 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
19904 (vect_transform_stmt): Handle SLP reductions.
19905 * tree-vectorizer.h (vectorizable_induction): Adjust.
19907 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19909 * config/rs6000/rs6000.c (make_resolver_func): Update
19910 init_lowered_empty_function call.
19912 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
19914 * doc/invoke.texi: Document the -fprofile-abs-path option.
19915 * common.opt (fprofile-abs-path): New option.
19916 * gcov-io.h (gcov_write_filename): Declare.
19917 * gcov-io.c (gcov_write_filename): New function.
19918 * coverage.c (coverage_begin_function): Use gcov_write_filename.
19919 * profile.c (output_location): Likewise.
19921 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19923 * shring-wrap.c: Revert accidental commit.
19925 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
19927 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
19929 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19931 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
19933 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
19934 profile in callgraph edge.
19935 * profile-count.h (apply_probability): If THIS is 0, then result is 0
19936 (apply_scale): Likewise.
19937 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
19938 Also scale profile when inlining function with zero profile.
19939 (initialize_cfun): Update exit block profile even when it is zero.
19940 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
19941 when profile is read.
19943 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
19945 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
19946 (CLONE_*): New constants to define the processors we can generate
19947 code for with the target_clone attribute.
19948 (rs6000_clone_map): New array to identify which clone processors
19949 the current program is running on.
19950 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
19951 target_clone attribute.
19952 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
19953 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
19954 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
19955 (cpu_expand_builtin): Add support for target_clone attribute.
19956 (rs6000_valid_attribute_p): Allow "default" attribute.
19957 (get_decl_name): New debug function to simplify printing the
19958 current function name in debugging statements.
19959 (rs6000_clone_priority): New functions to support the target_clone
19960 attribute, and be able to generate code to switch between ISA 2.05
19961 through ISA 3.0 (power6 through power9).
19962 (rs6000_compare_version_priority): Likewise.
19963 (rs6000_get_function_versions_dispatcher): Likewise.
19964 (make_resolver_func): Likewise.
19965 (add_condition_to_bb): Likewise.
19966 (dispatch_function_versions): Likewise.
19967 (rs6000_generate_version_dispatcher_body): Likewise.
19968 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
19969 (fusion_gpr_load_p): Fix a spacing issue.
19970 * doc/extend.texi (Common Function Attributes): Document that the
19971 PowerPC supports the target_clone attribute.
19973 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
19975 * config/arm/arm.h: explain F symbol found in description of ARM
19976 register allocation in its legend.
19978 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19980 * config/mips/frame-header-opt.c: Include profile-count.h.
19981 * config/riscv/riscv.c: Include profile-count.h
19983 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
19985 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
19987 (sm_set_flag_if_changed): Add bbs field.
19988 (execute_sm_if_changed_flag_set): Pass BBS.
19989 (execute_sm): Update.
19991 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19993 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
19996 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
19998 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
19999 (peephole2): New peephole2 to emit the above.
20000 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
20002 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20004 * config/aarch64/aarch64.c (define_peephole2 above
20005 *sub_<shift>_<mode>): New peephole.
20007 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20009 * config/i386/i386.c (make_resolver_func): Update.
20010 * Makefile.in: Add profile-count.h and profile-count.o
20011 * auto-profile.c (afdo_indirect_call): Update to new API.
20012 (afdo_set_bb_count): Update.
20013 (afdo_propagate_edge): Update.
20014 (afdo_propagate_circuit): Update.
20015 (afdo_calculate_branch_prob): Update.
20016 (afdo_annotate_cfg): Update.
20017 * basic-block.h: Include profile-count.h
20018 (struct edge_def): Turn count to profile_count.
20019 (struct basic_block_def): Likewie.
20020 (REG_BR_PROB_BASE): Move to profile-count.h
20021 (RDIV): Move to profile-count.h
20022 * bb-reorder.c (max_entry_count): Turn to profile_count.
20023 (find_traces): Update.
20024 (rotate_loop):Update.
20025 (connect_traces):Update.
20026 (sanitize_hot_paths):Update.
20027 * bt-load.c (migrate_btr_defs): Update.
20028 * cfg.c (RDIV): Remove.
20029 (init_flow): Use alloc_block.
20030 (alloc_block): Uninitialize count.
20031 (unchecked_make_edge): Uninitialize count.
20032 (check_bb_profile): Update.
20033 (dump_edge_info): Update.
20034 (dump_bb_info): Update.
20035 (update_bb_profile_for_threading): Update.
20036 (scale_bbs_frequencies_int): Update.
20037 (scale_bbs_frequencies_gcov_type): Update.
20038 (scale_bbs_frequencies_profile_count): New.
20039 * cfg.h (update_bb_profile_for_threading): Update.
20040 (scale_bbs_frequencies_profile_count): Declare.
20041 * cfgbuild.c (compute_outgoing_frequencies): Update.
20042 (find_many_sub_basic_blocks): Update.
20043 * cfgcleanup.c (try_forward_edges): Update.
20044 (try_crossjump_to_edge): Update.
20045 * cfgexpand.c (expand_gimple_tailcall): Update.
20046 (construct_exit_block): Update.
20047 * cfghooks.c (verify_flow_info): Update.
20048 (dump_bb_for_graph): Update.
20049 (split_edge): Update.
20050 (make_forwarder_block): Update.
20051 (duplicate_block): Update.
20052 (account_profile_record): Update.
20053 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
20054 (get_estimated_loop_iterations): Update.
20055 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
20056 (single_likely_exit): Update.
20057 * cfgloopmanip.c (scale_loop_profile): Update.
20059 (set_zero_probability): Update.
20060 (lv_adjust_loop_entry_edge): Update.
20061 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
20062 (purge_dead_edges): Update.
20063 (rtl_account_profile_record): Update.
20064 * cgraph.c (cgraph_node::create): Uninitialize count.
20065 (symbol_table::create_edge): Uninitialize count.
20066 (cgraph_update_edges_for_call_stmt_node): Update.
20067 (cgraph_edge::dump_edge_flags): Update.
20068 (cgraph_node::dump): Update.
20069 (cgraph_edge::maybe_hot_p): Update.
20070 * cgraph.h: Include profile-count.h
20071 (create_clone), create_edge, create_indirect_edge): Update.
20072 (cgraph_node): Turn count to profile_count.
20073 (cgraph_edge0: Likewise.
20074 (make_speculative, clone): Update.
20075 (create_edge): Update.
20076 (init_lowered_empty_function): Update.
20077 * cgraphclones.c (cgraph_edge::clone): Update.
20078 (duplicate_thunk_for_node): Update.
20079 (cgraph_node::create_clone): Update.
20080 * cgraphunit.c (cgraph_node::analyze): Update.
20081 (cgraph_node::expand_thunk): Update.
20082 * final.c (dump_basic_block_info): Update.
20083 * gimple-streamer-in.c (input_bb): Update.
20084 * gimple-streamer-out.c (output_bb): Update.
20085 * graphite.c (print_global_statistics): Update.
20086 (print_graphite_scop_statistics): Update.
20087 * hsa-brig.c: Include basic-block.h.
20088 * hsa-dump.c: Include basic-block.h.
20089 * hsa-gen.c (T sum_slice): Update.
20090 (convert_switch_statements):Update.
20091 * hsa-regalloc.c: Include basic-block.h.
20092 * ipa-chkp.c (chkp_produce_thunks): Update.
20093 * ipa-cp.c (struct caller_statistics): Update.
20094 (init_caller_stats): Update.
20095 (gather_caller_stats): Update.
20096 (ipcp_cloning_candidate_p): Update.
20097 (good_cloning_opportunity_p): Update.
20098 (get_info_about_necessary_edges): Update.
20099 (dump_profile_updates): Update.
20100 (update_profiling_info): Update.
20101 (update_specialized_profile): Update.
20102 (perhaps_add_new_callers): Update.
20103 (decide_about_value): Update.
20104 (ipa_cp_c_finalize): Update.
20105 * ipa-devirt.c (struct odr_type_warn_count): Update.
20106 (struct decl_warn_count): Update.
20107 (struct final_warning_record): Update.
20108 (possible_polymorphic_call_targets): Update.
20109 (ipa_devirt): Update.
20110 * ipa-fnsummary.c (redirect_to_unreachable): Update.
20111 * ipa-icf.c (sem_function::merge): Update.
20112 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
20113 * ipa-inline.c (compute_uninlined_call_time): Update.
20114 (compute_inlined_call_time): Update.
20115 (want_inline_small_function_p): Update.
20116 (want_inline_self_recursive_call_p): Update.
20117 (edge_badness): Update.
20118 (lookup_recursive_calls): Update.
20119 (recursive_inlining): Update.
20120 (inline_small_functions): Update.
20121 (dump_overall_stats): Update.
20122 (dump_inline_stats): Update.
20123 * ipa-profile.c (ipa_profile_generate_summary): Update.
20124 (ipa_propagate_frequency): Update.
20125 (ipa_profile): Update.
20126 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
20127 * ipa-utils.c (ipa_merge_profiles): Update.
20128 * loop-doloop.c (doloop_modify): Update.
20129 * loop-unroll.c (report_unroll): Update.
20130 (unroll_loop_runtime_iterations): Update.
20131 * lto-cgraph.c (lto_output_edge): Update.
20132 (lto_output_node): Update.
20133 (input_node): Update.
20134 (input_edge): Update.
20135 (merge_profile_summaries): Update.
20136 * lto-streamer-in.c (input_cfg): Update.
20137 * lto-streamer-out.c (output_cfg): Update.
20138 * mcf.c (create_fixup_graph): Update.
20139 (adjust_cfg_counts): Update.
20140 (sum_edge_counts): Update.
20141 * modulo-sched.c (sms_schedule): Update.
20142 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
20143 * predict.c (maybe_hot_count_p): Update.
20144 (probably_never_executed): Update.
20145 (dump_prediction): Update.
20146 (combine_predictions_for_bb): Update.
20147 (propagate_freq): Update.
20148 (handle_missing_profiles): Update.
20149 (counts_to_freqs): Update.
20150 (rebuild_frequencies): Update.
20151 (force_edge_cold): Update.
20152 * predict.h: Include profile-count.h
20153 (maybe_hot_count_p, counts_to_freqs): UPdate.
20154 * print-rtl-function.c: Do not include cfg.h
20155 * print-rtl.c: Include basic-block.h
20156 * profile-count.c: New file.
20157 * profile-count.h: New file.
20158 * profile.c (is_edge_inconsistent): Update.
20159 (correct_negative_edge_counts): Update.
20160 (is_inconsistent): Update.
20161 (set_bb_counts): Update.
20162 (read_profile_edge_counts): Update.
20163 (compute_frequency_overlap): Update.
20164 (compute_branch_probabilities): Update; Initialize and deinitialize
20166 (branch_prob): Update.
20167 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
20168 (edge_gcov_count): New.
20169 (bb_gcov_count): New.
20170 * shrink-wrap.c (try_shrink_wrapping): Update.
20171 * tracer.c (better_p): Update.
20172 * trans-mem.c (expand_transaction): Update.
20173 (ipa_tm_insert_irr_call): Update.
20174 (ipa_tm_insert_gettmclone_call): Update.
20175 * tree-call-cdce.c: Update.
20176 * tree-cfg.c (gimple_duplicate_sese_region): Update.
20177 (gimple_duplicate_sese_tail): Update.
20178 (gimple_account_profile_record): Update.
20179 (execute_fixup_cfg): Update.
20180 * tree-inline.c (copy_bb): Update.
20181 (copy_edges_for_bb): Update.
20182 (initialize_cfun): Update.
20183 (freqs_to_counts): Update.
20184 (copy_cfg_body): Update.
20185 (expand_call_inline): Update.
20186 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
20187 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
20188 (try_unroll_loop_completely): Update.
20189 (try_peel_loop): Update.
20190 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
20191 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
20192 * tree-ssa-loop-split.c (connect_loops): Update.
20193 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
20194 * tree-ssa-reassoc.c (branch_fixup): Update.
20195 * tree-ssa-tail-merge.c (replace_block_by): Update.
20196 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
20197 (compute_path_counts): Update.
20198 (update_profile): Update.
20199 (recompute_probabilities): Update.
20200 (update_joiner_offpath_counts): Update.
20201 (estimated_freqs_path): Update.
20202 (freqs_to_counts_path): Update.
20203 (clear_counts_path): Update.
20204 (ssa_fix_duplicate_block_edges): Update.
20205 (duplicate_thread_path): Update.
20206 * tree-switch-conversion.c (case_bit_test_cmp): Update.
20207 (struct switch_conv_info): Update.
20208 * tree-tailcall.c (decrease_profile): Update.
20209 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
20210 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
20211 * value-prof.c (check_counter): Update.
20212 (gimple_divmod_fixed_value): Update.
20213 (gimple_mod_pow2): Update.
20214 (gimple_mod_subtract): Update.
20215 (gimple_ic_transform): Update.
20216 (gimple_stringop_fixed_value): Update.
20217 * value-prof.h (gimple_ic): Update.
20219 2017-06-02 Carl Love <cel@us.ibm.com>
20221 * config/rs6000/rs6000-c: Add support for built-in functions
20222 vector double vec_doublee (vector signed int);
20223 vector double vec_doublee (vector unsigned int);
20224 vector double vec_doublee (vector float);
20225 vector double vec_doubleh (vector signed int);
20226 vector double vec_doubleh (vector unsigned int);
20227 vector double vec_doubleh (vector float);
20228 vector double vec_doublel (vector signed int);
20229 vector double vec_doublel (vector unsigned int);
20230 vector double vec_doublel (vector float);
20231 vector double vec_doubleo (vector signed int);
20232 vector double vec_doubleo (vector unsigned int);
20233 vector double vec_doubleo (vector float);.
20234 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
20235 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
20237 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
20238 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
20239 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
20241 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
20242 vec_doublel, vec_doubleh.
20243 * doc/extend.texi: Update the built-in documentation file for the
20244 new built-in functions.
20246 2017-06-02 David Malcolm <dmalcolm@redhat.com>
20249 * ipa-inline-analysis.c (free_growth_caches): Set
20250 edge_removal_hook_holder to NULL after removing it.
20252 2017-06-02 Sudakshina Das <sudi.das@arm.com>
20254 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
20255 comparision with zero.
20257 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20258 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20259 for early expansion of vec_min and vec_max builtins.
20260 (builtin_function_type): Add min/max unsigned variants to those
20261 identified as having unsigned arguments.
20263 2017-06-02 Olivier Hainque <hainque@adacore.com>
20265 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
20267 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20269 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
20270 Use VALL_F16 iterator rather than VALL.
20272 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20274 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
20275 Emit CBNZ inside loop when doing a strong exchange and comparing
20276 against zero. Generate the CC flags after the loop.
20278 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
20280 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
20281 (dl_section_ref): New.
20282 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
20283 On AIX, append an expression to subtract the size of the
20284 section length to dl_section_ref.
20286 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
20288 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
20289 for early expansion of vector absolute builtins.
20291 2017-06-02 Richard Biener <rguenther@suse.de>
20293 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
20294 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
20296 2017-06-02 Richard Biener <rguenther@suse.de>
20298 PR tree-optimization/80948
20299 * tree-tailcall.c (find_tail_calls): Track stmts to move in
20300 stmt order as well.
20302 2017-06-02 Richard Biener <rguenther@suse.de>
20304 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
20306 * tree-vect-stmts.c (process_use): Do not mark backedge defs
20307 for inductions as relevant.
20309 2017-06-02 Richard Biener <rguenther@suse.de>
20311 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
20312 (vectorizable_induction): ... this. Remove dead code.
20314 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
20316 * builtins. (expand_builtin_alloca): Remove second parameter and
20317 infer its value from the first parameter instead.
20318 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
20320 2017-06-02 Jakub Jelinek <jakub@redhat.com>
20322 PR rtl-optimization/80903
20323 * loop-doloop.c (add_test): Unshare sequence.
20325 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
20327 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
20329 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
20331 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
20333 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
20334 xlogue_layout::get_instance, logue_layout::xlogue_layout,
20335 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
20336 (xlogue_layout::get_stub_rtx): Make static.
20337 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
20338 (xlogue_layout::compute_stub_managed_regs): Rename to...
20339 (xlogue_layout::count_stub_managed_regs): ...this.
20340 (xlogue_layout::is_stub_managed_reg): New function.
20341 (xlogue_layout::m_stub_names): Rename to...
20342 (xlogue_layout::s_stub_names): ...this, make static.
20343 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
20344 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
20345 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
20346 xlogue_layout::s_stub_names): Instantiate statics.
20347 (stub_managed_regs): Remove.
20348 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
20349 (disable_call_ms2sysv_xlogues): Rename to...
20350 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
20351 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
20353 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
20354 change after reload_completed.
20355 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
20357 (ix86_expand_prologue): Likewise.
20358 (ix86_expand_epilogue): Likewise.
20359 (ix86_expand_split_stack_prologue): Likewise.
20360 (ix86_compute_frame_layout): Remove frame parameter ...
20361 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
20362 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
20364 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
20365 (ix86_frame): Move from here ...
20366 * config/i386/i386.h (ix86_frame): ... to here.
20367 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
20368 complete ix86_frame data structure instead. Remove some_ld_name.
20370 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20372 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
20373 symbols that hold a DECL_VALUE_EXPR.
20375 2017-06-01 Martin Jambor <mjambor@suse.cz>
20377 PR tree-optimization/80898
20378 * tree-sra.c (process_subtree_disqualification): Removed.
20379 (disqualify_candidate): Do not acll
20380 process_subtree_disqualification.
20381 (subtree_mark_written_and_enqueue): New function.
20382 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
20383 RHS has been disqualified and re-queue LHS if necessary. Apart
20384 from that, ignore disqualified RHS.
20386 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20388 * config/s390/s390.c (s390_emit_epilogue): Disable early return
20389 address fetch for z10 or later.
20391 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20393 * config/arc/arc.md (tst_movb): Add guard when splitting.
20395 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20397 * config/arc/arc.c (arc_can_eliminate): Test against
20398 arc_frame_pointer_needed.
20400 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20402 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
20403 to prevent store reordering.
20404 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
20405 (type): Add block type.
20406 (stack_tie): Define special instruction to be used in
20409 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20411 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
20412 constraint. It is not valid for the pattern.
20413 (noncommutative_binary_comparison): Likewise.
20415 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20417 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
20420 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20422 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
20423 be used by the reg-alloc.
20425 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20427 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
20428 reg-alloc when having mul64 or mul32x16 instructions.
20429 (mulsidi3): Likewise.
20430 (umulsidi3): Likewise.
20431 (mulsi32x16): New pattern.
20432 (mulsi64): Likewise.
20433 (mulsidi64): Likewise.
20434 (umulsidi64): Likewise.
20435 (MUL32x16_REG): Define.
20436 (mul64_600): Use MUL32x16_REG.
20437 (mac64_600): Likewise.
20438 (umul64_600): Likewise.
20439 (umac64_600): Likewise.
20441 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
20443 * config/arc/arc.md (mulsi3_700): Make it commutative.
20445 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
20447 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
20449 (*sign_extendsidi2_insn): Likewise for movstosw.
20451 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
20453 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
20454 the type of the input discriminant value. Convert the
20455 discriminant value of signedness vary.
20457 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
20459 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
20460 Add to -Wall section.
20462 2017-06-01 Richard Biener <rguenther@suse.de>
20464 PR middle-end/66313
20465 * fold-const.c (fold_plusminus_mult_expr): If the factored
20466 factor may be zero use a wrapping type for the inner operation.
20467 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
20468 and handle moved defs.
20469 (process_assignment): Properly guard the unary op case. Return a
20470 tri-state indicating that moving the stmt before the call may allow
20471 to continue. Pass through to_move.
20472 (find_tail_calls): Handle moving unrelated defs before
20475 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
20478 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
20479 splitter result in the canonical way.
20481 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
20483 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
20484 also for 32bit target. Update insn attributes.
20485 (zero-extendsidi2 splitter): Allow all registers for operand 1.
20487 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
20489 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
20490 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
20491 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
20492 (_mm_maskz_min_ss): New intrinsics.
20494 2017-05-31 Martin Liska <mliska@suse.cz>
20496 * tree-vect-loop.c (vect_create_epilog_for_reduction):
20497 Change comment style to one we normally use.
20498 (vectorizable_reduction): Likewise.
20499 (vectorizable_induction): Likewise.
20500 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
20501 (vectorizable_call): Likewise.
20502 (vectorizable_simd_clone_call): Likewise.
20503 (vectorizable_conversion): Likewise.
20504 (vectorizable_assignment): Likewise.
20505 (vectorizable_shift): Likewise.
20506 (vectorizable_operation): Likewise.
20507 (vectorizable_store): Likewise.
20508 (vectorizable_load): Likewise.
20509 * tree-vectorizer.h: Likewise.
20511 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
20513 * passes.c (emergency_dump_function): New.
20514 * tree-pass.h (emergency_dump_function): Declare.
20515 * plugin.c (plugins_internal_error_function): Remove.
20516 * plugin.h (plugins_internal_error_function): Remove declaration.
20517 * toplev.c (internal_error_function): New static function. Use it...
20518 (general_init): ...here.
20520 2017-05-31 Graham Markall <graham.markall@embecosm.com>
20522 * config/arc/arc.c (arc_print_operand): Handle constant operands.
20523 (arc_rtx_costs): Add costs for new patterns.
20524 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
20525 * config/arc/predicates.md: Add _1_2_3_operand predicate.
20527 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20529 * tree-ssa-strlen.c (get_next_strinfo): New function.
20530 (get_stridx_plus_constant): Use it.
20531 (zero_length_string): Likewise.
20532 (adjust_related_strinfos): Likewise.
20533 (adjust_last_stmt): Likewise.
20535 2017-05-31 Richard Biener <rguenther@suse.de>
20538 * config/i386/i386.c (ix86_expand_builtin): Remove assert
20539 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
20541 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
20543 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
20544 loop_vinfo argument and use of dependence distance vectors.
20545 Check instead whether the two references differ only in their
20546 initial value and assume that they have the same alignment if the
20547 difference is a multiple of the vector alignment.
20548 (vect_analyze_data_refs_alignment): Update call accordingly.
20550 2017-05-31 Martin Liska <mliska@suse.cz>
20553 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
20555 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20557 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
20558 (create_intersect_range_checks): Move from ...
20559 * tree-data-ref.c (create_intersect_range_checks_index)
20560 (create_intersect_range_checks): ... to here.
20561 (create_runtime_alias_checks): New function factored from ...
20562 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
20563 here. Call above function.
20564 * tree-data-ref.h (create_runtime_alias_checks): New function.
20566 2017-05-31 Bin Cheng <bin.cheng@arm.com>
20568 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
20569 segment length for dr_b and compute it in wide_int.
20571 2017-05-31 Richard Biener <rguenther@suse.de>
20573 PR tree-optimization/80906
20574 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
20575 and pass through iv_map.
20576 (copy_bb_and_scalar_dependences): Adjust.
20577 (translate_pending_phi_nodes): Likewise.
20578 (copy_loop_close_phi_args): Handle code-generating IVs instead
20581 2017-05-30 David Malcolm <dmalcolm@redhat.com>
20583 * diagnostic-color.c (color_dict): Add "type-diff".
20584 (parse_gcc_colors): Update comment.
20585 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
20586 -fdiagnostics-show-template-tree and -fno-elide-type.
20587 (GCC_COLORS): Add type-diff to example.
20589 (-fdiagnostics-show-template-tree): New.
20590 (-fno-elide-type): New.
20591 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
20592 the pp_format_decoder callback. Call any m_format_postprocessor's
20594 (pretty_printer::pretty_printer): Initialize
20595 m_format_postprocessor.
20596 (pretty_printer::~pretty_printer): Delete any
20597 m_format_postprocessor.
20598 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
20599 (class format_postprocessor): New class.
20600 (struct pretty_printer::format_decoder): Document the new parameters.
20601 (struct pretty_printer::m_format_postprocessor): New field.
20602 * tree-diagnostic.c (default_tree_printer): Update for new
20603 bool and const char ** params.
20604 * tree-diagnostic.h (default_tree_printer): Likewise.
20606 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
20608 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
20609 (lwa_operand): Delete rs6000_gen_cell_microcode test.
20610 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
20611 rs6000_gen_cell_microcode code.
20612 (rs6000_final_prescan_insn): Delete.
20613 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
20614 "warn-cell-microcode" entries.
20615 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
20616 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
20617 throughout. Change cc_reg_not_micro_cr0_operand to
20618 cc_reg_not_cr0_operand throughout.
20619 (*extendhi<mode>2_noload): Delete.
20620 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
20621 (mwarn-cell-microcode): Delete.
20622 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
20623 -mgen-cell-microcode and -mwarn-cell-microcode.
20625 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
20628 * config/i386/constraints.md (Yd): New constraint.
20630 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
20631 and (?Yd, r) alternatives. Update insn attributes.
20632 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
20633 and (?*Yd, r) alternatives. Update insn attributes.
20634 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
20636 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
20638 * gimplify.c (gimplify_modify_expr): Don't create a
20639 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
20642 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
20644 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
20646 2017-05-30 Richard Biener <rguenther@suse.de>
20648 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
20649 and reduc_def fields.
20650 (STMT_VINFO_REDUC_TYPE): New define.
20651 (STMT_VINFO_REDUC_DEF): Likewise.
20652 (vect_force_simple_reduction): Adjust prototype.
20653 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
20654 (vect_is_simple_reduction): Remove check_reduction argument.
20655 (vect_force_simple_reduction): Adjust and set
20656 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
20657 (vectorizable_reduction): Do not re-do reduction analysis
20658 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
20659 * tree-parloops.c (gather_scalar_reductions): Adjust.
20661 2017-05-30 Richard Biener <rguenther@suse.de>
20663 PR middle-end/80901
20664 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
20667 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
20669 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
20670 Introduce unknown_misalignment parameter and remove vf.
20671 (vect_peeling_hash_get_lowest_cost):
20672 Pass unknown_misalignment parameter.
20673 (vect_enhance_data_refs_alignment):
20674 Fix unsupportable data ref treatment.
20676 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20678 * tree-vect-data-refs.c (vect_get_data_access_cost):
20679 Workaround for SLP handling.
20680 (vect_enhance_data_refs_alignment):
20681 Compute costs for doing no peeling at all, compare to the best
20682 peeling costs so far and avoid peeling if cheaper.
20684 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20686 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
20687 Return peeling info and set costs to zero for unlimited cost
20689 (vect_enhance_data_refs_alignment): Also inspect all datarefs
20690 with unknown misalignment. Compute and costs for unknown
20691 misalignment, compare them to the costs for known misalignment
20692 and choose the cheapest for peeling.
20694 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20696 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
20697 (vect_get_peeling_costs_all_drs): Create function.
20698 (vect_peeling_hash_get_lowest_cost):
20699 Use vect_get_peeling_costs_all_drs.
20700 (vect_peeling_supportable): Create function.
20701 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
20703 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
20705 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
20706 DR_HAS_NEGATIVE_STEP.
20707 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
20708 (vect_enhance_data_refs_alignment): Use.
20709 (vect_duplicate_ssa_name_ptr_info): Use.
20710 * tree-vectorizer.h (dr_misalignment): Use.
20711 (known_alignment_for_access_p): Use.
20713 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
20716 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
20718 (has_section_name): New function.
20720 2017-05-30 Martin Liska <mliska@suse.cz>
20723 * auto-profile.c (get_function_decl_from_block): Fix
20726 2017-05-30 Richard Biener <rguenther@suse.de>
20728 PR middle-end/80876
20729 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
20731 2017-05-30 Martin Liska <mliska@suse.cz>
20733 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
20734 * dumpfile.h (struct dump_file_info): Remove ctors.
20736 2017-05-30 Martin Liska <mliska@suse.cz>
20738 * predict.def: Fix GNU coding style.
20740 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20742 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
20743 Mark 'to' argument with ATTRIBUTE_UNUSED.
20745 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
20747 * config/xtensa/xtensa.c (xtensa_emit_call): Use
20748 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
20749 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
20752 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
20754 * doc/install.texi (Options specification): Restore entry of
20755 --enable-sjlj-exceptions.
20757 2017-05-27 Michael Eager <eager@eagercon.com>
20760 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
20762 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
20764 * config/microblaze/microblaze.h
20765 (FIXED_REGISTERS): Update in macro.
20766 (CALL_USED_REGISTERS): Update in macro.
20768 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
20770 * doc/install.texi: Add links to macOS binary distributions.
20772 2017-05-27 Jakub Jelinek <jakub@redhat.com>
20776 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20778 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20780 2017-05-26 Martin Liska <mliska@suse.cz>
20782 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
20784 2017-05-26 Martin Liska <mliska@suse.cz>
20786 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
20787 always leading ';; '.
20788 (dump_bb_info): Likewise.
20789 (brief_dump_cfg): Likewise.
20790 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
20791 * dumpfile.c: Remove usage of TDF_VERBOSE.
20792 * dumpfile.h (enum dump_kind): Likewise.
20793 (dump_gimple_bb_header): Do not use TDF_COMMENT.
20794 * print-tree.c (debug_verbose): Remove.
20795 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
20796 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
20797 * tree-diagnostic.c (default_tree_printer): Replace
20798 TDF_DIAGNOSTIC with TDF_SLIM.
20800 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20802 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
20803 in parameter loop, rather than loop_vinfo.
20804 (create_intersect_range_checks): Ditto.
20805 (vect_create_cond_for_alias_checks): Update call to above functions.
20807 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20809 PR tree-optimization/80815
20810 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
20811 for merging runtime alias checks. Handle negative DR_STEPs.
20813 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20815 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
20817 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
20818 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
20819 out code pruning runtime alias checks.
20820 * tree-data-ref.c (prune_runtime_alias_test_list): New function
20821 factored out from above.
20822 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20824 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
20826 (prune_runtime_alias_test_list): New decalaration.
20828 2017-05-26 Bin Cheng <bin.cheng@arm.com>
20830 * tree-vect-data-refs.c (compare_tree): Rename and move ...
20831 * tree-data-ref.c (data_ref_compare_tree): ... to here.
20832 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
20833 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
20834 (operator==, comp_dr_with_seg_len_pair): Ditto.
20835 (vect_prune_runtime_alias_test_list): Ditto.
20837 2017-05-26 Martin Liska <mliska@suse.cz>
20840 * params.def: Bound partial-inlining-entry-probability param.
20842 2017-05-26 Marek Polacek <polacek@redhat.com>
20845 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
20848 2017-05-26 Richard Biener <rguenther@suse.de>
20850 PR tree-optimization/80842
20851 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
20854 2017-05-26 Richard Biener <rguenther@suse.de>
20856 PR tree-optimization/80844
20857 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
20859 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
20861 * doc/md.texi (Machine Constraints): Update x86 family
20862 machine constraints section to match 'config/i386/constraints.md'.
20864 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
20866 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
20868 2017-05-25 Nathan Sidwell <nathan@acm.org>
20870 * doc/invoke.texi (--enable-languages): Update documentation.
20872 2017-05-25 Martin Liska <mliska@suse.cz>
20874 * dumpfile.c: Add TDF_FOLDING.
20875 * dumpfile.h (enum dump_kind): Likewise.
20876 * genmatch.c (dt_simplify::gen_1): Use it.
20878 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20880 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
20882 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20884 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
20885 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
20887 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
20889 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
20890 * match.pd (X == C): Rewrite it here.
20891 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
20892 with_certain_nonzero_bits2): New predicates.
20893 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
20895 2017-05-24 Nathan Sidwell <nathan@acm.org>
20897 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
20900 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
20903 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20905 * config/powerpcspe: New port. Files are copied from the rs6000
20906 port, with "rs6000" in filenames replaced by "powerpcspe".
20908 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
20910 PR rtl-optimization/80754
20911 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
20913 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
20915 * config/sparc/sparc.md (length): Return the correct value for -mflat
20916 sibcalls to match output_sibcall.
20918 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
20922 * config/rs6000/rs6000.c (struct machine_function): Add new field
20924 (rs6000_get_separate_components): Init that field, use it.
20925 (rs6000_components_for_bb): Use the field.
20927 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
20929 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
20931 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
20933 PR middle-end/80823
20934 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
20936 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20939 * config/s390/s390.c (s390_check_qrst_address): Check incoming
20940 address against address_operand predicate.
20941 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
20943 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
20945 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
20946 parameters passed indirectly.
20948 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
20950 * config/i386/i386.md (*movdi_internal): Remove SSE4
20951 alternative 18 (?r, *v). Update insn attributes.
20952 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
20953 Update insn attributes.
20954 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
20955 Update insn attributes.
20956 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
20957 alternative 1 (r, v). Remove isa attribute.
20958 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
20959 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
20960 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
20962 2017-05-23 Tom de Vries <tom@codesourcery.com>
20964 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
20967 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
20969 * cgraphunit.c (symbol_table::process_new_functions): Update.
20970 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
20971 (inline_generate_summary): Rename to ...
20972 (ipa_fn_summary_generate): ... this one.
20973 (inline_read_summary): Rename to ...
20974 (ipa_fn_summary_read): ... this one.
20975 (inline_write_summary): Rename to ...
20976 (ipa_fn_summary_write): ... this one.
20977 (inline_free_summary): Rename to ...
20978 (ipa_free_fn_summary): ... this one.
20979 (pass_data_local_fn_summary, pass_local_fn_summary,
20980 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
20981 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
20982 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
20983 make_pass_ipa_fn_summary): New.
20984 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
20985 inline_write_summary, inline_free_summary): Remove.
20986 (ipa_free_fn_summary) : New.
20987 * ipa-inline.c (ipa_inline): Update.
20988 (pass_ipa_inline): Do not generate summaries.
20989 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
20991 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
20992 and add pass_ipa_fn_summary.
20993 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
20995 (make_pass_inline_parameters): Remove.
20997 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
20999 * omp-low.c (struct omp_context): Remove "default_kind" member.
21002 * omp-offload.c (execute_oacc_device_lower): Remove the
21003 parallelism dimensions function attributes for unparallelized
21004 OpenACC kernels constructs.
21006 2017-05-23 Martin Liska <mliska@suse.cz>
21008 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
21010 (cgraph_edge::make_speculative): Likewise.
21011 (cgraph_edge::resolve_speculation): Likewise.
21012 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
21013 (cgraph_node::dump): Likewise.
21014 * cgraph.h: Likewise.
21015 * cgraphunit.c (analyze_functions): Likewise.
21016 (symbol_table::compile): Likewise.
21017 * ipa-cp.c (print_all_lattices): Likewise.
21018 (determine_versionability): Likewise.
21019 (initialize_node_lattices): Likewise.
21020 (ipcp_verify_propagated_values): Likewise.
21021 (estimate_local_effects): Likewise.
21022 (update_profiling_info): Likewise.
21023 (create_specialized_node): Likewise.
21024 (perhaps_add_new_callers): Likewise.
21025 (decide_about_value): Likewise.
21026 (decide_whether_version_node): Likewise.
21027 (identify_dead_nodes): Likewise.
21028 (ipcp_store_bits_results): Likewise.
21029 * ipa-devirt.c (dump_targets): Likewise.
21030 (ipa_devirt): Likewise.
21031 * ipa-icf.c (sem_item::dump): Likewise.
21032 (sem_function::equals): Likewise.
21033 (sem_variable::equals): Likewise.
21034 (sem_item_optimizer::read_section): Likewise.
21035 (sem_item_optimizer::execute): Likewise.
21036 (congruence_class::dump): Likewise.
21037 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
21038 (dump_inline_summary): Likewise.
21039 (estimate_node_size_and_time): Likewise.
21040 (inline_analyze_function): Likewise.
21041 * ipa-inline-transform.c (inline_call): Likewise.
21042 * ipa-inline.c (report_inline_failed_reason): Likewise.
21043 (want_early_inline_function_p): Likewise.
21044 (edge_badness): Likewise.
21045 (update_edge_key): Likewise.
21046 (inline_small_functions): Likewise.
21047 * ipa-profile.c (ipa_profile): Likewise.
21048 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
21049 (ipa_make_edge_direct_to_target): Likewise.
21050 (remove_described_reference): Likewise.
21051 (ipa_impossible_devirt_target): Likewise.
21052 (propagate_controlled_uses): Likewise.
21053 (ipa_print_node_params): Likewise.
21054 (ipcp_transform_function): Likewise.
21055 * ipa-pure-const.c (pure_const_read_summary): Likewise.
21056 (propagate_pure_const): Likewise.
21057 * ipa-reference.c (generate_summary): Likewise.
21058 (read_write_all_from_decl): Likewise.
21059 (propagate): Likewise.
21060 (ipa_reference_read_optimization_summary): Likewise.
21061 * ipa-utils.c (ipa_merge_profiles): Likewise.
21062 * ipa.c (walk_polymorphic_call_targets): Likewise.
21063 (symbol_table::remove_unreachable_nodes): Likewise.
21064 (ipa_single_use): Likewise.
21065 * passes.c (execute_todo): Likewise.
21066 * predict.c (drop_profile): Likewise.
21067 * symtab.c (symtab_node::get_dump_name): New function.
21068 (symtab_node::dump_name): Likewise.
21069 (symtab_node::dump_asm_name): Likewise.
21070 (symtab_node::dump_references): Likewise.
21071 (symtab_node::dump_referring): Likewise.
21072 (symtab_node::dump_base): Likewise.
21073 (symtab_node::debug_symtab): Likewise.
21074 * tree-sra.c (convert_callers_for_node): Likewise.
21075 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
21076 * value-prof.c (init_node_map): Likewise.
21078 2017-05-23 Martin Liska <mliska@suse.cz>
21080 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
21081 and symtab_node::debug_symtab to symbol_table::debug.
21082 * cgraphunit.c (analyze_functions): Use the renamed function.
21083 (symbol_table::compile): Likewise.
21084 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
21085 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
21086 * passes.c (execute_todo): Likewise.
21087 * symtab.c (symbol_table::dump): New function.
21088 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
21090 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21092 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
21093 that nonconst implies exec.
21095 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
21097 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21098 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21099 (inline_edge_summary_vec): Turn into ...
21100 (ipa_call_summaries): ... this one.
21101 (redirect_to_unreachable, edge_set_predicate,
21102 evaluate_properties_for_edge, inline_summary_alloc,
21103 reset_ipa_call_summary, reset_inline_summary,
21104 inline_summary_t::duplicate): Update.
21105 (inline_edge_duplication_hook): Turn to ...
21106 (ipa_call_summary_t::duplicate): ... this one.
21107 (inline_edge_removal_hook): Turn to ...
21108 (ipa_call_summary_t::remove): ... this one.
21109 (dump_inline_edge_summary): Turn to ...
21110 (dump_ipa_call_summary): ... this one.
21111 (estimate_function_body_sizes): Update.
21112 (inline_update_callee_summaries): Update.
21113 (remap_edge_change_prob): Update.
21114 (remap_edge_summaries): Update.
21115 (inline_merge_summary): Update.
21116 (do_estimate_edge_time): Update.
21117 (inline_generate_summary): Update.
21118 (inline_read_section): Update.
21119 (inline_read_summary): Update.
21120 (inline_free_summary): Update.
21121 * ipa-inline.c (can_inline_edge_p): Update.
21122 (compute_inlined_call_time): Update.
21123 (want_inline_small_function_p): Update.
21124 (edge_badness): Update.
21125 (early_inliner): Update.
21126 * ipa-inline.h (inline_edge_summary): Turn to ...
21127 (ipa_call_summary): ... this one.
21128 (ipa_call_summary_t): New class.
21129 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21130 (ipa_call_summaries): New.
21131 (inline_edge_summary): Remove.
21132 (estimate_edge_growth): Update.
21133 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21134 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21135 * ipa-split.c (execute_split_functions): Update.
21136 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21138 2017-05-23 Tom de Vries <tom@codesourcery.com>
21140 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21141 attributes): Document rdrand effective target.
21143 2017-05-23 Tom de Vries <tom@codesourcery.com>
21145 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
21146 attributes): Sort alphabetically.
21148 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
21150 * config/avr/genmultilib.awk: Use gsub instead of gensub.
21152 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
21155 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
21156 V2DF/V2DI splat into two separate patterns, one that handles
21157 registers, and the other that only handles memory. Drop support
21158 for splatting from a GPR on ISA 2.07 and then splitting the
21159 splat into direct move and splat.
21160 (vsx_splat_<mode>_reg): Likewise.
21161 (vsx_splat_<mode>_mem): Likewise.
21163 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
21165 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
21167 2017-05-22 Jakub Jelinek <jakub@redhat.com>
21169 PR middle-end/80809
21170 * omp-low.c (finish_taskreg_remap): New function.
21171 (finish_taskreg_scan): If unit size of ctx->record_type
21172 is non-constant, unshare the size expression and replace
21173 decls in it with possible outer var refs.
21175 PR middle-end/80809
21176 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
21177 GOVD_SHARED rather than GOVD_PRIVATE with it.
21178 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
21179 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
21181 PR middle-end/80853
21182 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
21183 as last argument to build_outer_var_ref for pointer bases of array
21184 section reductions.
21186 2017-05-19 Martin Sebor <msebor@redhat.com>
21188 * print-tree.c (print_node): Print DECL_READ_P flag.
21190 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21192 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
21193 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
21194 * cgraph.c: Likewise.
21195 * cgraphunit.c: Likewise.
21196 * gengtype.c: Likewise.
21197 * ipa-cp.c: Likewise.
21198 * ipa-devirt.c: Likewise.
21199 * ipa-icf.c: Likewise.
21200 * ipa-predicate.c: Likewise.
21201 * ipa-profile.c: Likewise.
21202 * ipa-prop.c: Likewise.
21203 * ipa-split.c: Likewise.
21205 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
21206 edge_predicate_pool, dump_inline_hints,
21207 inline_summary::account_size_time, redirect_to_unreachable,
21208 edge_set_predicate, set_hint_predicate,
21209 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
21210 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
21211 inline_summary_t::remove, remap_hint_predicate_after_duplication,
21212 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
21213 ipa_call_summary_t::remove, initialize_growth_caches,
21214 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
21215 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
21216 mark_modified, unmodified_parm_1, unmodified_parm,
21217 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
21218 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
21219 compute_bb_predicates, will_be_nonconstant_expr_predicate,
21220 will_be_nonconstant_predicate, record_modified_bb_info,
21221 get_minimal_bb, record_modified, param_change_prob,
21222 phi_result_unknown_predicate, predicate_for_phi_result,
21223 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
21224 estimate_function_body_sizes, compute_inline_parameters,
21225 compute_inline_parameters_for_curren, pass_data_inline_parameters,
21226 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
21227 inline_update_callee_summaries, remap_edge_change_prob,
21228 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
21229 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
21230 inline_analyze_function, inline_summary_t::insert,
21231 inline_generate_summary, read_ipa_call_summary, inline_read_section,
21232 inline_read_summary, write_ipa_call_summary, inline_write_summary,
21233 inline_free_summary): Move to ipa-fnsummary.h
21234 (predicate_t): Remove.
21235 * ipa-fnsummary.c: New file.
21236 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
21237 (enum inline_hints_vals, inline_hints, agg_position_info,
21238 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
21239 inline_summaries, ipa_call_summary, ipa_call_summary_t,
21240 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
21241 dump_inline_summary, dump_inline_hints, inline_generate_summary,
21242 inline_read_summary, inline_write_summary, inline_free_summary,
21243 inline_analyze_function, initialize_inline_failed,
21244 inline_merge_summary, inline_update_overall_summary,
21245 compute_inline_parameters): Move to ipa-fnsummary.h
21246 * ipa-fnsummary.h: New file.
21247 * ipa-inline-transform.h: Include ipa-inline.h.
21248 * ipa-inline.c: LIkewise.
21250 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21252 * ipa-inline.c (edge_badness): Use inlined_time instead of
21253 inline_summaries->get.
21255 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21257 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
21259 2017-05-22 Nathan Sidwell <nathan@acm.org>
21261 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
21262 (fdump-lang): Document 'raw' option.
21263 * dumpfile.h (TDI_tu): Delete.
21264 * dumpfile.c (dump_files): Remove translation-unit.
21265 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
21267 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
21269 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
21270 command option from $(AWK) call.
21271 * config/avr/genmultilib.awk: Simplify and rewrite so that it
21272 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
21273 [FORMAT]: Remove handling of variable.
21274 * config/avr/t-multilib: Regenerate.
21276 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21278 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
21280 (dump_inline_summary): Do not print self_time.
21281 (estimate_function_body_sizes): Do not set self_time.
21282 (compute_inline_parameters): Likewise.
21283 (inline_read_section, inline_write_summary): Do not stream self_time.
21284 * ipa-inline.h (inline_summary): Drop self_time.
21286 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
21288 * ipa-inline-analysis.c (account_size_time): Rename to ...
21289 (inline_summary::account_size_time): ... this one.
21290 (reset_ipa_call_summary): Turn to ...
21291 (ipa_call_summary::reset): ... this one.
21292 (reset_inline_summary): Turn to ...
21293 (inline_summary::reset): ... this one.
21294 (inline_summary_t::remove): Update.
21295 (inline_summary_t::duplicate): Update.
21296 (ipa_call_summary_t::remove): Update.
21297 (dump_inline_summary): Update.
21298 (estimate_function_body_sizes): Update.
21299 (compute_inline_parameters): Update.
21300 (estimate_node_size_and_time): Update.
21301 (inline_merge_summary): Update.
21302 (inline_update_overall_summary): Update.
21303 (inline_read_section): Update.
21304 (inline_write_summary): Update.
21305 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
21306 add account_size_time and reset member functions.
21307 (ipa_call_summary): Add reset function.
21308 * ipa-predicate.h (predicate::operator &): Constify.
21310 2017-05-22 Richard Biener <rguenther@suse.de>
21312 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
21314 2017-05-19 Jason Merrill <jason@redhat.com>
21316 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
21318 2017-05-19 Marek Polacek <polacek@redhat.com>
21321 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
21322 TYPE_OVERFLOW_WRAPS checks.
21324 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
21326 * tree-core.h (enum omp_clause_default_kind): Add
21327 "OMP_CLAUSE_DEFAULT_PRESENT".
21328 * tree-pretty-print.c (dump_omp_clause): Handle it.
21329 * gimplify.c (enum gimplify_omp_var_data): Add
21330 "GOVD_MAP_FORCE_PRESENT".
21331 (gimplify_adjust_omp_clauses_1): Map it to
21332 "GOMP_MAP_FORCE_PRESENT".
21333 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
21335 * gimplify.c (oacc_default_clause): Clarify.
21337 2017-05-19 Nathan Sidwell <nathan@acm.org>
21339 LANG_HOOK_REGISTER_DUMPS
21340 * toplev.c (general_init): Call register dump lang hook.
21341 * doc/invoke.texi: Document -fdump-lang option family.
21342 * dumpfile.c (dump_files): Remove class dump here.
21343 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
21344 * dumpfile.h (tree_dump_index): Remove TDI_class.
21345 * langhooks-def.h (lhd_register_dumps): Declare.
21346 (LANG_HOOKS_REGISTER_DUMPS): Define.
21347 (LANG_HOOKS_INITIALIZER): Add it.
21348 * langhooks.c (lhd_register_dumps): Define.
21349 * langhooks.h (struct lang_hooks): Add register_dumps.
21351 2017-05-19 Nathan Sidwell <nathan@acm.org>
21353 * context.h (context::set_passes): New.
21354 * context.c (context::context): Do not create pass manager.
21355 * toplev.c (general_init): Create pass manager here.
21357 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
21359 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
21360 use this splitter if two add or or instructions would also work for
21361 the constant we want to generate.
21363 2017-05-19 Richard Biener <rguenther@suse.de>
21366 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
21367 predicate evaluation.
21369 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21371 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
21373 * ipa-inline.c (want_inline_small_function_p): Do not cast to
21376 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
21378 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
21379 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
21380 (inline_edge_summary_vec): Turn into ...
21381 (ipa_call_summaries): ... this one.
21382 (redirect_to_unreachable, edge_set_predicate,
21383 evaluate_properties_for_edge, inline_summary_alloc,
21384 reset_ipa_call_summary, reset_inline_summary,
21385 inline_summary_t::duplicate): Update.
21386 (inline_edge_duplication_hook): Turn to ...
21387 (ipa_call_summary_t::duplicate): ... this one.
21388 (inline_edge_removal_hook): Turn to ...
21389 (ipa_call_summary_t::remove): ... this one.
21390 (dump_inline_edge_summary): Turn to ...
21391 (dump_ipa_call_summary): ... this one.
21392 (estimate_function_body_sizes): Update.
21393 (inline_update_callee_summaries): Update.
21394 (remap_edge_change_prob): Update.
21395 (remap_edge_summaries): Update.
21396 (inline_merge_summary): Update.
21397 (do_estimate_edge_time): Update.
21398 (inline_generate_summary): Update.
21399 (inline_read_section): Update.
21400 (inline_read_summary): Update.
21401 (inline_free_summary): Update.
21402 * ipa-inline.c (can_inline_edge_p): Update.
21403 (compute_inlined_call_time): Update.
21404 (want_inline_small_function_p): Update.
21405 (edge_badness): Update.
21406 (early_inliner): Update.
21407 * ipa-inline.h (inline_edge_summary): Turn to ...
21408 (ipa_call_summary): ... this one.
21409 (ipa_call_summary_t): New class.
21410 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
21411 (ipa_call_summaries): New.
21412 (inline_edge_summary): Remove.
21413 (estimate_edge_growth): Update.
21414 * ipa-profile.c (ipa_propagate_frequency_1): Update.
21415 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
21416 * ipa-split.c (execute_split_functions): Update.
21417 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
21419 2017-05-19 Richard Biener <rguenther@suse.de>
21421 PR middle-end/80764
21422 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
21424 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
21426 * config/rs6000/rs6000.c (struct machine_function): Add field
21427 fpr_is_wrapped_separately.
21428 (rs6000_get_separate_components): Use 64 components. Handle the
21429 new FPR components.
21430 (rs6000_components_for_bb): Handle the FPR components.
21431 (rs6000_emit_prologue_components): Handle the FPR components.
21432 (rs6000_emit_epilogue_components): Handle the FPR components.
21433 (rs6000_set_handled_components): Handle the FPR components.
21434 (rs6000_emit_prologue): Don't output prologue code for those FPRs
21435 that are already separately shrink-wrapped.
21436 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
21437 that are already separately shrink-wrapped.
21439 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
21442 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
21445 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
21446 (define_peephole2 for Altivec d-form load): Add peepholes to catch
21447 cases where the register allocator uses a move and an offsettable
21448 memory operation to/from a FPR register on ISA 2.06/2.07.
21449 (define_peephole2 for Altivec d-form store): Likewise.
21451 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
21454 * config/i386/mmx.md (*mov<mode>_internal): Enable
21455 alternatives 11, 12, 13 and 14 also for 32bit targets.
21456 Remove alternatives 15, 16, 17 and 18.
21457 * config/i386/sse.md (vec_concatv2di): Change
21458 alternative (!x, *y) to (x, ?!*Yn).
21460 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
21462 * dumpfile.h (enum dump_kind): Remove stray comma.
21464 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21466 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
21467 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
21468 predicate::num_conditions
21469 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
21470 (CHANGED): turn into predicate::changed.
21471 (agg_position_info): Move to ipa-predicate.h
21472 (add_condition, predicate::add_clause, predicate::operator &=,
21473 predicate::or_with, predicate::evaluate, predicate::probability,
21474 dump_condition, dump_clause, predicate::dump,
21475 predicate::remap_after_duplication, predicate::remap_after_inlining,
21476 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
21477 (evaluate_conditions_for_known_args): Update.
21478 (set_cond_stmt_execution_predicate): Update.
21479 * ipa-inline.h: Include ipa-predicate.h
21480 (condition, inline_param_summary, conditions, agg_position_info,
21481 predicate): Move to ipa-predicate.h
21482 * ipa-predicate.c: New file.
21483 * ipa-predicate.h: New file.
21485 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
21487 * final.c (leaf_function_p): Check we are not in a sequence.
21489 2017-05-18 Martin Liska <mliska@suse.cz>
21491 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
21492 * dumpfile.c (dump_register): Use new enum dump_kind.
21493 (get_dump_file_name): Likewise.
21494 (dump_enable_all): Likewise.
21495 (dump_switch_p_1): Likewise.
21496 (enable_rtl_dump_file): Remove usage of TDF_RTL.
21497 * dumpfile.h (enum dump_kind): New enum type.
21498 (struct dump_file_info): Create constructor and
21499 format fields and comments.
21500 * passes.c (pass_manager::register_one_dump_file):
21502 * statistics.c (statistics_early_init): Likewise.
21503 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
21504 TDF_TREE with TDF_SLIM.
21505 (gather_memory_references_ref): Likewise.
21507 2017-05-18 Martin Liska <mliska@suse.cz>
21509 * vec.h (struct vnull): Use it.
21511 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
21513 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
21514 (true_predicate, false_predicate, true_predicate_p,
21515 false_predicate_p): Remove.
21516 (single_cond_predicate, not_inlined_predicate): Turn to member function
21518 (add_condition): Update.
21519 (add_clause): Turn to...
21520 (predicate::add_clause): ... this one; update; allow passing NULL
21522 (and_predicates): Turn to ...
21523 (predicate::operator &=): ... this one.
21524 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
21525 (or_predicates): Turn to ...
21526 (predicate::or_with): ... this one.
21527 (evaluate_predicate): Turn to ...
21528 (predicate::evaluate): ... this one.
21529 (predicate_probability): Turn to ...
21530 (predicate::probability): ... this one.
21531 (dump_condition): Update.
21532 (dump_predicate): Turn to ...
21533 (predicate::dump): ... this one.
21534 (account_size_time): Update.
21535 (edge_set_predicate): Update.
21536 (set_hint_predicate): UPdate.
21537 (evaluate_conditions_for_known_args): Update.
21538 (evaluate_properties_for_edge): Update.
21539 (remap_predicate_after_duplication): Turn to...
21540 (predicate::remap_after_duplication): ... this one.
21541 (remap_hint_predicate_after_duplication): Update.
21542 (inline_summary_t::duplicate): UPdate.
21543 (dump_inline_edge_summary): Update.
21544 (dump_inline_summary): Update.
21545 (set_cond_stmt_execution_predicate): Update.
21546 (set_switch_stmt_execution_predicate): Update.
21547 (compute_bb_predicates): Update.
21548 (will_be_nonconstant_expr_predicate): Update.
21549 (will_be_nonconstant_predicate): Update.
21550 (phi_result_unknown_predicate): Update.
21551 (predicate_for_phi_result): Update.
21552 (array_index_predicate): Update.
21553 (estimate_function_body_sizes): Update.
21554 (estimate_node_size_and_time): Update.
21555 (estimate_ipcp_clone_size_and_time): Update.
21556 (remap_predicate): Rename to ...
21557 (predicate::remap_after_inlining): ... this one.
21558 (remap_hint_predicate): Update.
21559 (inline_merge_summary): Update.
21560 (inline_update_overall_summary): Update.
21561 (estimate_size_after_inlining): Update.
21562 (read_predicate): Rename to ...
21563 (predicate::stream_in): ... this one.
21564 (read_inline_edge_summary): Update.
21565 (write_predicate): Rename to ...
21566 (predicate::stream_out): ... this one.
21567 (write_inline_edge_summary): Update.
21568 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
21569 (clause_t): Turn to uint32_t
21570 (predicate): Turn to class; implement constructor and operators
21572 (size_time_entry): Update.
21573 (inline_summary): Update.
21574 (inline_edge_summary): Update.
21576 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
21578 * fold-const.c (fold_binary_loc): Move transformation...
21579 * match.pd (C - X CMP X): ... here.
21581 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
21583 * config/sparc/sparc.c (sparc_option_override): Set function
21584 alignment for -mcpu=niagara7 to 64 to match the I$ line.
21585 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
21587 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
21589 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
21591 2017-05-18 Marek Polacek <polacek@redhat.com>
21594 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
21595 (pass_ubsan::execute): Call gimple_assign_single_p instead of
21596 gimple_assign_load_p.
21598 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
21600 PR middle-end/80692
21601 * real.c (do_compare): Give decimal_do_compare preference over
21602 comparing just the signs.
21604 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
21606 * doc/md.texi (Canonicalization of Instructions): Describe the
21607 canonical form of instructions that inherently set a condition
21610 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
21612 PR middle-end/80775
21613 * tree-cfg.c: Move deletion of unreachable case statements to after
21614 the merging of consecutive case labels.
21616 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
21618 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
21619 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
21620 restoring of callee-saved registers.
21622 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
21624 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
21625 * config/visium/visium.c (single_set_and_flags): Likewise.
21626 * config/visium/visium.md (Substitutions): Likewise.
21628 2017-05-17 Martin Liska <mliska@suse.cz>
21630 * cfg.c: Introduce dump_flags_t type and
21631 use it instead of int type.
21633 * cfghooks.c: Likewise.
21634 * cfghooks.h (struct cfg_hooks): Likewise.
21635 * cfgrtl.c: Likewise.
21636 * cfgrtl.h: Likewise.
21637 * cgraph.c (cgraph_node::get_body): Likewise.
21638 * coretypes.h: Likewise.
21639 * domwalk.c: Likewise.
21640 * domwalk.h: Likewise.
21641 * dumpfile.c (struct dump_option_value_info): Likewise.
21642 (dump_enable_all): Likewise.
21643 (dump_switch_p_1): Likewise.
21644 (opt_info_switch_p): Likewise.
21645 * dumpfile.h (enum tree_dump_index): Likewise.
21646 (struct dump_file_info): Likewise.
21647 * genemit.c: Likewise.
21648 * generic-match-head.c: Likewise.
21649 * gengtype.c (open_base_files): Likewise.
21650 * gimple-pretty-print.c: Likewise.
21651 * gimple-pretty-print.h: Likewise.
21652 * graph.c (print_graph_cfg): Likewise.
21653 * graphite-scop-detection.c (dot_all_sese): Likewise.
21654 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
21655 * loop-unroll.c (report_unroll): Likewise.
21656 * passes.c (pass_manager::register_one_dump_file): Likewise.
21657 * print-tree.c: Likewise.
21658 * statistics.c: Likewise.
21659 * tree-cfg.c: Likewise.
21660 * tree-cfg.h: Likewise.
21661 * tree-dfa.c: Likewise.
21662 * tree-dfa.h: Likewise.
21663 * tree-dump.c (dump_function): Likewise.
21664 * tree-dump.h (struct dump_info): Likewise.
21665 * tree-pretty-print.c: Likewise.
21666 * tree-pretty-print.h: Likewise.
21667 * tree-ssa-live.c: Likewise.
21668 * tree-ssa-live.h: Likewise.
21669 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
21670 * tree-vect-loop.c: Likewise.
21671 * tree-vect-slp.c: Likewise.
21673 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
21674 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21676 PR tree-optimization/80457
21677 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
21678 of all arguments to a statement as scalar_to_vec operations.
21679 (vectorizable_call): Adjust call to vect_model_simple_cost for
21681 (vectorizable_conversion): Likewise.
21682 (vectorizable_assignment): Likewise.
21683 (vectorizable_shift): Likewise.
21684 (vectorizable_operation): Likewise.
21685 (vectorizable_comparison): Likewise.
21686 (vect_is_simple_cond): Record the def types for operands.
21687 (vectorizable_condition): Likewise, call vect_model_simple_cost.
21688 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
21689 for statement argument count.
21691 2017-05-16 Carl Love <cel@us.ibm.com>
21693 * config/rs6000/rs6000-c: Add support for built-in functions
21694 vector unsigned long long vec_bperm (vector unsigned long long,
21695 vector unsigned char)
21696 vector signed long long vec_mule (vector signed int,
21698 vector unsigned long long vec_mule (vector unsigned int,
21699 vector unsigned int)
21700 vector signed long long vec_mulo (vector signed int,
21702 vector unsigned long long vec_mulo (vector unsigned int,
21703 vector unsigned int)
21704 vector signed char vec_sldw (vector signed char,
21705 vector signed char,
21707 vector unsigned char vec_sldw (vector unsigned char,
21708 vector unsigned char,
21710 vector signed short vec_sldw (vector signed short,
21711 vector signed short,
21713 vector unsigned short vec_sldw (vector unsigned short,
21714 vector unsigned short,
21716 vector signed int vec_sldw (vector signed int,
21719 vector unsigned int vec_sldw (vector unsigned int,
21720 vector unsigned int,
21722 vector signed long long vec_sldw (vector signed long long,
21723 vector signed long long,
21725 vector unsigned long long vec_sldw (vector unsigned long long,
21726 vector unsigned long long,
21728 * config/rs6000/rs6000-c: Add support for built-in functions
21729 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
21730 * config/rs6000/altivec.h: Add defintion for vec_sldw.
21731 * doc/extend.texi: Update the built-in documentation for the
21732 new built-in functions.
21734 2017-05-16 Marek Polacek <polacek@redhat.com>
21738 * tree.c (save_expr): Don't fold the expression.
21740 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
21742 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
21743 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
21744 and (?*y,m). Update insn attributes.
21746 2017-05-16 Martin Liska <mliska@suse.cz>
21748 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
21749 flags argument of print_gimple_stmt, print_gimple_expr,
21750 print_generic_stmt and print_generic_expr.
21751 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
21752 * coretypes.h: Likewise.
21753 * except.c (dump_eh_tree): Likewise.
21754 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
21755 * gimple-pretty-print.h: Likewise.
21756 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
21757 (backprop::push_to_worklist): Likewise.
21758 (backprop::pop_from_worklist): Likewise.
21759 (backprop::process_use): Likewise.
21760 (backprop::intersect_uses): Likewise.
21761 (note_replacement): Likewise.
21762 * gimple-ssa-store-merging.c
21763 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
21764 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
21765 (pass_store_merging::execute): Likewise.
21766 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
21767 (ssa_base_cand_dump_callback): Likewise.
21768 (dump_incr_vec): Likewise.
21769 (replace_refs): Likewise.
21770 (replace_mult_candidate): Likewise.
21771 (create_add_on_incoming_edge): Likewise.
21772 (create_phi_basis): Likewise.
21773 (insert_initializers): Likewise.
21774 (all_phi_incrs_profitable): Likewise.
21775 (introduce_cast_before_cand): Likewise.
21776 (replace_one_candidate): Likewise.
21777 * gimplify.c (gimplify_expr): Likewise.
21778 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
21779 (set_rename): Likewise.
21780 (rename_uses): Likewise.
21781 (copy_loop_phi_nodes): Likewise.
21782 (add_close_phis_to_merge_points): Likewise.
21783 (copy_loop_close_phi_args): Likewise.
21784 (copy_cond_phi_args): Likewise.
21785 (graphite_copy_stmts_from_block): Likewise.
21786 (translate_pending_phi_nodes): Likewise.
21787 * graphite-poly.c (print_pdr): Likewise.
21788 (dump_gbb_cases): Likewise.
21789 (dump_gbb_conditions): Likewise.
21790 (print_scop_params): Likewise.
21791 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
21792 (build_cross_bb_scalars_use): Likewise.
21793 (gather_bbs::before_dom_children): Likewise.
21794 * hsa-dump.c (dump_hsa_immed): Likewise.
21795 * ipa-cp.c (print_ipcp_constant_value): Likewise.
21796 (get_replacement_map): Likewise.
21797 * ipa-inline-analysis.c (dump_condition): Likewise.
21798 (estimate_function_body_sizes): Likewise.
21799 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
21800 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
21801 * ipa-prop.c (ipa_dump_param): Likewise.
21802 (ipa_print_node_jump_functions_for_edge): Likewise.
21803 (ipa_modify_call_arguments): Likewise.
21804 (ipa_modify_expr): Likewise.
21805 (ipa_dump_param_adjustments): Likewise.
21806 (ipa_dump_agg_replacement_values): Likewise.
21807 (ipcp_modif_dom_walker::before_dom_children): Likewise.
21808 * ipa-pure-const.c (check_stmt): Likewise.
21809 (pass_nothrow::execute): Likewise.
21810 * ipa-split.c (execute_split_functions): Likewise.
21811 * omp-offload.c (dump_oacc_loop_part): Likewise.
21812 (dump_oacc_loop): Likewise.
21813 * trans-mem.c (tm_log_emit): Likewise.
21814 (tm_memopt_accumulate_memops): Likewise.
21815 (dump_tm_memopt_set): Likewise.
21816 (dump_tm_memopt_transform): Likewise.
21817 * tree-cfg.c (gimple_verify_flow_info): Likewise.
21818 (print_loop): Likewise.
21819 * tree-chkp-opt.c (chkp_print_addr): Likewise.
21820 (chkp_gather_checks_info): Likewise.
21821 (chkp_get_check_result): Likewise.
21822 (chkp_remove_check_if_pass): Likewise.
21823 (chkp_use_outer_bounds_if_possible): Likewise.
21824 (chkp_reduce_bounds_lifetime): Likewise.
21825 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
21826 (chkp_mark_completed_bounds): Likewise.
21827 (chkp_register_incomplete_bounds): Likewise.
21828 (chkp_mark_invalid_bounds): Likewise.
21829 (chkp_maybe_copy_and_register_bounds): Likewise.
21830 (chkp_build_returned_bound): Likewise.
21831 (chkp_get_bound_for_parm): Likewise.
21832 (chkp_build_bndldx): Likewise.
21833 (chkp_get_bounds_by_definition): Likewise.
21834 (chkp_generate_extern_var_bounds): Likewise.
21835 (chkp_get_bounds_for_decl_addr): Likewise.
21836 * tree-chrec.c (chrec_apply): Likewise.
21837 * tree-data-ref.c (dump_data_reference): Likewise.
21838 (dump_subscript): Likewise.
21839 (dump_data_dependence_relation): Likewise.
21840 (analyze_overlapping_iterations): Likewise.
21841 * tree-inline.c (expand_call_inline): Likewise.
21842 (tree_function_versioning): Likewise.
21843 * tree-into-ssa.c (dump_defs_stack): Likewise.
21844 (dump_currdefs): Likewise.
21845 (dump_names_replaced_by): Likewise.
21846 (dump_update_ssa): Likewise.
21847 (update_ssa): Likewise.
21848 * tree-object-size.c (pass_object_sizes::execute): Likewise.
21849 * tree-parloops.c (build_new_reduction): Likewise.
21850 (try_create_reduction_list): Likewise.
21851 (ref_conflicts_with_region): Likewise.
21852 (oacc_entry_exit_ok_1): Likewise.
21853 (oacc_entry_exit_single_gang): Likewise.
21854 * tree-pretty-print.h: Likewise.
21855 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
21856 (get_scalar_evolution): Likewise.
21857 (add_to_evolution): Likewise.
21858 (get_loop_exit_condition): Likewise.
21859 (analyze_evolution_in_loop): Likewise.
21860 (analyze_initial_condition): Likewise.
21861 (analyze_scalar_evolution): Likewise.
21862 (instantiate_scev): Likewise.
21863 (number_of_latch_executions): Likewise.
21864 (gather_chrec_stats): Likewise.
21865 (final_value_replacement_loop): Likewise.
21866 (scev_const_prop): Likewise.
21867 * tree-sra.c (dump_access): Likewise.
21868 (disqualify_candidate): Likewise.
21869 (create_access): Likewise.
21870 (reject): Likewise.
21871 (maybe_add_sra_candidate): Likewise.
21872 (create_access_replacement): Likewise.
21873 (analyze_access_subtree): Likewise.
21874 (analyze_all_variable_accesses): Likewise.
21875 (sra_modify_assign): Likewise.
21876 (initialize_constant_pool_replacements): Likewise.
21877 (find_param_candidates): Likewise.
21878 (decide_one_param_reduction): Likewise.
21879 (replace_removed_params_ssa_names): Likewise.
21880 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
21881 * tree-ssa-copy.c (dump_copy_of): Likewise.
21882 (copy_prop_visit_cond_stmt): Likewise.
21883 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
21884 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
21885 (record_equivalences_from_stmt): Likewise.
21886 * tree-ssa-dse.c (compute_trims): Likewise.
21887 (delete_dead_call): Likewise.
21888 (delete_dead_assignment): Likewise.
21889 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
21890 (forward_propagate_into_cond): Likewise.
21891 (pass_forwprop::execute): Likewise.
21892 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
21893 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
21895 (move_computations_worker): Likewise.
21896 (execute_sm): Likewise.
21897 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
21898 (remove_exits_and_undefined_stmts): Likewise.
21899 (remove_redundant_iv_tests): Likewise.
21900 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
21901 (adjust_iv_update_pos): Likewise.
21902 * tree-ssa-math-opts.c (bswap_replace): Likewise.
21903 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
21904 (value_replacement): Likewise.
21905 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
21906 * tree-ssa-pre.c (print_pre_expr): Likewise.
21907 (get_representative_for): Likewise.
21908 (create_expression_by_pieces): Likewise.
21909 (insert_into_preds_of_block): Likewise.
21910 (eliminate_insert): Likewise.
21911 (eliminate_dom_walker::before_dom_children): Likewise.
21912 (eliminate): Likewise.
21913 (remove_dead_inserted_code): Likewise.
21914 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
21915 * tree-ssa-reassoc.c (get_rank): Likewise.
21916 (eliminate_duplicate_pair): Likewise.
21917 (eliminate_plus_minus_pair): Likewise.
21918 (eliminate_not_pairs): Likewise.
21919 (undistribute_ops_list): Likewise.
21920 (eliminate_redundant_comparison): Likewise.
21921 (update_range_test): Likewise.
21922 (optimize_range_tests_var_bound): Likewise.
21923 (optimize_vec_cond_expr): Likewise.
21924 (rewrite_expr_tree): Likewise.
21925 (rewrite_expr_tree_parallel): Likewise.
21926 (linearize_expr): Likewise.
21927 (break_up_subtract): Likewise.
21928 (linearize_expr_tree): Likewise.
21929 (attempt_builtin_powi): Likewise.
21930 (attempt_builtin_copysign): Likewise.
21931 (transform_stmt_to_copy): Likewise.
21932 (transform_stmt_to_multiply): Likewise.
21933 (dump_ops_vector): Likewise.
21934 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
21935 (print_scc): Likewise.
21936 (set_ssa_val_to): Likewise.
21937 (visit_reference_op_store): Likewise.
21938 (visit_use): Likewise.
21939 (sccvn_dom_walker::before_dom_children): Likewise.
21940 (run_scc_vn): Likewise.
21941 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
21943 (expr_hash_elt::print): Likewise.
21944 (const_and_copies::pop_to_marker): Likewise.
21945 (const_and_copies::record_const_or_copy_raw): Likewise.
21946 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
21947 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
21948 (dump_predicates): Likewise.
21949 (find_uninit_use): Likewise.
21950 (warn_uninitialized_phi): Likewise.
21951 (pass_late_warn_uninitialized::execute): Likewise.
21952 * tree-ssa.c (verify_vssa): Likewise.
21953 (verify_ssa): Likewise.
21954 (maybe_optimize_var): Likewise.
21955 * tree-vrp.c (dump_value_range): Likewise.
21956 (dump_all_value_ranges): Likewise.
21957 (dump_asserts_for): Likewise.
21958 (register_edge_assert_for_2): Likewise.
21959 (vrp_visit_cond_stmt): Likewise.
21960 (vrp_visit_switch_stmt): Likewise.
21961 (vrp_visit_stmt): Likewise.
21962 (vrp_visit_phi_node): Likewise.
21963 (simplify_cond_using_ranges_1): Likewise.
21964 (fold_predicate_in): Likewise.
21965 (evrp_dom_walker::before_dom_children): Likewise.
21966 (evrp_dom_walker::push_value_range): Likewise.
21967 (evrp_dom_walker::pop_value_range): Likewise.
21968 (execute_early_vrp): Likewise.
21970 2017-05-16 Richard Biener <rguenther@suse.de>
21972 * dwarf2out.c (loc_list_from_tree_1): Do not create
21973 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
21975 2017-05-16 Richard Biener <rguenther@suse.de>
21977 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
21979 (note_variable_value_in_expr): If we resolved the decl ref
21980 do not push to the stack.
21982 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
21984 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
21985 operations in fast-math mode.
21986 (vaddq_f16): Likewise.
21987 (vmul_f16): Likewise.
21988 (vmulq_f16): Likewise.
21989 (vsub_f16): Likewise.
21990 (vsubq_f16): Likewise.
21991 * config/arm/neon.md (add<mode>3): New.
21993 (fma:<VH:mode>3): New. Also remove outdated comment.
21996 2017-05-16 Martin Liska <mliska@suse.cz>
22000 * ipa-devirt.c (warn_types_mismatch): Fix typo.
22001 (odr_types_equivalent_p): Likewise.
22003 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
22005 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
22007 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
22010 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
22011 non-interunit SSE move alternatives with '?'.
22012 (zero-extendsidi peephole2): New peephole to skip intermediate
22013 general register in SSE zero-extend sequence.
22015 2017-05-15 Jeff Law <law@redhat.com>
22017 * reorg.c (relax_delay_slots): Create a new variable to hold
22018 the temporary target rather than clobbering TARGET_LABEL.
22020 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
22021 missing argument to extract_bit_field call.
22022 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
22024 2017-05-15 Martin Liska <mliska@suse.cz>
22027 * gcc.c (process_command): Do not allow empty argument of -o option.
22029 2017-05-15 Renlin Li <renlin.li@arm.com>
22031 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
22032 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
22033 * config/aarch64/constraints.md (Usf): Add long call check.
22034 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
22035 (call_value): Likewise.
22036 (sibcall): Likewise.
22037 (sibcall_value): Likewise.
22039 (call_value_insn): New.
22040 (sibcall_insn): Update rtx pattern.
22041 (sibcall_value_insn): Likewise.
22042 (call_internal): Remove.
22043 (call_value_internal): Likewise.
22044 (sibcall_internal): Likewise.
22045 (sibcall_value_internal): Likewise.
22046 (call_reg): Likewise.
22047 (call_symbol): Likewise.
22048 (call_value_reg): Likewise.
22049 (call_value_symbol): Likewise.
22051 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
22054 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
22056 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
22058 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
22059 compatible with CCGOCmode and with CCZmode.
22061 2017-05-14 Martin Sebor <msebor@redhat.com>
22063 PR middle-end/77671
22064 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
22065 (gimple_fold_builtin_snprintf): Same.
22066 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
22067 (gimple_fold_builtin_snprintf): Same.
22068 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
22069 of character types.
22070 (is_call_safe): New function.
22071 (try_substitute_return_value): Call it.
22072 (try_simplify_call): New function.
22073 (pass_sprintf_length::handle_gimple_call): Call it.
22075 2017-05-14 Martin Sebor <msebor@redhat.com>
22077 PR middle-end/80669
22078 * builtins.c (expand_builtin_stpncpy): Simplify.
22080 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
22082 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
22083 * config/i386/i386.h
22084 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
22085 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
22086 (struct machine_function): Add new members call_ms2sysv,
22087 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
22088 (struct machine_frame_state): New fields sp_realigned and
22089 sp_realigned_offset.
22090 * config/i386/i386.c
22091 (enum xlogue_stub): New enum.
22092 (enum xlogue_stub_sets): New enum.
22093 (class xlogue_layout): New class.
22094 (struct ix86_frame): New fields stack_realign_allocate_offset,
22095 stack_realign_offset and outlined_save_offset. Modify comments to
22096 detail stack layout when using out-of-line stubs.
22097 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
22098 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
22099 -mcall-ms2sysv-xlogues.
22100 (stub_managed_regs): New static variable.
22101 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
22102 registers managed by out-of-line stub.
22103 (disable_call_ms2sysv_xlogues): New function.
22104 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
22105 m->call_ms2sysv when appropriate and compute frame layout for
22107 (sp_valid_at, fp_valid_at): New inline functions.
22108 (choose_basereg): New function.
22109 (choose_baseaddr): Add align parameter, use choose_basereg and modify
22111 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
22112 Use align parameter of choose_baseaddr to generated aligned SSE movs
22114 (pro_epilogue_adjust_stack): Modify to track
22115 machine_frame_state::sp_realigned.
22116 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
22117 (ix86_nsaved_sseregs): Likewise.
22118 (ix86_emit_save_regs): Likewise.
22119 (ix86_emit_save_regs_using_mov): Likewise.
22120 (ix86_emit_save_sse_regs_using_mov): Likewise.
22121 (get_scratch_register_on_entry): Likewise.
22122 (gen_frame_set): New function.
22123 (gen_frame_load): Likewise.
22124 (gen_frame_store): Likewise.
22125 (emit_outlined_ms2sysv_save): Likewise.
22126 (emit_outlined_ms2sysv_restore): Likewise.
22127 (ix86_expand_prologue): Modify stack re-alignment code and call
22128 emit_outlined_ms2sysv_save when appropriate.
22129 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
22130 parameter rtx_insn *insn, which allows the function to be used to only
22131 generate the notes.
22132 (ix86_expand_epilogue): Modify validity checks of frame and stack
22133 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
22134 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
22135 * config/i386/predicates.md
22136 (save_multiple): New predicate.
22137 (restore_multiple): Likewise.
22138 * config/i386/sse.md
22139 (save_multiple<mode>): New pattern.
22140 (save_multiple_realign<mode>): Likewise.
22141 (restore_multiple<mode>): Likewise.
22142 (restore_multiple_and_return<mode>): Likewise.
22143 (restore_multiple_leave_return<mode>): Likewise.
22144 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
22146 2017-05-14 Julia Koval <julia.koval@intel.com>
22148 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
22149 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
22150 (__builtin_ia32_xsetbv): New builtins.
22151 * config/i386/i386.c (ix86_expand_special_args_builtin):
22153 (ix86_expand_builtin): Special expand for new intrinsics.
22154 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
22155 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
22156 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
22158 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22160 * cfganal.c (inverted_post_order_compute): Change argument type
22162 * cfganal.h (inverted_post_order_compute): Adjust prototype.
22163 * df-core.c (rest_of_handle_df_initialize): Adjust.
22164 (rest_of_handle_df_finish): Likewise.
22165 (df_analyze_1): Likewise.
22166 (df_analyze): Likewise.
22167 (loop_inverted_post_order_compute): Change argument to be a vec *.
22168 (df_analyze_loop): Adjust.
22169 (df_get_n_blocks): Likewise.
22170 (df_get_postorder): Likewise.
22171 * df.h (struct df_d): Change field to be a vec.
22172 * lcm.c (compute_laterin): Adjust.
22173 (compute_available): Likewise.
22174 * lra-lives.c (lra_create_live_ranges_1): Likewise.
22175 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
22176 * tree-ssa-pre.c (compute_antic): Likewise.
22178 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22180 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
22181 (depth_first_search::depth_first_search): Change structure init
22182 function to this constructor.
22183 (depth_first_search::add_bb): Rename function to this member.
22184 (depth_first_search::execute): Likewise.
22185 (flow_dfs_compute_reverse_finish): Adjust.
22187 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22189 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
22190 (longest_simple_path): Likewise.
22191 * shrink-wrap.c (spread_components): Likewise.
22192 (disqualify_problematic_components): Likewise.
22193 (emit_common_heads_for_components): Likewise.
22194 (emit_common_tails_for_components): Likewise.
22195 (insert_prologue_epilogue_for_components): Likewise.
22197 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22199 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
22202 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22204 * df-core.c (df_set_blocks): Start using auto_bitmap.
22205 (df_compact_blocks): Likewise.
22206 * df-problems.c (df_rd_confluence_n): Likewise.
22207 * df-scan.c (df_insn_rescan_all): Likewise.
22208 (df_process_deferred_rescans): Likewise.
22209 (df_update_entry_block_defs): Likewise.
22210 (df_update_exit_block_uses): Likewise.
22211 (df_entry_block_bitmap_verify): Likewise.
22212 (df_exit_block_bitmap_verify): Likewise.
22213 (df_scan_verify): Likewise.
22214 * lra-constraints.c (lra_constraints): Likewise.
22215 (undo_optional_reloads): Likewise.
22216 (lra_undo_inheritance): Likewise.
22217 * lra-remat.c (calculate_gen_cands): Likewise.
22218 (do_remat): Likewise.
22219 * lra-spills.c (assign_spill_hard_regs): Likewise.
22220 (spill_pseudos): Likewise.
22221 * tree-ssa-pre.c (bitmap_set_and): Likewise.
22222 (bitmap_set_subtract_values): Likewise.
22224 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22226 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
22227 management with auto_bitmap.
22228 (fix_inter_tick): Likewise.
22229 (fix_recovery_deps): Likewise.
22230 * ira.c (add_store_equivs): Likewise.
22231 (find_moveable_pseudos): Likewise.
22232 (split_live_ranges_for_shrink_wrap): Likewise.
22233 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
22234 (rtx_reuse_manager::seen_def_p): Likewise.
22235 (rtx_reuse_manager::set_seen_def): Likewise.
22236 * print-rtl.h (class rtx_reuse_manager): Likewise.
22238 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22240 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
22242 (migrate_btr_def): Likewise.
22243 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
22244 * df-core.c (loop_post_order_compute): Likewise.
22245 (loop_inverted_post_order_compute): Likewise.
22246 * hsa-common.h: Likewise.
22247 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
22248 * init-regs.c (initialize_uninitialized_regs): Likewise.
22249 * ipa-inline.c (resolve_noninline_speculation): Likewise.
22250 (inline_small_functions): Likewise.
22251 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
22252 * ira.c (combine_and_move_insns): Likewise.
22253 (build_insn_chain): Likewise.
22254 * loop-invariant.c (find_invariants): Likewise.
22255 * lower-subreg.c (propagate_pseudo_copies): Likewise.
22256 * predict.c (tree_predict_by_opcode): Likewise.
22257 (predict_paths_leading_to): Likewise.
22258 (predict_paths_leading_to_edge): Likewise.
22259 (estimate_loops_at_level): Likewise.
22260 (estimate_loops): Likewise.
22261 * shrink-wrap.c (try_shrink_wrapping): Likewise.
22262 (spread_components): Likewise.
22263 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
22264 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
22265 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
22266 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
22267 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
22268 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
22269 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
22270 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
22271 (mark_threaded_blocks): Likewise.
22272 (thread_through_all_blocks): Likewise.
22273 * tree-ssa.c (verify_ssa): Likewise.
22274 (execute_update_addresses_taken): Likewise.
22275 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
22277 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22279 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
22281 (post_order_compute): Likewise.
22282 (inverted_post_order_compute): Likewise.
22283 (pre_and_rev_post_order_compute_fn): Likewise.
22285 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22287 * genrecog.c (int_set::int_set): Explicitly construct our
22288 auto_vec base class.
22289 * vec.h (auto_vec::auto_vec): New constructor.
22291 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22293 * bitmap.h (class auto_bitmap): New constructor taking
22294 bitmap_obstack * argument.
22296 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
22298 * bitmap.h (class auto_bitmap): Change type of m_bits to
22299 bitmap_head, and adjust ctor / dtor and member operators.
22301 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
22303 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
22304 when returned register mode doesn't match original mode.
22306 2017-05-12 Jeff Law <law@redhat.com>
22307 Jakub Jelinek <jakub@redhat.com>
22309 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
22310 we look for cc setter after the compare-elim changes.
22311 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
22312 within the vector to match what compare-elim now expects.
22313 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
22314 (xorsi3_flags, one_cmplsi2_flags): Likewise.
22316 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
22317 after the compare-elim changes.
22318 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
22319 the vector to match what compare-elim now expects.
22320 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
22321 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
22322 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
22323 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
22324 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
22326 * config/visium/visium.c (single_set_and_flags): Fix where
22327 we look for cc setter after the compare-elim changes.
22328 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
22329 with the vector to match what compare-elim now expects.
22330 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
22331 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
22332 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
22333 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
22334 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
22335 (neg<mode>2_insn_set_overflow): Likewise.
22337 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
22339 PR middle-end/79794
22340 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
22341 maybe_expand_insn call, set ops[0].target. If still set after call,
22342 set alt_rtl. Add extra arg to recursive calls.
22343 (extract_bit_field): Add alt_rtl argument. Pass to
22345 * expmed.h (extract_bit_field): Fix prototype.
22346 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
22347 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
22348 to extract_bit_field_calls.
22349 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
22350 Pass alt_rtl to extract_bit_field calls.
22351 * calls.c (store_unaligned_arguments_into_psuedos)
22352 load_register_parameters): Pass extra NULL to extract_bit_field calls.
22353 * optabs.c (maybe_legitimize_operand): Clear op->target when call
22355 * optabs.h (struct expand_operand): Add target bitfield.
22357 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22359 * compare-elim.c (try_eliminate_compare): Canonicalize
22360 operation with embedded compare to
22361 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
22362 (set (reg) (operation)].
22364 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
22366 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
22369 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
22370 cost of adding a carry flag for ADC instruction.
22371 [case MINUS]: Ignore the cost of subtracting a carry flag
22372 for SBB instruction.
22374 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
22376 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
22378 * config/rs6000/bmiintrin.h: New file.
22379 * config/rs6000/bmi2intrin.h: New file.
22380 * config/rs6000/x86intrin.h: New file.
22382 2017-05-12 Jeff Law <law@redhat.com>
22384 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
22387 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
22389 PR middle-end/80707
22390 * tree-cfg.c: Remove cfg edges of unreachable case statements.
22392 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22394 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22395 early expansion of vector divide builtins.
22396 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
22397 builtins identified as having unsigned arguments.
22399 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22401 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
22402 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
22403 expansion of vector logical operations (and, andc, or, xor,
22406 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22408 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
22409 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
22411 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
22413 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
22414 early GIMPLE expansion of vector multiplies.
22416 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
22418 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
22419 TARGET_HAVE_MOVT conditional.
22420 (movt splitter): Likewise.
22422 2017-05-12 Richard Biener <rguenther@suse.de>
22424 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
22425 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22426 Fold all stmts not inplace.
22428 2017-05-12 Richard Biener <rguenther@suse.de>
22430 PR tree-optimization/80713
22431 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
22432 inserted_exprs bit for not removed stmts.
22434 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
22436 PR middle-end/69921
22437 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
22438 parallelized" attribute for parallelized OpenACC kernels.
22439 * omp-offload.c (execute_oacc_device_lower): Use it.
22441 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
22442 Set "oacc kernels" attribute.
22443 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
22444 parameter. Adjust all users.
22445 (oacc_fn_attrib_kernels_p): Remove function.
22446 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
22447 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
22448 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
22449 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
22450 assert "oacc kernels" attribute is set.
22452 2017-05-11 Carl Love <cel@us.ibm.com>
22454 * config/rs6000/rs6000-c: Add support for built-in functions
22455 vector unsigned char vec_popcnt (vector signed char)
22456 vector unsigned char vec_popcnt (vector unsigned char)
22457 vector unsigned short vec_popcnt (vector signed short)
22458 vector unsigned short vec_popcnt (vector unsigned short)
22459 vector unsigned int vec_popcnt (vector signed int)
22460 vector unsigned int vec_popcnt (vector unsigned int)
22461 vector unsigned long long vec_popcnt (vector signed long long)
22462 vector unsigned long long vec_popcnt (vector unsigned long long)
22463 vector signed long long vec_slo (vector signed long long,
22464 vector signed char)
22465 vector signed long long vec_slo (vector signed long long,
22466 vector unsigned char)
22467 vector unsigned long long vec_slo (vector unsigned long long,
22468 vector signed char)
22469 vector unsigned long long vec_slo (vector unsigned long long,
22470 vector unsigned char)
22471 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
22472 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
22473 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
22474 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
22475 * doc/extend.texi: Update the built-in documentation file for the
22476 new built-in functions.
22478 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22480 * attribs.h (sorted_attr_string): Move machine independent
22481 functions for target clone support from the i386 port to common
22482 code. Rename ix86_function_versions to common_function_versions.
22483 Rename make_name to make_unique_name.
22484 (common_function_versions): Likewise.
22485 (make_unique_name): Likewise.
22486 (make_dispatcher_decl): Likewise.
22487 (is_function_default_version): Likewise.
22488 * attribs.c (attr_strcmp): Likewise.
22489 (sorted_attr_string): Likewise.
22490 (common_function_versions): Likewise.
22491 (make_unique_name): Likewise.
22492 (make_dispatcher_decl): Likewise.
22493 (is_function_default_version): Likewise.
22494 * config/i386/i386.c (attr_strcmp): Likewise.
22495 (sorted_attr_string): Likewise.
22496 (ix86_function_versions): Likewise.
22497 (make_name): Likewise.
22498 (make_dispatcher_decl): Likewise.
22499 (is_function_default_version): Likewise.
22500 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
22502 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22505 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
22506 Account for direct move costs for vec_construct of integer
22509 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
22512 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
22513 (UNSPEC_STX_ATOMIC): Ditto.
22514 (loaddi_via_sse): New insn.
22515 (storedi_via_sse): Ditto.
22516 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
22517 Update corresponding peephole2 patterns.
22518 (atomic_storedi_fpu): Ditto.
22520 2017-05-11 Julia Koval <julia.koval@intel.com>
22522 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
22523 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
22525 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
22526 (__builtin_ia32_rsqrt14ss_mask): New builtins.
22527 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
22529 2017-05-11 Nathan Sidwell <nathan@acm.org>
22531 * graphite-poly.c: Include dumpfile.h.
22533 * dumpfle.h (dump_function): Declare here ...
22534 * tree-dump.h (dump_function): ... not here.
22535 * dumpfile.c: #include tree-cfg.h.
22536 (dump_function): Move here from ...
22537 * tree-dump.c (dump_function): ... here.
22538 * gimplify.c: #include splay-tree.h, not tree-dump.h.
22539 * graphite-poly.c: Don't include tree-dump.h.
22540 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
22541 * print-tree.c: Likewise.
22542 * stor-layout.c: Likewise.
22543 * tree-nested.c: Likewise.
22545 * dumpfile.c (dump_start): Use TDF_FLAGS.
22546 (dump_enable_all): Fix TDF_KIND check thinko.
22548 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
22550 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
22551 array entries to represent two legal parameterizations of the
22552 overloaded __builtin_cmpb function, as represented by the
22553 P6_OV_BUILTIN_CMPB constant.
22554 (altivec_resolve_overloaded_builtin): Add special case handling
22555 for the __builtin_cmpb function, as represented by the
22556 P6_OV_BUILTIN_CMPB constant.
22557 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
22558 (BU_P6_64BIT_2): New macro.
22559 (BU_P6_OVERLOAD_2): New macro
22560 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
22561 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
22562 (CMPB): Add overload support to represent both 32-bit and 64-bit
22563 compare-bytes function.
22564 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
22565 support for TARGET_CMPB.
22566 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
22567 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
22568 documentation of the __builtin_cmpb overloaded built-in function.
22570 2017-05-11 Richard Biener <rguenther@suse.de>
22572 PR tree-optimization/80705
22573 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
22574 bases are not vectorizable.
22576 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22578 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
22579 when counting register pressure.
22581 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22583 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
22584 (iv_ca_more_deps): Renamed to ...
22585 (iv_ca_compare_deps): ... this.
22586 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
22588 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22590 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
22592 (determine_group_iv_costs): ... here.
22593 (find_inv_vars_cb): Record inv var if it's not recorded before.
22595 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22597 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
22598 (get_shiftadd_cost): Ditto.
22600 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22602 * tree-ssa-address.c: Include header file.
22603 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
22605 (add_to_parts): Refactor.
22606 (addr_to_parts): New parameter. Update use of move_hint_to_base.
22607 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
22610 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22612 PR tree-optimization/53090
22613 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
22615 (extract_cond_operands): Detect condition with IV on both sides
22616 and return COMP_IV_EXPR_2.
22617 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
22618 (rewrite_use_compare): Simplify by removing call to function
22619 extract_cond_operands.
22621 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22623 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
22624 (extract_cond_operands): Detect condition comparing against non-
22625 invariant bound and return appropriate enum value.
22626 (find_interesting_uses_cond): Update use of extract_cond_operands.
22627 Handle its return value accordingly.
22628 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
22630 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22632 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
22633 nonlinear iv_use computation in loop invariant sensitive way.
22635 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22637 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
22638 (find_iv_candidates): Call relate_compare_use_with_all_cands.
22640 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22642 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
22643 (dump_cand): Support iv_cand.inv_exprs.
22644 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
22646 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
22649 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22651 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
22653 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
22654 as local function. Include necessary header files.
22655 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
22657 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22659 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
22661 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22663 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
22664 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
22665 RSHIFT_EXPR and BIT_NOT_EXPR.
22667 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22669 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
22670 (adjust_setup_cost): New parameter supporting round up adjustment.
22671 (struct address_cost_data): Delete.
22672 (force_expr_to_var_cost): Don't bound cost with spill_cost.
22673 (split_address_cost, ptr_difference_cost): Delete.
22674 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
22675 (struct ainc_cost_data): New struct.
22676 (get_address_cost_ainc): New function.
22677 (get_address_cost, get_computation_cost): Reimplement.
22678 (determine_group_iv_cost_address): Record inv_expr for all uses of
22680 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
22681 (iv_ca_has_deps): Reimplemented to ...
22682 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
22684 (iv_ca_extend): Call iv_ca_more_deps.
22686 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22688 * tree-ssa-address.c (struct mem_address): Move to header file.
22689 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
22690 * tree-ssa-address.h (struct mem_address): Move from C file.
22691 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
22693 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22695 * tree-affine.h (aff_combination_type): New interface.
22696 (aff_combination_zero_p): Remove static.
22697 (aff_combination_const_p): New interface.
22698 (aff_combination_singleton_var_p): New interfaces.
22700 2017-05-11 Richard Biener <rguenther@suse.de>
22702 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
22703 Skip unreachable blocks and destinations.
22704 (eliminate): Move stmt removal and fixup ...
22705 (fini_eliminate): ... here. Skip inserted exprs.
22706 (pass_pre::execute): Move fini_pre after fini_eliminate.
22707 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
22708 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
22709 PRE to get rid of dead code that has invalid SSA form and
22710 split critical edges again.
22712 2017-05-11 Bin Cheng <bin.cheng@arm.com>
22714 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
22716 2017-05-11 Richard Biener <rguenther@suse.de>
22718 * passes.c (execute_function_todo): Verify loops if they are
22719 said to be up-to-date.
22720 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
22721 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
22723 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
22726 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
22727 handle calling assemble_external ourself.
22730 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
22731 modes with zero size. Enhance comment.
22733 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22735 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
22736 built-ins for vec_xl and vec_xst with short and char pointer
22739 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22741 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
22742 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
22743 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
22744 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
22745 (_mm_maskz_min_round_ss): New intrinsics.
22746 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
22747 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
22748 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
22749 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
22750 (__builtin_ia32_minss_mask_round): New builtins.
22751 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22752 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22753 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
22755 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
22756 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
22758 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
22761 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
22763 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
22764 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
22765 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
22766 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
22767 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
22768 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
22769 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
22770 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22771 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
22772 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
22773 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
22774 (__builtin_ia32_mulss_mask_round): New builtins.
22775 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
22776 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
22777 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
22779 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
22780 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
22782 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
22785 2017-05-10 Julia Koval <julia.koval@intel.com>
22787 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
22788 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
22789 (_mm256_setr_m128i): New intrinsics.
22791 2017-05-10 Julia Koval <julia.koval@intel.com>
22793 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
22794 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
22795 (_mm_maskz_rcp14_ss): New intrinsics.
22796 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
22797 (__builtin_ia32_rcp14ss_mask): New builtins.
22798 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
22800 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
22802 PR tree-optimization/51513
22803 * tree-cfg.c (gimple_seq_unreachable_p): New function.
22804 (assert_unreachable_fallthru_edge_p): Use it.
22805 (group_case_labels_stmt): Likewise.
22806 * tree-cfg.h: Prototype it.
22807 * stmt.c: Include cfghooks.h and tree-cfg.h.
22808 (emit_case_dispatch_table) <gap_label>: New local variable.
22809 Use it to fill dispatch table gaps.
22810 Test for default_label before updating probabilities.
22811 (expand_case) <default_label>: Remove unneeded initialization.
22812 Test for unreachable default case statement and remove its edge.
22813 Set default_label accordingly.
22814 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
22816 2017-05-10 Carl Love <cel@us.ibm.com>
22818 * config/rs6000/rs6000-c: Add support for built-in functions
22819 vector signed char vec_neg (vector signed char)
22820 vector signed short int vec_neg (vector short int)
22821 vector signed int vec_neg (vector signed int)
22822 vector signed long long vec_neg (vector signed long long)
22823 vector float vec_neg (vector float)
22824 vector double vec_neg (vector double)
22825 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
22827 * config/rs6000/altivec.h: Add define for vec_neg
22828 * doc/extend.texi: Update the built-in documentation for the
22829 new built-in functions.
22831 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22833 PR tree-optimization/77644
22834 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
22836 2017-05-10 Nathan Sidwell <nathan@acm.org>
22838 * dumpfile.h (TDI_lang_all): New.
22839 (TDF_KIND): New. Renumber others
22840 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
22842 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
22844 (get_dump_file_name): Adjust suffix generation.
22845 (dump_enable_all): Use TDF_KIND.
22846 * doc/invoke.texi (-fdump-lang-all): Document.
22848 * dumpfile.h: Tabify.
22850 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
22853 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
22854 Move member access before delete.
22856 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
22858 * tree-inline.c (expand_call_inline): Split block at stmt
22861 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
22864 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
22865 are now unused after splitting mov{sf,sd}_hardfloat.
22866 (f32_lr2): Likewise.
22867 (f32_lm): Likewise.
22868 (f32_lm2): Likewise.
22869 (f32_li): Likewise.
22870 (f32_li2): Likewise.
22871 (f32_lv): Likewise.
22872 (f32_sr): Likewise.
22873 (f32_sr2): Likewise.
22874 (f32_sm): Likewise.
22875 (f32_sm2): Likewise.
22876 (f32_si): Likewise.
22877 (f32_si2): Likewise.
22878 (f32_sv): Likewise.
22879 (f32_dm): Likewise.
22880 (f32_vsx): Likewise.
22881 (f32_av): Likewise.
22882 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
22883 For movsf, order stores so the VSX stores occur before the GPR
22884 store which encourages the register allocator to use a traditional
22885 FPR instead of a GPR. For movsd, order the stores so that the GPR
22886 store comes before the VSX stores to allow the power6 to work.
22887 This is due to the power6 not having a 32-bit integer store
22888 instruction from a FPR.
22889 (movsf_hardfloat): Likewise.
22890 (movsd_hardfloat): Likewise.
22892 2017-05-09 Martin Sebor <msebor@redhat.com>
22894 PR translation/80280
22895 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
22898 PR translation/80280
22899 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
22900 data member added in r247778.
22901 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
22903 2017-05-09 Nathan Sidwell <nathan@acm.org>
22905 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
22907 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
22910 2017-05-09 Marek Polacek <polacek@redhat.com>
22912 * doc/invoke.texi: Fix typo.
22914 2017-05-09 Richard Biener <rguenther@suse.de>
22916 * tree-vrp.c (vrp_val_is_max): Adjust comment.
22917 (vrp_val_is_min): Likewise.
22918 (set_value_range_to_value): Likewise.
22919 (set_value_range_to_nonnegative): Likewise.
22920 (gimple_assign_nonzero_p): Likewise.
22921 (gimple_stmt_nonzero_p): Likewise.
22922 (vrp_int_const_binop): Likewise. Remove unreachable case.
22923 (adjust_range_with_scev): Adjust comments.
22924 (compare_range_with_value): Likewise.
22925 (extract_range_from_phi_node): Likewise.
22926 (test_for_singularity): Likewise.
22928 2017-05-09 Richard Biener <rguenther@suse.de>
22930 * tree-vrp.c (get_single_symbol): Add assert that we don't
22931 get overflowed constants as invariant part.
22932 (compare_values_warnv): Add comment before the TREE_NO_WARNING
22933 checks. Use wi::cmp instead of recursing for integer constants.
22934 (compare_values): Just ignore whether we assumed undefined
22935 overflow instead of failing the compare.
22936 (extract_range_for_var_from_comparison_expr): Add comment before the
22937 TREE_NO_WARNING sets.
22938 (test_for_singularity): Likewise.
22939 (extract_range_from_comparison): Do not disable optimization
22940 when we assumed undefined overflow.
22941 (extract_range_basic): Remove init of unused var.
22943 2017-05-09 Richard Biener <rguenther@suse.de>
22945 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
22946 (extract_range_from_multiplicative_op_1): Adjust.
22947 (extract_range_from_binary_expr_1): Use int_const_binop.
22949 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
22952 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
22953 rs6000_store_data_bypass_p in seven define_bypass directives and
22954 in several comments.
22955 * config/rs6000/rs6000-protos.h: Add prototype for
22956 rs6000_store_data_bypass_p function.
22957 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
22958 function implements slightly different (rs6000-specific) semantics
22959 than store_data_bypass_p, returning false rather than aborting
22960 with assertion error when arguments do not satisfy the
22961 requirements of store data bypass.
22962 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
22963 rs6000_store_data_bypass_p.
22965 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
22967 * config/xtensa/xtensa-protos.h
22968 (xtensa_initial_elimination_offset): New declaration.
22969 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
22970 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
22971 macro definition, add case for FRAME_POINTER_REGNUM when
22972 FRAME_GROWS_DOWNWARD.
22973 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
22974 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
22975 xtensa_initial_elimination_offset.
22977 2017-05-08 Nathan Sidwell <nathan@acm.org>
22979 * doc/invoke.texi: Alphabetize -fdump options.
22981 2017-05-08 Martin Sebor <msebor@redhat.com>
22983 PR translation/80280
22984 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
22986 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
22988 * target.def (compute_frame_layout): New optional target hook.
22989 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
22990 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
22991 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
22993 * reload1.c (verify_initial_elim_offsets): Likewise.
22994 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
22995 (use_simple_return_p): Call arm_compute_frame_layout if needed.
22996 (arm_get_frame_offsets): Split up into this ...
22997 (arm_compute_frame_layout): ... and this function.
22999 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
23001 * config/aarch64/constraints.md (Usa): New constraint.
23002 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
23004 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
23006 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
23007 with_multilib_list after it has been checked.
23009 2017-05-08 Richard Biener <rguenther@suse.de>
23011 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
23012 (bitmap_set_subtract_values): Likewise.
23014 2017-05-08 Richard Biener <rguenther@suse.de>
23016 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
23017 (gimple_assign_nonzero): ... this and remove strict_overflow_p
23019 (gimple_stmt_nonzero_warnv_p): Rename to ...
23020 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
23022 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
23023 (extract_range_basic): Adjust, do not disable propagation on
23024 strict overflow sensitive simplification.
23025 (vrp_visit_cond_stmt): Likewise.
23027 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
23029 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
23030 body size unconditionally.
23032 2017-05-07 Jeff Law <law@redhat.com>
23035 2017-05-06 Jeff Law <law@redhat.com>
23036 PR tree-optimization/78496
23037 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23040 PR tree-optimization/78496
23041 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23042 (simplify_stmt_using_ranges): Call it.
23043 (vrp_dom_walker::before_dom_children): Extract equivalences
23044 from an ASSERT_EXPR with an equality comparison against a
23047 2017-05-06 Jeff Law <law@redhat.com>
23049 PR tree-optimization/78496
23050 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
23053 PR tree-optimization/78496
23054 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
23055 (simplify_stmt_using_ranges): Call it.
23056 (vrp_dom_walker::before_dom_children): Extract equivalences
23057 from an ASSERT_EXPR with an equality comparison against a
23060 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23062 * lra-constraints.c (lra_copy_reg_equiv): New function.
23063 (split_reg): Use it to copy equivalence information from the
23064 original register to the spill register.
23066 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
23068 PR rtl-optimization/75964
23069 * simplify-rtx.c (simplify_const_relational_operation): Remove
23070 invalid handling of comparisons of integer ABS.
23072 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
23074 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
23075 initialize to zero.
23076 (init_regs): Remove declaration.
23077 (function_arg_advance_32): Initialize error_p as boolean variable.
23079 2017-05-05 Nathan Sidwell <nathan@acm.org>
23081 * store-motion.c (remove_reachable_equiv_notes): Reformat long
23082 lines. Use for (;;).
23084 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23086 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
23087 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
23088 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
23089 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
23090 VF=2 that require versioning.
23092 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23094 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
23097 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23099 * diagnostic.h (diagnostic_override_option_index): Convert from
23100 macro to inline function.
23102 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23104 * diagnostic.c (last_module_changed_p): New function.
23105 (set_last_module): New function.
23106 (diagnostic_report_current_module): Convert macro usage to
23107 the above functions.
23108 * diagnostic.h (diagnostic_context::last_module): Strengthen
23109 from const line_map * to const line_map_ordinary *.
23110 (diagnostic_last_module_changed): Delete macro.
23111 (diagnostic_set_last_module): Delete macro.
23113 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23115 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
23116 with diagnostic_report_diagnostic.
23117 (diagnostic_n_impl_richloc): Likewise.
23118 * diagnostic.h (report_diagnostic): Delete macro.
23119 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
23120 with diagnostic_report_diagnostic.
23121 * substring-locations.c (format_warning_va): Likewise.
23123 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23125 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
23126 save/restor of format_spec. Move option-printing code to...
23127 (print_option_information): ...this new function, and
23128 reimplement by simply printing to the pretty_printer,
23129 rather than appending to the format string.
23131 2017-05-05 David Malcolm <dmalcolm@redhat.com>
23133 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
23134 handling logic into...
23135 (update_effective_level_from_pragmas): ...this new function.
23137 2017-05-04 Andrew Waterman <andrew@sifive.com>
23139 * config/riscv/riscv.opt (mstrict-align): New option.
23140 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
23141 (SLOW_UNALIGNED_ACCESS): Define.
23142 (riscv_slow_unaligned_access): Declare.
23143 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
23145 (riscv_slow_unaligned_access): New variable.
23146 (rocket_tune_info): Set slow_unaligned_access to true.
23147 (optimize_size_tune_info): Set slow_unaligned_access to false.
23148 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
23149 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
23150 (riscv_option_override): Set riscv_slow_unaligned_access.
23151 * doc/invoke.texi: Add -mstrict-align to RISC-V.
23153 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
23155 * config/riscv/riscv.md: Unify indentation.
23157 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
23162 * config/rs6000/rs6000.md (u code attribute): Add FIX and
23164 (extendsi<mode>2): Add support for doing sign extension via
23165 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
23166 don't have ISA 3.0 instructions.
23167 (extendsi<mode>2 splitter): Likewise.
23168 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
23169 generate the normal insns since SImode can now go in vector
23170 registers. Disallow the special UNSPECs needed for previous
23171 machines to hide SImode being used. Add new insns
23172 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
23173 (fix_trunc<mode>si2_stfiwx): Likewise.
23174 (fix_trunc<mode>si2_internal): Likewise.
23175 (fixuns_trunc<mode>si2): Likewise.
23176 (fixuns_trunc<mode>si2_stfiwx): Likewise.
23177 (fctiw<u>z_<mode>_smallint): Likewise.
23178 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
23179 of floating point to 32-bit integer from doing a direct move to
23180 the GPR registers to do a store.
23181 (fctiwz_<mode>): Break long line.
23183 2017-05-05 Bin Cheng <bin.cheng@arm.com>
23185 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
23186 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
23187 (addr_list, addr_offset_valid_p): New.
23188 (split_address_groups): Check offset validity with above function.
23189 (gt-tree-ssa-loop-ivopts.h): Include header file.
23191 2017-05-05 Nathan Sidwell <nathan@acm.org>
23193 * config.gcc (arm*-*-*): Add missing 'fi'.
23195 2017-05-05 Steve Ellcey <sellcey@cavium.com>
23197 * doc/invoke.texi (-fopt-info): Explicitly say order of options
23198 included in -fopt-info does not matter.
23199 * doc/optinfo.texi (-fopt-info): Fix description of default
23200 behavour. Explicitly say order of options included in -fopt-info
23203 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
23205 * config.gcc: Allow combinations of aprofile and rmprofile values for
23206 --with-multilib-list.
23207 * config/arm/t-multilib: New file.
23208 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
23209 variables. Remove setting of ISA and floating-point ABI in
23210 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
23211 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
23212 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
23213 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
23215 * config/arm/t-rmprofile: Likewise except for the matches changes.
23216 * doc/install.texi (--with-multilib-list): Document the combination of
23217 aprofile and rmprofile values and warn about pitfalls in doing that.
23219 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23221 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
23222 (movdi_aarch64): Likewise.
23224 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23226 PR tree-optimization/80632
23227 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
23229 (build_arrays): Initialize it for virtual phis.
23230 (fix_phi_nodes): Use it for virtual phis.
23232 PR tree-optimization/80558
23233 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
23234 [x, y] op z into [x op, y op z] for op & or | if conditions
23237 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
23238 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23241 * config/arm/arm.md (use_literal_pool): Remove.
23242 (64-bit immediate split): No longer takes cost into consideration
23243 if arm_disable_literal_pool is enabled.
23244 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
23245 used when arm_disable_literal_pool is enabled.
23246 (arm_max_const_double_inline_cost): Remove use of
23247 arm_disable_literal_pool.
23248 (push_minipool_fix): Add assert.
23249 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
23250 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
23251 (no_literal_pool_sf_immediate): New.
23253 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23255 PR tree-optimization/80613
23256 * tree-ssa-dce.c (propagate_necessity): Remove cases for
23257 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
23259 2017-05-05 Richard Biener <rguenther@suse.de>
23261 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
23263 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
23265 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
23266 of this flag from insn conditions due to removal from r247495.
23268 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
23270 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
23272 (arm_early_store_addr_dep_ptr): Likewise.
23273 * config/arm/aarch-common-protos.h
23274 (arm_early_load_addr_dep_ptr): Add prototype.
23275 (arm_early_store_addr_dep_ptr): Likewise.
23276 * config/arm/cortex-a53.md: Add new bypasses.
23278 2017-05-05 Jakub Jelinek <jakub@redhat.com>
23280 * tree.c (next_type_uid): Change type to unsigned.
23281 (type_hash_canon): Decrement back next_type_uid if
23282 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
23283 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
23286 2017-05-04 Martin Sebor <msebor@redhat.com>
23288 * builtins.c: Fix a trivial typo in a comment.
23290 PR middle-end/79234
23291 * builtins.c (check_sizes): Adjust to handle reading past the end.
23292 Avoid printing excessive upper bound of ranges. Use %E to print
23293 tree nodes instead of converting them to %wu.
23294 (expand_builtin_memchr): New function.
23295 (compute_dest_size): Rename...
23296 (compute_objsize): ...to this.
23297 (expand_builtin_memcpy): Adjust.
23298 (expand_builtin_mempcpy): Adjust.
23299 (expand_builtin_strcat): Adjust.
23300 (expand_builtin_strcpy): Adjust.
23301 (check_strncat_sizes): Adjust.
23302 (expand_builtin_strncat): Adjust.
23303 (expand_builtin_strncpy): Adjust and simplify.
23304 (expand_builtin_memset): Adjust.
23305 (expand_builtin_bzero): Adjust.
23306 (expand_builtin_memcmp): Adjust.
23307 (expand_builtin): Handle memcmp.
23308 (maybe_emit_chk_warning): Check strncat just once.
23310 2017-05-04 Martin Sebor <msebor@redhat.com>
23312 PR preprocessor/79214
23313 PR middle-end/79222
23314 PR middle-end/79223
23315 * builtins.c (check_sizes): Add inlining context and issue
23316 warnings even when -Wno-system-headers is set.
23317 (check_strncat_sizes): Same.
23318 (expand_builtin_strncat): Same.
23319 (expand_builtin_memmove): New function.
23320 (expand_builtin_stpncpy): Same.
23321 (expand_builtin): Handle memmove and stpncpy.
23323 2017-05-04 Bin Cheng <bin.cheng@arm.com>
23325 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
23326 which is not used any more.
23328 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23330 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
23332 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23334 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
23335 (cortexa53_tunings): Likewise.
23336 (cortexa57_tunings): Likewise.
23337 (cortexa72_tunings): Likewise.
23338 (cortexa73_tunings): Likewise.
23340 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23342 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
23343 Set loop alignment to 8.
23345 2017-05-04 Martin Sebor <msebor@redhat.com>
23347 PR translation/80280
23348 * builtins.c (expand_builtin_object_size): Add missing quoting to
23349 %D and like directives.
23350 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
23351 (hsa_type_for_tree_type): Same.
23352 (verify_function_arguments): Same.
23353 * symtab.c (symbol_table::change_decl_assembler_name): Same.
23354 * varasm.c (get_section): Same.
23357 2017-05-04 Martin Sebor <msebor@redhat.com>
23359 PR translation/80280
23360 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
23362 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
23364 * config/aarch64/aarch64.c (generic_addrcost_table):
23365 Change HI/TI mode setting.
23367 2017-05-04 Martin Jambor <mjambor@suse.cz>
23369 PR tree-optimization/80622
23370 * tree-sra.c (comes_initialized_p): New function.
23371 (build_accesses_from_assign): Only set write lazily when
23372 comes_initialized_p is false.
23373 (analyze_access_subtree): Use comes_initialized_p.
23374 (propagate_subaccesses_across_link): Assert !comes_initialized_p
23375 instead of testing for PARM_DECL.
23377 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23379 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
23380 constraint on operand 0 to allow more general addressing modes.
23381 Adjust output template.
23382 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
23384 * config/aarch64/aarch64-protos.h
23385 (aarch64_address_valid_for_prefetch_p): Declare prototype.
23386 * config/aarch64/constraints.md (Dp): New address constraint.
23387 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
23390 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
23392 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
23393 update use of estimate_ipcp_clone_size_and_time.
23394 (estimate_local_effects): Update use of
23395 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
23396 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
23397 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
23398 Return nonspecialized time.
23400 2017-05-04 Richard Biener <rguenther@suse.de>
23402 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
23403 for the last VUSE which def dominates the PHI. Directly call
23405 (get_continuation_for_phi_1): Remove.
23407 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
23409 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
23410 to explain the use of truncating division. Cap the number of
23411 iterations to the maximum given by nb_iterations_upper_bound,
23414 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
23416 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
23417 * configure: Regenerate.
23418 * config.in: Regenerate.
23419 * config/i386/driver-mingw32.c: new file.
23420 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
23421 * config.host: Link driver-mingw32.o on MinGW host.
23422 * doc/install.texi: Document new --enable-mingw-wildcard configure
23425 2017-05-04 Marek Polacek <polacek@redhat.com>
23427 PR tree-optimization/80612
23428 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
23430 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23431 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
23433 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
23434 (movt splitter): Likewise.
23435 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
23436 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
23437 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
23438 block for Thumb-1 with MOVT.
23439 (thumb2_legitimate_address_p): Move code block ...
23440 (can_avoid_literal_pool_for_label_p): ... into this new function.
23441 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
23443 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
23444 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
23445 "M-profile targets with the MOVT instruction".
23447 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
23449 * config/arm/arm-builtins.c (arm_init_builtins): Rename
23450 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
23451 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
23453 2017-05-04 Martin Liska <mliska@suse.cz>
23455 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
23456 variable cond_code.
23458 2017-05-04 Richard Biener <rguenther@suse.de>
23460 * tree.c (array_at_struct_end_p): Handle arrays at struct
23461 end with flexarrays more conservatively. Refactor and treat
23462 arrays of arrays or aggregates more strict. Fix
23463 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
23464 * tree.c (array_at_struct_end_p): Adjust prototype.
23465 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
23466 * gimple-fold.c (get_range_strlen): Likewise.
23467 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
23469 2017-05-04 Richard Biener <rguenther@suse.de>
23471 PR tree-optimization/31130
23472 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
23474 (supports_overflow_infinity): Likewise.
23475 (is_negative_overflow_infinity): Likewise.
23476 (is_positive_overflow_infinity): Likewise.
23477 (is_overflow_infinity): Likewise.
23478 (stmt_overflow_infinity): Likewise.
23479 (overflow_infinity_range_p): Likewise.
23480 (usable_range_p): Remove as always returning true.
23481 (make_overflow_infinity): Remove.
23482 (negative_overflow_infinity): Likewise.
23483 (positive_overflow_infinity): Likewise.
23484 (avoid_overflow_infinity): Likewise.
23485 (set_value_range): Adjust accordingly.
23486 (set_value_range_to_nonnegative): Likewise, remove now unused
23487 overflow_infinity arg.
23488 (vrp_operand_equal_p): Adjust.
23489 (update_value_range): Likewise.
23490 (range_int_cst_singleton_p): Likewise.
23491 (operand_less_p): Likewise.
23492 (compare_values_warnv): Likewise.
23493 (extract_range_for_var_from_comparison_expr): Likewise.
23494 (vrp_int_const_binop): Likewise.
23495 (zero_nonzero_bits_from_vr): Likewise.
23496 (extract_range_from_multiplicative_op_1): Likewise.
23497 (extract_range_from_binary_expr_1): Likewise.
23498 (extract_range_from_unary_expr): Likewise.
23499 (extract_range_from_comparison): Likewise.
23500 (extract_range_basic): Likewise.
23501 (adjust_range_with_scev): Likewise.
23502 (compare_ranges): Likewise.
23503 (compare_range_with_value): Likewise.
23504 (dump_value_range): Likewise.
23505 (test_for_singularity): Likewise, remove strict_overflow_p parameter
23507 (simplify_cond_using_ranges): Adjust.
23509 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
23511 * brig-builtins.def: Added a builtin for class_f64.
23512 * builtin-types.def: Added a builtin type needed by class_f64.
23514 2017-05-03 Jason Merrill <jason@redhat.com>
23516 * timevar.def: Add TV_CONSTEXPR.
23518 2017-05-03 David Malcolm <dmalcolm@redhat.com>
23520 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
23522 2017-05-03 Martin Jambor <mjambor@suse.cz>
23524 * ipa-prop.c (ipa_update_after_lto_read): Removed.
23525 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
23526 * ipa-cp.c (ipcp_propagate_stage): Do not call
23527 ipa_update_after_lto_read.
23528 * ipa-inline.c (ipa_inline): Likewise.
23530 2017-05-03 Martin Jambor <mjambor@suse.cz>
23532 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
23533 tag. Added a default constructor and a destructor.
23534 (ipa_edge_args_sum_t): New class;
23535 (ipa_edge_args_sum): Declare.
23536 (ipa_edge_args_vector): Remove declaration.
23537 (IPA_EDGE_REF): Use ipa_edge_args_sum.
23538 (ipa_free_edge_args_substructures): Remove declaration.
23539 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
23540 (ipa_edge_args_info_available_for_edge_p): Likewise.
23541 * ipa-prop.c (ipa_edge_args_vector): Removed.
23542 (edge_removal_hook_holder): Likewise.
23543 (edge_duplication_hook_holder): Likewise.
23544 (ipa_edge_args_sum): New variable.
23545 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
23546 ipa_edge_args_vector.
23547 (ipa_free_edge_args_substructures): Likewise.
23548 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
23549 ipa_edge_args_vector.
23550 (ipa_edge_removal_hook): Turned into method
23551 ipa_edge_args_sum_t::remove.
23552 (ipa_edge_duplication_hook): Turned into method
23553 ipa_edge_args_sum_t::duplicate.
23554 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
23555 registering edge hooks.
23556 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
23557 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
23558 ipa_edge_args_sum instead of ipa_edge_args_vector.
23559 * ipa-profile.c (ipa_profile): Likewise.
23561 2017-05-03 Martin Jambor <mjambor@suse.cz>
23563 * symbol-summary.h (function_summary): New method exists.
23564 (function_summary::symtab_removal): Deallocate through release.
23565 (call_summary): New class.
23566 (gt_ggc_mx): New overload.
23567 (gt_pch_nx): Likewise.
23568 (gt_pch_nx): Likewise.
23570 2017-05-03 Jeff Law <law@redhat.com>
23572 PR tree-optimization/78496
23573 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
23574 from simplify_cond_using_ranges. Split off code to walk
23575 backwards through casts into ...
23576 (simplify_cond_using_ranges_2): New function.
23577 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
23578 (execute_vrp): After identifying jump threads, call
23579 simplify_cond_using_ranges_2.
23581 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
23584 * ipa-inline.h (inline_summary): Add ctor.
23585 (create_ggc): Do not use ggc_cleared_alloc.
23587 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
23588 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23590 * gcc.c (handle_braces): Support escaping in switch matching
23592 * doc/invoke.texi (Spec Files): Document it.
23593 Remove superfluous @code markup in items.
23595 2017-05-03 David Malcolm <dmalcolm@redhat.com>
23597 * diagnostic-show-locus.c (struct column_range): New struct.
23598 (get_affected_columns): New function.
23599 (get_printed_columns): New function.
23600 (struct correction): New struct.
23601 (correction::ensure_capacity): New function.
23602 (correction::ensure_terminated): New function.
23603 (struct line_corrections): New struct.
23604 (line_corrections::~line_corrections): New dtor.
23605 (line_corrections::add_hint): New function.
23606 (layout::print_trailing_fixits): Reimplement in terms of the new
23608 (selftest::test_overlapped_fixit_printing): New function.
23609 (selftest::diagnostic_show_locus_c_tests): Call it.
23611 2017-05-03 Nathan Sidwell <nathan@acm.org>
23613 Canonicalize canonical type hashing
23614 * tree.h (type_hash_canon_hash): Declare.
23615 * tree.c (type_hash_list, attribute_hash_list): Move into
23616 type_hash_canon_hash.
23617 (build_type_attribute_qual_variant): Break out hash code calc into
23618 type_hash_canon_hash.
23619 (type_hash_canon_hash): New. Generic type hash computation.
23620 (build_range_type_1, build_array_type_1, build_function_type,
23621 build_method_type_directly, build_offset_type, build_complex_type,
23622 make_vector_type): Call it.
23624 2017-05-03 Richard Biener <rguenther@suse.de>
23626 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
23627 When all DRs have unknown misaligned do not always peel
23628 when there is a store but apply the same costing model as if
23629 there were only loads.
23631 2017-05-03 Richard Biener <rguenther@suse.de>
23634 PR tree-optimization/80492
23635 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
23636 compare_base_decls returning dont-know properly.
23638 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
23640 * config/arm/iterators.md (CCSI): New mode iterator.
23641 (arch): New mode attribute.
23642 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
23643 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
23644 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
23645 code iterator for success result mode.
23646 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
23647 the corresponding new insn generators.
23649 2017-05-03 Bin Cheng <bin.cheng@arm.com>
23652 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23653 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23655 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
23657 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
23658 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
23659 (DDR_A): Wrap DDR argument in brackets.
23660 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
23661 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
23662 (DDR_REVERSED_P): Likewise.
23664 2017-05-03 Jakub Jelinek <jakub@redhat.com>
23666 PR tree-optimization/79472
23667 * tree-switch-conversion.c (struct switch_conv_info): Add
23668 contiguous_range and default_case_nonstandard fields.
23669 (collect_switch_conv_info): Compute contiguous_range and
23670 default_case_nonstandard fields, don't clear final_bb if
23671 contiguous_range and only the default case doesn't have the required
23673 (check_all_empty_except_final): Set default_case_nonstandard instead
23674 of failing if contiguous_range and the default case doesn't have empty
23676 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
23677 and only the default case doesn't have the required constants. Skip
23679 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
23680 if default_case_nonstandard.
23681 (build_constructors): Build constant 1 just once. Assert that default
23682 values aren't inserted in between cases if contiguous_range. Skip
23684 (build_arrays): Skip virtual phis.
23685 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
23686 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
23687 Handle virtual phis.
23688 (gen_inbound_check): Handle default_case_nonstandard case.
23689 (process_switch): Adjust check_final_bb caller. Call
23690 gather_default_values with the first non-default case instead of
23691 default case if default_case_nonstandard.
23693 2017-05-02 Nathan Sidwell <nathan@acm.org>
23695 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
23696 check. Fix formatting.
23698 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
23700 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
23701 errors when comparing specialized and unspecialized times.
23703 2017-05-02 David Malcolm <dmalcolm@redhat.com>
23705 * diagnostic-show-locus.c
23706 (layout::should_print_annotation_line_p): Make private.
23707 (layout::print_annotation_line): Make private.
23708 (layout::annotation_line_showed_range_p): Make private.
23709 (layout::show_ruler): Make private.
23710 (layout::print_source_line): Make private. Pass in line and
23711 line_width, rather than calling location_get_source_line. Drop
23713 (layout::print_leading_fixits): New method.
23714 (layout::print_any_fixits): Rename to...
23715 (layout::print_trailing_fixits): ...this, and make private.
23716 Don't print newline fixits.
23717 (diagnostic_show_locus): Move logic for printing one row into...
23718 (layout::print_line): ...this new function. Move the
23719 location_get_source_line call and error-handling from
23720 print_source_line to here. Call print_leading_fixits, and rename
23721 print_any_fixits to print_trailing_fixits.
23722 (selftest::test_fixit_insert_containing_newline): Update now that
23723 newlines are partially supported.
23724 (selftest::test_fixit_insert_containing_newline_2): New test.
23725 (selftest::test_fixit_replace_containing_newline): Update comments.
23726 (selftest::diagnostic_show_locus_c_tests): Call the new test.
23727 * edit-context.c (class added_line): New class.
23728 (class edited_line): Describe newline handling in comment.
23729 (edited_line::actually_edited_p): New method.
23730 (edited_line::print_content): Delete redundant decl.
23731 (edited_line::m_predecessors): New field.
23732 (edited_file::print_content): Call edited_line::print_content.
23733 (edited_file::print_diff): Update to support newlines.
23734 (edited_file::print_diff_hunk): Likewise.
23735 (edited_file::print_run_of_changed_lines): New function.
23736 (edited_file::print_diff_line): Convert to...
23737 (print_diff_line): ...this.
23738 (edited_file::get_effective_line_count): New function.
23739 (edited_line::edited_line): Initialize new field m_predecessors.
23740 (edited_line::~edited_line): Clean up m_predecessors.
23741 (edited_line::apply_fixit): Handle newlines.
23742 (edited_line::get_effective_line_count): New function.
23743 (edited_line::print_content): New function.
23744 (edited_line::print_diff_lines): New function.
23745 (selftest::test_applying_fixits_insert_containing_newline): New
23747 (selftest::test_applying_fixits_replace_containing_newline): New
23749 (selftest::insert_line): New function.
23750 (selftest::test_applying_fixits_multiple_lines): Add example of
23752 (selftest::edit_context_c_tests): Call the new tests.
23754 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23756 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
23757 parameter cand. Update dump information.
23758 (get_computation_cost): Update uses.
23760 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23762 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
23763 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
23764 (get_computation_at, rewrite_use_address): Update use of
23765 get_computation_aff.
23767 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23769 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
23770 (get_computation): Delete.
23771 (get_computation_cost): Implement like get_computation_cost_at.
23772 Use get_computation_at.
23773 (get_computation_cost_at): Delete.
23774 (rewrite_use_nonlinear_expr): Use get_computation_at.
23775 (rewrite_use_compare, remove_unused_ivs): Ditto.
23777 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23779 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
23781 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23783 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
23784 (ivopts_global_cost_for_size): Rename parameter and update uses.
23785 (iv_ca_recount_cost): Update uses.
23786 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
23787 candidates seperately in n_invs and n_cands.
23788 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
23790 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23792 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
23793 (find_inv_vars_cb): New.
23794 (find_depends): Renamed to ...
23795 (find_inv_vars): ... this.
23796 (add_candidate_1, force_var_cost): Call find_inv_vars.
23797 (split_address_cost, determine_group_iv_cost_cond): Ditto.
23799 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23801 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
23802 inv_vars. Add inv_exprs.
23803 (struct iv_cand): Rename depends_on to inv_vars.
23804 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
23805 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
23806 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
23807 (dump_cand): Dump inv_vars.
23808 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
23809 (record_invariant, find_depends, add_candidate_1): Ditto.
23810 (set_group_iv_cost, force_var_cost): Ditto.
23811 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
23812 (get_computation_cost_at, get_computation_cost): Ditto.
23813 (determine_group_iv_cost_generic): Ditto.
23814 (determine_group_iv_cost_address): Ditto.
23815 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
23816 (determine_group_iv_costs): Ditto.
23817 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
23818 (iv_ca_set_remove_invariants): Renamed to ...
23819 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
23820 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
23821 (iv_ca_set_add_invariants): Renamed to ...
23822 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
23823 (iv_ca_set_cp): Use iv_ca_set_add_invs.
23824 (iv_ca_has_deps): Support inv_vars and inv_exprs.
23825 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
23826 (create_new_ivs): Remove useless dump.
23828 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23830 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
23832 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
23833 (iv_ca_set_no_cp, create_new_iv): Ditto.
23835 2017-05-02 Bin Cheng <bin.cheng@arm.com>
23837 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
23839 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23841 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
23842 function tree_check2.
23844 2017-05-02 Martin Liska <mliska@suse.cz>
23846 * doc/gcov.texi: Add missing preposition.
23847 * gcov.c (function_info::function_info): Properly fill up
23848 all member variables.
23850 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23852 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
23854 2017-05-02 Tamar Christina <tamar.christina@arm.com>
23856 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
23858 2017-05-02 Martin Liska <mliska@suse.cz>
23861 * lto-streamer-in.c (lto_read_tree_1): Remove
23862 LTO_STREAMER_DEBUG.
23863 * lto-streamer.c (struct tree_hash_entry): Likewise.
23864 (struct tree_entry_hasher): Likewise.
23865 (tree_entry_hasher::hash): Likewise.
23866 (tree_entry_hasher::equal): Likewise.
23867 (lto_streamer_init): Likewise.
23868 (lto_orig_address_map): Likewise.
23869 (lto_orig_address_get): Likewise.
23870 (lto_orig_address_remove): Likewise.
23871 * lto-streamer.h: Likewise.
23872 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
23873 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
23875 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
23877 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
23878 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
23879 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
23880 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
23881 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
23882 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
23883 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
23884 (mm_maskz_sub_ss): New intrinsics.
23885 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23886 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
23887 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
23888 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
23889 (__builtin_ia32_subss_mask_round): New builtins.
23890 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
23891 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
23892 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
23894 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
23895 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
23897 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
23900 2017-05-02 Martin Jambor <mjambor@suse.cz>
23902 PR tree-optimization/78687
23903 * tree-sra.c (access): New field parent.
23904 (process_subtree_disqualification): New function.
23905 (disqualify_candidate): Call it.
23906 (build_accesses_from_assign): Reset write flag if creating an
23908 (build_access_subtree): Fill in parent field and also prpagate
23909 down grp_write flag.
23910 (create_artificial_child_access): New parameter set_grp_write, set
23911 grp_write to its value.
23912 (propagate_subaccesses_across_link): Also propagate grp_write flag
23914 (propagate_all_subaccesses): Push the closest parent back to work
23915 queue if add_access_to_work_queue returned true.
23917 2017-05-02 Richard Biener <rguenther@suse.de>
23919 * common.opt (fstrict-overflow): Alias negative to fwrapv.
23920 * doc/invoke.texi (fstrict-overflow): Remove all traces of
23921 -fstrict-overflow documentation.
23922 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
23923 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
23924 flag_strict_overflow.
23925 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
23926 * lto-opts.c (lto_write_options): Do not stream it.
23927 * lto-wrapper.c (merge_and_complain): Do not handle it.
23928 * opts.c (default_options_table): Do not set -fstrict-overflow.
23929 (finish_options): Likewise do not clear it when sanitizing.
23930 * simplify-rtx.c (simplify_const_relational_operation): Do not
23931 test flag_strict_overflow.
23933 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23935 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
23936 using enabled attribute.
23937 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
23938 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
23939 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
23940 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
23941 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
23942 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
23943 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
23944 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
23945 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
23946 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
23948 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
23950 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
23952 2017-05-02 Richard Biener <rguenther@suse.de>
23954 PR tree-optimization/80591
23956 2017-04-10 Richard Biener <rguenther@suse.de>
23958 * tree-ssa-structalias.c (find_func_aliases): Properly handle
23961 2017-05-02 Richard Biener <rguenther@suse.de>
23963 PR tree-optimization/80549
23964 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
23965 (cleanup_tree_cfg_noloop): Create forwarders to known loop
23966 headers if they do not have a preheader.
23968 2017-05-02 Martin Liska <mliska@suse.cz>
23971 * common.opt: Fix typo.
23972 * doc/invoke.texi: Likewise.
23974 2017-05-01 Jan Beulich <jbeulich@suse.com>
23976 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
23977 swapping, add (x,x,m,x,n) alternative.
23979 2017-05-01 Nathan Sidwell <nathan@acm.org>
23981 * calls.c (combine_pending_stack_adjustment_and_call): Remove
23982 unnecessary unadjusted_alignment check.
23984 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
23987 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
23989 * gimplify.c (gimplify_cilk_detach): New function.
23990 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
23991 * tree-core.h: Document EXPR_CILK_SPAWN.
23992 * tree.h (EXPR_CILK_SPAWN): Define.
23994 2017-05-01 David Malcolm <dmalcolm@redhat.com>
23996 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
23997 to use new fixit_hint representation, using the "replace" logic.
23998 (get_line_span_for_fixit_hint): Likewise.
23999 (layout::print_any_fixits): Likewise.
24000 (selftest::test_one_liner_many_fixits): Rename to...
24001 (selftest::test_one_liner_many_fixits_1): ...this, and update
24002 comment and expected output to reflect that the multiple fix-it
24003 hints are now consolidated into one insertion.
24004 (selftest::test_one_liner_many_fixits_2): New test.
24005 (selftest::test_diagnostic_show_locus_one_liner): Update for
24007 (selftest::test_fixit_consolidation): Update for fix-it API
24009 * diagnostic.c (print_parseable_fixits): Likewise.
24010 * edit-context.c (edited_line::m_line_events): Convert from
24011 auto_vec <line_event *> to auto_vec <line_event>.
24012 (class line_event): Convert from abstract base class to a concrete
24013 class, taking over the role of replace_event.
24014 (class insert_event): Delete.
24015 (class replace_event): Rename to class line_event. Convert to
24017 (edit_context::add_fixits): Reimplement.
24018 (edit_context::apply_insert): Delete.
24019 (edit_context::apply_replace): Rename to...
24020 (edit_context::apply_fixit): ...this. Convert to half-open range.
24021 (edited_file::apply_insert): Delete.
24022 (edited_file::apply_replace): Rename to...
24023 (edited_file::apply_fixit): ...this.
24024 (edited_line::~edited_line): Drop deletion of events.
24025 (edited_line::apply_insert): Delete.
24026 (edited_line::apply_replace): Rename to...
24027 (edited_line::apply_fixit): ...this. Convert to half-open range.
24028 Update for change to type of m_line_events.
24029 * edit-context.h (edit_context::apply_insert): Delete.
24030 (edit_context::apply_replace): Rename to...
24031 (edit_context::apply_fixit): ...this.
24033 2017-05-01 Martin Sebor <msebor@redhat.com>
24035 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
24038 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
24041 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
24042 __get_cpuid_max returns 0.
24043 (__get_cpuid_count): Ditto.
24045 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
24047 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
24048 replacement expression is another instance of one of its arguments.
24050 2017-05-01 Jakub Jelinek <jakub@redhat.com>
24053 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
24054 check for stack push/pop autoinc.
24055 * config/i386/i386.c (ix86_agi_dependent): Return false
24056 if the only reason why modified_in_p returned true is that
24057 addr is SP based and set_insn is a push or pop.
24059 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24061 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
24064 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
24067 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
24068 (account_size_time): Use two predicates - exec_pred and
24070 (evaluate_conditions_for_known_args): Compute both clause and
24072 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
24073 (inline_summary_t::duplicate): Update.
24074 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
24076 (compute_inline_parameters): Likewise.
24077 (estimate_edge_size_and_time): Update caluclation of time.
24078 (estimate_node_size_and_time): Compute both time and nonspecialized
24080 (estimate_ipcp_clone_size_and_time): Update.
24081 (inline_merge_summary): Update.
24082 (do_estimate_edge_time): Update.
24083 (do_estimate_edge_size): Update.
24084 (do_estimate_edge_hints): Update.
24085 (inline_read_section, inline_write_summary): Stream both new predicates.
24086 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
24088 (compute_inlined_call_time): Cleanup.
24089 (big_speedup_p): Update.
24090 (edge_badness): Update.
24091 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
24092 (size_time_entry): Replace predicate by exec_predicate and
24093 nonconst_predicate.
24094 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
24095 (estimate_edge_time): Return also nonspec_time.
24096 (reset_edge_growth_cache): Update.
24098 2017-04-29 Jakub Jelinek <jakub@redhat.com>
24100 PR rtl-optimization/80491
24101 * ifcvt.c (noce_process_if_block): When looking for x setter
24102 with missing else_bb, don't check only the insn right before
24103 cond_earliest, but look for the last insn that x is modified in
24104 within the same bb.
24106 PR rtl-optimization/80491
24107 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
24109 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
24111 PR tree-optimization/80487
24112 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
24114 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24116 PR tree-optimization/79697
24117 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
24118 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
24119 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
24121 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
24122 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
24124 2017-04-28 Martin Sebor <msebor@redhat.com>
24126 PR tree-optimization/80523
24127 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
24128 (init_target_to_host_charmap, target_to_host, target_strtol10): New
24130 (maybe_warn, format_directive, parse_directive): Use new functions.
24131 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
24133 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
24135 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
24137 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
24139 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
24140 target_header_dir): Set correctly.
24141 * configure: Regenerated.
24142 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24143 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24144 instead of SYSTEM_HEADER_DIR.
24146 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
24148 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
24149 (estimate_local_effects): Likewise.
24150 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
24151 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
24152 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
24153 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
24154 do_estimate_edge_time, estimate_edge_time): Likewise.
24155 * ipa-inline-analysis.c (estimate_node_size_and_time,
24156 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
24157 (estimate_time_after_inlining): Remove.
24159 2017-04-28 Martin Liska <mliska@suse.cz>
24161 * doc/gcov.texi: Enhance documentation of gcov.
24163 2017-04-28 Martin Liska <mliska@suse.cz>
24165 * doc/gcov.texi: Sort options in alphabetic order.
24166 * doc/gcov-dump.texi: Likewise.
24167 * doc/gcov-tool.texi: Likewise.
24168 * gcov.c (print_usage): Likewise.
24169 * gcov-dump.c (print_usage): Likewise.
24170 * gcov-tool.c (print_merge_usage_message): Likewise.
24171 (print_rewrite_usage_message): Likewise.
24172 (print_overlap_usage_message): Likewise.
24174 2017-04-28 Martin Liska <mliska@suse.cz>
24176 PR gcov-profile/53915
24177 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
24179 2017-04-28 Martin Liska <mliska@suse.cz>
24181 PR gcov-profile/79891
24182 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
24183 is marked by compiler as living on a line.
24184 (get_cycles_count): Remove usage of the union.
24185 (output_intermediate_file): Likewise.
24186 (find_source): Fix GNU coding style.
24187 (accumulate_line_counts): Remove old non-all block mode.
24188 (output_lines): Remove usage of the union.
24189 * profile.c (output_location): Include all BBs, even if
24190 belonging to a same line (and file) as a previous BB.
24192 2017-04-28 Martin Liska <mliska@suse.cz>
24194 * gcov.c (process_args): Handle new argument 'w'.
24195 (read_graph_file): Assign ID to BBs.
24196 (output_branch_count): Display BB # if verbose flag is set.
24197 (output_lines): Likewise for arcs.
24198 (print_usage): Add '--verbose' option help.
24199 * doc/gcov.texi: Document --verbose (-w) option.
24201 2017-04-28 Martin Liska <mliska@suse.cz>
24203 * gcov.c (struct block_location_info): New struct.
24204 (process_file): Fill up the new structure.
24205 (read_graph_file): Replace usage of encoding by the newly added
24207 (add_line_counts): Likewise.
24208 (accumulate_line_counts): Remove usage of the union.
24209 (function_info::function_info): New function.
24210 (function_info::~function_info): Likewise.
24211 (process_file): Call delete instead of release_function.
24212 (release_function): Release the function.
24213 (release_structures): Call delete instead of release_function.
24214 (solve_flow_graph): Replace usage of num_blocks.
24215 (find_exception_blocks): Likewise.
24216 (output_lines): Fix GNU coding style.
24218 2017-04-28 Martin Liska <mliska@suse.cz>
24221 * coverage.c (coverage_remove_note_file): New function.
24222 * coverage.h: Declare the function.
24223 * toplev.c (finalize): Clean if an error has been seen.
24225 2017-04-28 Martin Liska <mliska@suse.cz>
24227 PR gcov-profile/80031
24228 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
24229 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
24230 * gcov.c (read_graph_file): Read just number of blocks.
24231 * profile.c (branch_prob): Do not stream 0 flags per a basic
24234 2017-04-28 Martin Liska <mliska@suse.cz>
24236 * gcov-dump.c (tag_*): Add new argument to declarations.
24237 (dump_gcov_file): Likewise.
24238 (tag_blocks): Add and use new argument depth.
24239 (tag_arcs): Likewise.
24240 (tag_lines): Likewise.
24241 (tag_counters): Likewise.
24242 (tag_summary): Likewise.
24243 (dump_working_sets): Use depth to do a proper indentation.
24245 2017-04-28 Jakub Jelinek <jakub@redhat.com>
24248 * cgraph.h (symtab_node::debug_symtab): No longer inline.
24249 * symtab.c (symtab_node::debug_symtab): Move definition here.
24251 2017-04-28 Richard Biener <rguenther@suse.de>
24253 * lto-streamer.h (LTO_major_version): Bump to 7.
24255 2017-04-28 Richard Biener <rguenther@suse.de>
24257 * tree-vrp.c (assert_info): New struct.
24258 (add_assert_info): New helper.
24259 (register_edge_assert_for_2): Refactor to add asserts to a vector
24261 (register_edge_assert_for_1): Likewise.
24262 (register_edge_assert_for): Likewise.
24263 (finish_register_edge_assert_for): New helper actually registering
24264 asserts where live on edge.
24265 (find_conditional_asserts): Adjust.
24266 (find_switch_asserts): Likewise.
24267 (evrp_dom_walker::try_find_new_range): Generalize.
24268 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
24270 2017-04-27 Marek Polacek <polacek@redhat.com>
24273 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
24274 arg10 and arg11 to itype.
24276 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
24278 * doc/extend.texi (Object Size Checking): Improve grammar.
24280 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
24283 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
24284 that the logic for permitting reciprocal estimates matches that
24287 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24290 * tree.c (type_cache_hasher::equal): Only compare
24291 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
24292 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
24293 non-aggregate element types.
24294 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
24295 about the flag on ARRAY_TYPEs in the comment, formatting fix.
24297 2017-04-27 Richard Biener <rguenther@suse.de>
24299 PR middle-end/80533
24300 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
24301 stripping ARRAY_REFs from MEM_EXPR make sure we're not
24302 keeping a reference to a trailing array.
24304 2017-04-27 Richard Biener <rguenther@suse.de>
24306 PR middle-end/80539
24307 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
24308 being in loop-closed SSA form conservatively.
24309 (chrec_fold_multiply_poly_poly): Likewise.
24311 2017-04-27 Tamar Christina <tamar.christina@arm.com>
24313 PR middle-end/79665
24314 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
24315 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
24317 2017-04-27 Jakub Jelinek <jakub@redhat.com>
24320 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
24321 (aarch64_function_arg_alignment): Return unsigned int again, but still
24322 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
24323 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24324 Don't emit -Wpsabi note.
24325 (aarch64_function_arg_boundary): Likewise.
24326 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24329 2017-04-26 Nathan Sidwell <nathan@acm.org>
24331 * tree.h (crc32_unsigned_n): Declare.
24332 (crc32_unsigned, crc32_unsigned): Make inline.
24333 * tree.c (crc32_unsigned_bits): Replace with ...
24334 (crc32_unsigned_n): ... this.
24335 (crc32_unsigned, crc32_byte): Remove.
24336 (crc32_string): Remove unnecessary braces.
24338 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24340 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
24341 * ipa-inline-analysis.c (MAX_TIME): Remove.
24342 (account_size_time): Use sreal for time.
24343 (dump_inline_summary): Update.
24344 (estimate_function_body_sizes): Update.
24345 (estimate_edge_size_and_time): Update.
24346 (estimate_calls_size_and_time): Update.
24347 (estimate_node_size_and_time): Update.
24348 (inline_merge_summary): Update.
24349 (inline_update_overall_summary): Update.
24350 (estimate_time_after_inlining): Update.
24351 (inline_read_section): Update.
24352 (inline_write_summary): Update.
24353 * ipa-inline.c (compute_uninlined_call_time): Update.
24354 (compute_inlined_call_time): Update.
24355 (recursive_inlining): Update.
24356 (inline_small_functions): Update.
24357 (dump_overall_stats): Update.
24358 * ipa-inline.h: Include sreal.h.
24359 (size_time_entry): Turn time to sreal.
24360 (inline_summary): Turn self_time nad time to sreal.
24362 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
24364 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
24366 (sreal::stream_out, sreal::stream_in): New.
24367 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
24369 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24371 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
24374 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
24377 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
24379 (dimode_scalar_chain::compute_convert_gain): Ditto.
24380 (dimode_scalar_chain::make_vector_copies): Ditto.
24381 (dimode_scalar_chain::convert_reg): Ditto.
24382 (dimode_scalar_chain::convert_insn): Ditto.
24383 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
24384 (VI248_AVX512BW_1): New mode iterator.
24385 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
24386 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
24389 2017-04-25 Martin Sebor <msebor@redhat.com>
24391 PR tree-optimization/80497
24392 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
24393 constants are representable in HOST_WIDE_INT.
24394 (parse_directive): Ditto.
24396 2017-04-25 Martin Sebor <msebor@redhat.com>
24399 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
24400 (new_zero_array): Adjust signature.
24401 (dom_info::dom_init): Used unsigned rather that size_t.
24402 (dom_info::dom_info): Same.
24404 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24405 Jakub Jelinek <jakub@redhat.com>
24408 * config/arm/arm.c: Include gimple.h.
24409 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24410 returns negative, increment ncrn only if it returned positive.
24411 (arm_needs_doubleword_align): Return int instead of bool,
24412 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
24413 members, but if there is any such non-FIELD_DECL
24414 > PARM_BOUNDARY aligned decl, return -1 instead of false.
24415 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
24416 returns negative, increment nregs only if it returned positive.
24417 (arm_setup_incoming_varargs): Likewise.
24418 (arm_function_arg_boundary): Emit -Wpsabi note if
24419 arm_needs_doubleword_align returns negative, return
24420 DOUBLEWORD_ALIGNMENT only if it returned positive.
24422 2017-04-25 Marek Polacek <polacek@redhat.com>
24425 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
24426 first argument to type.
24428 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
24431 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
24432 type checks to test for compatibility instead of equality.
24434 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24435 Jakub Jelinek <jakub@redhat.com>
24438 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
24440 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
24441 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
24442 the alignment computation, but return their maximum in warn_alignment.
24443 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
24444 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
24446 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
24447 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
24450 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24452 * config/arc/simdext.md (dmpyh): Fix typo.
24454 2017-04-25 Richard Biener <rguenther@suse.de>
24456 PR tree-optimization/80492
24457 * alias.c (compare_base_decls): Handle registers with asm
24458 specification conservatively.
24459 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
24460 compare_base_decls returning dont-know properly.
24462 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24464 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
24465 (legitimate_offset_address_p): New function.
24466 (arc_legitimate_address_p): Use above function.
24468 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24470 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
24472 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24474 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
24475 ACCH registers whenever they are available.
24477 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24479 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
24480 double regs fix when not used.
24482 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24484 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
24486 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
24487 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
24489 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24491 * config/arc/arc.c (arc_output_addsi): Check for h-register class
24492 when emitting short ADD instructions.
24494 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
24496 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
24498 (cmpsi_cc_c_insn): Likewise.
24499 (cbranchsi4_scratch): Compute proper instruction length using
24500 compact_hreg_operand.
24501 * config/arc/predicates.md (compact_hreg_operand): New predicate.
24503 2017-04-25 Richard Biener <rguenther@suse.de>
24505 PR middle-end/80509
24506 * passes.c (pass_manager::pass_manager): Initialize
24507 m_name_to_pass_map.
24509 2017-04-25 Richard Biener <rguenther@suse.de>
24511 PR tree-optimization/79201
24512 * tree-ssa-sink.c (statement_sink_location): Handle calls.
24514 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24517 * config/s390/vector.md: Split MEM->GPR vector moves for
24518 non-s_operand addresses.
24520 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
24523 * config/s390/predicates.md (reload_const_wide_int_operand): New
24525 * config/s390/s390.md ("movti"): Remove d/P alternative.
24526 ("movti_bigconst"): New pattern definition.
24528 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24531 * s390-protos.h (s390_expand_cs_hqi): Removed.
24532 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
24533 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
24534 modes as well as CCZ1mode and CCZmode.
24535 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
24536 signature of s390_emit_compare_and_swap.
24537 (s390_expand_cs_hqi): Likewise, make static.
24538 (s390_expand_cs_tdsi): Generate an explicit compare before trying
24539 compare-and-swap, in some cases.
24540 (s390_expand_cs): Wrapper function.
24541 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
24543 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
24544 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
24545 patterns for small and large integers. Forbid symref memory operands.
24546 Move expander to s390.c. Require cc register.
24547 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
24548 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
24549 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
24550 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
24551 symref memory operands. Remove CC mode and call s390_match_ccmode
24553 ("atomic_exchange<mode>"): Allow and implement all integer modes.
24555 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24557 * config/s390/s390.md (define_peephole2): New peephole to help
24558 combining the load-and-test pattern with volatile memory.
24560 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
24562 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
24563 with CCZmode for TARGET_Z196.
24565 2017-04-25 Jakub Jelinek <jakub@redhat.com>
24567 PR rtl-optimization/80501
24568 * combine.c (make_compound_operation_int): Set subreg_code to SET
24569 even for AND with mask of the sign bit of mode.
24571 PR rtl-optimization/80500
24572 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
24573 sum's initial value.
24575 2017-04-25 Julian Brown <julian@codesourcery.com>
24576 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24578 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
24580 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
24582 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
24584 2017-04-25 Julian Brown <julian@codesourcery.com>
24585 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24587 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
24588 (thunderx2t99_sha): New Reservation.
24590 2017-04-25 Julian Brown <julian@codesourcery.com>
24591 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
24593 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
24594 type for 1-element load.
24596 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
24598 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
24600 2017-04-24 Martin Jambor <mjambor@suse.cz>
24602 PR tree-optimization/80293
24603 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
24604 char arrays not totally scalarizable if it is false.
24605 (analyze_all_variable_accesses): Pass correct value in the new
24606 parameter. Add a statistics counter.
24608 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
24610 PR middle-end/79931
24611 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
24613 2017-04-24 Richard Biener <rguenther@suse.de>
24615 PR tree-optimization/80494
24616 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
24617 out for complex types.
24619 2017-04-24 Richard Biener <rguenther@suse.de>
24621 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
24622 * tree-ssa-sccvn.c (print_scc): Print SCC size.
24623 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
24624 (DFS): Adjust and never fail.
24625 (sccvn_dom_walker::fail): Remove.
24626 (sccvn_dom_walker::before_dom_children): Adjust.
24627 (run_scc_vn): Likewise and never fail.
24628 * tree-ssa-pre.c (pass_pre::execute): Adjust.
24629 (pass_fre::execute): Likewise.
24631 2017-04-24 Richard Biener <rguenther@suse.de>
24633 PR tree-optimization/79725
24634 * tree-ssa-sink.c (statement_sink_location): Return whether
24635 failure reason was zero uses. Move that check later.
24636 (sink_code_in_bb): Deal with zero uses by removing the stmt
24639 2017-04-24 Richard Biener <rguenther@suse.de>
24642 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
24643 pointer-based references.
24645 2017-04-24 Richard Biener <rguenther@suse.de>
24648 * pass_manager.h (pass_manager::operator new): Remove.
24649 (pass_manager::operator delete): Likewise.
24650 * passes.c (pass_manager::operator new): Remove.
24651 (pass_manager::operator delete): Likewise.
24652 (pass_manager::pass_manager): Zero individual pass members.
24654 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
24657 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
24658 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
24659 Check "XEXP (src, 1)" operand here.
24660 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
24661 Check "XEXP (src, 1)" operand here.
24662 (dimode_scalar_chain::make_vector_copies): Detect count register
24663 of a shift instruction. Zero extend count register from QImode
24664 to DImode to satisfy vector shift pattern count operand predicate.
24665 Substitute vector shift count operand with a DImode copy.
24666 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
24669 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
24671 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
24672 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
24673 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
24674 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
24675 (UNSPEC_NOREX_MEM): Remove definition.
24677 2017-04-21 Richard Biener <rguenther@suse.de>
24679 PR tree-optimization/79547
24680 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
24681 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
24682 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
24683 without any constraints.
24685 2017-04-21 Richard Biener <rguenther@suse.de>
24687 PR tree-optimization/78847
24688 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
24690 2017-04-21 Richard Biener <rguenther@suse.de>
24692 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
24693 (build_distinct_type_copy): Likewise.
24694 (build_variant_type_copy): Likewise.
24695 * tree.c (build_qualified_type): Pass down mem-stat info.
24696 (build_distinct_type_copy): Likewise.
24697 (build_variant_type_copy): Likewise.
24699 2017-04-21 Richard Biener <rguenther@suse.de>
24701 PR tree-optimization/80237
24702 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
24704 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
24705 for a simplified result.
24707 2016-04-21 Richard Biener <rguenther@suse.de>
24709 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
24710 sth as strict as a simple_iv but a chrec without symbols and an
24711 operand defined in the loop we are peeling (and not some subloop).
24712 (propagate_constants_for_unrolling): Propagate all constants.
24714 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24717 * config/i386/i386.c (print_reg): Remove assert for disalowed
24718 regno values, call output_operand_lossage instead.
24720 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
24723 * config/i386/constraints.md (Yc): New register constraint.
24724 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
24725 Use Yc constraint for alternative 2 of operand 0. Remove
24726 preferred_for_speed attribute.
24728 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24730 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
24731 lastprivate clauses in SIMT case.
24733 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
24735 * doc/invoke.texi (-Wextra-semi): Document new warning option.
24737 2017-04-20 Richard Biener <rguenther@suse.de>
24739 PR tree-optimization/57796
24740 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
24741 as N scalar stores.
24742 (vect_model_load_cost): Cost gathers as N scalar loads.
24744 2017-04-20 Richard Biener <rguenther@suse.de>
24746 * ggc-page.c (ggc_allocated_p): Rename to ...
24747 (safe_lookup_page_table_entry): ... this and return the lookup
24749 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
24751 2017-04-20 Richard Biener <rguenther@suse.de>
24753 PR tree-optimization/80453
24754 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
24755 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
24756 from the conditions.
24757 (vn_phi_eq): Pass them down.
24758 (vn_phi_lookup): Record them.
24759 (vn_phi_insert): Likewise.
24761 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
24763 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
24764 uninitialized variable warning to avoid buffer overrun.
24766 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
24769 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
24770 is suppressed for '{ 0 }' in C.
24772 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24774 * BASE-VER: Set to 8.0.0.
24776 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
24778 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
24779 priority .init_array and .fini_array section with SECTION_NOTYPE
24782 2017-04-20 Jakub Jelinek <jakub@redhat.com>
24784 PR middle-end/80423
24785 * tree.h (build_array_type): Add typeless_storage default argument.
24786 * tree.c (type_cache_hasher::equal): Also compare
24787 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
24788 (build_array_type): Add typeless_storage argument, set
24789 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
24791 (build_nonshared_array_type): Adjust build_array_type_1 caller.
24792 (build_array_type): Likewise. Add typeless_storage argument.
24794 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24795 Jakub Jelinek <jakub@redhat.com>
24797 PR tree-optimization/80426
24798 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
24799 operation on symbolic operands, also compute the overflow for the
24800 invariant part when the operation degenerates into a negation.
24802 2017-04-19 Jakub Jelinek <jakub@redhat.com>
24805 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
24806 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
24809 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
24811 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
24814 * config/avr/avr.c (tree.h): Include it.
24815 (cgraph.h): Include it.
24816 (avr_encode_section_info): Don't warn for uninitialized progmem
24817 variable if it's just an alias.
24819 2017-04-19 Richard Biener <rguenther@suse.de>
24822 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
24823 when needed by AutoPGO.
24825 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
24828 * doc/lto.texi: Remove an extra 'that'.
24830 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
24832 PR rtl-optimization/80429
24833 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
24834 are only used in debug insns.
24836 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
24837 Vladimir Makarov <vmakarov@redhat.com>
24839 * config/sparc/predicates.md (input_operand): Add comment. Return
24840 true for any memory operand when LRA is in progress.
24841 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
24843 2017-04-18 Jeff Law <law@redhat.com>
24846 * mips.md ({return,simple_return}_internal): Do not overwrite
24849 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24851 PR tree-optimization/80443
24852 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
24853 instead of adding 1, subtract -1 and similarly instead of subtracting
24856 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
24858 PR rtl-optimization/80357
24859 * haifa-sched.c (tmp_bitmap): New variable.
24860 (model_recompute): Handle duplicate use records.
24861 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
24862 (free_global_sched_pressure_data): Free it.
24864 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
24867 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
24868 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
24869 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
24870 instead of SYSTEM_HEADER_DIR.
24872 2017-04-18 Jeff Law <law@redhat.com>
24874 PR middle-end/80422
24875 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
24876 predecessors after walking up the insn chain.
24878 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24881 * dwarf2out.c (modified_type_die): Try harder not to emit internal
24882 sizetype type into debug info.
24884 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
24887 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
24888 unneeded test for TARGET_UPPER_REGS_SF.
24889 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
24891 2017-04-18 Jakub Jelinek <jakub@redhat.com>
24894 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
24895 instead of gsi_after_labels.
24897 2017-04-18 Jeff Law <law@redhat.com>
24899 * regcprop.c (maybe_mode_change): Avoid creating copies of the
24903 2017-04-13 Jeff Law <law@redhat.com>
24904 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24905 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24907 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
24910 * config/avr/avr.c (intl.h): Include it.
24911 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
24913 2017-04-18 Martin Liska <mliska@suse.cz>
24915 PR gcov-profile/78783
24916 * gcov-tool.c (gcov_output_files): Validate that destination
24917 file is either removed by the tool or by a user.
24919 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24920 Guy Benyei <guybe@mellanox.com>
24922 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
24923 block, and do not negate it, the stored id is already negative.
24925 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
24927 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
24929 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
24932 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
24933 masks of options that should be turned off if the VSX vector
24934 options are turned off.
24935 (OTHER_P8_VECTOR_MASKS): Likewise.
24936 (OTHER_VSX_VECTOR_MASKS): Likewise.
24937 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
24938 rs6000_disable_incompatible_switches to validate no type switches
24940 (rs6000_incompatible_switch): New function to disallow turning on
24941 other vector options if -mno-vsx, -mno-power8-vector, or
24942 -mno-power9-vector are specified.
24944 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24946 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
24948 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24950 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
24951 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
24952 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
24953 (ARG_POINTER_CFA_OFFSET): Likewise.
24955 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
24957 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
24958 conditions to take advantage of various optimizations.
24960 2017-04-13 Jeff Law <law@redhat.com>
24962 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
24963 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
24964 (zero_extendsidi2_dext): Likewise.
24966 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24969 * fold-const.c (fold_ternary_loc): Revert
24970 use op0 instead of fold_convert_loc (loc, type, arg0) part of
24973 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
24975 PR rtl-optimization/80343
24976 * lra-remat.c (update_scratch_ops): Assign original hard reg to
24977 new scratch pseudo.
24979 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
24982 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
24983 to ubsan_encode_value.
24985 2017-04-13 Jeff Law <law@redhat.com>
24987 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
24988 appearing in DEBUG_INSNs.
24990 2017-04-13 Martin Liska <mliska@suse.cz>
24992 PR gcov-profile/80413
24993 * gcov-io.c (gcov_write_string): Copy to buffer just when
24994 allocated size is greater than zero.
24996 2017-04-13 Jakub Jelinek <jakub@redhat.com>
24999 * dwarf2out.c (decls_for_scope): Ignore declarations of
25000 current_function_decl in BLOCK_NONLOCALIZED_VARS.
25002 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
25005 * ipa-visibility.c (non_local_p): Fix typos.
25006 (localize_node): When localizing symbol in same comdat group,
25007 dissolve the group only when we know external symbols are going
25009 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
25011 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25013 PR tree-optimization/79390
25014 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
25015 order does not result in usable sequence, retry with reversed operand
25021 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
25022 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
25023 op0 instead of fold_convert_loc (loc, type, arg0).
25025 2017-04-12 Jeff Law <law@redhat.com>
25027 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
25028 has a delay slot in the generated code.
25030 * config/cris/cris.md (cris_preferred_reload_class): Return
25031 GENNONACR_REGS rather than GENERAL_REGS.
25033 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25036 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
25037 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
25038 signedness of the result type.
25040 2017-04-12 Richard Biener <rguenther@suse.de>
25041 Jeff Law <law@redhat.com>
25043 PR tree-optimization/80359
25044 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
25045 trim stores to TARGET_MEM_REFs.
25047 2017-04-12 Richard Biener <rguenther@suse.de>
25049 PR tree-optimization/79390
25050 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
25051 threading case even more.
25053 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
25056 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
25057 for quad_address_p for TImode, instead of just not indexed_address.
25059 2017-04-12 Richard Biener <rguenther@suse.de>
25060 Bernd Edlinger <bernd.edlinger@hotmail.de>
25062 PR middle-end/79671
25063 * alias.c (component_uses_parent_alias_set_from): Handle
25064 TYPE_TYPELESS_STORAGE.
25065 (get_alias_set): Likewise.
25066 * tree-core.h (tree_type_common): Add typeless_storage flag.
25067 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
25068 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
25069 for types containing members with TYPE_TYPELESS_STORAGE.
25070 (place_field): Likewise.
25071 (layout_type): Likewise for ARRAY_TYPE.
25072 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
25073 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
25074 TYPE_TYPELESS_STORAGE.
25075 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
25077 2017-04-12 Jakub Jelinek <jakub@redhat.com>
25080 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
25081 first argument to type.
25083 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25087 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
25088 CONST0_RTX (mode) rather than const0_rtx where appropriate.
25089 (rs6000_expand_binop_builtin): Likewise.
25090 (rs6000_expand_ternop_builtin): Likewise; also add missing
25091 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
25092 vshasigma built-ins.
25093 * doc/extend.texi: Document that vec_xxpermdi's third argument
25094 must be a constant.
25096 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
25098 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
25099 Use shift_const cost parameter when calculating gain of STV shifts.
25101 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
25103 PR rtl-optimization/70478
25104 * lra-constraints.c (process_alt_operands): Check memory for
25105 disfavoring memory insn operand.
25107 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25109 PR middle-end/80100
25110 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
25111 left shift in unsigned HOST_WIDE_INT type.
25113 PR rtl-optimization/80385
25114 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
25115 (not (neg X)) into (plus X -1) for complex or non-integral modes.
25118 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
25119 if they have any depend clauses.
25121 2017-04-11 Martin Liska <mliska@suse.cz>
25124 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
25125 * ipa-split.c (split_function): Create a local comdat symbol
25126 if caller is in a comdat group.
25128 2017-04-11 Martin Liska <mliska@suse.cz>
25131 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
25134 2017-04-11 Martin Sebor <msebor@redhat.com>
25136 PR middle-end/80364
25137 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
25138 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
25140 (directive::set_width, directive::set_precision, format_character):
25142 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
25145 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
25148 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
25149 conflict, set target->arch_name instead of target->cpu_name.
25151 2017-04-11 Richard Biener <rguenther@suse.de>
25153 PR tree-optimization/80374
25154 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
25155 build_zero_cst, remove fold_convertible_p check again.
25157 2017-04-11 Martin Liska <mliska@suse.cz>
25160 * ubsan.c (instrument_object_size): Do not instrument register
25163 2017-04-11 Jakub Jelinek <jakub@redhat.com>
25166 * config/i386/i386-builtin-types.def
25167 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
25168 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
25169 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
25170 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
25171 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
25172 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
25173 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
25174 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
25175 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
25176 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
25177 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
25178 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
25179 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
25180 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
25181 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
25182 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
25183 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
25184 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
25185 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
25186 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
25187 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
25188 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
25189 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
25190 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
25191 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
25192 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
25193 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
25194 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
25195 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
25196 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
25197 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
25198 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
25199 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
25200 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
25201 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
25202 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
25203 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
25204 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
25205 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
25206 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
25207 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
25208 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
25209 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
25210 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
25211 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
25212 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
25214 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
25215 flag to second_arg_count, handle 4 argument function type _COUNT
25216 aliases, handle second_arg_count on second argument rather than last.
25218 2017-04-10 Jeff Law <law@redhat.com>
25220 PR tree-optimization/80374
25221 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
25222 record anything if we can not convert integer_zero_node to the
25225 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
25228 * config/rs6000/rs6000.c (rs6000_option_override_internal):
25229 Enhance special handling given to the TARGET_P9_MINMAX option in
25230 relation to certain other options.
25232 2017-04-10 Bin Cheng <bin.cheng@arm.com>
25234 PR tree-optimization/80153
25235 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
25236 remove POINTER_PLUS_EXPR's base part directly, rather than through
25239 2017-04-10 Richard Biener <rguenther@suse.de>
25240 Bin Cheng <bin.cheng@arm.com>
25242 PR tree-optimization/80153
25243 * tree-affine.c (aff_combination_to_tree): Get base pointer from
25244 the first element of pointer type aff_tree. Build result expr in
25246 (add_elt_to_tree): Convert to type unconditionally. Remove other
25247 fold_convert calls.
25248 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
25249 (rewrite_use_nonlinear_expr): Check invariant using iv information.
25251 2017-04-10 Richard Biener <rguenther@suse.de>
25253 * tree-ssa-structalias.c (find_func_aliases): Properly handle
25256 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
25258 PR rtl-optimization/70478
25259 * lra-constraints.c (curr_small_class_check): New.
25260 (update_and_check_small_class_inputs): New.
25261 (process_alt_operands): Update curr_small_class_check. Disfavor
25262 alternative insn memory operands. Check available regs for small
25265 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25268 * config/mips/mips.opt (-mvirt): Update description.
25269 * doc/invoke.texi (-mvirt): Likewise.
25271 2017-04-10 Richard Biener <rguenther@suse.de>
25273 PR middle-end/80362
25274 * fold-const.c (fold_binary_loc): Look at unstripped ops when
25275 looking for NEGATE_EXPR in -A / -B to A / B folding.
25277 2017-04-10 Martin Liska <mliska@suse.cz>
25279 PR gcov-profile/80224
25280 * gcov.c (print_usage): Fix usage string.
25281 (get_gcov_intermediate_filename): Remove.
25282 (output_gcov_file): Use both for normal and intermediate format.
25283 (generate_results): Do not initialize special file for
25284 intermediate format.
25286 2017-04-10 Richard Biener <rguenther@suse.de>
25288 PR tree-optimization/80304
25289 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
25292 2017-04-10 Nathan Sidwell <nathan@acm.org>
25295 * config/rs6000/rs6000.c (rs6000_vector_type): New.
25296 (rs6000_init_builtins): Use it.
25298 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25300 * config/arm/arm.md (<mrc>): Add mode to SET source.
25301 (<mrrc>): Likewise.
25303 2017-04-10 Richard Biener <rguenther@suse.de>
25305 PR middle-end/80344
25306 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
25308 2017-04-10 Jakub Jelinek <jakub@redhat.com>
25311 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
25312 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
25313 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
25314 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
25315 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
25316 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
25317 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
25318 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
25319 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
25320 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
25321 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
25322 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
25323 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
25324 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
25325 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
25326 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
25327 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
25328 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
25329 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
25330 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
25331 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
25332 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
25333 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
25335 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
25337 PR rtl-optimization/70478
25338 * lra-constraints.c: Reverse the last patch.
25340 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
25342 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
25343 Add comment for WCHAR_T.
25345 2017-04-08 Martin Liska <mliska@suse.cz>
25348 2017-04-07 Martin Liska <mliska@suse.cz>
25351 * ipa-split.c (split_function): Add function part to a same comdat
25354 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25357 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
25359 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
25361 * rs6000/rs6000.c (vec_load_pendulum): Rename...
25362 (vec_pairing): ...to this.
25363 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
25364 (rs6000_sched_init): Adjust for name change.
25365 (struct rs6000_sched_context): Likewise.
25366 (rs6000_init_sched_context): Likewise.
25367 (rs6000_set_sched_context): Likewise.
25369 2017-04-07 Jakub Jelinek <jakub@redhat.com>
25375 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
25377 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
25378 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
25379 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
25381 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
25383 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
25385 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25387 PR rtl-optimization/70703
25388 * ira-color.c (update_conflict_hard_regno_costs): Use
25389 int64_t instead of HOST_WIDE_INT.
25391 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
25393 PR rtl-optimization/70478
25394 * lra-constraints.c (process_alt_operands): Disfavor alternative
25395 insn memory operands.
25397 2017-04-07 Jeff Law <law@redhat.com>
25399 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
25400 CALL and NOTE_INSN_CALL_ARG_LOCATION.
25402 2017-04-07 Martin Liska <mliska@suse.cz>
25405 * config/aarch64/aarch64.c (aarch64_process_target_attr):
25406 Show error message instead of an ICE.
25408 2017-04-07 Martin Liska <mliska@suse.cz>
25411 * ipa-split.c (split_function): Add function part to a same comdat
25414 2017-04-07 Richard Biener <rguenther@suse.de>
25416 PR middle-end/80341
25417 * tree.c (get_unwidened): Also handle ! for_type case for
25419 * convert.c (do_narrow): Split out from ...
25420 (convert_to_integer_1): ... here. Do not pass final truncation
25421 type to get_unwidened for TRUNC_DIV_EXPR.
25423 2017-04-07 Richard Biener <rguenther@suse.de>
25425 * tree-affine.c (wide_int_ext_for_comb): Take type rather
25427 (aff_combination_const): Adjust.
25428 (aff_combination_scale): Likewise.
25429 (aff_combination_add_elt): Likewise.
25430 (aff_combination_add_cst): Likewise.
25431 (aff_combination_convert): Likewise.
25432 (add_elt_to_tree): Likewise. Remove unused argument.
25433 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
25435 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
25437 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
25439 * config/arm/arm.c (arm_default_short_enums): Use
25440 ARM_DEFAULT_SHORT_ENUMS.
25441 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
25443 2017-04-06 Jakub Jelinek <jakub@redhat.com>
25446 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
25447 members with redundant out-of-class redeclaration.
25449 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25452 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
25453 * config/i386/i386.md (*zero_extendsidi2):
25454 Add (?*x,*x) and (?*v,*v) alternatives.
25456 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
25459 * config/i386/i386.c (ix86_expand_builtin)
25460 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
25461 mode from insn data. Convert operands to insn operand mode.
25462 Copy operands that don't satisfy insn predicate to a register.
25464 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
25466 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
25469 2017-04-06 Richard Biener <rguenther@suse.de>
25471 PR tree-optimization/80334
25472 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
25473 preserve alignment of accesses.
25475 2017-04-06 Richard Biener <rguenther@suse.de>
25477 PR tree-optimization/80262
25478 * tree-sra.c (build_ref_for_offset): Preserve address-space
25480 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
25481 Drop useless address-space information on MEM_REF offsets.
25483 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
25485 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
25487 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25489 PR rtl-optimization/70703
25490 * ira-color.c (update_conflict_hard_regno_costs): Use
25491 HOST_WIDE_INT instead of long.
25493 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
25496 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
25497 not defined for x86_64 target. Add -mmmx target option when __SSE2__
25499 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
25500 for x86_64 target. Handle -m3dnowa option.
25502 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
25504 PR rtl-optimization/70703
25505 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
25506 (update_conflict_hard_regno_costs): Use long instead of unsigned
25507 arithmetic for cost calculation.
25509 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25510 Bernd Edlinger <bernd.edlinger@hotmail.de>
25513 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
25516 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
25519 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
25520 ptr_mode with Pmode throughout.
25521 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
25522 into probe_stack_range and use DImode.
25524 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
25527 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
25528 call_eh_return is true.
25530 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
25532 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
25533 Initialize last_match_fntype_index.
25535 2017-04-05 Jakub Jelinek <jakub@redhat.com>
25538 * tree-nvr.c: Include internal-fn.h.
25539 (pass_return_slot::execute): Ignore internal calls without
25542 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25543 Richard Biener <rguenther@suse.de>
25546 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
25547 captures used multiple times, except for the last use.
25548 * generic-match-head.c: Include gimplify.h.
25550 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25552 PR tree-optimization/79390
25553 * target.h (struct noce_if_info): Declare.
25554 * targhooks.h (default_noce_conversion_profitable_p): Declare.
25555 * target.def (noce_conversion_profitable_p): New target hook.
25556 * ifcvt.h (struct noce_if_info): New type, moved from ...
25557 * ifcvt.c (struct noce_if_info): ... here.
25558 (noce_conversion_profitable_p): Renamed to ...
25559 (default_noce_conversion_profitable_p): ... this. No longer
25561 (noce_try_store_flag_constants, noce_try_addcc,
25562 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
25563 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
25564 instead of noce_conversion_profitable_p.
25565 * config/i386/i386.c: Include ifcvt.h.
25566 (ix86_option_override_internal): Don't override
25567 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
25568 (ix86_noce_conversion_profitable_p): New function.
25569 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
25570 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
25571 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
25572 * doc/tm.texi: Regenerated.
25574 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25576 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
25579 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
25582 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
25583 instructions for small multiply cores.
25585 2017-04-04 Jeff Law <law@redhat.com>
25587 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
25589 (mips_expand_vec_perm_const): Initialize elements in orig_perm
25590 that are not set by the loop over the elements.
25592 2017-04-04 Jakub Jelinek <jakub@redhat.com>
25595 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
25596 int mode, convert_modes it to mode as unsigned, otherwise use
25597 lowpart_subreg to mode rather than SImode.
25598 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
25599 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
25600 Use DImode instead of SImode for the shift count operand.
25601 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
25604 2017-04-04 Richard Biener <rguenther@suse.de>
25606 PR middle-end/80281
25607 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
25608 arithmetic done for the negate or the plus. Simplify.
25609 (A - (-B) -> A + B): Likewise.
25610 * fold-const.c (split_tree): Make sure to not negate pointers.
25612 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
25614 PR rtl-optimization/60818
25615 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
25616 a compare of comparisons with the thing compared if this results
25617 in a different machine mode.
25619 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
25621 * alias.c (base_alias_check): Fix typo in comment.
25622 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
25623 * cgraphunit.c (symbol_table::compile): Likewise.
25624 * collect2.c (maybe_run_lto_and_relink): Likewise.
25625 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
25626 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
25627 * config/avr/avr.c (avr_map_op_t): Likewise.
25628 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
25629 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
25630 * config/epiphany/epiphany.md (movcc): Likewise.
25631 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
25632 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
25634 * config/mips/mips.c (mips_save_restore_reg): Likewise.
25635 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
25636 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
25637 * config/sh/sh.c (sh_rtx_costs): Likewise.
25638 * fold-const.c (fold_truth_andor): Likewise.
25639 * genautomata.c (collapse_flag): Likewise.
25640 * gengtype.h (struct type::u::s): Likewise.
25641 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
25642 * input.c (FORMAT_AMOUNT): Likewise.
25643 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
25644 (known_aggs_to_agg_replacement_list): Likewise.
25645 * ipa-inline-analysis.c: Likewise.
25646 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
25647 * ipa-polymorphic-call.c
25648 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
25649 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
25650 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
25652 * modulo-sched.c (apply_reg_moves): Likewise.
25653 * omp-expand.c (build_omp_regions_1): Likewise.
25654 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
25655 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
25656 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
25657 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
25658 * value-prof.c: Likewise.
25659 * var-tracking.c (val_reset): Likewise.
25661 2017-04-03 Richard Biener <rguenther@suse.de>
25663 PR tree-optimization/80275
25664 * fold-const.c (split_address_to_core_and_offset): Handle
25667 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
25669 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
25670 descriptors is at least equal to that of functions.
25672 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
25674 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
25676 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
25679 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
25680 (mov<IMOD4:mode>): New expander.
25681 (*mov<IMOD4:mode>_internal): New insn and split pattern.
25683 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
25685 PR rtl-optimization/79405
25686 * fwprop.c (propagations_left): New variable.
25687 (forward_propagate_into): Decrement it.
25688 (fwprop_init): Initialize it.
25689 (fw_prop): If the variable has reached zero, stop propagating.
25690 (fwprop_addr): Ditto.
25692 2017-03-31 Jakub Jelinek <jakub@redhat.com>
25695 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
25696 a FUNCTION_DECL, pass it as decl instead of origin to
25699 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
25701 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
25704 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
25707 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
25708 TARGET_VSX_SMALL_INTEGER.
25710 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25712 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
25713 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
25715 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
25717 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
25718 extraction from odd-numbered MSA register.
25720 2017-03-31 Jakub Jelinek <jakub@redhat.com>
25722 PR middle-end/80173
25723 * expmed.c (store_bit_field_1): Don't attempt to create
25724 a word subreg out of hard registers wider than word if they
25725 have HARD_REGNO_NREGS of 1 for their mode.
25727 PR middle-end/80163
25728 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
25729 conversions to integer types wider than word and pointer.
25732 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
25733 (rtx_equal_for_cselib_p): Pass 0 to it.
25734 * cselib.c (cselib_hasher::equal): Likewise.
25735 (rtx_equal_for_cselib_1): Add depth argument. If depth
25736 is 128, don't look up VALUE locs and punt. Increment
25737 depth in recursive calls when walking VALUE locs.
25739 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
25741 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
25742 (make_gcov_file_name): Use the canonical path name for generating
25744 (read_line): Fix handling of files with ascii null bytes.
25746 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
25748 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
25749 to initialise a vector register instead
25750 of using a const_int.
25752 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25754 PR translation/80189
25755 * gimplify.c (omp_default_clause): Use %qs instead of %s in
25756 diagnostic messages.
25758 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
25761 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
25762 (dfp_diex_<mode>): Update mode of operand 1.
25763 * doc/extend.texi (dxex, dxexq): Document change to return type.
25764 (diex, diexq): Document change to argument type.
25766 2017-03-30 Martin Jambor <mjambor@suse.cz>
25769 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
25770 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
25771 it reflects the signature changes performed at the callee side.
25772 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
25773 to cgraph_build_function_type_skip_args.
25774 (build_function_decl_skip_args): Adjust call to the above function.
25776 2017-03-30 Jakub Jelinek <jakub@redhat.com>
25779 * config/i386/sse.md
25780 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
25781 register as dest whenever it is a MEM not rtx_equal_p to the
25782 corresponding dup operand, and when forcing into reg move the
25783 reg into the memory afterwards.
25784 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
25785 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
25786 for the force_reg mode.
25787 (avx512vl_vextractf128<mode>): Use register as dest either
25788 always when a MEM, or when it is a MEM not rtx_equal_p to the
25789 corresponding dup operand, or even not when it is a CONST_VECTOR
25790 depending on the mode and lo vs. hi.
25791 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
25793 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
25794 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
25795 Likewise. Require that operands[2] is even.
25796 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
25797 Remove extraneous parens. Require that operands[2] is a multiple
25799 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
25800 operands[0] is a MEM if <mask_applied>, the predicates/constraints
25801 disallow memory then.
25803 2017-03-30 Richard Biener <rguenther@suse.de>
25805 PR tree-optimization/77498
25806 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
25807 to non-constants over backedges.
25809 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
25811 PR rtl-optimization/80233
25812 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
25813 as last_combined_insn. Do not test for BARRIER_P separately.
25815 2017-03-29 Andreas Schwab <schwab@suse.de>
25818 * calls.c (prepare_call_address): Convert funexp to Pmode before
25819 copying to temp reg.
25821 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25823 PR tree-optimization/80158
25824 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
25825 Handle possible future case of more than one alternate
25827 (replace_rhs_if_not_dup): Likewise.
25828 (replace_one_candidate): Likewise.
25830 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
25832 PR rtl-optimization/80193
25833 * ira.c (ira): Do not check allocation for LRA.
25835 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
25837 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
25838 (nvptx_output_simt_exit): Declare.
25839 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
25840 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
25841 (init_softstack_frame): Move initialization of crtl->is_leaf to...
25842 (nvptx_declare_function_name): ...here. Emit declaration of local
25843 memory space buffer for omp_simt_enter insn.
25844 (nvptx_output_unisimt_switch): New.
25845 (nvptx_output_softstack_switch): New.
25846 (nvptx_output_simt_enter): New.
25847 (nvptx_output_simt_exit): New.
25848 * config/nvptx/nvptx.h (struct machine_function): New fields
25849 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
25850 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
25851 (UNSPECV_SIMT_EXIT): Ditto.
25852 (omp_simt_enter_insn): New insn.
25853 (omp_simt_enter): New expansion.
25854 (omp_simt_exit): New insn.
25855 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
25857 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
25858 (expand_GOMP_SIMT_ENTER_ALLOC): New.
25859 (expand_GOMP_SIMT_EXIT): New.
25860 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
25861 (GOMP_SIMT_ENTER_ALLOC): Ditto.
25862 (GOMP_SIMT_EXIT): Ditto.
25863 * target-insns.def (omp_simt_enter): New insn.
25864 (omp_simt_exit): Ditto.
25865 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
25867 (lower_rec_simd_input_clauses): Implement SIMT privatization.
25868 (lower_rec_input_clauses): Likewise.
25869 (lower_lastprivate_clauses): Handle SIMT privatization.
25871 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
25872 (ompdevlow_adjust_simt_enter): New.
25873 (find_simtpriv_var_op): New.
25874 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
25875 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
25877 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
25878 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
25879 (copy_decl_for_dup_finish): Ditto.
25881 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
25883 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
25886 * config/i386/i386.c (ix86_option_override_internal): Always
25887 allow -mpreferred-stack-boundary=3 for 64-bit targets.
25889 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25891 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
25893 2017-03-28 Bin Cheng <bin.cheng@arm.com>
25895 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
25896 mark new edge's irreducible flag accordign to it.
25897 (vect_do_peeling): Check loop preheader edge's irreducible flag
25898 and pass it to function slpeel_add_loop_guard.
25900 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
25902 PR tree-optimization/80218
25903 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
25904 Update block frequencies and counts.
25906 2017-03-28 Richard Biener <rguenther@suse.de>
25908 PR tree-optimization/78644
25909 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
25910 of a simplification result we may not use it at all.
25912 2017-03-28 Richard Biener <rguenther@suse.de>
25915 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
25916 without arguments, generate default definition of a SSA name.
25918 2017-03-28 Richard Biener <rguenther@suse.de>
25920 PR middle-end/80222
25921 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
25922 TYPE_REF_CAN_ALIAS_ALL references.
25923 * fold-const.c (fold_indirect_ref_1): Likewise.
25925 2017-03-28 Martin Liska <mliska@suse.cz>
25928 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
25929 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
25931 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25932 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
25934 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
25935 (EXTRA_SPECS): Define.
25936 (SUBTARGET_EXTRA_SPECS): Likewise.
25937 (SUBTARGET_CPP_SPEC): Likewise.
25938 * config/arc/elf.h (EXTRA_SPECS): Renamed to
25939 SUBTARGET_EXTRA_SPECS.
25940 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
25942 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
25944 * config/arc/simdext.md (vst64_insn): Update pattern.
25945 (vld32wh_insn): Likewise.
25946 (vld32wl_insn): Likewise.
25947 (vld64_insn): Likewise.
25948 (vld32_insn): Likewise.
25950 2017-03-28 Marek Polacek <polacek@redhat.com>
25953 * fold-const.c (fold_comparison): Use protected_set_expr_location
25954 instead of SET_EXPR_LOCATION.
25956 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
25958 * tree.c (add_expr): Avoid name lookup warning.
25960 2017-03-27 Jeff Law <law@redhat.com>
25962 PR tree-optimization/80216
25963 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
25964 function name. Limit recursion depth.
25965 (record_temporary_equivalences): Corresponding changes.
25967 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
25969 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
25972 2017-03-27 Jakub Jelinek <jakub@redhat.com>
25975 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
25977 * cfgcleanup.c (reg_note_cfa_p): New array.
25978 (insns_have_identical_cfa_notes): New function.
25979 (old_insns_match_p): Don't cross-jump in between /f
25980 and non-/f instructions. If both i1 and i2 are frame related,
25981 verify all CFA notes, their order and content.
25983 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
25986 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
25987 HImode and SImode with zero extend to DImode to one insn.
25988 (bswap<mode>2_extenddi): Likewise.
25989 (bswapsi2_extenddi): Likewise.
25990 (bswaphi2_extendsi): Likewise.
25991 (bswaphi2): Combine bswap HImode and SImode into one insn.
25992 Separate memory insns from swapping register.
25993 (bswapsi2): Likewise.
25994 (bswap<mode>2): Likewise.
25995 (bswaphi2_internal): Delete, no longer used.
25996 (bswapsi2_internal): Likewise.
25997 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
25998 store, and gpr<-gpr swap insns.
25999 (bswap<mode>2_store): Likewise.
26000 (bswaphi2_reg): Register only splitter, combine with the splitter.
26001 (bswaphi2 splitter): Likewise.
26002 (bswapsi2_reg): Likewise.
26003 (bswapsi2 splitter): Likewise.
26004 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
26005 the insns into load, store, and register/register insns.
26006 (bswapdi2_ldbrx): Likewise.
26007 (bswapdi2_load): Likewise.
26008 (bswapdi2_store): Likewise.
26009 (bswapdi2_reg): Likewise.
26011 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
26013 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
26014 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
26016 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
26019 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
26021 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26022 special handling for target option conflicts between dform
26023 options (-mpower9-dform, -mpower9-dform-vector,
26024 -mpower9-dform-scalar) and -mno-direct-move.
26026 2017-03-27 Richard Biener <rguenther@suse.de>
26028 PR tree-optimization/80181
26029 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
26031 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26033 * config/arc/predicates.md (move_double_src_operand): Replace the
26034 call to move_double_src_operand with a call to address_operand.
26036 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26038 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
26039 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
26040 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
26042 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26044 * config/arc/predicates.md (long_immediate_loadstore_operand):
26045 Consider scaled addresses cases.
26047 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
26049 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
26050 restored when in interrupt.
26051 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
26052 doesn't have delay slot.
26054 2017-03-27 Richard Biener <rguenther@suse.de>
26057 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
26058 inlined thunk clones.
26060 2017-03-27 Jakub Jelinek <jakub@redhat.com>
26063 * asan.c (instrument_derefs): Copy over last operand from
26064 original COMPONENT_REF to the new COMPONENT_REF with
26065 DECL_BIT_FIELD_REPRESENTATIVE.
26066 * ubsan.c (instrument_object_size): Likewise.
26068 2017-03-27 Richard Biener <rguenther@suse.de>
26070 PR tree-optimization/80170
26071 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
26072 sure DR/SCEV didnt fold in constants we do not see when looking
26073 at the reference base alignment.
26075 2017-03-27 Richard Biener <rguenther@suse.de>
26077 PR middle-end/80171
26078 * gimple-fold.c (fold_ctor_reference): Properly guard against
26079 NULL return value from canonicalize_constructor_val.
26081 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
26084 * config/i386/i386.c (ix86_expand_builtin)
26085 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
26086 flags reg setting and flags reg using instructions.
26087 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
26088 clobbering instructions to zero extend op2.
26090 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
26092 * doc/install.texi (Configuration) <--with-aix-soname>:
26093 Update link to AIX ld.
26095 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
26097 PR rtl-optimization/80160
26098 PR rtl-optimization/80159
26099 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
26100 reg_alternate_class into account.
26102 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
26105 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
26106 to consider in curr_insn_transform.
26108 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26110 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
26111 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
26112 and emit_mode_inner.
26114 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26116 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
26117 argument to the overloaded builtin variants. Use the new flag to
26118 deprecate certain builtin variants.
26119 * config/s390/s390-builtin-types.def: Add new builtin types.
26120 * config/s390/s390-builtins.h: Support new flags field for
26121 overloaded builtins.
26122 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
26123 (s390_macro_to_expand): Enable vector float data type.
26124 (s390_cpu_cpp_builtins_internal): Indicate support of the new
26125 builtins by incrementing the __VEC__ version number.
26126 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
26128 (s390_resolve_overloaded_builtin): Emit error messages depending
26129 on the builtin flags.
26130 * config/s390/s390.c (s390_expand_builtin): Support additional
26131 flags argument. Change error message to match the messages
26132 emitted in s390-c.c.
26133 * config/s390/s390.md: New UNSPEC_* constants.
26134 (op_type): Add new instruction types.
26135 * config/s390/vecintrin.h: Add new builtins and test data class
26137 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
26138 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
26139 (VEC_INEXACT, VEC_NOINEXACT): New constants.
26140 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
26141 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
26142 ("vec_mergel<mode>"): V_HW -> VEC_HW.
26144 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
26145 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
26146 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
26147 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
26149 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
26150 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
26151 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
26152 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
26154 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
26155 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
26156 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
26157 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
26158 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
26159 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
26160 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
26162 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
26163 ("vec_scatter_element<V_HW_4:mode>_DI")
26164 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
26165 ("vec_fpint<mode>", "vflls")
26166 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
26167 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
26168 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
26169 ("*vec_cmphe<mode>_cc"): ... these.
26171 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
26172 mode constant instead of magic value.
26174 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26176 * config/s390/s390.c (s390_expand_vec_compare): Support other
26177 vector floating point modes than just V2DF.
26178 (s390_expand_vcond): Likewise.
26179 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
26180 (s390_cannot_change_mode_class): Prevent mode changes between TF
26181 and V1TF in vector registers.
26182 * config/s390/s390.md (DF, SF): New mode attributes.
26183 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
26184 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
26185 SFmode support for VRs.
26186 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
26188 (VFT, VF_HW): New mode iterators.
26189 (vw, sdx): New mode attributes.
26190 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
26191 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
26192 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
26193 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
26194 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
26195 also the new vector floating point modes. Renaming to ...
26197 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
26198 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
26199 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
26200 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
26201 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
26202 ("vec_unordered<mode>"): ... these.
26204 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
26205 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
26206 ("*vec_extendv2df"): New insn definitions.
26208 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26210 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
26211 ("mulditi3_2", "*muldi3_sign"): New patterns.
26212 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
26213 rename the pattern definition.
26215 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26217 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
26219 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
26221 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26223 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
26224 instruction if possible.
26225 * config/s390/vector.md (vec_halfnumelts): New mode
26227 ("*vec_vllezlf<mode>"): New pattern.
26229 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26231 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
26232 ("popcountv4si2", "popcountv2di2"): Rename to ...
26233 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
26234 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
26236 ("popcount<mode>2_vxe"): New pattern.
26238 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26240 * common/config/s390/s390-common.c (processor_flags_table): Add
26242 * config.gcc: Add arch12.
26243 * config/s390/driver-native.c (s390_host_detect_local_cpu):
26244 Default to arch12 for unknown CPU model numbers.
26245 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
26246 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
26247 PROCESSOR_max sanity check.
26248 * config/s390/s390-opts.h (enum processor_type): Add
26250 * config/s390/s390.c (processor_table): Add arch12.
26251 (s390_expand_builtin): Add check for B_VXE flag.
26252 (s390_issue_rate): Add PROCESSOR_ARCH12.
26253 (s390_get_sched_attrmask): Likewise.
26254 (s390_get_unit_mask): Likewise.
26255 (s390_sched_score): Enable z13 scheduling for arch12.
26256 (s390_sched_reorder): Likewise.
26257 (s390_sched_variable_issue): Likewise.
26258 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
26260 (s390_tune_attr): Use z13 scheduling also for arch12.
26261 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
26262 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
26263 (TARGET_VXE_P): New macros.
26264 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
26265 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
26266 * config/s390/s390.opt: Add arch12 as processor_type.
26268 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26270 * config/s390/s390.md
26271 ("fixuns_truncdddi2", "fixuns_trunctddi2")
26272 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
26273 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
26275 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
26276 Rename expanders to ...
26278 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
26279 ("fixuns_truncdddi2_emu"): ... these.
26281 ("fixuns_trunc<mode>si2_emu"): New expander.
26283 ("*fixuns_truncdfdi2_z13"): Rename to ...
26284 ("*fixuns_truncdfdi2_vx"): ... this.
26286 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26288 * config/s390/2964.md: Remove the single element vector compare
26289 instructions which are no longer used.
26290 * config/s390/s390.c (s390_select_ccmode): Remove handling of
26292 (s390_canonicalize_comparison): Remove handling of DFmode
26294 (s390_expand_vec_compare_scalar): Remove function.
26295 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
26296 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
26298 ("*cmp<mode>_ccs"): Add wfcdb instruction.
26300 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26302 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
26304 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
26305 will anyway by matched by mov<mode>_64dfp.
26307 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26309 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
26310 vlef/vstef. Add missing operand to vleif.
26312 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26314 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
26315 pair for all vector types with 64 bit elements.
26316 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
26317 * config/s390/vector.md (V_HW_64): ... here.
26318 (V_128_NOSINGLE): New mode iterator.
26319 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
26320 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
26321 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
26322 ("*vec_load_pairv2di"): Change to ...
26323 ("*vec_load_pair<mode>"): ... this one.
26325 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26327 * config/s390/constraints.md: Add comments.
26328 (jKK): Reject element sizes > 8 bytes.
26329 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
26331 * config/s390/s390.md: Add the s_operand checks formerly in
26332 s390_split_ok_p to various splitters where they are still
26334 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
26335 for 128 bit vectors. Plus two splitters.
26337 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26339 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
26342 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26345 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
26346 error if the boundary argument is not constant.
26348 2017-03-24 Jakub Jelinek <jakub@redhat.com>
26350 PR rtl-optimization/80112
26351 * loop-doloop.c (doloop_condition_get): Don't check condition
26352 if cmp isn't SET with IF_THEN_ELSE src.
26354 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26356 PR tree-optimization/80158
26357 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
26358 replacing a candidate statement, also replace it for the
26359 candidate's alternate interpretation.
26360 (replace_rhs_if_not_dup): Likewise.
26361 (replace_one_candidate): Likewise.
26363 2017-03-24 Richard Biener <rguenther@suse.de>
26365 PR tree-optimization/80167
26366 * graphite-isl-ast-to-gimple.c
26367 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
26369 (translate_isl_ast_to_gimple::get_rename): Likewise.
26371 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
26373 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
26374 handling of certain combinations of target options, including the
26375 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
26376 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
26378 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26381 * config/arm/arm.md (*load_multiple): Add reload_completed to
26382 matching condition.
26384 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26385 Richard Biener <rguenth@suse.de>
26387 PR tree-optimization/79908
26388 PR tree-optimization/80136
26389 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26390 been cast away, gimplify_and_add suffices.
26392 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
26394 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
26396 2017-03-23 Richard Biener <rguenther@suse.de>
26398 PR tree-optimization/80032
26399 * gimplify.c (gimple_push_cleanup): Forced unconditional
26400 cleanups still have to go to the conditional_cleanups
26403 2017-03-22 Jakub Jelinek <jakub@redhat.com>
26405 PR tree-optimization/80072
26406 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
26408 (next_operand_entry_id): Change type to unsigned int.
26409 (sort_by_operand_rank): Make sure to return the right return value
26410 even if unsigned fields are bigger than INT_MAX.
26411 (struct oecount): Change cnt and id type to unsigned int.
26412 (oecount_hasher::equal): Formatting fix.
26413 (oecount_cmp): Make sure to return the right return value
26414 even if unsigned fields are bigger than INT_MAX.
26415 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
26418 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
26419 TREE_READONLY on result if writing it more than once.
26422 * doc/invoke.texi (-fsanitize=thread): Document that with
26423 -fnon-call-exceptions atomics are not able to throw
26427 * tsan.c: Include tree-eh.h.
26428 (instrument_builtin_call): Call maybe_clean_eh_stmt or
26429 maybe_clean_or_replace_eh_stmt where needed.
26430 (instrument_memory_accesses): Add cfg_changed argument.
26431 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
26432 if it returned true.
26433 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
26435 PR rtl-optimization/63191
26436 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
26437 wrapper function, moved the whole old content into ...
26438 (ix86_delegitimize_address_1): ... this. New inline function.
26439 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
26440 true as last argument instead of ix86_delegitimize_address.
26442 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26444 * config/aarch64/aarch64.c (generic_branch_cost): Copy
26445 cortexa57_branch_cost.
26447 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
26449 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
26451 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26454 * doc/md.texi (Constraints): Document wA constraint.
26455 * config/rs6000/constraints.md (wA): New.
26456 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
26457 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
26458 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
26459 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
26461 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
26464 * gimplify.c (is_oacc_declared): New function.
26465 (oacc_default_clause): Use it to set default flags for acc declared
26466 variables inside parallel regions.
26467 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
26468 declared variables.
26469 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
26470 declare attribute to any decl as necessary.
26472 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
26475 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
26476 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
26477 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
26478 (arm_arch_lpae): This.
26479 * config/arm/arm.c (arm_arch7ve): Rename into ...
26480 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
26481 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
26484 2017-03-22 Martin Liska <mliska@suse.cz>
26487 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
26488 error message instead of an ICE.
26490 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26492 * doc/extend.texi (6.11 Additional Floating Types): Revise.
26494 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
26496 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
26498 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
26501 2017-03-21 Martin Sebor <msebor@redhat.com>
26503 * doc/extend.texi: Use "cannot" instead of "can't."
26504 * doc/hostconfig.texi: Same.
26505 * doc/install.texi: Same.
26506 * doc/invoke.texi: Same.
26507 * doc/loop.texi: Same.
26508 * doc/md.texi: Same.
26509 * doc/objc.texi: Same.
26510 * doc/rtl.texi: Same.
26511 * doc/tm.texi: Same.
26512 * doc/tm.texi.in: Same.
26513 * doc/trouble.texi: Same.
26515 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
26518 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
26519 (collect_checksum_attributes): Set it.
26520 (die_checksum_ordered): Use it.
26522 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26524 PR tree-optimization/79908
26525 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
26526 change: For a VA_ARG whose LHS has been cast away, use
26527 force_gimple_operand to construct the side effects.
26529 2017-03-21 David Malcolm <dmalcolm@redhat.com>
26531 PR translation/80001
26532 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
26533 more amenable to translation.
26534 (oacc_loop_auto_partitions): Likewise.
26536 2017-03-21 Marek Polacek <polacek@redhat.com>
26537 Martin Sebor <msebor@redhat.com>
26539 PR tree-optimization/80109
26540 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
26541 on INTEGRAL_TYPE_P.
26543 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26544 Segher Boessenkool <segher@kernel.crashing.org>
26547 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
26548 check reg_used_between_p between insn and one of succ or succ2
26549 depending on if succ is artificial insn not inserted into insn
26552 2017-03-21 Martin Liska <mliska@suse.cz>
26554 PR gcov-profile/80081
26555 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
26556 * doc/gcc.texi: Include gcov-dump stuff.
26557 * doc/gcov-dump.texi: New file.
26559 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
26561 PR rtl-optimization/79150
26562 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
26563 conditional jump, if the jump is the last insn of the loop.
26565 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26566 Richard Biener <rguenth@suse.de>
26568 PR tree-optimization/79908
26569 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
26570 been cast away, use force_gimple_operand to construct the side
26573 2017-03-21 Martin Liska <mliska@suse.cz>
26575 PR libfortran/79956
26576 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
26579 2017-03-21 Brad Spengler <spender@grsecurity.net>
26582 * plugin.c (htab_hash_plugin): New function.
26583 (add_new_plugin): Use it and adjust.
26584 (parse_plugin_arg_opt): Adjust.
26585 (init_one_plugin): Likewise.
26587 2017-03-21 Richard Biener <rguenther@suse.de>
26589 PR tree-optimization/80032
26590 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
26591 if set force the cleanup to happen unconditionally.
26592 (gimplify_target_expr): Push inserted clobbers with force_uncond
26593 to avoid them being removed by control-dependent DCE.
26595 2017-03-21 Richard Biener <rguenther@suse.de>
26597 PR tree-optimization/80122
26598 * tree-inline.c (copy_bb): Do not expans va-arg packs or
26599 va_arg_pack_len when the inlined call stmt requires pack
26601 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
26603 2017-03-21 Jakub Jelinek <jakub@redhat.com>
26606 * tsan.c (instrument_builtin_call): If the memory model argument
26607 is not a constant, assume it is valid.
26610 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
26613 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
26615 PR rtl-optimization/79910
26616 * combine.c (can_combine_p): Do not allow combining an I0 or I1
26617 if its dest is used by an insn before I2 (other than the combined
26618 insns themselves, which are properly handled already).
26620 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
26623 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26625 * combine.c (record_used_regs): New static function.
26626 (try_combine): Handle situations where there is an additional
26627 instruction between I2 and I3 which needs to have a LOG_LINK
26631 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26633 * combine.c (try_combine): Delete redundant i1 test. Call
26634 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26636 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
26639 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
26640 alternatives 13/14.
26642 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
26644 PR tree-optimization/80054
26645 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
26646 the optimization if a PHI or any of its arguments is not dominated
26647 by the candidate's basis. Use gphi* rather than gimple* as
26649 (replace_profitable_candidates): Clean up a gimple* variable that
26650 should be a gphi* variable.
26652 2017-03-20 Martin Sebor <msebor@redhat.com>
26655 * doc/extend.texi (attribute constructor): Document present limitation.
26657 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
26660 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
26661 __POWER9_VECTOR__ #ifdef control, change template definition to
26662 use Power9-specific built-in function.
26663 (vec_any_eq): Likewise.
26664 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
26665 to control outcomes from this test.
26666 (vector_ae_<mode>p): For VEC_F modes, likewise.
26668 2017-03-20 Ian Lance Taylor <iant@google.com>
26670 * config/i386/i386.c (ix86_function_regparm): Save an extra
26671 register for -fsplit-stack with DECL_STATIC_CHAIN.
26673 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
26676 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
26677 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
26679 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
26681 * config/riscv/riscv.c (riscv_print_operand): Use "fence
26683 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
26686 2017-03-20 Marek Polacek <polacek@redhat.com>
26689 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
26691 2017-03-20 Richard Biener <rguenther@suse.de>
26693 PR tree-optimization/80113
26694 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
26695 allocate extra SSA name for PHI def.
26696 (add_close_phis_to_outer_loops): Likewise.
26697 (add_close_phis_to_merge_points): Likewise.
26698 (copy_loop_close_phi_args): Likewise.
26699 (copy_cond_phi_nodes): Likewise.
26701 2017-03-20 Martin Liska <mliska@suse.cz>
26703 PR middle-end/79753
26704 * tree-chkp.c (chkp_build_returned_bound): Do not build
26705 returned bounds for a LHS that's not a BOUNDED_P type.
26707 2017-03-20 Martin Liska <mliska@suse.cz>
26711 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
26712 COMPLEX_CST and VECTOR_CST.
26714 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
26717 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
26718 target operand. A new splitter adds the clobber statement in case
26719 the target operand is dead anyway.
26721 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
26723 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
26724 to age-old versions of binutils and glibc.
26726 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
26728 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
26730 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26732 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
26734 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
26736 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
26737 requirement for binutils 2.13.
26739 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
26741 * combine.c (try_combine): Delete redundant i1 test. Call
26742 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
26744 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
26746 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
26747 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
26749 <riscv64-*-elf>: Re-arrange section
26750 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
26751 <riscv32-*-linux>: Likewise.
26752 <riscv64-*-elf>: Likewise
26753 <riscv64-*-linux>: Likewise.
26755 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
26758 * aarch64.opt(verbose-cost-dump): Fix typo.
26760 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
26763 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
26764 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
26766 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
26768 * reload.c (find_reloads): When reloading a nonoffsettable address,
26769 use RELOAD_OTHER for it and its address reloads.
26771 PR rtl-optimization/79910
26772 * combine.c (record_used_regs): New static function.
26773 (try_combine): Handle situations where there is an additional
26774 instruction between I2 and I3 which needs to have a LOG_LINK
26777 2017-03-17 Jeff Law <law@redhat.com>
26779 PR tree-optimization/71437
26780 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
26781 conditional in the hash table first.
26782 (vrp_dom_walker::before_dom_children): Extract condition from
26783 ASSERT_EXPR. Record condition, its inverion and any implied
26784 conditions as well.
26786 2017-03-17 Marek Polacek <polacek@redhat.com>
26787 Markus Trippelsdorf <markus@trippelsdorf.de>
26789 PR tree-optimization/80079
26790 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
26793 2017-03-17 Richard Biener <rguenther@suse.de>
26795 PR middle-end/80075
26796 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
26797 Properly verify the LHS before the RHS possibly claims to be
26799 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
26802 2017-03-17 Martin Jambor <mjambor@suse.cz>
26804 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
26805 (List of -O2 options): Likewise.
26806 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
26809 2017-03-17 Tom de Vries <tom@codesourcery.com>
26811 * gcov-dump.c (print_usage): Print bug_report_url.
26813 2017-03-17 Richard Biener <rguenther@suse.de>
26815 PR middle-end/80050
26816 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
26817 (parser::peek): Likewise.
26819 2017-03-17 Richard Biener <rguenther@suse.de>
26821 PR tree-optimization/80048
26822 * sese.c (free_sese_info): Properly release rename_map and
26823 copied_bb_map elements.
26825 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
26827 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
26828 Add linked-list forward and backlinks. Insert on
26829 construction, remove on destruction.
26830 (class pass_store_merging): Add m_stores_head field.
26831 (pass_store_merging::terminate_and_process_all_chains):
26832 Iterate over m_stores_head list.
26833 (pass_store_merging::terminate_all_aliasing_chains):
26835 (pass_store_merging::execute): Check for debug stmts first.
26836 Push new chains onto the m_stores_head stack.
26838 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
26841 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
26842 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
26843 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
26845 2017-03-16 Jeff Law <law@redhat.com>
26847 PR tree-optimization/71437
26848 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
26849 member function. Implementation moved into after_dom_children
26850 member function and into the threader's thread_outgoing_edges
26852 (dom_opt_dom_walker::after_dom_children): Simplify by moving
26853 some code into new thread_outgoing_edges.
26854 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
26855 definition. Simplify marker handling (do it here). Assume we always
26856 have the available expression and the const/copies tables.
26857 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
26859 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
26860 * tree-vrp.c (equiv_stack): No longer file scoped.
26861 (vrp_dom_walker): New class.
26862 (vrp_dom_walker::before_dom_children): New member function.
26863 (vrp_dom_walker::after_dom_children): Likewise.
26864 (identify_jump_threads): Setup domwalker. Use it rather than
26865 walking edges in a random order by hand. Simplify setup/finalization.
26866 (finalize_jump_threads): Remove.
26867 (vrp_finalize): Do not call identify_jump_threads here.
26868 (execute_vrp): Do it here instead and call thread_through_all_blocks
26871 PR tree-optimization/71437
26872 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
26874 (simplify_stmt_for_jump_threading): Add basic_block argument. All
26876 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
26877 (dom_opt_dom_walker::thread_across_edge): Remove
26878 handle_dominating_asserts argument. All callers changed.
26879 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
26880 changes. Remove calls to lhs_of_dominating_assert. Other
26881 uses of handle_dominating_asserts turn into unconditional code
26882 (simplify_control_stmt_condition_1): Likewise.
26883 (simplify_control_stmt_condition): Likewise.
26884 (thread_through_normal_block, thread_across_edge): Likewise.
26885 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
26886 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
26887 object if it is not an SSA_NAME.
26888 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
26889 before calling into the VRP specific simplifiers.
26890 (identify_jump_threads): Remove handle_dominating_asserts
26893 2017-03-16 Jakub Jelinek <jakub@redhat.com>
26896 * tree-diagnostic.c (default_tree_printer): No longer static.
26897 * tree-diagnostic.h (default_tree_printer): New prototype.
26899 2017-03-16 Tamar Christina <tamar.christina@arm.com>
26901 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
26902 Change ins into fmov.
26904 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
26906 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
26907 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
26908 Use h_con constraint for operand 1.
26909 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
26910 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
26912 2017-03-15 Jeff Law <law@redhat.com>
26914 PR tree-optimization/71437
26915 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
26916 (record_temporary_equivalences): Use it.
26918 PR tree-optimization/71437
26919 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
26920 tree-ssa-scopedtables.
26921 (lookup_avail_expr, build_and_record_new_cond): Likewise.
26922 (record_conditions, record_cond, vuse_eq): Likewise.
26923 (record_edge_info): Adjust to API tweak of record_conditions.
26924 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
26925 (record_temporary_equivalences, optimize_stmt): Likewise.
26926 (eliminate_redundant_computations): Likewise.
26927 (record_equivalences_from_stmt): Likewise.
26928 * tree-ssa-scopedtables.c: Include options.h and params.h.
26929 (vuse_eq): New function, moved from tree-ssa-dom.c
26930 (build_and_record_new_cond): Likewise.
26931 (record_conditions): Likewise. Accept vector of conditions rather
26932 than edge_equivalence structure for first argument.
26933 for the first argument.
26934 (avail_exprs_stack::lookup_avail_expr): New member function, moved
26935 from tree-ssa-dom.c.
26936 (avail_exprs_stack::record_cond): Likewise.
26937 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
26938 from tree-ssa-dom.c.
26939 (avail_exprs_stack): Add new member functions lookup_avail_expr
26941 (record_conditions): Declare.
26943 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
26946 * lra-constraints.c (process_alt_operands): Increase reject for
26947 reloading an input/output operand.
26949 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
26952 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
26953 insns to convert from signed/unsigned char/short to IEEE 128-bit
26955 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
26957 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
26960 * config/i386/i386.c (ix86_vector_duplicate_value): Create
26961 subreg of inner mode for values already in registers.
26963 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
26965 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
26966 iteration reg is used after the loop.
26968 2017-03-14 Martin Sebor <msebor@redhat.com>
26970 PR tree-optimization/79800
26971 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
26972 precision in negative-positive range.
26973 (format_floating): Call non-const overload with adjusted precision.
26975 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
26978 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
26981 2017-03-14 Martin Sebor <msebor@redhat.com>
26983 PR middle-end/80020
26984 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
26985 * builtins.def (aligned_alloc): Use it.
26988 * Makefile.in (GTFILES): Add calls.c.
26989 * calls.c: Include "gt-calls.h".
26991 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
26993 PR rtl-optimization/79728
26994 * regs.h (struct target_regs): New field
26995 x_contains_allocatable_regs_of_mode.
26996 (contains_allocatable_regs_of_mode): New macro.
26997 * reginfo.c (init_reg_sets_1): Initialize it, and change
26998 contains_reg_of_mode so it includes global regs as well.
26999 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
27000 rather than contains_regs_of_mode.
27002 2017-03-14 Martin Liska <mliska@suse.cz>
27004 * doc/invoke.texi: Document options that can't be combined with
27005 -fcheck-pointer-bounds.
27007 2017-03-14 Martin Liska <mliska@suse.cz>
27009 PR middle-end/79831
27010 * doc/invoke.texi (-Wchkp): Document the option.
27012 2017-03-14 Martin Liska <mliska@suse.cz>
27014 * Makefile.in: Install gcov-dump.
27016 2017-03-14 Martin Liska <mliska@suse.cz>
27018 * multiple_target.c (expand_target_clones): Bail out for
27019 an invalid attribute.
27021 2017-03-14 Richard Biener <rguenther@suse.de>
27023 * alias.c (struct alias_set_entry): Pack properly.
27024 * cfgloop.h (struct loop): Likewise.
27025 * cse.c (struct set): Likewise.
27026 * ipa-utils.c (struct searchc_env): Likewise.
27027 * loop-invariant.c (struct invariant): Likewise.
27028 * lra-remat.c (struct cand): Likewise.
27029 * recog.c (struct change_t): Likewise.
27030 * rtl.h (struct address_info): Likewise.
27031 * symbol-summary.h (function_summary): Likewise.
27032 * tree-loop-distribution.c (struct partition): Likewise.
27033 * tree-object-size.c (struct object_size_info): Likewise.
27034 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
27035 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
27036 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
27037 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
27038 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
27039 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
27040 (struct _stmt_vec_info): Likewise.
27042 2017-03-14 Martin Liska <mliska@suse.cz>
27045 * multiple_target.c (create_dispatcher_calls): Check that
27046 a target can create a function dispatcher.
27048 2017-03-14 Martin Liska <mliska@suse.cz>
27051 * multiple_target.c (expand_target_clones): Drop local.local
27052 flag for default implementation.
27054 2017-03-14 Richard Biener <rguenther@suse.de>
27056 PR tree-optimization/80030
27057 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
27059 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
27061 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
27062 gcc_fallthrough() instead of __attribute__((fallthrough));
27064 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27066 * doc/gcc.texi: Remove "up" link to (DIR).
27067 * doc/gccint.texi: Ditto.
27069 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
27071 * doc/install.texi (Specific) <avr>: Remove reference to
27074 2017-03-13 Jeff Law <law@redhat.com>
27076 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
27077 attribute rather than comments.
27079 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
27080 match_scratch operand is highest.
27082 2017-03-13 Martin Liska <mliska@suse.cz>
27084 PR middle-end/78339
27085 * ipa-pure-const.c (warn_function_noreturn): If the declarations
27086 is a CHKP clone, use original declaration.
27088 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27090 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
27091 (arc_conditional_register_usage): Use a different allocation order
27092 when optimizing for size.
27093 * common/config/arc/arc-common.c (arc_option_optimization_table):
27094 Section anchors default on when optimizing for size.
27096 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27098 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
27100 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27102 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
27103 * config/arc/arc.md (cpu_facility): Add cd variant.
27104 (*movqi_insn): Add code density variant.
27105 (*movhi_insn): Likewise.
27106 (*movqi_insn): Likewise.
27107 (*addsi3_mixed): Likewise.
27108 (subsi3_insn): Likewise.
27110 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27112 * config/arc/arc.md (movsi_cond_exec): Update constraint.
27114 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
27116 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
27117 expressions with MINUS and UNARY ops.
27119 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27122 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
27124 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
27125 between vec_select and vector argument.
27126 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
27127 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
27128 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
27129 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
27130 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
27131 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
27133 2017-03-13 Richard Biener <rguenther@suse.de>
27136 * params.def (vect-max-peeling-for-alignment): Fix typo.
27138 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27140 * doc/install.texi (Specific) <mips-*-*>: Remove description of
27141 issue that only occurred with binutils below 2.18.
27143 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27145 * doc/install.texi (Specific) <cris-axis-elf>: No longer
27146 refer to binutils 2.11/2.12 minimum.
27148 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
27150 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
27151 ftp.kernel.org and simplify binutils requirement.
27153 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
27155 * doc/invoke.texi (Warning Options): Fix spelling of link-time
27157 (Optimize Options): Ditto. Also remove redundancy.
27159 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27161 PR translation/79848
27162 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
27164 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
27165 to G_ to avoid double translation.
27167 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27169 PR translation/79923
27170 * auto-profile.c (get_combined_location): Convert leading
27171 character of diagnostics to lower case and remove trailing period.
27172 (read_profile): Likewise for various diagnostics.
27173 * config/arm/arm.c (arm_option_override): Remove trailing period
27174 from various diagnostics.
27175 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
27176 (msp430_expand_delay_cycles): Likewise.
27178 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27181 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
27182 full command-line argument, rather than just "str".
27183 (aarch64_validate_march): Likewise.
27184 (aarch64_validate_mtune): Likewise.
27186 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
27188 PR rtl-optimization/78911
27189 * lra-assigns.c (must_not_spill_p): New function.
27190 (spill_for): Use it.
27192 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27194 PR tree-optimization/79981
27195 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
27196 ATOMIC_COMPARE_EXCHANGE ifn result.
27197 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
27198 IFN_ATOMIC_COMPARE_EXCHANGE.
27200 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27203 * opts.c (parse_sanitizer_options): Add missing question mark to
27204 "did you mean" message.
27206 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27208 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
27210 (VMULEUH_UNS): Likewise.
27211 (VMULOUB_UNS): Likewise.
27212 (VMULOUH_UNS): Likewise.
27213 * config/rs6000/rs6000.c (builtin_function_type): Remove
27214 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
27216 2017-03-10 David Malcolm <dmalcolm@redhat.com>
27219 * read-rtl-function.c (function_reader::read_rtx_operand): Update
27220 x with result of extra_parsing_for_operand_code_0.
27221 (function_reader::extra_parsing_for_operand_code_0): Convert
27222 return type from void to rtx, returning x. When reading
27223 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
27224 larger size containing struct block_symbol.
27226 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
27228 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
27229 -mfloat128-hardware without -m64.
27231 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
27234 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
27235 entries to the case statement that marks unsigned arguments to
27236 overloaded functions.
27238 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
27240 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
27241 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
27243 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
27246 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
27247 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
27249 2017-03-10 Martin Liska <mliska@suse.cz>
27253 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
27254 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
27257 2017-03-10 Olivier Hainque <hainque@adacore.com>
27259 * tree-switch-conversion (array_value_type): Start by resetting
27260 candidate type to it's main variant.
27262 2017-03-10 Jakub Jelinek <jakub@redhat.com>
27264 PR rtl-optimization/79909
27265 * combine.c (try_combine): Use simplify_replace_rtx on individual
27266 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
27267 of the whole CALL_INSN_FUNCTION_USAGE.
27269 PR tree-optimization/79972
27270 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
27271 get_range_info on SSA_NAMEs. Formatting fixes.
27273 2017-03-10 Richard Biener <rguenther@suse.de>
27274 Jakub Jelinek <jakub@redhat.com>
27276 PR tree-optimization/77975
27277 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
27278 edge to be constant.
27279 (get_val_for): For constant x return it. Formatting fix.
27280 (loop_niter_by_eval): Avoid pointless looping if the next iteration
27281 would use the same bases as the current one.
27283 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27285 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
27286 instead of vec_select for V1TImode.
27287 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
27289 (VSX_LE_128): Add V1TI to this mode iterator.
27290 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
27291 (*vsx_le_perm_store_<mode>): Likewise.
27292 (pre-reload splitter for VSX stores): Likewise.
27293 (post-reload splitter for VSX stores): Likewise.
27294 (*vsx_xxpermdi2_le_<mode>): Likewise.
27295 (*vsx_lxvd2x2_le_<mode>): Likewise.
27296 (*vsx_stxvd2x2_le_<mode>): Likewise.
27298 2017-03-09 Michael Eager <eager@eagercon.com>
27300 Correct failures with --enable-checking=yes,rtl.
27302 * config/microblaze/microblaze.c (microblaze_expand_shift):
27303 Replace GET_CODE test with CONST_INT_P and INTVAL test with
27304 test for const0_rtx.
27305 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
27306 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
27308 2017-03-09 Richard Biener <rguenther@suse.de>
27310 PR tree-optimization/79977
27311 * graphite-scop-detection.c (scop_detection::merge_sese):
27312 Handle the case of extra exits to blocks dominating the entry.
27314 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
27316 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
27319 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
27321 PR rtl-optimization/79949
27322 * lra-constraints.c (process_alt_operands): Check memory when
27323 trying to predict a cycle. Print about the overall increase.
27325 2017-03-09 Richard Biener <rguenther@suse.de>
27327 PR middle-end/79971
27328 * gimple-expr.c (useless_type_conversion_p): Preserve
27329 TYPE_SATURATING for fixed-point types.
27331 2017-03-09 Richard Biener <rguenther@suse.de>
27334 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
27335 alignment of BLKmode params.
27337 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27340 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
27341 (VALL_NO_V2Q): Likewise.
27343 * config/aarch64/aarch64-simd.md
27344 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
27346 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
27347 VALL_NO_V2Q mode iterator.
27348 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
27350 2017-03-09 Martin Liska <mliska@suse.cz>
27352 PR tree-optimization/79631
27353 * tree-chkp-opt.c (chkp_is_constant_addr): Call
27354 tree_int_cst_sign_bit just for INTEGER constants.
27356 2017-03-09 Martin Liska <mliska@suse.cz>
27360 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
27363 2017-03-09 Marek Polacek <polacek@redhat.com>
27366 * tree.c (inchash::add_expr): Handle TREE_VEC.
27368 2017-03-09 Martin Liska <mliska@suse.cz>
27371 (chkp_narrow_size_and_offset): New function.
27372 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
27373 (void chkp_parse_bit_field_ref): New function.
27374 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
27375 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
27377 2017-03-09 Martin Liska <mliska@suse.cz>
27380 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
27381 (chkp_find_bounds_1): Remove gcc_unreachable.
27383 2017-03-09 Jakub Jelinek <jakub@redhat.com>
27386 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
27387 BUILT_IN_SYNC*, determine the access type from the size suffix and
27388 always build a MEM_REF with that type. Handle forgotten
27389 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
27392 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
27393 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
27394 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
27395 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
27396 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
27397 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
27398 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
27399 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
27400 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
27401 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
27402 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
27403 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
27404 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
27405 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
27406 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
27407 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
27408 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
27409 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
27410 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
27411 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
27412 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
27413 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
27414 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
27415 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
27416 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
27417 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
27418 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
27419 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
27420 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
27421 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
27422 definitions outside of __OPTIMIZE__ guarded section.
27425 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
27426 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
27427 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
27428 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
27431 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27433 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
27434 ("vfenez<mode>"): Add missing constraints.
27436 2017-03-08 Martin Sebor <msebor@redhat.com>
27439 * config/nds32/nds32.c (nds32_option_override):
27440 Fix misspelled diagnostic.
27442 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27445 * gimplify.c (gimplify_omp_for): Replace index var in outer
27446 taskloop statement with an artificial variable and add
27447 OMP_CLAUSE_PRIVATE clause for it.
27449 2017-03-08 Richard Biener <rguenther@suse.de>
27451 PR tree-optimization/79955
27452 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
27453 for accesses that are completely outside of the variable.
27455 2017-03-08 Andrew Haley <aph@redhat.com>
27457 PR tree-optimization/79943
27458 * tree-ssa-loop-split.c (compute_new_first_bound): When
27459 calculating the new upper bound, (END-BEG) should be added, not
27462 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27464 * config/avr/avr.md (setmemhi): Make sure match_dup
27465 operand number comes before match_scratch.
27467 2017-03-08 Richard Biener <rguenther@suse.de>
27469 PR tree-optimization/79920
27470 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
27471 with ncopies == 1 to ...
27472 (vect_transform_slp_perm_load): ... here. Properly compute
27473 all element loads by iterating VF times over the group. Do
27474 not handle ncopies (computed in a broken way) in
27475 vect_create_mask_and_perm.
27477 2017-03-08 Jakub Jelinek <jakub@redhat.com>
27480 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
27481 is a uniform vector, use uniform_vector_p return value instead of
27482 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
27484 2017-03-07 Marek Polacek <polacek@redhat.com>
27486 PR middle-end/79809
27487 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
27488 (alloca_call_type): Likewise.
27490 2017-03-07 Martin Liska <mliska@suse.cz>
27492 * gcov.c (process_args): Put comment to correct location.
27494 2017-03-07 Martin Liska <mliska@suse.cz>
27496 PR middle-end/68270
27497 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
27498 Use array_at_struct_end_p instead of DECL_CHAIN (field).
27499 (chkp_narrow_bounds_for_field): Likewise.
27500 (chkp_parse_array_and_component_ref): Pass one more argument to
27503 2017-03-07 Richard Biener <rguenther@suse.de>
27505 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
27508 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
27510 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
27511 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
27513 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27516 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
27517 to end of description.
27518 (PARAM_MAX_STORES_TO_MERGE): Likewise.
27520 2017-03-07 Jakub Jelinek <jakub@redhat.com>
27522 PR rtl-optimization/79901
27523 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
27525 (*avx512f_<code><mode>3<mask_name>): ... this.
27526 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
27527 iterator instead of VI8_AVX2_AVX512BW.
27529 PR rtl-optimization/79901
27530 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
27531 min/max expander, expand it using expand_vec_cond_expr.
27534 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
27537 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27540 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
27541 to void * for PCH reasons.
27542 * dwarf2out.c (output_loc_operands, output_die): Cast
27543 v.val_vec.array to unsigned char *.
27545 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
27548 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
27551 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
27553 PR rtl-optimization/79571
27554 * lra-constraints.c (process_alt_operands): Calculate static
27555 reject and subtract it from overall when only addresses will be
27558 2017-03-06 Julia Koval <julia.koval@intel.com>
27561 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
27562 incoming stack boundary to 128 for 64-bit targets.
27564 2017-03-06 Richard Biener <rguenther@suse.de>
27566 PR tree-optimization/79894
27567 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
27568 to NULL after folding it.
27570 2017-03-06 Richard Biener <rguenther@suse.de>
27572 PR tree-optimization/79824
27573 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
27574 check disabling peeling for gaps.
27576 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
27578 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
27579 attributes): Document gettimeofday.
27581 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
27583 * config/s390/s390.c (s390_option_override_internal): Set
27584 PARAM_MIN_VECT_LOOP_BOUND
27586 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
27588 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
27589 * config/s390/s390.md: Likewise.
27591 2017-03-06 Jakub Jelinek <jakub@redhat.com>
27594 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
27595 (<avx2_avx512>_perm<mode>): Rename to ...
27596 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
27598 (<avx512>_perm<mode>_mask): Rename to ...
27599 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
27601 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
27602 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
27603 instead of VI8F_256_512.
27604 (avx512f_perm<mode>): New define_expand.
27605 (avx512f_perm<mode>_mask): Likewise.
27606 (avx512f_perm<mode>_1<mask_name>): New define_insn.
27607 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
27609 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27611 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
27612 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
27614 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
27616 2017-03-06 Martin Liska <mliska@suse.cz>
27619 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
27620 when having a SSA NAME w/o VAR_DECL assigned to it.
27622 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27624 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
27625 msa_dpsub_<su>_d): Fix MODE for vec_select.
27627 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
27629 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
27631 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
27633 2017-03-06 Richard Biener <rguenther@suse.de>
27635 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
27636 * plugin.c (register_plugin_info): Likewise.
27637 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
27639 2017-03-05 Jakub Jelinek <jakub@redhat.com>
27641 * config/i386/sse.md (sse_storehps, sse_storelps,
27642 avx_<castmode><avxsizesuffix>_<castmode>,
27643 avx512f_<castmode><avxsizesuffix>_<castmode>,
27644 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
27645 in condition that at least one operand is not a MEM.
27647 2017-03-03 Jakub Jelinek <jakub@redhat.com>
27649 PR middle-end/79805
27650 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
27651 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
27653 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
27654 gimple_call_nothrow_p flag based on whether original builtin can throw.
27655 If it can, emit following stmts on the fallthrough edge.
27656 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
27657 don't create new bb if inserting just debug stmts on the edge, try to
27658 insert them on the fallthru bb or just reset debug stmts.
27660 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
27663 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
27664 restore recog_data (including the operand rtxes inside it) around
27665 the call to get_insn_template.
27667 2017-03-03 Martin Sebor <msebor@redhat.com>
27669 PR tree-optimization/79699
27670 * context.c (context::~context): Free MPFR caches to avoid
27671 a memory leak on program exit.
27673 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
27675 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
27676 Use wide_int::ulow () instead of .elt (0).
27678 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
27680 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
27681 (*pushxf): Limit oF constraint to 32bit targets and add oC
27682 constraint for 64bit targets.
27683 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
27684 (*pushdf): Change rmF constraint to rmC.
27686 2017-03-03 Martin Liska <mliska@suse.cz>
27688 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
27689 Remove unused variable.
27691 2017-03-03 Jakub Jelinek <jakub@redhat.com>
27694 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
27695 is a memory operand, increase num_memory.
27696 (ix86_expand_args_builtin): Likewise.
27698 2017-03-03 Jan Hubicka <jh@suse.cz>
27701 * ipa-devirt.c (maybe_record_node): Properly handle
27702 __cxa_pure_virtual visibility.
27704 2017-03-03 Martin Liska <mliska@suse.cz>
27706 PR tree-optimization/79803
27707 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
27709 (pass_loop_prefetch::execute): Disabled optimization if an
27710 assumption about L1 cache size is not met.
27712 2017-03-03 Martin Liska <mliska@suse.cz>
27714 PR rtl-optimization/79574
27715 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
27716 (hash_scan_set): Likewise.
27717 (dump_hash_table): Likewise.
27718 (hoist_code): Likewise.
27720 2017-03-03 Richard Biener <rguenther@suse.de>
27722 * fixed-value.c (fixed_from_string): Restore use of elt (1)
27723 in place of uhigh ().
27724 (fixed_convert_from_real): Likewise.
27726 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
27729 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
27731 2017-03-03 Richard Biener <rguenther@suse.de>
27733 PR middle-end/79818
27734 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
27735 TYPE_OVERFLOW_UNDEFINED check.
27737 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27739 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
27741 (vector_ae_<mode>_p): Likewise.
27742 (vector_nez_<mode>_p): Likewise.
27743 (vector_ne_v2di_p): Likewise.
27744 (vector_ae_v2di_p): Likewise.
27745 (vector_ne_<mode>_p): Likewise.
27746 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
27748 (vsx_tsqrt<mode>2_fe): Likewise.
27750 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
27753 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
27755 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27757 PR rtl-optimization/79780
27758 * cprop.c (one_cprop_pass): When second and further conditional trap
27759 in a single basic block is turned into an unconditional trap, turn it
27760 into a deleted note to avoid RTL verification failures.
27762 2017-03-02 Richard Biener <rguenther@suse.de>
27764 * fold-const.c (const_binop): Use ulow () instead of elt (0).
27766 2017-03-02 Richard Biener <rguenther@suse.de>
27768 PR tree-optimization/79345
27770 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
27771 param and abort the walk, returning -1 if it is hit.
27772 (walk_aliased_vdefs): Take a limit param and pass it on.
27773 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
27774 defaulting to 0 and return a signed int.
27775 * tree-ssa-uninit.c (struct check_defs_data): New struct.
27776 (check_defs): New helper.
27777 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
27778 about uninitialized memory.
27779 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
27780 bogus uninitialized warning.
27781 (fixed_convert_from_real): Likewise.
27783 2017-03-02 Bin Cheng <bin.cheng@arm.com>
27785 PR tree-optimization/66768
27786 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
27787 iv_use if base object can't be determined.
27789 2017-03-02 Jakub Jelinek <jakub@redhat.com>
27791 PR tree-optimization/79345
27792 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
27793 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
27794 (get_pattern_stats): Initialize it.
27795 * genemit.c (gen_expand): Verify match_scratch numbers come after
27796 match_operand/match_dup numbers.
27797 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
27798 match_scratch numbers.
27799 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
27801 * config/s390/s390.md (trunctdsd2): Likewise.
27803 2017-03-02 Richard Biener <rguenther@suse.de>
27805 * wide-int.h (wide_int_storage::operator=): Implement in terms
27808 2017-03-02 Richard Biener <rguenther@suse.de>
27810 PR tree-optimization/79777
27811 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
27812 the to insert expression to sth existing.
27814 2017-03-01 Martin Sebor <msebor@redhat.com>
27816 PR middle-end/79692
27817 * gimple-ssa-sprintf.c
27818 (directive::known_width_and_precision): New function.
27819 (format_integer): Use it.
27820 (get_mpfr_format_length): Consider the full range of precision
27821 when computing %g output with the # flag. Set the likely byte
27822 count to 3 rather than 1 when precision is indeterminate.
27823 (format_floating): Correct the lower bound of precision.
27825 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27827 * doc/invoke.texi: Document default code model for 64-bit Linux.
27829 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
27832 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
27833 udiv rather than div since input pattern is unsigned.
27835 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
27837 * config/i386/i386.c (print_reg): Warn for values of
27838 unsupported size in integer register.
27840 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
27843 * config/rs6000/predicates.md (current_file_function_operand): Do
27844 not allow self calls to be local if the function is replaceable.
27846 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
27849 * config/rs6000/altivec.h (vec_ctz and others): Change the
27850 preprocessor macro that controls conditional compilation from
27851 _ARCH_PWR9 to __POWER9_VECTOR__.
27852 (vec_all_ne): Change parameterization of __altivec_scalar_pred
27853 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
27854 control (instead of _ARCH_PWR9 control) so that template
27855 definition uses power9-specific function.
27856 (vec_any_eq): Likewise.
27857 (vec_all_ne): Change macro definition to use a power9-specific
27858 expansion under #ifdef __POWER9_VECTOR__ control (instead of
27859 _ARCH_PWR9 control).
27860 (vec_any_eq) Likewise.
27861 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
27862 expansion for CMPNEF to remove support for xvcmpnesp instruction.
27863 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
27864 support for xvcmpnedp instruction.
27865 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
27866 macro expansion so that Power9 implementation of vec_all_ne does
27867 not use the AltiVec predicate framework.
27868 (VCMPNEH_P): Likewise.
27869 (VCMPNEW_P): Likewise.
27870 (VCMPNED_P): Likewise.
27871 (VCMPNEFP_P): Likewise.
27872 (VCMPNEDP_P): Likewise.
27873 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
27874 implementation of vec_any_eq to not use AltiVec predicate
27876 (VCMPAEH_P): Likewise.
27877 (VCMPAEW_P): Likewise.
27878 (VCMPAED_P): Likewise.
27879 (VCMPAEFP_P): Likewise.
27880 (VCMPAEDP_P): Likewise.
27881 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
27882 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
27883 not use the AltiVec predicate framework.
27884 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
27885 of vec_any_eq to not use AltiVec predicate framework.
27886 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
27887 support for predefined __POWER9_VECTOR__ macro to indicate that
27888 Power9 instruction selection is enabled.
27889 (altivec_overloaded_builtins): Remove extraneous
27890 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
27891 function argument types RS6000_BTI_bool_V16QI and
27892 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
27893 entry for overloaded function argument types RS6000_BTI_bool_V4SI
27894 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
27895 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
27896 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
27897 Power9 for implementations of vec_cmpne. Change the signature for
27898 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
27899 (representing vec_all_ne) to remove the previously described first
27900 argument of type RS6000_BTI_INTSI, as this was an artifact of
27901 reliance on the AltiVec predicate framework, which is no longer
27902 used in the implementation of these functions. Add
27903 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
27904 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
27905 since, unlike the AltiVec predicate framework implementation, we
27906 do not share function descriptors between vec_alle and vec_anyeq.
27907 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
27908 set of modes that receive special treatment even when
27909 TARGET_P9_VECTOR is true. The special treatment emits code that
27910 does not depend on Power9 instructions.
27911 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
27912 define_expand to not rely on AltiVec predicate framework.
27913 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27915 (vector_ne_v2di_p): Change this define_expand to not rely on
27916 AltiVec predicate framework.
27917 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
27919 (vector_ne_<mode>_p): Change this define_expand to not rely on
27920 AltiVec predicate framework.
27921 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
27923 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
27924 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
27925 define_insn pattern.
27926 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
27927 define_insn pattern because the xvcmpne<VSs>. instruction is not
27929 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
27930 instruction is not supported.
27932 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27934 * config/nvptx/nvptx.c: Include intl.h.
27936 2017-03-01 Martin Jambor <mjambor@suse.cz>
27939 * ipa-prop.h (ipa_bits): Removed field known.
27940 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
27941 to pointers. Adjusted their comments to warn about their sharing.
27942 (ipcp_transformation_summary): Change bits to a vector of pointers.
27943 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
27944 (ipa_get_ipa_bits_for_value): Declare.
27945 * tree-vrp.h (value_range): Mark as GTY((for_user)).
27946 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
27947 (ipa_bits_hash_table): Likewise.
27948 (ipa_vr_ggc_hash_traits): Likewise.
27949 (ipa_vr_hash_table): Likewise.
27950 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
27951 being pointers and vr_known being removed.
27952 (ipa_set_jf_unknown): Likewise.
27953 (ipa_get_ipa_bits_for_value): New function.
27954 (ipa_set_jfunc_bits): Likewise.
27955 (ipa_get_value_range): New overloaded functions.
27956 (ipa_set_jfunc_vr): Likewise.
27957 (ipa_compute_jump_functions_for_edge): Use the above functions to
27958 construct bits and vr parts of jump functions.
27959 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
27960 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27962 (ipcp_grow_transformations_if_necessary): Also allocate
27963 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
27965 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
27966 them. Fix too long lines.
27967 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
27968 vr_known being removed.
27969 (ipa_read_jump_function): Use new setter functions to construct bits
27970 and vr parts of jump functions or set them to NULL.
27971 (write_ipcp_transformation_info): Adjust for bits being pointers.
27972 (read_ipcp_transformation_info): Likewise.
27973 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
27975 Include gt-ipa-prop.h.
27976 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
27978 (ipcp_store_bits_results): Likewise.
27979 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
27980 Do not write to existing jump functions but use a temporary instead.
27982 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27985 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
27986 attempt to use its first operand as BIT_FIELD_REF base.
27988 2017-03-01 Richard Biener <rguenther@suse.de>
27990 PR middle-end/79721
27991 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
27992 interpolating formula in wrapping arithmetic.
27993 (chrec_apply): Convert chrec_evaluate return value to wanted type.
27995 2017-03-01 Jakub Jelinek <jakub@redhat.com>
27997 PR tree-optimization/79734
27998 * tree-vect-generic.c (expand_vector_condition): Optimize
27999 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
28000 Handle VEC_COND_EXPR where comparison has different inner width from
28001 type's inner width.
28003 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
28005 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
28006 markup, and similar issues. Remove @opindex entries for things
28007 that aren't options. Add missing -mmpy-option entries.
28009 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28011 PR tree-optimization/79737
28012 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
28013 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
28014 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
28015 instead of byte_size. Formatting fix.
28016 (shift_bytes_in_array_right): Formatting fix.
28018 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
28021 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
28022 condition on optimize for the leaf function test.
28024 2017-02-28 Martin Liska <mliska@suse.cz>
28027 * read-rtl-function.c (function_reader::handle_unknown_directive):
28028 Bail out when one uses -flto.
28030 2017-02-28 Martin Liska <mliska@suse.cz>
28032 * common.opt: Replace space with tabular for options of <number>
28034 * config/i386/i386.opt: Show <number> value for
28035 -mlarge-data-threshold.
28036 * opts.c (print_filtered_help): Do not display number in hexadecimal
28039 2017-02-28 Martin Liska <mliska@suse.cz>
28041 * common.opt: Fix --help=option -Q for options which are of
28044 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
28046 * config/i386/i386.c (print_reg): Error out for values
28047 of 8-bit size in invalid integer register.
28049 2017-02-28 Martin Sebor <msebor@redhat.com>
28051 PR tree-optimization/79691
28052 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
28054 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28057 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
28058 gcc_unreachable with output_operand_lossage.
28060 2017-02-28 Richard Biener <rguenther@suse.de>
28062 PR tree-optimization/79740
28063 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
28065 (visit_nary_op): Insert the nary into the hashtable if we
28066 pattern-matched sth.
28067 * tree-ssa-pre.c (eliminate_insert): Robustify.
28069 2017-02-28 Richard Biener <rguenther@suse.de>
28071 PR middle-end/79731
28072 * fold-const.c (decode_field_reference): Reject out-of-bound
28075 2017-02-28 Jakub Jelinek <jakub@redhat.com>
28077 * config/i386/i386.c: Include intl.h.
28078 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
28079 instead of just cond ? "..." : "...".
28080 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
28081 * coverage.c (read_counts_file): Likewise.
28082 * omp-offload.c: Include intl.h.
28083 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
28084 of just cond ? "..." : "...".
28085 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
28086 of just cond ? "..." : "...".
28088 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
28091 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
28093 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
28094 'tune for' CPU name.
28095 * config/arm/arm-cpu-data.h: Regenerated.
28097 2017-02-28 Richard Biener <rguenther@suse.de>
28099 PR tree-optimization/79732
28100 * tree-inline.c (expand_call_inline): Do not shadow var.
28102 2017-02-28 Richard Biener <rguenther@suse.de>
28104 PR tree-optimization/79723
28105 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
28106 address-space properly.
28108 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
28110 * doc/optinfo.texi (Optimization groups): Fix option used for
28112 * doc/invoke.texi (-fopt-info): Document "omp".
28113 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
28114 (OPTGROUP_ALL): Add OPTGROUP_OMP.
28115 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
28116 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
28117 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
28119 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
28121 * dumpfile.c (optgroup_options): Instead of "openmp", associate
28122 OPTGROUP_OMP with "omp".
28124 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
28127 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
28128 for arithmetic shift of unsigned V2DI.
28130 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
28132 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
28133 arc/linux.h headers.
28134 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
28135 (LINK_SPEC): Likewise.
28136 (ARC_TLS_EXTRA_START_SPEC): Likewise.
28137 (EXTRA_SPECS): Likewise.
28138 (STARTFILE_SPEC): Likewise.
28139 (ENDFILE_SPEC): Likewise.
28140 (LIB_SPEC): Likewise.
28141 (TARGET_SDATA_DEFAULT): Likewise.
28142 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
28143 (MULTILIB_DEFAULTS): Likewise.
28144 (DWARF2_UNWIND_INFO): Likewise.
28145 * config/arc/big.h: New file.
28146 * config/arc/elf.h: Likewise.
28147 * config/arc/linux.h: Likewise.
28148 * config/arc/t-uClibc: Remove.
28150 2017-02-27 Bin Cheng <bin.cheng@arm.com>
28152 PR tree-optimization/77536
28153 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
28154 (tree_transform_and_unroll_loop): Use above function to compute the
28155 estimated niter of unrolled loop and use it when scaling profile.
28156 Also use count info rather than frequency if it's non-zero.
28157 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
28158 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
28159 (vect_transform_loop): Call above function.
28161 2017-02-27 Richard Biener <rguenther@suse.de>
28163 PR tree-optimization/45397
28164 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
28165 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
28166 (visit_nary_op): Add pattern matching for CSEing sign-changed
28167 or truncated operations with wider ones.
28169 2017-02-27 Richard Biener <rguenther@suse.de>
28171 PR tree-optimization/79690
28172 * tree-vect-stmts.c (vectorizable_store): Use vector type
28173 built from the DR with address-space.
28175 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
28177 * doc/invoke.texi (Optimize Options): Refine the description
28178 of asan-use-after-return.
28180 2017-02-25 Alan Modra <amodra@gmail.com>
28182 PR rtl-optimization/79584
28183 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
28184 base, not ad->base_term, the reg within base. Remove assertion
28185 that ad->base == ad->base_term. Replace gen_int_mode using
28186 bogus mode with const0_rtx.
28188 2017-02-25 Jakub Jelinek <jakub@redhat.com>
28190 PR middle-end/79396
28191 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
28192 FMA_EXPR like tcc_binary or tcc_unary.
28194 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
28197 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
28199 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
28200 (output_loc_operands): Handle DW_OP_call_ref and
28201 DW_OP_GNU_variable_value.
28202 (struct variable_value_struct): New type.
28203 (struct variable_value_hasher): Likewise.
28204 (variable_value_hash): New variable.
28205 (string_types): Remove.
28206 (copy_loc_descr): New function.
28207 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
28208 (prepend_loc_descr_to_each): New function.
28209 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
28210 instead of add_loc_descr_to_each if the first argument is single
28211 location list and the second has multiple.
28212 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
28213 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
28214 when looking for variable value which doesn't have other location info.
28215 (loc_list_from_tree): Formatting fix.
28216 (gen_array_type_die): Simplify DW_AT_string_length handling.
28217 (adjust_string_types): Remove.
28218 (gen_subprogram_die): Don't call adjust_string_types nor test/set
28219 string_types. Call resolve_variable_values.
28220 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
28221 (resolve_addr_in_expr): Likewise. Add A argument.
28222 (copy_deref_exprloc): Remove deref argument. Adjust for the
28223 original expression being DW_OP_GNU_variable_value with optionally
28224 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
28225 optionally after it.
28226 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
28227 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
28228 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
28229 (variable_value_hasher::hash, variable_value_hasher::equal): New
28231 (resolve_variable_value_in_expr, resolve_variable_value,
28232 resolve_variable_values, note_variable_value_in_expr,
28233 note_variable_value): New functions.
28234 (dwarf2out_early_finish): Call note_variable_value on all toplevel
28237 2017-02-24 Jakub Jelinek <jakub@redhat.com>
28240 * opts.h (handle_generated_option): Add GENERATED_P argument.
28241 * opts-common.c (handle_option): Adjust function comment.
28242 (handle_generated_option): Add GENERATED_P argument, pass it to
28244 (control_warning_option): Pass false to handle_generated_option
28246 * opts.c (maybe_default_option): Pass true to handle_generated_option
28248 * optc-gen.awk: Likewise.
28250 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28252 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
28253 a REG, look at the REG it is a SUBREG of.
28254 (splitter for cmpeqsi_t): Ditto.
28256 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28258 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
28259 the special USEs with the pattern of the insn, not the insn itself.
28261 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
28264 * doc/invoke.texi: Document -mload-store-pairs.
28266 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28267 Sandra Loosemore <sandra@codesourcery.com>
28269 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
28270 argument isn't a CONST_INT.
28271 (nios2_alternate_compare_const): Assert op is a CONST_INT.
28272 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
28273 (nios2_validate_compare): Bypass alternate compare logic if *op2
28274 is not a CONST_INT.
28275 (ldstwm_operation_p): Return false if first_base is not a REG or
28276 if first_offset is not a CONST_INT.
28278 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28280 * config/cris/cris.md: Use correct operand in a define_peephole2.
28282 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28284 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
28286 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
28288 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
28289 this_insn if it is an INSN or JUMP_INSN.
28290 (force_offsettable): Look at base, not at addr.
28291 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
28292 on things that aren't necessarily CONST_INTs.
28294 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
28296 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
28297 -mfpmath=sse is the default also for x86-32 targets with SSE2
28298 instruction set when @option{-ffast-math} is enabled
28300 2017-02-24 Jeff Law <law@redhat.com>
28302 PR rtl-optimizatoin/79286
28303 * ira.c (update_equiv_regs): Drop may_trap_p exception to
28306 2017-02-24 Richard Biener <rguenther@suse.de>
28308 PR tree-optimization/79389
28309 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
28312 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
28314 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
28315 function comment to reflect reality.
28316 (loop_exits_before_overflow): Fix typo in function description.
28318 2017-02-24 Richard Biener <rguenther@suse.de>
28320 PR tree-optimization/79389
28321 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
28322 properly that a threading opportunity exists. Detect conditional
28323 copy/constant propagation opportunities.
28325 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
28327 * config/visium/visium.md (type): Add trap.
28328 (b): New mode attribute.
28329 (*btst): Rename into...
28330 (*btst<mode>): ...this and adjust.
28331 (*cbranchsi4_btst_insn): Rename into...
28332 (*cbranch<mode>4_btst_insn): ...this and adjust.
28333 (trap): New define_insn.
28335 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28337 PR tree-optimization/79389
28338 * ifcvt.c (struct noce_if_info): Add rev_cond field.
28339 (noce_reversed_cond_code): New function.
28340 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
28341 reversed_comparison_code. Formatting fix.
28342 (noce_try_store_flag): Test rev_cond != NULL in addition to
28343 reversed_comparison_code.
28344 (noce_try_store_flag_constants): Likewise.
28345 (noce_try_store_flag_mask): Likewise.
28346 (noce_try_addcc): Use rev_cond if non-NULL instead of
28347 reversed_comparison_code.
28348 (noce_try_cmove_arith): Likewise. Formatting fixes.
28349 (noce_try_minmax, noce_try_abs): Clear rev_cond.
28350 (noce_find_if_block): Initialize rev_cond.
28351 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
28352 instead of false as last argument never attempt to reverse it
28355 2017-02-23 Bin Cheng <bin.cheng@arm.com>
28357 PR tree-optimization/79663
28358 * tree-predcom.c (combine_chains): Process refs in reverse order
28359 only for ZERO length chains, and add explaining comment.
28361 2017-02-23 Jeff Law <law@redhat.com>
28363 PR tree-optimization/79578
28364 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
28365 in call to operand_equal_p.
28367 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
28370 * config/i386/cpuid.h: Fix another undefined behavior.
28372 2017-02-23 Richard Biener <rguenther@suse.de>
28374 PR tree-optimization/79683
28375 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
28376 vector types for data-refs.
28378 2017-02-23 Martin Liska <mliska@suse.cz>
28380 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
28382 2017-02-23 Jakub Jelinek <jakub@redhat.com>
28384 PR middle-end/79665
28385 * internal-fn.c (get_range_pos_neg): Moved to ...
28386 * tree.c (get_range_pos_neg): ... here. No longer static.
28387 * tree.h (get_range_pos_neg): New prototype.
28388 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
28389 are known to be in between 0 and signed maximum inclusive, try to
28390 expand both unsigned and signed divmod and use the cheaper one from
28393 2017-02-22 Jeff Law <law@redhat.com>
28395 PR tree-optimization/79578
28396 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
28397 to compare base operands.
28399 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
28402 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
28403 gpc_reg_operand instead of fpr_reg_operand.
28405 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
28407 * config/mips/mips.c (mips_return_in_memory): Force FP
28408 vector types to be returned in memory for o32 ABI.
28410 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28412 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
28413 instead of DW_TAG_member for static data member declarations and don't
28414 set no_linkage_name for static inline data members.
28415 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
28418 2017-02-22 Martin Liska <mliska@suse.cz>
28420 * doc/invoke.texi: Replace inequality signs with square brackets
28423 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28425 PR tree-optimization/68644
28426 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
28428 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
28431 * lra-constraints.c (simplify_operand_subreg): Handle
28432 WORD_REGISTER_OPERATIONS targets.
28434 2017-02-22 Jakub Jelinek <jakub@redhat.com>
28437 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
28438 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
28439 elimination by swapping fld*.
28441 2017-02-22 Richard Biener <rguenther@suse.de>
28443 PR tree-optimization/79673
28444 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
28445 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
28446 irrelevant address-space qualifiers and avoiding a
28447 ADDR_SPACE_CONVERT_EXPR from fold_convert.
28449 2017-02-22 Richard Biener <rguenther@suse.de>
28451 PR tree-optimization/79666
28452 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
28453 to not symbolically negate if that may introduce undefined
28456 2017-02-22 Martin Liska <mliska@suse.cz>
28459 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
28460 * data-streamer-out.c (streamer_write_gcov_count_stream):
28462 * value-prof.c (stream_out_histogram_value): Make assert more
28463 precise based on type of counter.
28465 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
28468 * config/i386/i386.md (standard_x87sse_constant_load splitter):
28469 Use nonimmediate_operand instead of memory_operand for operand 1.
28470 (float-extend standard_x87sse_constant_load splitter): Ditto.
28472 2017-02-21 Jeff Law <law@redhat.com>
28474 PR tree-optimization/79621
28475 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
28476 blocks with edges to themselves.
28478 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28481 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
28482 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
28483 Use gimple_call_builtin_p.
28486 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
28487 on temporarily removed DEBUG_INSNs.
28489 PR tree-optimization/79649
28490 * tree-loop-distribution.c (classify_partition): Give up on
28491 non-generic address space loads/stores.
28493 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
28495 * doc/loop.texi (Loop manipulation): Remove nonexistent
28496 tree_ssa_loop_version from the documentation.
28497 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
28499 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28502 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
28503 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
28504 * config/rs6000/rs6000.c: Include except.h.
28505 (rs6000_expand_split_stack_prologue): Call
28506 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
28508 2017-02-21 Martin Jambor <mjambor@suse.cz>
28511 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
28512 have been analyzed.
28514 2017-02-21 Martin Jambor <mjambor@suse.cz>
28516 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
28517 for backward compatibility only.
28518 * doc/invoke.texi (Option Summary): Remove all references to
28519 -fipa-cp-alignment.
28521 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
28525 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28527 * lra-constraints.c (curr_insn_transform): Handle
28528 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28530 2017-02-21 Martin Liska <mliska@suse.cz>
28532 * config/i386/i386.opt: Replace -masm-dialect with -masm.
28534 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
28536 PR translation/79638
28537 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
28539 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
28542 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
28543 (arm_function_ok_for_sibcall): Return false for an indirect call by
28544 descriptor if all the argument registers are used.
28545 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
28546 alignment of the function.
28548 2017-02-21 Jakub Jelinek <jakub@redhat.com>
28550 PR tree-optimization/61441
28551 * simplify-rtx.c (simplify_const_unary_operation): For
28552 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
28553 the sNaN unmodified.
28555 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
28557 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
28558 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
28559 instead of SYSTEM_HEADER_DIR.
28561 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
28562 Martin Liška <mliska@suse.cz>
28564 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
28565 Fix typos and grammar, use active voice, and clarify.
28567 2017-02-20 Marek Polacek <polacek@redhat.com>
28569 PR middle-end/79537
28570 * gimplify.c (gimplify_expr): Handle unused *&&L;.
28573 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
28575 2017-02-20 Jakub Jelinek <jakub@redhat.com>
28578 * config/i386/i386.c (ix86_expand_builtin): Handle
28579 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
28580 ix86_builtins_isa[fcode].isa as a requirement of those
28581 flags and any other flag in the bitmask.
28582 (ix86_init_mmx_sse_builtins): Use 0 instead of
28583 ~OPTION_MASK_ISA_64BIT as mask.
28584 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
28585 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
28586 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
28587 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
28589 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28592 * lra-constraints.c (split_reg): Check requested split mode
28593 is supported by the register.
28595 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28597 * lra-constraints.c (simplify_operand_subreg): Remove early
28600 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28603 * lra-constraints.c (curr_insn_transform): Tighten condition
28604 for converting SUBREG reloads from OP_OUT to OP_INOUT.
28606 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
28609 * lra-constraints.c (curr_insn_transform): Handle
28610 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
28612 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
28615 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
28617 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
28619 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
28622 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
28625 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28627 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
28630 2017-02-18 Jakub Jelinek <jakub@redhat.com>
28632 * final.c (last_columnnum, override_columnnum): New variables.
28633 (final_start_function): Set last_columnnum, pass it to begin_prologue
28634 hook and pass 0 to dwarf2out_begin_prologue.
28635 (final_scan_insn): Update override_columnnum. Pass last_columnnum
28636 to source_line debug hook.
28637 (notice_source_line): Compute last_columnnum and for debug_column_info
28638 return true on column changes.
28639 * debug.h (struct gcc_debug_hooks): Add column argument to
28640 source_line and begin_prologue hooks.
28641 (debug_nothing_int_charstar_int_bool): Remove prototype.
28642 (debug_nothing_int_int_charstar,
28643 debug_nothing_int_int_charstar_int_bool): New prototypes.
28644 (dwarf2out_begin_prologue): Add column argument.
28645 * debug.c (do_nothing_debug_hooks): Adjust source_line and
28646 begin_prologue hooks.
28647 (debug_nothing_int_charstar_int_bool): Remove.
28648 (debug_nothing_int_int_charstar,
28649 debug_nothing_int_int_charstar_int_bool): New functions.
28650 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
28651 through to dwarf2out_source_line.
28652 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
28653 (dwarf2out_source_line): Add column argument, emit it if requested.
28654 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
28656 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
28657 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
28658 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
28659 through to dwarf2out_begin_prologue.
28660 (vmsdbgout_source_line): Add column argument, pass it through to
28661 dwarf2out_source_line.
28662 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
28663 dbxout_source_line caller.
28664 (dbxout_source_line): Add column argument.
28666 * common.opt (gno-column-info, gcolumn-info): New options.
28667 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
28668 (check_die): Also test for multiple DW_AT_decl_column attributes.
28669 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
28670 DW_AT_decl_column if requested.
28671 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
28673 (gen_variable_die): Likewise.
28674 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
28675 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
28678 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
28679 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
28680 (ix86_handle_option): Handle OPT_m3dnowa.
28681 * doc/invoke.texi (-m3dnowa): Document.
28682 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
28683 -m3dnowa instead of -m3dnow -march=athlon.
28686 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
28687 instead of gcc_assert for K, r and R code checks. Formatting fixes.
28689 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28692 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
28693 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
28694 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
28695 generator for vsx_xxpermdi_<mode>_be.
28696 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
28697 force big-endian semantics.
28698 (vsx_xxpermdi_<mode>_be): New define_expand with same
28699 implementation as previous version of vsx_xxpermdi_<mode>.
28701 2017-02-17 Jakub Jelinek <jakub@redhat.com>
28703 PR tree-optimization/79327
28704 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
28705 variable, its initialization and use.
28707 2017-02-17 Julia Koval <julia.koval@intel.com>
28709 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
28710 (OPTION_MASK_ISA_PKU_UNSET): New.
28711 (ix86_handle_option): Handle -mrdpid.
28712 * config/i386/cpuid.h (bit_RDPID): New.
28713 * config/i386/driver-i386.c (host_detect_local_cpu):
28714 Detect RDPID feature.
28715 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
28716 * config/i386/i386-c.c (ix86_target_macros_internal):
28718 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
28719 (ix86_valid_target_attribute_inner_p): Add "rdpid".
28720 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
28721 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
28722 * config/i386/i386.md (define_insn "rdpid"): New.
28723 * config/i386/i386.opt Add -mrdpid.
28724 * config/i386/immintrin.h (_rdpid_u32): New.
28726 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
28728 PR rtl-optimization/79541
28729 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
28730 instead of transforming it into USE.
28732 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
28734 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
28735 If HONOR_SNANS (SFmode) force the input to a register.
28736 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
28737 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
28738 an frsp or similar insn.
28740 2017-02-17 Martin Liska <mliska@suse.cz>
28742 PR rtl-optimization/79577
28743 * params.def (selsched-max-sched-times): Increase minimum to 1.
28745 2017-02-17 Martin Liska <mliska@suse.cz>
28747 PR rtl-optimization/79574
28748 * gcse.c (want_to_gcse_p): Prevent integer overflow.
28750 2017-02-17 Martin Liska <mliska@suse.cz>
28752 PR tree-optimization/79529
28753 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
28754 ssa_defined_default_def_p to handle cases which are implicitly
28756 * tree-ssa.c (ssa_defined_default_def_p): New function.
28757 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
28758 which are implicitly defined.
28759 * tree-ssa.h (ssa_defined_default_def_p): Declare.
28761 2017-02-17 Richard Biener <rguenther@suse.de>
28763 PR middle-end/79576
28764 * params.def (max-ssa-name-query-depth): Limit to 10.
28766 2017-02-17 Richard Biener <rguenther@suse.de>
28768 PR tree-optimization/79552
28769 * tree-ssa-structalias.c (visit_loadstore): Properly verify
28772 2017-02-17 Richard Biener <rguenther@suse.de>
28775 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
28777 2017-02-17 Marek Polacek <polacek@redhat.com>
28779 PR middle-end/79536
28780 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
28781 (fold_negate_expr): New wrapper.
28783 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
28785 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
28786 Correct terminology and de-emphasize pre-standard behavior.
28788 2017-02-16 Alan Modra <amodra@gmail.com>
28790 PR rtl-optimization/79286
28791 * ira.c (def_dominates_uses): New function.
28792 (update_equiv_regs): Don't create an equivalence for insns that
28793 may trap where the register def does not dominate the use.
28795 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
28797 PR rtl-optimization/78127
28798 * lra.c (lra): Call lra_eliminate before finish the loop after
28801 2017-02-16 Richard Biener <rguenther@suse.de>
28803 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
28805 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
28806 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
28807 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
28808 (isl_val_int_from_wi): New function.
28809 (extract_affine_gmp): Rename to ...
28810 (extract_affine_wi): ... this, take a widest_int.
28811 (extract_affine_int): Just wrap extract_affine_wi.
28812 (add_param_constraints): Use isl_val_int_from_wi.
28813 (add_loop_constraints): Likewise, and extract_affine_wi.
28815 2017-02-15 Jeff Law <law@redhat.com>
28817 PR middle-end/79521
28818 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
28819 ira_init_register_move_cost_if_necessary.
28821 2017-02-15 Martin Sebor <msebor@redhat.com>
28823 PR middle-end/32003
28824 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
28825 removed in a prior commit.
28827 2017-02-15 Bin Cheng <bin.cheng@arm.com>
28829 PR tree-optimization/79347
28830 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
28831 counters during peeling.
28833 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
28835 * Makefile.in (site.exp): Remove "set ISLVER".
28837 2017-02-15 Jakub Jelinek <jakub@redhat.com>
28840 * real.c (real_from_integer): Call real_convert even for decimal.
28842 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
28845 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
28847 2017-02-14 Andrew Pinski <apinski@cavium.com>
28849 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
28850 cores and change the partno/implementer to be correct.
28851 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
28852 the 'B" as the implementer.
28853 * config/aarch64/aarch64-tune.md: Regenerate.
28855 2017-02-14 Carl Love <cel@us.ibm.com>
28857 * config/rs6000/rs6000.c: Add case statement entry to make the
28858 xvcvuxdsp built-in argument unsigned.
28859 * config/rs6000/vsx.md: Fix the source and return operand types so they
28860 match the instruction definitions from the ISA document. Fix typo
28861 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
28864 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
28867 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
28868 member early_clobber_alts.
28869 * lra-lives.c (reg_early_clobber_p): New.
28870 (process_bb_lives): Use it.
28871 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
28872 (debug_operand_data): Initialize early_clobber_alts.
28873 (setup_operand_alternative): Set up early_clobber_alts.
28874 (collect_non_operand_hard_regs): Ditto. Pass early clobber
28875 alternatives to new_insn_reg.
28876 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
28878 (lra_update_insn_regno_info): Pass the new arg.
28880 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28882 PR middle-end/79505
28883 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
28884 (new_oacc_loop_raw): Don't clear already cleared fields.
28887 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
28888 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
28889 _mm512_prefetch_i64gather_ps): New inline functions and macros.
28891 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
28894 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
28896 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
28899 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
28900 the extra instruction to the right place to store 128-bit constant
28903 2017-02-14 Martin Sebor <msebor@redhat.com>
28905 PR middle-end/79448
28906 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
28907 warning for strings of unknown length.
28909 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
28911 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
28913 2017-02-14 Jeff Law <law@redhat.com>
28916 * ira-costs.c (scan_one_insn): Initialize register move costs
28917 for pseudos seen in USE/CLOBBER insns.
28919 PR tree-optimization/79095
28920 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
28921 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
28922 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
28923 if the operands are known to be not equal, then the resulting range
28925 (intersect_ranges): If the new range is ~[0,0] and the old range is
28926 wide, then prefer ~[0,0].
28927 * tree-vrp.c (overflow_comparison_p_1): New function.
28928 (overflow_comparison_p): New function.
28929 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
28930 if NAME is used in an overflow test.
28931 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
28932 overflow check that can be expressed as an equality test, then adjust
28933 ops to be that equality test.
28935 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28937 * config/s390/s390-builtin-types.def: Remove flags argument.
28938 * config/s390/s390.c (s390_init_builtins): Likewise.
28940 2017-02-14 Martin Liska <mliska@suse.cz>
28942 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
28943 vector. Fix trailing white spaces.
28945 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
28947 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
28950 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28952 PR rtl-optimization/68664
28953 * config/arm/arm.c (arm_sched_can_speculate_insn):
28954 New function. Declare prototype.
28955 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28957 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28959 PR rtl-optimization/68664
28960 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
28962 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
28964 2017-02-14 Amit Pawar <amit.pawar@amd.com>
28966 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
28967 max skip bytes for function, loop and jump.
28969 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
28971 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
28972 ABS_EXPR for gimple dump.
28974 2017-02-14 Jakub Jelinek <jakub@redhat.com>
28977 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
28979 PR tree-optimization/79408
28980 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
28981 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
28982 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
28983 also if rhs1 is INTEGER_CST.
28985 2017-02-14 Richard Biener <rguenther@suse.de>
28987 PR middle-end/79432
28988 * tree-into-ssa.c (insert_phi_nodes): When the function can
28989 have abnormal edges rewrite SSA names with broken use-def
28990 dominance out of SSA and register them for PHI insertion.
28992 2017-02-13 Martin Sebor <msebor@redhat.com>
28994 PR middle-end/79496
28995 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
28996 clearing info.nowrite flag when snprintf size argument is a range.
28998 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29000 * cprop.c (cprop_jump): Add missing space in string literal.
29001 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
29002 (get_constraint_for_component_ref): Likewise.
29003 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
29004 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
29005 * lra-constraints.c (process_alt_operands): Likewise.
29006 * ipa-inline.c (inline_small_functions): Likewise.
29007 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
29008 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
29009 * trans-mem.c (diagnose_tm_1_op): Likewise.
29010 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
29011 (grid_parallel_clauses_gridifiable): Likewise.
29013 * config/nvptx/mkoffload.c (process): Add space in between
29016 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
29017 "MOD4_SSE_REGS" and "ALL_REGS".
29019 * spellcheck.c (test_data): Add , in between "foo" and "food".
29021 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29024 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
29025 boundary crossing check and subsequent code generation agree.
29027 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29029 * config/aarch64/aarch64.c (has_memory_op): Delete.
29030 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
29033 2017-02-13 Jakub Jelinek <jakub@redhat.com>
29035 PR rtl-optimization/79388
29036 PR rtl-optimization/79450
29037 * combine.c (distribute_notes): When removing TEM_INSN for which
29038 corresponding dest has last value recorded, invalidate that last
29041 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29043 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
29044 of explicit '@'. Add missing assembly comment marker on branch costs
29047 2017-02-13 Nathan Sidwell <nathan@acm.org>
29049 * gengtype-lex.l (<in_struct>): Add '/'.
29051 2017-02-13 Martin Liska <mliska@suse.cz>
29054 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
29056 2017-02-13 Richard Biener <rguenther@suse.de>
29058 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
29060 * configure: Re-generate.
29061 * config.in: Likewise.
29062 * graphite-dependences.c: Simplify as if
29063 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
29064 * graphite-isl-ast-to-gimple.c: Likewise.
29065 * graphite-optimize-isl.c: Likewise.
29066 * graphite-poly.c: Likewise.
29067 * graphite-sese-to-poly.c: Likewise.
29068 * graphite.h: Likewise.
29069 * toplev.c: Include isl/version.h and use isl_version () for
29070 printing the ISL version.
29071 * doc/install.texi: Update ISL requirement.
29073 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29075 * doc/standards.texi (Standards): Update reference to
29078 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
29080 * doc/extend.texi (Named Address Spaces): sourceware.org now
29082 * doc/install.texi (Binaries): Ditto.
29085 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29087 * doc/cpp.texi: Replace "stringify"/"stringification" with C
29088 standard terminology "stringize"/"stringizing" throughout.
29089 * doc/cppinternals.texi: Likewise.
29091 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
29093 * doc/extend.texi: Fix some spelling mistakes and typos.
29094 * doc/invoke.texi: Likewise.
29096 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29099 * params.def (inline-min-speedup) Change from 10 to 8.
29101 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29103 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
29106 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
29109 * ipa-inline-analysis.c (get_minimal_bb): New function.
29110 (record_modified): Use it.
29111 (remap_edge_change_prob): Handle also ancestor functions.
29113 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
29115 * doc/contrib.texi (Contributors): Remove broken link into
29116 the Mauve CVS repository.
29118 2017-02-11 Jakub Jelinek <jakub@redhat.com>
29120 PR middle-end/79454
29121 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
29122 result computation whenever lhs doesn't have vector mode, not
29123 just when it has BLKmode.
29125 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29127 * doc/makefile.texi (profiledbootstrap): Refer to the
29128 installation instructions only in textual form.
29130 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29133 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
29135 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
29137 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
29138 (Specific): Update mingw-w64 reference.
29140 (Specific): Remove broken link to Renesas RX processor.
29142 2017-02-10 Richard Biener <rguenther@suse.de>
29144 * toplev.c (process_options): Do not mention obsolete graphite
29145 options when printing sorry message about missing graphite support.
29146 Mention -floop-nest-optimize.
29148 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
29150 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
29151 (vtst_p16): Likewise.
29152 (vtstq_p8): Likewise.
29153 (vtstq_p16): Likewise.
29155 (vtstq_p64): Likewise.
29156 * config/arm/arm_neon.h (vgetq_lane_p64): New.
29157 (vset_lane_p64): New.
29158 (vsetq_lane_p64): New.
29160 2017-02-10 Jakub Jelinek <jakub@redhat.com>
29162 PR tree-optimization/79411
29163 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
29164 stmt operands are SSA_NAMEs used in abnormal phis.
29165 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
29168 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29171 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
29174 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
29176 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
29178 2017-02-09 Jakub Jelinek <jakub@redhat.com>
29180 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
29184 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
29185 not arbitrary TREE_CONSTANT.
29188 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
29189 "omp declare target link" attribute unless is_global_var.
29190 * omp-offload.c (find_link_var_op): Likewise.
29192 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
29193 Chung-Lin Tang <cltang@codesourcery.com>
29195 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
29197 (gimplify_adjust_omp_clauses): Don't delete TILE.
29198 (gimplify_omp_for): Deal with TILE.
29199 * internal-fn.c (expand_GOACC_TILE): New function.
29200 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
29202 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
29203 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
29205 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
29206 avoid DIV for outermost collapse var.
29207 (expand_oacc_for): Insert tile element loop as needed. Adjust.
29208 Remove out of date comments, fix whitespace.
29209 * omp-general.c (omp_extract_for_data): Deal with tiling.
29210 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
29211 adjust OLF_DIM_BASE value.
29212 (struct omp_for_data): Add tiling field.
29213 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
29214 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
29215 for auto loops. Remove default auto determining, moved to
29216 oacc_loop_fixed_partitions.
29217 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
29218 stmts, add e_mask field.
29219 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
29220 (oacc_thread_numbers): Use oacc_dim_call.
29221 (oacc_xform_tile): New.
29222 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
29223 (finish_oacc_loop): Adjust for ifns vector.
29224 (oacc_loop_discover_walk): Append loop abstraction sites to list,
29225 add case for GOACC_TILE fns.
29226 (oacc_loop_xform_loop): Delete.
29227 (oacc_loop_process): Iterate over call list directly, and add
29228 handling for GOACC_TILE fns.
29229 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
29231 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
29232 vector partitioning to outer loops. Assign 2 partitions to loops
29233 when available. Add TILE handling.
29234 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
29235 (execite_oacc_device_lower): Process GOACC_TILE fns,
29236 ignore unknown specs.
29237 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
29238 * tree.c (omp_clause_num_ops): Adjust TILE ops.
29239 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
29241 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
29243 * configure.ac (ACX_BUGURL): Update.
29244 * configure: Regenerate.
29246 2017-02-09 Richard Biener <rguenther@suse.de>
29248 PR tree-optimization/69823
29249 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
29250 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
29252 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29254 * config/arc/arc-c.def: Add __NPS400__ definition.
29255 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
29256 (TARGET_NPS400): Define.
29258 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
29260 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
29262 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
29263 pointer, arch_info.
29264 (arc_cpu_types): Fill the arch_info field with a pointer into the
29265 arc_arch_types table.
29266 (arc_selected_cpu): Declare.
29267 * config/arc/arc.c (arc_selected_cpu): Make global.
29268 (arc_selected_arch): Delete.
29269 (arc_base_cpu): Delete.
29270 (arc_override_options): Remove references to deleted variables,
29271 update access to arch information.
29272 (ARC_OPT): Update access to arch information.
29273 (ARC_OPTX): Likewise.
29274 * config/arc/arc.h (arc_base_cpu): Remove declaration.
29275 (TARGET_ARC600): Update access to arch information.
29276 (TARGET_ARC601): Likewise.
29277 (TARGET_ARC700): Likewise.
29278 (TARGET_EM): Likewise.
29279 (TARGET_HS): Likewise.
29280 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
29283 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
29286 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
29287 condition/operands for integer GE/LE/GEU/LEU operations.
29289 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
29291 PR translation/79397
29292 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
29295 2017-02-08 Martin Jambor <mjambor@suse.cz>
29298 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
29299 whether allocation happened.
29300 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
29301 nothing was allocated.
29303 2017-02-08 Jakub Jelinek <jakub@redhat.com>
29305 PR tree-optimization/79408
29306 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
29307 constant, but SSA_NAME with a known integer range, use the minimum
29308 of that range instead of op1 to determine if modulo can be replaced
29309 with its first operand.
29311 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29313 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
29315 2017-02-08 Richard Biener <rguenther@suse.de>
29317 PR tree-optimization/71824
29318 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29319 Check all loops contained in the merged region.
29321 2017-02-07 Andrew Pinski <apinski@cavium.com>
29323 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
29325 2017-02-07 Andrew Pinski <apinski@cavium.com>
29327 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
29328 (thunderxt88): Likewise.
29329 (thunderxt81): Disable LSE and change v8.1 to v8.
29330 (thunderxt83): Likewise.
29332 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29333 Richard Biener <rguenther@suse.de>
29335 PR middle-end/79399
29336 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
29337 type from int to size_t.
29338 * ira-costs.c (struct_costs_size): Change type from int to size_t.
29340 2017-02-07 Jakub Jelinek <jakub@redhat.com>
29342 PR rtl-optimization/79386
29343 * cprop.c (bypass_conditional_jumps): Initialize
29344 bypass_last_basic_block already before splitting bbs after
29345 unconditional traps...
29346 (bypass_conditional_jumps): ... rather than here.
29349 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
29350 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
29351 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
29352 fix -masm=intel patterns.
29354 2017-02-07 Richard Biener <rguenther@suse.de>
29356 PR tree-optimization/79256
29357 PR middle-end/79278
29358 * builtins.c (get_object_alignment_2): Use min_align_of_type
29359 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
29360 and ADJUST_FIELD_ALIGN.
29362 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
29364 * doc/tm.texi: Regenerate.
29365 * stor-layout.c (layout_decl): Adjust.
29366 (update_alignment_for_field): Likewise.
29367 (place_field): Likewise.
29368 (min_align_of_type): Likewise.
29369 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
29370 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
29371 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
29372 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
29373 * config/frv/frv.c (frv_adjust_field_align): Likewise.
29374 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
29375 * config/i386/i386.c (x86_field_alignment): Likewise.
29376 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
29377 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
29378 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
29379 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
29380 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
29381 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
29385 2017-01-30 Richard Biener <rguenther@suse.de>
29387 PR tree-optimization/79256
29388 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
29389 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
29392 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
29394 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
29395 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
29396 builtins to SImode and emit a zero-extend, if necessary.
29398 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29400 * docs/invoke.texi (RISC-V Options): Alphabetize.
29402 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29404 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
29407 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
29409 * config/riscv/riscv.c: New file.
29410 * common/config/riscv/riscv-common.c: Likewise.
29411 * config.gcc: Likewise.
29412 * config/riscv/constraints.md: Likewise.
29413 * config/riscv/elf.h: Likewise.
29414 * config/riscv/generic.md: Likewise.
29415 * config/riscv/linux.h: Likewise.
29416 * config/riscv/multilib-generator: Likewise.
29417 * config/riscv/peephole.md: Likewise.
29418 * config/riscv/pic.md: Likewise.
29419 * config/riscv/predicates.md: Likewise.
29420 * config/riscv/riscv-builtins.c: Likewise.
29421 * config/riscv/riscv-c.c: Likewise.
29422 * config/riscv/riscv-ftypes.def: Likewise.
29423 * config/riscv/riscv-modes.def: Likewise.
29424 * config/riscv/riscv-opts.h: Likewise.
29425 * config/riscv/riscv-protos.h: Likewise.
29426 * config/riscv/riscv.h: Likewise.
29427 * config/riscv/riscv.md: Likewise.
29428 * config/riscv/riscv.opt: Likewise.
29429 * config/riscv/sync.md: Likewise.
29430 * config/riscv/t-elf-multilib: Likewise.
29431 * config/riscv/t-linux: Likewise.
29432 * config/riscv/t-linux-multilib: Likewise.
29433 * config/riscv/t-riscv: Likewise.
29434 * configure.ac: Likewise.
29435 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
29436 Waterman as RISC-V maintainers.
29437 * doc/install.texi: Add RISC-V entries.
29438 * doc/invoke.texi: Add RISC-V options section.
29439 * doc/md.texi: Add RISC-V constraints section.
29440 * configure: Regenerated.
29442 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
29445 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
29446 false values to be constant vectors with all 0 or all 1 bits set.
29447 (vcondu<mode><mode>): Likewise.
29448 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
29450 (fpmask_comparison_operator): Update comment.
29451 (vecint_comparison_operator): New predicate.
29452 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
29453 vector conditionals when the true and false values are constant
29454 vectors with all 0 bits or all 1 bits set.
29456 2017-02-06 Martin Sebor <msebor@redhat.com>
29458 PR tree-optimization/79376
29459 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
29461 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
29463 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
29464 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
29465 to simplify split condition.
29467 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29469 * omp-expand.c (oxpand_omp_atomic_fetch_op,
29470 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
29473 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
29475 PR rtl-optimization/68664
29476 * target.def (can_speculate_insn): New hook.
29477 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
29478 * doc/tm.texi: Regenerate.
29479 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
29480 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
29481 (rs6000_sched_can_speculate_insn): New function.
29483 2017-02-06 Jakub Jelinek <jakub@redhat.com>
29485 PR tree-optimization/79284
29486 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
29487 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
29488 vectorizable_mask_load_store, vectorizable_operation,
29489 vect_is_simple_cond, get_same_sized_vectype): Use it instead
29490 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
29491 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
29492 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
29493 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
29494 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
29495 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
29496 is_gimple_assign (stmt). Replace another such test with
29497 is_gimple_assign (stmt).
29499 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29502 * config/avr/avr.c (rtl-iter.h): Include it.
29503 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
29504 (avr_legitimate_combined_insn): ...and implementation.
29506 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
29508 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
29509 * config/s390/s390.c (s390_const_operand_ok)
29510 (s390_canonicalize_comparison, s390_extract_part)
29511 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
29512 (s390_contiguous_bitmask_p, s390_rtx_costs)
29513 (legitimize_pic_address): Likewise.
29514 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
29515 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
29516 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
29517 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
29518 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
29520 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
29522 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
29523 REGNO($0) == REGNO($1).
29525 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29527 * config/s390/linux.h(SIZE_TYPE): Add comment.
29529 2017-02-06 Julian Brown <julian@codesourcery.com>
29530 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29531 Virendra Pathak <virendra.pathak@broadcom.com>
29533 * config/aarch64/aarch64-cores.def: Change the scheduler
29535 * config/aarch64/aarch64.md: Include thunderx2t99.md.
29536 * config/aarch64/thunderx2t99.md: New file.
29538 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29540 * doc/standards.texi (Go Language): Update link to language
29543 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29545 * tree-eh.c (lower_resx): Sanitize profile.
29546 (cleanup_empty_eh_move_lp): Likewise.
29548 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
29551 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
29553 * cfgloopmanip.h (loop_version): Update prototype.
29554 * modulo-sched.c (sms_schedule): Update call of loop_version.
29555 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
29556 * tree-parloops.c (gen_parallel_loop): Likewise.
29557 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
29558 * tree-ssa-loop-split.c (split_loop): Likewise.
29559 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
29560 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
29562 2017-02-05 Martin Liska <mliska@suse.cz>
29565 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
29567 (print_operand_address): Initialize a struct to zero.
29569 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29571 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
29572 garbage collector only in textual form.
29574 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
29576 * doc/extend.texi (x86 specific memory model extensions for
29577 transactional memory): Simplify a phrase.
29579 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
29582 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
29583 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
29584 (atomic_storedi_1): Likewise.
29586 2017-02-04 Jakub Jelinek <jakub@redhat.com>
29588 PR tree-optimization/79338
29589 * tree-parloops.c (gather_scalar_reductions): Don't call
29590 vect_analyze_loop_form for loop->inner before destroying loop's
29593 2017-02-03 Martin Sebor <msebor@redhat.com>
29595 PR tree-optimization/79327
29596 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
29597 when precision has resulted in leading zeros.
29598 (format_integer): Adjust the likely counter to assume an unknown
29599 argument that may be zero is non-zero.
29601 2017-02-03 Jason Merrill <jason@redhat.com>
29604 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
29605 avoid copying non-taken branch.
29607 2017-02-03 Jakub Jelinek <jakub@redhat.com>
29609 PR tree-optimization/79340
29610 * tree-vect-loop.c (vectorizable_reduction): Release
29611 vec_defs elements after safe_splicing them into other vectors.
29614 PR tree-optimization/79327
29615 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
29616 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
29618 (format_integer): Use wide_int_to_tree instead of build_int_cst
29619 + to_?hwi. If argmin is NULL, just set argmin and argmax to
29620 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
29621 of shortest and longest sequence.
29623 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
29625 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
29626 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
29628 2017-02-03 Walter Lee <walt@tilera.com>
29631 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
29632 after initial stackframe link reg save.
29633 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
29635 2017-02-03 Jakub Jelinek <jakub@redhat.com>
29638 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
29639 wu for stxssp alternative.
29641 2017-02-03 Martin Sebor <msebor@redhat.com>
29643 PR tree-optimization/79352
29644 * gimple-fold.c (get_range_strlen): Add argument.
29645 (get_range_strlen): Change return type to bool.
29646 (get_maxval_strlen): Pass in a dummy argument.
29647 * gimple-fold.h (get_range_strlen): Change return type to bool.
29648 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
29649 * tree.h (array_at_struct_end_p): Add argument.
29650 * tree.c (array_at_struct_end_p): Handle it.
29652 2017-02-03 Martin Liska <mliska@suse.cz>
29655 * multiple_target.c (create_dispatcher_calls): Redirect edge
29656 from a caller of a dispatcher.
29657 (expand_target_clones): Make the clones local.
29658 (ipa_target_clone): Do both target clones and resolvers.
29659 (ipa_dispatcher_calls): Remove the pass.
29660 (pass_dispatcher_calls::gate): Likewise.
29661 (make_pass_dispatcher_calls): Likewise.
29662 * passes.def (pass_target_clone): Put as very first IPA early
29665 2017-02-03 Martin Liska <mliska@suse.cz>
29667 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
29668 in case of a function with ifunc attribute.
29670 2017-02-03 Martin Liska <mliska@suse.cz>
29672 * cgraph.c (cgraph_node::dump): Dump function version info.
29673 * symtab.c (symtab_node::dump_base): Add missing new line.
29675 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
29677 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
29678 (ifcombine_ifandif): Use it.
29680 2017-02-03 Martin Liska <mliska@suse.cz>
29682 * doc/invoke.texi: Document default value for
29683 use-after-scope-direct-emission-threshold.
29685 2017-02-03 Martin Liska <mliska@suse.cz>
29687 PR tree-optimization/79339
29688 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
29689 (format_floating): Likewise.
29691 2017-02-03 Martin Liska <mliska@suse.cz>
29694 * ipa-prop.c (ipa_node_params_t::insert): Remove current
29696 (ipa_node_params_t::remove): Likewise.
29697 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
29698 initialization from removed ipa_node_params_t::insert.
29699 (ipa_node_params::~ipa_node_params): Move from removed
29700 ipa_node_params_t::release.
29701 * symbol-summary.h (symbol_summary::m_released): New member.
29702 Do not release a summary twice. Do not allow to call finalizer
29703 for types of a summary that live in GGC memory.
29705 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
29707 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
29710 2017-02-02 Martin Sebor <msebor@redhat.com>
29712 PR middle-end/79275
29713 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
29714 (format_string): Tighten up the range of output for non-constant
29715 strings and correct the expected range for wide non-constant strings.
29717 2017-02-02 Martin Sebor <msebor@redhat.com>
29719 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
29721 PR middle-end/32003
29722 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
29724 (-fdump-tree-@var): Add to index and document how to come up
29725 with pass-specific option and dump file names.
29726 (-fdump-passes): Clarify where to look for output.
29728 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
29730 PR middle-end/77445
29731 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
29732 statistics of the analyzed path; allow threading for speed when
29733 any of BBs along the path are optimized for speed.
29735 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
29737 PR middle-end/78468
29738 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
29739 settings of the virtual registers.
29742 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
29744 * explow.c (get_dynamic_stack_size): Take known alignment of stack
29745 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
29748 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29750 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
29751 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
29753 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29755 * config/s390/s390.md: Add missing comments with the expanded
29757 * config/s390/vector.md: Likewise.
29758 * config/s390/vx-builtins.md: Likewise.
29760 2017-02-02 Jakub Jelinek <jakub@redhat.com>
29763 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
29764 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
29765 conditions on a single line.
29767 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29769 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29770 __S390_VX__ to __VX__.
29772 2017-02-01 Andrew Pinski <apinski@cavium.com>
29774 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
29775 stmt_info to record_stmt_cost.
29776 (vect_get_known_peeling_cost): Pass stmt_info if known to
29778 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
29779 cpu_vector_cost field into
29780 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
29781 field into vec_int_stmt_cost and vec_fp_stmt_cost.
29782 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
29783 splitting of scalar_stmt_cost and vec_stmt_cost.
29784 (thunderx_vector_cost): Likewise.
29785 (cortexa57_vector_cost): LIkewise.
29786 (exynosm1_vector_cost): Likewise.
29787 (xgene1_vector_cost): Likewise.
29788 (thunderx2t99_vector_cost): Improve after the splitting of the two
29790 (aarch64_builtin_vectorization_cost): Update for the splitting of
29791 scalar_stmt_cost and vec_stmt_cost.
29793 2017-02-01 Torvald Riegel <triegel@redhat.com>
29794 Richard Henderson <rth@redhat.com>
29796 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
29797 conditional on existance of a fast atomic load.
29798 * optabs-query.c (can_atomic_load_p): New function.
29799 * optabs-query.h (can_atomic_load_p): Declare it.
29800 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
29801 no fast atomic load is available for the particular size of access.
29802 (expand_atomic_compare_and_swap): Likewise.
29803 (expand_atomic_load): Likewise.
29804 (expand_atomic_store): Likewise.
29805 (expand_atomic_fetch_op): Likewise.
29806 * testsuite/lib/target-supports.exp
29807 (check_effective_target_sync_int_128): Remove x86 because it provides
29808 no fast atomic load.
29809 (check_effective_target_sync_int_128_runtime): Likewise.
29811 2017-02-01 Richard Biener <rguenther@suse.de>
29813 * graphite.c: Include tree-vectorizer.h for find_loop_location.
29814 (graphite_transform_loops): Provide opt-info for optimized nests.
29815 * tree-parloop.c (parallelize_loops): Provide opt-info for
29816 parallelized loops.
29818 2017-02-01 Richard Biener <rguenther@suse.de>
29820 PR middle-end/79315
29821 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
29822 was not set before.
29824 2017-02-01 Richard Biener <rguenther@suse.de>
29826 PR tree-optimization/71824
29827 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
29828 Verify the loops are valid in the merged SESE region.
29829 (scop_detection::can_represent_loop_1): Check analyzing the
29830 evolution of the number of iterations in the region succeeds.
29832 2017-01-31 Ian Lance Taylor <iant@golang.org>
29834 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
29835 REG_ARGS_SIZE note to 32-bit push insns and call insn.
29837 2017-01-31 David Malcolm <dmalcolm@redhat.com>
29839 PR preprocessor/79210
29840 * input.c (get_substring_ranges_for_loc): Replace line_width
29841 assertion with error-handling.
29843 2017-01-31 Richard Biener <rguenther@suse.de>
29845 PR tree-optimization/77318
29846 * graphite-sese-to-poly.c (extract_affine): Fix assert.
29847 (create_pw_aff_from_tree): Take loop parameter.
29848 (add_condition_to_pbb): Pass loop of the condition to
29849 create_pw_aff_from_tree.
29851 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29853 * config/s390/s390.c (s390_asan_shadow_offset): New function.
29854 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
29856 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
29860 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
29862 (convert_int_to_float128): Likewise.
29863 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
29864 (convert_int_to_float128): Likewise.
29865 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
29866 (UNSPEC_IEEE128_CONVERT): Likewise.
29867 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
29868 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
29869 Use local variables for IBM extended format.
29870 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
29871 (fix_trunc<mode>si2_fprs): Likewise.
29872 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
29873 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
29874 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
29875 to know that we can now have integers of all sizes in vector
29877 (fix<uns>_<mode>di2_hw): Likewise.
29878 (float<uns>_<mode>si2_hw): Likewise.
29879 (fix_<mode>si2_hw): Likewise.
29880 (fixuns_<mode>si2_hw): Likewise.
29881 (float<uns>_<mode>di2_hw): Likewise.
29882 (float_<mode>di2_hw): Likewise.
29883 (float_<mode>si2_hw): Likewise.
29884 (floatuns_<mode>di2_hw): Likewise.
29885 (floatuns_<mode>si2_hw): Likewise.
29886 (xscvqp<su>wz_<mode>): Delete, no longer used.
29887 (xscvqp<su>dz_<mode>): Likewise.
29888 (xscv<su>dqp_<mode>): Likewise.
29889 (ieee128_mfvsrd_64bit): Likewise.
29890 (ieee128_mfvsrd_32bit): Likewise.
29891 (ieee128_mfvsrwz): Likewise.
29892 (ieee128_mtvsrw): Likewise.
29893 (ieee128_mtvsrd_64bit): Likewise.
29894 (ieee128_mtvsrd_32bit): Likewise.
29896 2017-01-31 Martin Liska <mliska@suse.cz>
29899 * ipa-prop.c (ipa_free_all_node_params): Call release method
29900 instead of ~sumbol_summary to not to trigger double times
29903 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
29905 PR tree-optimization/71691
29906 * bitmap.h (class auto_bitmap): New.
29907 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
29908 is_maybe_undefined instead of ssa_undefined_value_p.
29910 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
29912 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
29913 __S390_ARCH_LEVEL__ to __ARCH__.
29915 2017-01-31 Jakub Jelinek <jakub@redhat.com>
29917 PR tree-optimization/79267
29918 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
29919 if should_remove_lhs_p is true.
29921 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
29924 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
29925 (add_alignment_attribute): New.
29926 (base_type_die): Add alignment attribute.
29927 (subrange_type_die): Likewise.
29928 (modified_type_die): Likewise.
29929 (gen_array_type_die): Likewise.
29930 (gen_descr_array_type_die: Likewise.
29931 (gen_enumeration_type_die): Likewise.
29932 (gen_subprogram_die): Likewise.
29933 (gen_variable_die): Likewise.
29934 (gen_field_die): Likewise.
29935 (gen_ptr_to_mbr_type_die): Likewise.
29936 (gen_struct_or_union_type_die): Likewise.
29937 (gen_subroutine_type_die): Likewise.
29938 (gen_typedef_die): Likewise.
29939 (base_type_cmp): Compare alignment attribute.
29941 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
29944 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
29945 (setb_unsigned) New pattern for setb with CCUNS.
29946 * config/rs6000/rs6000.c (expand_block_compare): Use a different
29947 subfc./subfe sequence to avoid overflow problems. Generate a
29948 shorter sequence with cmpld/setb for power9.
29949 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
29950 for generating subfc. instruction.
29951 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
29952 now uses this instruction.
29954 2017-01-30 Ian Lance Taylor <iant@google.com>
29957 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
29958 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
29960 2017-01-30 Martin Sebor <msebor@redhat.com>
29962 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
29963 Move constant to the right of a relational operator.
29964 (get_mpfr_format_length, format_character, format_string): Ditto.
29965 (should_warn_p, maybe_warn): Same.
29967 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
29969 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
29972 * asan.c (get_translation_unit_decl): Remove function.
29973 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
29975 2017-01-30 Martin Liska <mliska@suse.cz>
29977 PR gcov-profile/79259
29978 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
29979 -fprofile-generate.
29981 2017-01-30 Martin Liska <mliska@suse.cz>
29984 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
29985 Initialize variables with NULL value.
29987 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
29990 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
29992 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
29994 2017-01-30 Richard Biener <rguenther@suse.de>
29996 PR tree-optimization/79276
29997 * tree-vrp.c (process_assert_insertions): Properly adjust common
29998 when removing a duplicate.
30000 * gcc.dg/torture/pr79276.c: New testcase.
30002 2017-01-30 Richard Biener <rguenther@suse.de>
30004 PR tree-optimization/79256
30005 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
30006 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
30008 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
30010 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
30013 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
30014 ("*r<noxa>sbg_<mode>_sll_bitmask")
30015 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
30016 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
30017 Use contiguous_bitmask_nowrap_operand.
30019 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
30022 * config/rs6000/altivec.h (vec_xl): Revise #define.
30023 (vec_xst): Likewise.
30025 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
30027 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
30029 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
30031 PR rtl-optimization/79194
30032 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
30033 traps before call to bypass_conditional_jumps.
30035 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30037 PR tree-optimization/71374
30038 * lra-constraints.c (check_conflict_input_operands): New.
30039 (match_reload): Use it.
30041 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
30044 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
30045 account to calculate conflict_set.
30047 2017-01-27 Bin Cheng <bin.cheng@arm.com>
30049 PR rtl-optimization/78559
30050 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
30051 other_insn in combine.
30053 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
30055 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
30056 uint16_type_node for BT_UINT16.
30058 2017-01-27 David Malcolm <dmalcolm@redhat.com>
30060 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
30061 "RTL Tests" to menu.
30062 (GIMPLE Tests): New node.
30063 (RTL Tests): New node.
30065 2017-01-27 Richard Biener <rguenther@suse.de>
30067 PR tree-optimization/79245
30068 * tree-loop-distribution.c (distribute_loop): Apply cost
30069 modeling also to detected patterns.
30071 2017-01-27 Richard Biener <rguenther@suse.de>
30073 PR tree-optimization/71433
30074 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
30075 (compare_assert_loc): New function.
30076 (process_assert_insertions): Sort and optimize assert locations
30077 to remove duplicates and push down identical assertions on
30078 edges to their destination block.
30080 2017-01-27 Richard Biener <rguenther@suse.de>
30082 PR tree-optimization/79244
30083 * tree-vrp.c (remove_range_assertions): Forcefully propagate
30084 out SSA names even if abnormal.
30086 2017-01-27 Jakub Jelinek <jakub@redhat.com>
30088 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
30089 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
30090 instead of MPFR_RNDN.
30092 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
30095 * arm.c (arm_option_override): Don't call build_target_option_node
30096 until after doing all option overrides.
30097 (arm_valid_target_attribute_tree): Likewise.
30099 2017-01-27 Martin Liska <mliska@suse.cz>
30101 * doc/invoke.texi (-fprofile-arcs): Document profiling support
30102 for {cd}tors and C++ {cd}tors.
30104 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
30106 * config/s390/s390.md ("*setmem_long_and")
30107 ("*setmem_long_and_31z"): Use zero_extend instead of and.
30109 2017-01-26 Martin Sebor <msebor@redhat.com>
30111 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
30114 2017-01-26 Martin Sebor <msebor@redhat.com>
30116 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
30117 HAVE_DFmode before using XFmode or DFmode.
30118 (parse_directive): Avoid using the z length modifier to avoid
30119 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
30121 PR middle-end/78703
30122 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
30123 to accept adjustment as an array.
30124 (get_int_range): New function.
30125 (struct directive): Make width and prec arrays.
30126 (directive::set_width, directive::set_precision): Call get_int_range.
30127 (format_integer, format_floating): Handle width and precision ranges.
30128 (format_string, parse_directive): Same.
30130 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30133 * dwarf2out.c (generate_skeleton_bottom_up): For children with
30134 comdat_type_p set, just clone them, but keep the children in the
30138 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
30139 which have direct callers with -fvar-tracking-assignments enabled
30141 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
30142 inside of type units.
30144 2017-01-26 Martin Sebor <msebor@redhat.com>
30146 PR middle-end/78703
30147 * gimple-ssa-sprintf.c (struct result_range): Add likely and
30149 (struct format_result): Replace number_chars, number_chars_min,
30150 and number_chars_max with a single member of struct result_range.
30152 (format_result::operator+=): Adjust.
30153 (struct fmtresult): Remove bounded. Handle likely and unlikely
30155 (fmtresult::adjust_for_width_or_precision): New function.
30156 (fmtresult:type_max_digits): New function.
30157 (bytes_remaining): Handle likely and unlikely counters.
30158 (min_bytes_remaining): Remove.
30159 (format_percent): Simplify.
30160 (format_integer, format_floating): Set likely and unlikely counters.
30161 (get_string_length, format_character, format_string): Same.
30162 (format_plain, should_warn_p): New function.
30163 (maybe_warn): Call should_warn_p. Update diagnostic messages
30164 and handle those for all directives, including plain strings.
30165 (format_directive): Handle likely and unlikely counters.
30166 Remove unnecessary quoting from diagnostics. Add an informational
30168 (add_bytes): Remove.
30169 (pass_sprintf_length::compute_format_length): Simplify.
30170 (try_substitute_return_value): Handle likely and unlikely counters.
30172 2017-01-26 Carl Love <cel@us.ibm.com>
30174 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
30175 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
30177 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
30180 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
30181 endianess for subregs into account.
30182 * lra-constraints.c (lra_constraints): Do risky transformations
30183 always on the first iteration.
30184 * lra-lives.c (check_pseudos_live_through_calls): Add arg
30185 last_call_used_reg_set.
30186 (process_bb_lives): Define and use last_call_used_reg_set.
30187 * lra.c (lra): Always continue after lra_constraints on the first
30190 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
30192 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
30194 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
30196 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30198 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
30199 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
30200 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
30201 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
30202 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
30203 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
30204 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
30205 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
30206 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
30208 2017-01-26 Marek Polacek <polacek@redhat.com>
30211 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
30212 for the third operand.
30214 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30216 PR middle-end/79236
30217 * omp-low.c (struct omp_context): Add simt_stmt field.
30218 (scan_omp_for): Return omp_context *.
30219 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
30220 context to the _simt_ SIMD stmt.
30221 (lower_omp_for): For combined SIMD with sibling _simt_
30222 SIMD, make sure to use the same decls in _looptemp_
30223 clauses as in the sibling.
30225 2017-01-26 David Sherwood <david.sherwood@arm.com>
30227 PR middle-end/79212
30228 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
30231 2017-01-26 Jakub Jelinek <jakub@redhat.com>
30234 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
30235 emit fld b; fld a; if possible.
30237 * brig-builtins.def: Update copyright years.
30238 * config/arm/arm_acle_builtins.def: Update copyright years.
30240 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
30243 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
30244 constraint instead of o for the stxsd instruction.
30246 2017-01-25 Carl Love <cel@us.ibm.com>
30248 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
30249 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
30251 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
30253 * doc/invoke.texi (C++ Dialect Options): Fix typo.
30255 2017-01-25 Richard Biener <rguenther@suse.de>
30257 PR tree-optimization/69264
30258 * target.def (vector_alignment_reachable): Improve documentation.
30259 * doc/tm.texi: Regenerate.
30260 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
30262 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
30263 earlier changes with respect to TYPE_USER_ALIGN.
30264 (vector_alignment_reachable_p): Likewise. Improve dumping.
30266 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30269 * config/arm/arm.md (xordi3): Force constant operand into a register
30272 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30274 * doc/invoke.texi (-fstore-merging): Correct default optimization
30275 levels at which it is enabled.
30276 (-O): Move -fstore-merging from list to...
30279 2017-01-25 Richard Biener <rguenther@suse.de>
30282 * omp-expand.c: Include debug.h.
30283 (expand_omp_taskreg): Make sure to generate early debug before
30284 outlining anything from a function.
30285 (expand_omp_target): Likewise.
30286 (grid_expand_target_grid_body): Likewise.
30288 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
30291 * asan.c (get_translation_unit_decl): New function.
30292 (asan_add_global): Extract modules file name from globals
30293 TRANSLATION_UNIT_DECL name.
30295 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
30298 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
30299 for long calls with APCS frame and VFP.
30301 2017-01-24 David Malcolm <dmalcolm@redhat.com>
30303 * cfg.c (original_copy_tables_initialized_p): New function.
30304 * cfg.h (original_copy_tables_initialized_p): New decl.
30305 * cfgrtl.c (relink_block_chain): Guard the call to
30306 free_original_copy_tables with a call to
30307 original_copy_tables_initialized_p.
30308 * cgraph.h (symtab_node::native_rtl_p): New decl.
30309 * cgraphunit.c (symtab_node::native_rtl_p): New function.
30310 (symtab_node::needed_p): Don't assert for early assembly output
30311 for __RTL functions.
30312 (cgraph_node::finalize_function): Set "force_output" for __RTL
30314 (cgraph_node::analyze): Bail out early for __RTL functions.
30315 (analyze_functions): Update assertion to support __RTL functions.
30316 (cgraph_node::expand): Bail out early for __RTL functions.
30317 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
30319 * function.h (struct function): Update comment for field
30321 * gimple-expr.c: Include "tree-pass.h".
30322 (gimple_has_body_p): Return false for __RTL functions.
30323 * Makefile.in (OBJS): Add run-rtl-passes.o.
30324 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
30326 (gcc::pass_manager::get_clean_slate): New accessor.
30327 * passes.c: Include "insn-addr.h".
30328 (should_skip_pass_p): Add logging. Update logic for running
30329 "expand" to be compatible with both __GIMPLE and __RTL. Guard
30330 property-provider override so it is only done for gimple passes.
30332 (skip_pass): New function.
30333 (execute_one_pass): Call skip_pass when skipping passes.
30334 * read-md.c (md_reader::read_char): Support filtering
30335 the input to a subset of line numbers.
30336 (md_reader::md_reader): Initialize fields
30337 m_first_line and m_last_line.
30338 (md_reader::read_file_fragment): New function.
30339 * read-md.h (md_reader::read_file_fragment): New decl.
30340 (md_reader::m_first_line): New field.
30341 (md_reader::m_last_line): New field.
30342 * read-rtl-function.c (function_reader::create_function): Only
30343 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
30344 curr_properties. Set DECL_INITIAL to a dummy block.
30345 (read_rtl_function_body_from_file_range): New function.
30346 * read-rtl-function.h (read_rtl_function_body_from_file_range):
30348 * run-rtl-passes.c: New file.
30349 * run-rtl-passes.h: New file.
30351 2017-01-24 Jeff Law <law@redhat.com>
30353 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
30356 2017-01-24 Bin Cheng <bin.cheng@arm.com>
30358 PR tree-optimization/79159
30359 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
30360 (record_nonwrapping_iv): Improve boundary using above function if no
30361 value range information.
30363 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
30364 Martin Jambor <mjambor@suse.cz>
30366 * brig-builtins.def: New file.
30367 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
30368 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
30369 (DEF_HSAIL_SAT_BUILTIN): Likewise.
30370 (DEF_HSAIL_INTR_BUILTIN): Likewise.
30371 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
30372 * builtin-types.def (BT_INT8): New.
30373 (BT_INT16): Likewise.
30374 (BT_UINT8): Likewise.
30375 (BT_UINT16): Likewise.
30376 (BT_FN_ULONG): Likewise.
30377 (BT_FN_UINT_INT): Likewise.
30378 (BT_FN_UINT_ULONG): Likewise.
30379 (BT_FN_UINT_LONG): Likewise.
30380 (BT_FN_UINT_PTR): Likewise.
30381 (BT_FN_ULONG_PTR): Likewise.
30382 (BT_FN_INT8_FLOAT): Likewise.
30383 (BT_FN_INT16_FLOAT): Likewise.
30384 (BT_FN_UINT32_FLOAT): Likewise.
30385 (BT_FN_UINT16_FLOAT): Likewise.
30386 (BT_FN_UINT8_FLOAT): Likewise.
30387 (BT_FN_UINT64_FLOAT): Likewise.
30388 (BT_FN_UINT16_UINT32): Likewise.
30389 (BT_FN_UINT32_UINT16): Likewise.
30390 (BT_FN_UINT16_UINT16_UINT16): Likewise.
30391 (BT_FN_INT_PTR_INT): Likewise.
30392 (BT_FN_UINT_PTR_UINT): Likewise.
30393 (BT_FN_LONG_PTR_LONG): Likewise.
30394 (BT_FN_ULONG_PTR_ULONG): Likewise.
30395 (BT_FN_VOID_UINT64_UINT64): Likewise.
30396 (BT_FN_UINT8_UINT8_UINT8): Likewise.
30397 (BT_FN_INT8_INT8_INT8): Likewise.
30398 (BT_FN_INT16_INT16_INT16): Likewise.
30399 (BT_FN_INT_INT_INT): Likewise.
30400 (BT_FN_UINT_FLOAT_UINT): Likewise.
30401 (BT_FN_FLOAT_UINT_UINT): Likewise.
30402 (BT_FN_ULONG_UINT_UINT): Likewise.
30403 (BT_FN_ULONG_UINT_PTR): Likewise.
30404 (BT_FN_ULONG_ULONG_ULONG): Likewise.
30405 (BT_FN_UINT_UINT_UINT): Likewise.
30406 (BT_FN_VOID_UINT_PTR): Likewise.
30407 (BT_FN_UINT_UINT_PTR: Likewise.
30408 (BT_FN_UINT32_UINT64_PTR): Likewise.
30409 (BT_FN_INT_INT_UINT_UINT): Likewise.
30410 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
30411 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
30412 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
30413 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
30414 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
30415 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
30416 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
30417 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
30418 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
30419 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
30420 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
30421 * doc/frontends.texi: List BRIG FE.
30422 * doc/install.texi (Testing): Add BRIG tesring requirements.
30423 * doc/invoke.texi (Overall Options): Mention BRIG.
30424 * doc/standards.texi (Standards): Doucment BRIG HSA version.
30426 2017-01-24 Richard Biener <rguenther@suse.de>
30428 PR translation/79208
30429 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
30431 2017-01-24 Martin Jambor <mjambor@suse.cz>
30434 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
30435 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
30436 and known_contexts.
30438 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
30440 PR middle-end/79123
30441 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
30442 casts from signed to unsigned really don't have a range.
30444 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
30446 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
30447 GMP_RNDx for compatiblity.
30449 2017-01-24 Martin Liska <mliska@suse.cz>
30452 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
30453 that would prevent us to call alloca with -1 as argument.
30455 2017-01-24 Jakub Jelinek <jakub@redhat.com>
30457 * dwarf2out.c (output_compilation_unit_header, output_file_names):
30458 Avoid -Wformat-security warning.
30460 2017-01-23 Andrew Pinski <apinski@cavium.com>
30462 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
30465 2017-01-23 Martin Sebor <msebor@redhat.com>
30467 PR middle-end/78703
30468 * gimple-ssa-sprintf.c (warn_level): New global.
30469 (format_integer): Use it here and throughout the rest of the file.
30470 Use the same switch to compute sign as base.
30471 (maybe_warn): New function.
30472 (format_directive): Factor out warnings into maybe_warn.
30473 Add debugging output. Use warn_level.
30474 (add_bytes): Use warn_level.
30475 (pass_sprintf_length::compute_format_length): Add debugging output.
30476 (try_substitute_return_value): Same.
30477 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
30479 PR middle-end/78703
30480 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
30481 (struct fmtresult, format_integer, format_floating): Adjust.
30482 (fmtresult::fmtresult): Set max correctly in two argument ctor.
30483 (get_string_length, format_string,format_directive): Same.
30484 (pass_sprintf_length::compute_format_length): Same.
30485 (try_substitute_return_value): Simplify slightly.
30487 PR middle-end/78703
30488 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
30489 (fmtresult::operator+=): Outlined.
30490 (struct fmtresult): Add ctors.
30491 (struct conversion_spec): Rename...
30492 (struct directive): ...to this. Add and remove data members.
30493 (directive::set_width, directive::set_precision): New functions.
30494 (format_percent): Use fmtresult ctor.
30495 (get_width_and_precision): Remove.
30496 (format_integer): Make naming changes. Avoid computing width and
30498 (format_floating): Same. Adjust indentation.
30499 (format_character, format_none): New functions.
30500 (format_string): Moved character handling to format_character.
30501 (format_directive): Remove arguments, change return type.
30502 (parse_directive): New function.
30503 (pass_sprintf_length::compute_format_length): Move directive
30504 parsing to parse_directive.
30506 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30508 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
30509 (assign_assembler_name_if_needed): ... this.
30510 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
30511 (assign_assembler_name_if_needed): ... this.
30512 (free_lang_data_in_cgraph): Adjust callers.
30513 * cgraphunit.c (cgraph_node::analyze): Likewise.
30514 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
30517 2017-01-23 Richard Biener <rguenther@suse.de>
30519 PR tree-optimization/79088
30520 PR tree-optimization/79188
30521 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
30522 resetting loop bounds after last path deletion. Reset loop
30523 bounds of the target loop, make code match the comments.
30524 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
30525 Make sure loops need no fixups.
30527 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
30529 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
30530 exponent support with double type for first argument.
30531 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
30532 type returned by __builtin_vec_extract_sig,
30533 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
30534 functions from "vector int" to "vector unsigned int" or from
30535 "vector long long int" to "vector unsigned long long int".
30536 Changed type returned by __builtin_vec_extract_exp,
30537 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
30538 functions from "vector int" to "vector unsigned int" or from
30539 "vector long long int" to "vector unsigned long long int".
30540 Changed return type of __builtin_vec_test_data_class,
30541 __builtin_vec_test_data_class_sp, and
30542 __builtin_vec_test_data_class_dp from "vector int" to
30543 "vector bool int" or from "vector long long int" to "vector bool
30544 long long int" and changed second argument type from "unsigned
30545 int" to "int". Added new overloaded function forms "vector float
30546 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
30547 "vector float __builtin_vec_insert_exp_sp (vector float, vector
30548 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
30549 double, vector unsigned long long int)" and "vector double
30550 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
30551 long int)". Changed return type of
30552 __builtin_scalar_test_data_class and
30553 __builtin_scalar_test_data_class_sp and
30554 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
30555 int" and changed second argument from "unsigned int" to "int".
30556 Changed type returned by __builtin_scalar_test_neg,
30557 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
30558 from "int" to "bool int". Added new overloaded function form
30559 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
30560 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
30561 exponent double-precision with floating point first argument.
30562 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
30563 documentation of scalar_test_data_class, scalar_test_neg,
30564 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
30565 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
30566 vec_test_data_class built-in functions to reflect refinements in
30567 their type signatures.
30569 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
30571 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
30573 (aarch64_elf_asm_destructor): Likewise.
30575 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
30577 PR rtl-optimization/78634
30578 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
30579 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
30580 * ifcvt.c (noce_try_cmove): Add missing cost check.
30582 PR rtl-optimization/71724
30583 * combine.c (if_then_else_cond): Look for situations where it is
30584 beneficial to undo the work of one of the recursive calls.
30586 2017-01-23 Bin Cheng <bin.cheng@arm.com>
30588 PR tree-optimization/70754
30589 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
30590 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
30591 combined stmt before it if not NULL.
30592 (combine_chains): Process refs reversely and compute dominance point
30595 2017-01-23 Martin Liska <mliska@suse.cz>
30597 PR tree-optimization/79196
30598 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
30599 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
30601 (strlen_optimize_stmt): Call the renamed function.
30603 2017-01-23 Michael Matz <matz@suse.de>
30605 PR tree-optimization/78384
30606 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
30608 2017-01-23 Richard Biener <rguenther@suse.de>
30610 PR tree-optimization/79186
30611 * tree-vrp.c (register_new_assert_for): Make sure we've seen
30612 both incoming edges before moving an assert.
30614 2017-01-23 Martin Jambor <mjambor@suse.cz>
30616 * ipa-prop.c (load_from_param_1): Removed.
30617 (load_from_unmodified_param): Bits from load_from_param_1 put back
30619 (load_from_param): Removed.
30620 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
30621 with stmt. Reverted back to use of load_from_unmodified_param.
30623 2017-01-23 Martin Jambor <mjambor@suse.cz>
30626 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
30627 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
30628 field a pointer to garbage collected vector, mark lattices and
30629 ipcp_orig_node with GTY((skip)).
30630 (ipa_get_param_count): Adjust to descriptors being a pointer.
30631 (ipa_get_param): Likewise.
30632 (ipa_get_type): Likewise.
30633 (ipa_get_param_move_cost): Likewise.
30634 (ipa_set_param_used): Likewise.
30635 (ipa_get_controlled_uses): Likewise.
30636 (ipa_set_controlled_uses): Likewise.
30637 (ipa_is_param_used): Likewise.
30638 (ipa_node_params_t): Move into garbage collector. New methods insert
30640 (ipa_node_params_sum): Annotate wth GTY(()).
30641 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
30643 (ipa_load_from_parm_agg): Adjust declaration.
30644 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
30645 * ipa-profile.c (ipa_profile): Likewise.
30646 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
30647 (ipa_populate_param_decls): Make descriptors parameter garbage
30649 (ipa_dump_param): Adjust to descriptors being a pointer.
30650 (ipa_alloc_node_params): Likewise.
30651 (ipa_initialize_node_params): Likewise.
30652 (load_from_param_1): Make descriptors parameter garbage collected.
30653 (load_from_unmodified_param): Likewise.
30654 (load_from_param): Likewise.
30655 (ipa_load_from_parm_agg): Likewise.
30656 (ipa_node_params::~ipa_node_params): Removed.
30657 (ipa_free_all_node_params): Remove call to delete operator.
30658 (ipa_node_params_t::insert): New.
30659 (ipa_node_params_t::remove): Likewise.
30660 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
30661 copy known_csts and known_contexts vectors.
30662 (ipa_read_node_info): Adjust to descriptors being a pointer.
30663 (ipcp_modif_dom_walker): Make m_descriptors field garbage
30665 (ipcp_transform_function): Make descriptors variable garbage
30668 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
30670 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
30671 * config/i386/avx512dqintrin.h: Ditto.
30672 * config/i386/avx512fintrin.h: Ditto.
30673 * config/i386/i386.c: Handle new builtins.
30674 * config/i386/i386-builtin.def: Add new builtins.
30675 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
30676 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
30678 2017-01-23 Jakub Jelinek <jakub@redhat.com>
30679 Martin Liska <mliska@suse.cz>
30681 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
30682 * asan.c (asan_expand_poison_ifn): Support stores and use
30683 appropriate ASAN report function.
30684 * internal-fn.c (expand_ASAN_POISON_USE): New function.
30685 * internal-fn.def (ASAN_POISON_USE): Declare.
30686 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
30687 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
30688 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
30689 ASAN_POISON calls w/o LHS.
30690 * tree-ssa.c (execute_update_addresses_taken): Create clobber
30691 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
30692 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
30693 * gimplify.c (asan_poison_variables): Add attribute
30694 use_after_scope_memory to variables that really needs to live
30696 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
30697 having the attribute.
30699 2017-01-23 Martin Liska <mliska@suse.cz>
30701 * asan.c (create_asan_shadow_var): New function.
30702 (asan_expand_poison_ifn): Likewise.
30703 * asan.h (asan_expand_poison_ifn): New declaration.
30704 * internal-fn.c (expand_ASAN_POISON): Likewise.
30705 * internal-fn.def (ASAN_POISON): New builtin.
30706 * sanopt.c (pass_sanopt::execute): Expand
30707 asan_expand_poison_ifn.
30708 * tree-inline.c (copy_decl_for_dup_finish): Make function
30710 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
30711 * tree-ssa.c (is_asan_mark_p): New function.
30712 (execute_update_addresses_taken): Rewrite local variables
30713 (identified just by use-after-scope as addressable) into SSA.
30715 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30717 * doc/install.texi (Specific): opensource.apple.com uses https
30718 now. Remove trailing slash.
30720 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
30722 * README.Portability: Remove note on an Irix compatibility issue.
30724 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
30726 * gcov.c (INCLUDE_ALGORITHM): Define.
30727 (INCLUDE_VECTOR): Define.
30728 No longer include <vector> and <algorithm> directly.
30730 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
30732 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
30734 * doc/invoke.texi (Code Gen Options): Ditto.
30736 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
30739 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
30741 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
30743 rtl-optimization/79125
30744 * cprop.c (local_cprop_pass): Handle cases where we make an
30745 unconditional trap.
30747 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
30751 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
30752 read from, for big endian.
30754 2017-01-20 Jiong Wang <jiong.wang@arm.com>
30756 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
30757 register pauth builtins for LP64 only.
30759 2017-01-20 Marek Polacek <polacek@redhat.com>
30762 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
30765 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30767 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
30769 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
30770 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
30771 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
30773 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
30776 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
30777 in big-endian mode when they are not a single duplicated value.
30779 2017-01-20 Richard Biener <rguenther@suse.de>
30781 * BASE-VER: Bump to 7.0.1.
30783 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
30785 * omp-low.c (omplow_simd_context): New struct. Use it...
30786 (lower_rec_simd_input_clauses): ...here and...
30787 (lower_rec_input_clauses): ...here to hold common data. Adjust all
30788 references to idx, lane, max_vf, is_simt.
30790 2017-01-20 Graham Markall <graham.markall@embecosm.com>
30792 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
30795 2017-01-20 Martin Jambor <mjambor@suse.cz>
30797 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
30798 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
30800 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
30801 (GTFILES): Rename hsa.c to hsa-common.c.
30802 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
30803 * hsa-dump.c: Likewise.
30804 * hsa-gen.c: Likewise.
30805 * hsa-regalloc.c: Likewise.
30806 * ipa-hsa.c: Likewise.
30807 * omp-expand.c: Likewise.
30808 * omp-low.c: Likewise.
30809 * toplev.c: Likewise.
30811 2017-01-20 Marek Polacek <polacek@redhat.com>
30814 * doc/invoke.texi: Document -Wduplicated-branches.
30815 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
30816 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
30817 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
30818 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
30819 return 0 only when not OEP_LEXICOGRAPHIC.
30820 (fold_build_cleanup_point_expr): Use the expression
30821 location when building CLEANUP_POINT_EXPR.
30822 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
30823 * tree.c (add_expr): Handle error_mark_node.
30825 2017-01-20 Martin Liska <mliska@suse.cz>
30828 * tree-profile.c (init_ic_make_global_vars): Do not call
30830 (gimple_init_gcov_profiler): Likewise.
30832 2017-01-20 Martin Liska <mliska@suse.cz>
30835 * cgraph.h (maybe_create_reference): Remove argument and
30837 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
30839 * ipa-cp.c (create_specialized_node): Likewise.
30840 * symtab.c (symtab_node::maybe_create_reference): Handle
30841 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
30843 2017-01-20 Martin Liska <mliska@suse.cz>
30845 * read-rtl-function.c (function_reader::create_function): Use
30846 build_decl instread of build_decl_stat.
30848 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
30850 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
30851 * config/i386/avx512dqintrin.h: Ditto.
30852 * config/i386/avx512fintrin.h: Ditto.
30853 * config/i386/i386-builtin-types.def: Add new types.
30854 * config/i386/i386.c: Handle new types.
30855 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
30856 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
30857 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
30858 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
30859 (__builtin_ia32_kshiftridi): New.
30860 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
30862 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
30866 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
30867 define to rs6000_init_stack_protect_guard.
30868 (rs6000_init_stack_protect_guard): New function.
30870 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30871 Yunqiang Su <yunqiang.su@imgtec.com>
30873 * config.gcc (supported_defaults): Add madd4.
30874 (with_madd4): Add validation.
30875 (all_defaults): Add madd4.
30876 * config/mips/mips.opt (mmadd4): New option.
30877 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
30879 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
30881 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
30882 (ISA_HAS_FUSED_MADD4): Likewise.
30883 * doc/invoke.texi (-mmadd4): Document the new option.
30884 * doc/install.texi (--with-madd4): Document the new option.
30886 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30888 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
30889 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
30890 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
30891 (aarch64_init_pauth_hint_builtins): New.
30892 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
30893 (aarch64_expand_builtin): Expand new builtins.
30895 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30897 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
30898 * combine-stack-adj.c (no_unhandled_cfa): Handle
30899 REG_CFA_TOGGLE_RA_MANGLE.
30900 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
30901 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
30902 info for return address signing.
30903 (aarch64_expand_epilogue): Likewise.
30905 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30907 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
30908 * config/aarch64/aarch64-protos.h
30909 (aarch64_return_address_signing_enabled): New declaration.
30910 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
30912 (aarch64_expand_prologue): Sign return address before it's pushed onto
30914 (aarch64_expand_epilogue): Authenticate return address fetched from
30916 (aarch64_override_options): Sanity check for ILP32 and ISA level.
30917 (aarch64_attributes): New function attributes for "sign-return-address".
30918 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
30919 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
30920 ("*do_return"): Generate combined instructions according to key index.
30921 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
30922 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
30924 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
30925 * config/aarch64/aarch64.opt (msign-return-address=): New.
30926 * doc/extend.texi (AArch64 Function Attributes): Documents
30927 "sign-return-address=".
30928 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
30930 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
30932 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
30933 overall option summary.
30935 2017-01-19 Jiong Wang <jiong.wang@arm.com>
30937 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
30938 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
30939 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
30940 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
30942 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
30944 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
30945 -mpower9-minmax by default for -mcpu=power9.
30946 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
30947 128-bit floating point.
30949 2017-01-20 Alan Modra <amodra@gmail.com>
30951 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
30952 optimizing for size.
30954 2017-01-20 Alan Modra <amodra@gmail.com>
30957 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
30958 for strcmp and strncmp from corresponding builtin decl.
30960 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30962 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
30963 instead of i386/rtems-64.h.
30964 * config/i386/rtems-64.h: Remove.
30966 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
30970 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
30972 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
30974 2017-01-19 Tamar Christina <tamar.christina@arm.com>
30976 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
30977 Change int to HOST_WIDE_INT.
30978 * config/aarch64/aarch64-protos.h
30979 (aarch64_simd_gen_const_vector_dup): Likewise.
30980 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
30982 2017-01-19 David Malcolm <dmalcolm@redhat.com>
30984 * langhooks-def.h (lhd_type_for_size): New decl.
30985 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
30986 * langhooks.c (lhd_type_for_size): New function, taken from
30989 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
30991 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
30992 define_bypass for CR latency.
30993 (power9-cracked-alu): Update bypass latency and remove power9-branch.
30994 (power9-alu2): Add define_bypass for CR latency.
30996 (power9-mul): Update insn latency.
30997 (power9-mul-compare): Update insn latency, bypass latency and remove
31000 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31002 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
31004 * config/aarch64/aarch64.md
31005 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
31006 aarch64_nopcrelative_literal_loads.
31007 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
31009 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
31011 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
31012 TARGET_LOONGSON_3A.
31013 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
31015 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
31018 * config.gcc (supported_defaults): Add lxc1-sxc1.
31019 (with_lxc1_sxc1): Add validation.
31020 (all_defaults): Add lxc1-sxc1.
31021 * config/mips/mips.opt (mlxc1-sxc1): New option.
31022 * config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
31024 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
31025 __mips_no_lxc1_sxc1.
31026 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
31027 * doc/invoke.texi (-mlxc1-sxc1): Document the new option.
31028 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
31030 2017-01-19 Richard Biener <rguenther@suse.de>
31032 PR tree-optimization/72488
31033 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
31034 sure to restore SSA info.
31035 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
31037 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
31039 PR rtl-optimization/79121
31040 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
31041 of the inner type when shifting an extended value.
31043 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31046 * symtab.c (symtab_node::equal_address_to): Fix comparing of
31047 interposable aliases.
31049 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
31052 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
31053 Use the evmergelohi instruction.
31054 (mov_si<mode>_e500_subreg4_2_le): Likewise.
31055 (mov_sitf_e500_subreg8_2_be): Likewise.
31056 (mov_sitf_e500_subreg12_2_le): Likewise.
31057 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
31058 (mov_si<mode>_e500_subreg4_2_be): Likewise.
31059 (mov_sitf_e500_subreg8_2_le): Likewise.
31060 (mov_sitf_e500_subreg12_2_be): Likewise.
31062 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31064 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
31065 attribute from vecsimple to vecperm.
31066 (altivec_vbpermq2): Likewise.
31068 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31071 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
31073 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31074 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
31075 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
31076 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
31077 case where N arg is SIZE_MAX.
31078 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
31079 (cmpstrsi): Add pattern.
31081 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
31083 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
31084 __builtin_vec_revb builtins.
31085 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
31086 built-in functions to support generation of the ISA 3.0 XXBR<x>
31087 vector byte reverse instructions.
31088 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
31089 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
31090 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
31091 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
31092 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
31093 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
31094 (P9V_BUILTIN_VEC_REVB): Likewise.
31095 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
31096 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
31097 (p9_xxbrq_v16qi): Likewise.
31098 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
31099 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
31100 (p9_xxbrh_v8hi): Likewise.
31101 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
31102 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
31103 vec_revb built-in functions.
31105 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
31107 PR rtl-optimization/78952
31108 * config/i386/i386.md (any_extract): New code iterator.
31109 (*insvqi_2): Use any_extract for source operand.
31110 (*insvqi_3): Use any_shiftrt for source operand.
31112 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
31114 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
31116 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
31118 2017-01-18 Matthias Klose <doko@ubuntu.com>
31120 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
31122 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31124 * config/rs6000/altivec.h (vec_bperm): Change #define.
31125 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
31126 (altivec_vbpermq2): New define_insn.
31127 (altivec_vbpermd): Likewise.
31128 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
31129 function interface.
31130 (VBPERMD): Likewise.
31131 (VBPERM): New polymorphic function interface.
31132 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
31133 Add entries for P9V_BUILTIN_VEC_VBPERM.
31134 * doc/extend.texi: Add interfaces for vec_bperm.
31136 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
31138 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
31139 first letter of error messages.
31140 (s390_resolve_overloaded_builtin): Likewise.
31141 * config/s390/s390.c (s390_expand_builtin): Likewise.
31142 (s390_invalid_arg_for_unprototyped_fn): Likewise.
31143 (s390_valid_target_attribute_inner_p): Likewise.
31144 * config/s390/s390.md ("tabort"): Likewise.
31146 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
31148 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
31149 (ISA_AVOID_DIV_HILO): New macro.
31150 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
31151 (ISA_HAS_DDIV): Likewise.
31153 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31155 * doc/invoke.texi (fabi-version): Correct number of occurrences.
31157 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31159 * doc/invoke.texi (fabi-version): Spelling fix.
31161 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31164 * doc/invoke.texi (fabi-version): Mention mangling fix for
31167 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
31170 * doc/invoke.texi (fabi-version): Document discriminator mangling.
31172 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
31175 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
31176 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
31178 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
31180 (stack_protect_test): Ditto.
31181 * config/rs6000/rs6000.opt (mstack-protector-guard=,
31182 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
31184 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
31185 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
31186 -mstack-protector-guard-offset=.
31187 (RS/6000 and PowerPC Options): Ditto.
31189 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31191 * config/i386/i386.h (MASK_CLASS_P): New define.
31192 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
31193 there are no registers from different register sets also when
31194 mask registers are used. Update function comment.
31195 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
31196 to (*k/*r) and (*k/*km) alternatives.
31198 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
31200 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
31201 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
31202 (EH_RETURN_HANDLER_RTX): New define.
31203 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
31204 Force frame pointer in EH return functions.
31205 (aarch64_expand_epilogue): Add barrier for eh_return.
31206 (aarch64_final_eh_return_addr): Remove.
31207 (aarch64_eh_return_handler_rtx): New function.
31208 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
31210 (aarch64_eh_return_handler_rtx): New prototype.
31212 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31214 * config/rs6000/altivec.h (vec_rlmi): New #define.
31215 (vec_vrlnm): Likewise.
31216 (vec_rlnm): Likewise.
31217 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
31218 (UNSPEC_VRLNM): Likewise.
31219 (VIlong): New mode iterator.
31220 (altivec_vrl<VI_char>mi): New define_insn.
31221 (altivec_vrl<VI_char>nm): Likewise.
31222 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
31224 (VRLDNM): Likewise.
31225 (RLNM): New polymorphic function entry.
31226 (VRLWMI): New monomorphic function entry.
31227 (VRLDMI): Likewise.
31228 (RLMI): New polymorphic function entry.
31229 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
31230 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
31231 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
31234 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31237 * dwarf2out.c (field_byte_offset): Restore the
31238 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
31239 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
31240 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
31243 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31246 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
31248 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31251 * dwarf2out.c (add_data_member_location_attribute): For constant
31252 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
31253 instead of DW_AT_data_member_location, DW_AT_bit_offset and
31254 DW_AT_byte_size attributes.
31256 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
31258 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
31259 after forcing to constant memory when the code model is medium.
31261 2017-01-17 Julia Koval <julia.koval@intel.com>
31264 * config/i386/avx512fintrin.h
31265 (_mm512_i32gather_ps): Change __addr type to void const*.
31266 (_mm512_mask_i32gather_ps): Ditto.
31267 (_mm512_i32gather_pd): Ditto.
31268 (_mm512_mask_i32gather_pd): Ditto.
31269 (_mm512_i64gather_ps): Ditto.
31270 (_mm512_mask_i64gather_ps): Ditto.
31271 (_mm512_i64gather_pd): Ditto.
31272 (_mm512_mask_i64gather_pd): Ditto.
31273 (_mm512_i32gather_epi32): Ditto.
31274 (_mm512_mask_i32gather_epi32): Ditto.
31275 (_mm512_i32gather_epi64): Ditto.
31276 (_mm512_mask_i32gather_epi64): Ditto.
31277 (_mm512_i64gather_epi32): Ditto.
31278 (_mm512_mask_i64gather_epi32): Ditto.
31279 (_mm512_i64gather_epi64): Ditto.
31280 (_mm512_mask_i64gather_epi64): Ditto.
31281 (_mm512_i32scatter_ps): Change __addr type to void*.
31282 (_mm512_mask_i32scatter_ps): Ditto.
31283 (_mm512_i32scatter_pd): Ditto.
31284 (_mm512_mask_i32scatter_pd): Ditto.
31285 (_mm512_i64scatter_ps): Ditto.
31286 (_mm512_mask_i64scatter_ps): Ditto.
31287 (_mm512_i64scatter_pd): Ditto.
31288 (_mm512_mask_i64scatter_pd): Ditto.
31289 (_mm512_i32scatter_epi32): Ditto.
31290 (_mm512_mask_i32scatter_epi32): Ditto.
31291 (_mm512_i32scatter_epi64): Ditto.
31292 (_mm512_mask_i32scatter_epi64): Ditto.
31293 (_mm512_i64scatter_epi32): Ditto.
31294 (_mm512_mask_i64scatter_epi32): Ditto.
31295 (_mm512_i64scatter_epi64): Ditto.
31296 (_mm512_mask_i64scatter_epi64): Ditto.
31297 * config/i386/avx512pfintrin.h
31298 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
31299 (_mm512_mask_prefetch_i32gather_ps): Ditto.
31300 (_mm512_mask_prefetch_i64gather_pd): Ditto.
31301 (_mm512_mask_prefetch_i64gather_ps): Ditto.
31302 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
31303 (_mm512_prefetch_i32scatter_ps): Ditto.
31304 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
31305 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
31306 (_mm512_prefetch_i64scatter_pd): Ditto.
31307 (_mm512_prefetch_i64scatter_ps): Ditto.
31308 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
31309 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
31310 * config/i386/avx512vlintrin.h
31311 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
31312 (_mm_mmask_i32gather_ps): Ditto.
31313 (_mm256_mmask_i32gather_pd): Ditto.
31314 (_mm_mmask_i32gather_pd): Ditto.
31315 (_mm256_mmask_i64gather_ps): Ditto.
31316 (_mm_mmask_i64gather_ps): Ditto.
31317 (_mm256_mmask_i64gather_pd): Ditto.
31318 (_mm_mmask_i64gather_pd): Ditto.
31319 (_mm256_mmask_i32gather_epi32): Ditto.
31320 (_mm_mmask_i32gather_epi32): Ditto.
31321 (_mm256_mmask_i32gather_epi64): Ditto.
31322 (_mm_mmask_i32gather_epi64): Ditto.
31323 (_mm256_mmask_i64gather_epi32): Ditto.
31324 (_mm_mmask_i64gather_epi32): Ditto.
31325 (_mm256_mmask_i64gather_epi64): Ditto.
31326 (_mm_mmask_i64gather_epi64): Ditto.
31327 (_mm256_i32scatter_ps): Change __addr type to void*.
31328 (_mm256_mask_i32scatter_ps): Ditto.
31329 (_mm_i32scatter_ps): Ditto.
31330 (_mm_mask_i32scatter_ps): Ditto.
31331 (_mm256_i32scatter_pd): Ditto.
31332 (_mm256_mask_i32scatter_pd): Ditto.
31333 (_mm_i32scatter_pd): Ditto.
31334 (_mm_mask_i32scatter_pd): Ditto.
31335 (_mm256_i64scatter_ps): Ditto.
31336 (_mm256_mask_i64scatter_ps): Ditto.
31337 (_mm_i64scatter_ps): Ditto.
31338 (_mm_mask_i64scatter_ps): Ditto.
31339 (_mm256_i64scatter_pd): Ditto.
31340 (_mm256_mask_i64scatter_pd): Ditto.
31341 (_mm_i64scatter_pd): Ditto.
31342 (_mm_mask_i64scatter_pd): Ditto.
31343 (_mm256_i32scatter_epi32): Ditto.
31344 (_mm256_mask_i32scatter_epi32): Ditto.
31345 (_mm_i32scatter_epi32): Ditto.
31346 (_mm_mask_i32scatter_epi32): Ditto.
31347 (_mm256_i32scatter_epi64): Ditto.
31348 (_mm256_mask_i32scatter_epi64): Ditto.
31349 (_mm_i32scatter_epi64): Ditto.
31350 (_mm_mask_i32scatter_epi64): Ditto.
31351 (_mm256_i64scatter_epi32): Ditto.
31352 (_mm256_mask_i64scatter_epi32): Ditto.
31353 (_mm_i64scatter_epi32): Ditto.
31354 (_mm_mask_i64scatter_epi32): Ditto.
31355 (_mm256_i64scatter_epi64): Ditto.
31356 (_mm256_mask_i64scatter_epi64): Ditto.
31357 (_mm_i64scatter_epi64): Ditto.
31358 (_mm_mask_i64scatter_epi64): Ditto.
31359 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
31360 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
31361 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
31362 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
31363 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
31364 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
31365 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
31366 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
31367 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
31368 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
31369 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
31370 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
31371 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
31372 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
31373 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
31374 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
31375 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
31376 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
31377 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
31378 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
31379 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
31380 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
31381 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
31382 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
31383 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
31384 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
31385 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
31386 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
31387 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
31388 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
31389 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
31390 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
31391 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
31392 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
31393 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
31394 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
31395 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
31396 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
31397 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
31398 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
31399 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
31400 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
31401 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
31402 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
31403 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
31404 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
31405 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
31406 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
31407 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
31408 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
31409 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
31410 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
31411 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
31412 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
31413 definitions accordingly.
31415 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
31416 Kuan-Lin Chen <kuanlinchentw@gmail.com>
31419 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
31422 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
31425 * ira-conflicts.c (ira_build_conflicts): Update total conflict
31426 hard regs for inner regno.
31428 2017-01-17 Martin Liska <mliska@suse.cz>
31431 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
31432 assumption and add comment.
31434 2017-01-17 Nathan Sidwell <nathan@acm.org>
31436 * ipa-visibility.c (localize_node): New function, broken out of ...
31437 (function_and_variable_visibility): ... here. Call it.
31439 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
31441 PR middle-end/77445
31442 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
31443 correctly set frequency of oudgoing edge.
31444 (duplicate_thread_path): Fix profile updating.
31446 2017-01-17 Jakub Jelinek <jakub@redhat.com>
31449 * configure.ac: Add GCC_BASE_VER.
31450 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
31451 version from BASE-VER file.
31452 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
31453 (gcc.o): Depend on $(BASEVER).
31454 * common.opt (dumpfullversion): New option.
31455 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
31456 * doc/invoke.texi: Document -dumpfullversion.
31457 * doc/install.texi: Document --with-gcc-major-version-only.
31458 * configure: Regenerated.
31460 2017-01-17 Richard Biener <rguenther@suse.de>
31462 PR tree-optimization/71433
31463 * tree-vrp.c (register_new_assert_for): Merge same asserts
31464 on all incoming edges.
31465 (process_assert_insertions_for): Handle insertions at the
31468 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
31470 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
31471 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
31473 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
31476 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
31479 2017-01-17 Alan Modra <amodra@gmail.com>
31482 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
31483 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
31484 symbolic stack limit when pic.
31486 2017-01-16 Martin Sebor <msebor@redhat.com>
31488 PR tree-optimization/78608
31489 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
31491 2017-01-16 Jeff Law <law@redhat.com>
31494 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
31495 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
31496 for several include directories that may be relative to sysroot.
31497 * config/i386/x-mingw32 (gplus_includedir): Define.
31498 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
31499 (native_system_includedir): Likewise.
31500 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
31501 override if TARGET_SYSTEM_ROOT is defined.
31502 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
31504 PR tree-optimization/79090
31505 PR tree-optimization/33562
31506 PR tree-optimization/61912
31507 PR tree-optimization/77485
31508 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
31509 and computed trims into the dump file.
31511 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
31513 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
31515 2017-01-16 Jakub Jelinek <jakub@redhat.com>
31518 * gimplify.c (gimplify_init_constructor): If want_value and
31519 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
31523 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
31524 sequence. Formatting fixes.
31525 (doloop_optimize): Formatting fixes.
31528 * gcc.c (debug_level_greater_than_spec_func): New function.
31529 (static_spec_functions): Add debug-level-gt spec function.
31530 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
31532 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31533 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
31534 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
31535 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
31536 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
31537 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
31539 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
31541 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
31542 QImode fixups to general and mask registers only.
31544 2017-01-16 Carl Love <cel@us.ibm.com>
31546 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
31547 for built-in functions
31548 vector signed char vec_nabs (vector signed char)
31549 vector signed short vec_nabs (vector signed short)
31550 vector signed int vec_nabs (vector signed int)
31551 vector signed long long vec_nabs (vector signed long long)
31552 vector float vec_nabs (vector float)
31553 vector double vec_nabs (vector double)
31554 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
31556 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
31557 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
31558 * doc/extend.texi: Update the documentation file for the new built-in
31561 2017-01-16 Martin Sebor <msebor@redhat.com>
31563 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
31566 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31568 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
31569 UNSPEC_VSX__XXSPLTD to require special splat handling.
31571 2017-01-16 David Malcolm <dmalcolm@redhat.com>
31574 * system.h: Poison strndup.
31576 2017-01-16 Alan Modra <amodra@gmail.com>
31579 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
31582 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
31584 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
31586 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
31588 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
31589 call recog here. Assert that INSN_CODE (insn) is non-negative.
31591 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
31594 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
31596 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
31597 in the currently scheduled RTL fragment.
31599 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
31601 PR rtl-optimization/78751
31602 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
31605 2017-01-14 Jeff Law <law@redhat.com>
31607 PR tree-optimization/79090
31608 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
31609 variable length stores.
31610 (compute_trims): Delete dead assignment to *trim_tail.
31611 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
31614 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
31616 PR rtl-optimization/78626
31617 PR rtl-optimization/78727
31618 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
31619 of a block, and split such blocks after everything else is finished.
31621 2017-01-14 Alan Modra <amodra@gmail.com>
31624 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
31625 target legitimate_combined_insn.
31626 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
31627 (rs6000_legitimate_combined_insn): New function.
31628 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
31630 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
31631 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
31632 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
31634 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
31636 * doc/frontends.texi (G++ and GCC): Remove references to Java.
31638 2017-01-13 Jeff Law <law@redhat.com>
31640 PR tree-optimization/33562
31641 PR tree-optimization/61912
31642 PR tree-optimization/77485
31643 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
31645 (delete_dead_assignment): Likewise.
31646 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
31647 statement to delete_dead_call and delete_dead_assignment.
31649 2017-01-13 David Malcolm <dmalcolm@redhat.com>
31652 * substring-locations.c (format_warning_va): Strengthen case 1 so
31653 that both endpoints of the substring must be within the format
31654 range for just the substring to be printed.
31656 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
31658 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
31659 * config/i386/i386.c (ix86_target_string): Add missing options
31660 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
31661 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
31662 flags_other and ix86_target_other to flags2_other. Display unknown
31664 (ix86_valid_target_attribute_inner_p): Add missing options and
31665 reorder options by implied ISAs, as in ix86_target_string.
31667 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
31669 * hash-table.h (hash_table::too_empty_p): New function.
31670 (hash_table::expand): Use it.
31671 (hash_table::traverse): Likewise.
31672 (hash_table::empty_slot): Use sizeof (value_type) instead of
31673 sizeof (PTR) to convert bytes to elements. Shrink the table
31674 if the current size is excessive for the current number of
31677 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
31679 * ira-costs.c (record_reg_classes): Break from the inner loop
31680 early once alt_fail is known to be true. Update outer loop
31681 handling accordingly.
31683 2017-01-13 Jeff Law <law@redhat.com>
31685 * tree-ssa-dse.c (decrement_count): New function.
31686 (increment_start_addr, maybe_trim_memstar_call): Likewise.
31687 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
31688 when we know the partially dead statement is a mem* function.
31690 PR tree-optimization/61912
31691 PR tree-optimization/77485
31692 * tree-ssa-dse.c: Include expr.h.
31693 (maybe_trim_constructor_store): New function.
31694 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
31696 PR tree-optimization/33562
31697 PR tree-optimization/61912
31698 PR tree-optimization/77485
31699 * doc/invoke.texi: Document new dse-max-object-size param.
31700 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
31701 * tree-ssa-dse.c: Include params.h.
31702 (dse_store_status): New enum.
31703 (initialize_ao_ref_for_dse): New, partially extracted from
31705 (valid_ao_ref_for_dse, normalize_ref): New.
31706 (setup_live_bytes_from_ref, compute_trims): Likewise.
31707 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
31708 (maybe_trim_partially_dead_store): Likewise.
31709 (maybe_trim_complex_store): Likewise.
31710 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
31711 Track what bytes live from the original store. Return tri-state
31712 for dead, partially dead or live.
31713 (dse_dom_walker): Add constructor, destructor and new private members.
31714 (delete_dead_call, delete_dead_assignment): New extracted from
31716 (dse_optimize_stmt): Make a member of dse_dom_walker.
31717 Use initialize_ao_ref_for_dse.
31719 PR tree-optimization/33562
31720 PR tree-optimization/61912
31721 PR tree-optimization/77485
31722 * sbitmap.h (bitmap_count_bits): Prototype.
31723 (bitmap_clear_range, bitmap_set_range): Likewise.
31724 * sbitmap.c (bitmap_clear_range): New function.
31725 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
31727 2017-01-13 Martin Liska <mliska@suse.cz>
31730 * function.c (set_cfun): Add new argument force.
31731 * function.h (set_cfun): Likewise.
31732 * ipa-inline-transform.c (inline_call): Use the function when
31733 strict alising from is dropped for function we inline to.
31735 2017-01-13 Richard Biener <rguenther@suse.de>
31737 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
31738 for dumping GIMPLE INTEGER_CSTs.
31740 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31742 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
31743 to 201112L since C++17.
31745 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
31748 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
31749 if -fsanitize=kernel-address is present.
31751 2017-01-13 Richard Biener <rguenther@suse.de>
31753 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
31754 as _Literal ( type ) number in case usual suffixes do not
31755 preserve all information.
31757 2017-01-13 Richard Biener <rguenther@suse.de>
31759 PR tree-optimization/77283
31760 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
31761 and ssa-iterators.h.
31762 (is_feasible_trace): Implement a cost model based on joiner
31765 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
31768 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
31769 char or short to __float128/_Float128 directly.
31771 2017-01-12 Martin Sebor <msebor@redhat.com>
31773 to -Wformat-overflow.
31774 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
31775 (min_bytes_remaining): Same.
31776 (get_string_length): Same.
31777 (format_string): Same.
31778 (format_directive): Same.
31780 (pass_sprintf_length::handle_gimple_call): Same.
31782 2017-01-12 Jakub Jelinek <jakub@redhat.com>
31784 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
31785 info.nowrite calls with no lhs that can't throw. Return bool
31786 whether gsi_remove has been called or not.
31787 (pass_sprintf_length::handle_gimple_call): Return bool whether
31788 try_substitute_return_value called gsi_remove. Formatting fix.
31789 (pass_sprintf_length::execute): Don't use gsi_remove if
31790 handle_gimple_call returned true.
31793 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
31794 be removed due to side-effects, don't remove following barrier nor
31795 turn the successor edge into fallthru edge.
31797 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31800 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
31801 element-reversing loads and stores as not swappable.
31803 2017-01-12 Nathan Sidwell <nathan@acm.org>
31804 Nicolai Stange <nicstange@gmail.com>
31806 * combine.c (try_combine): Don't ignore result of overlap checking
31807 loop. Combine overlap & asm check into single loop.
31809 2017-01-12 Richard Biener <rguenther@suse.de>
31811 * tree-pretty-print.c (dump_generic_node): Provide -gimple
31812 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
31814 2017-01-12 Richard Biener <rguenther@suse.de>
31816 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
31817 and TS_TARGET_OPTION directly derive from TS_BASE.
31818 * tree-core.h (tree_optimization_option): Derive from tree_base.
31819 (tree_target_option): Likewise.
31821 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
31823 * config/i386/i386.c (memory_address_length): Increase len
31824 only when rip_relative_addr_p returns false.
31826 2017-01-11 Julia Koval <julia.koval@intel.com>
31828 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
31829 (OPTION_MASK_ISA_SGX_SET): New.
31830 (ix86_handle_option): Handle OPT_msgx.
31831 * config.gcc: Added sgxintrin.h.
31832 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
31833 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
31834 * config/i386/i386.c (ix86_target_string): Add -msgx.
31836 (ix86_option_override_internal): Handle new options.
31837 (ix86_valid_target_attribute_inner_p): Add sgx.
31838 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
31839 * config/i386/i386.opt: Add msgx.
31840 * config/i386/sgxintrin.h: New file.
31841 * config/i386/x86intrin.h: Add sgxintrin.h.
31843 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31846 * fold-const.c (maybe_nonzero_address): Return 1 for function
31848 (tree_single_nonzero_warnv_p): Don't handle function local objects
31852 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
31855 2017-01-11 David Malcolm <dmalcolm@redhat.com>
31858 * opts.c: Include "spellcheck.h"
31859 (struct string_fragment): New struct.
31860 (struct edit_distance_traits<const string_fragment &>): New
31862 (get_closest_sanitizer_option): New function.
31863 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
31865 2017-01-11 Jakub Jelinek <jakub@redhat.com>
31867 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
31869 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
31870 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
31871 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
31872 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
31873 for initial die_offset if dwarf_split_debug_info.
31874 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
31875 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
31877 (output_skeleton_debug_sections): Formatting fix. Use
31878 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
31879 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
31881 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
31883 * config/arm/cortex-a53.md: Add bypasses for
31884 cortex_a53_r2f_cvt.
31885 (cortex_a53_r2f): Only use for transfers.
31886 (cortex_a53_f2r): Likewise.
31887 (cortex_a53_r2f_cvt): Add reservation for conversions.
31888 (cortex_a53_f2r_cvt): Likewise.
31890 2017-01-11 Tamar Christina <tamar.christina@arm.com>
31892 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
31893 to all inlined functions, change static to extern.
31895 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
31898 * config/arm/arm.c (legitimize_pic_address): Handle reference to
31900 (arm_assemble_integer): Likewise.
31902 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
31904 * config.gcc: Use new awk script to check CPU, FPU and architecture
31905 parameters for --with-... options.
31906 * config/arm/parsecpu.awk: New file
31907 * config/arm/arm-cpus.in: New file.
31908 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
31910 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
31912 * config/arm/t-arm: Update dependency rules.
31913 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
31914 of processing .def files.
31915 * config/arm/genopt.sh: Deleted.
31916 * config/arm/gentune.sh: Deleted.
31917 * config/arm/arm-cores.def: Deleted.
31918 * config/arm/arm-arches.def: Deleted.
31919 * config/arm/arm-fpus.def: Deleted.
31920 * config/arm/arm-tune.md: Regenerated.
31921 * config/arm/arm-tables.opt: Regenerated.
31922 * config/arm/arm-cpu.h: New generated file.
31923 * config/arm/arm-cpu-data.h: New generated file.
31924 * config/arm/arm-cpu-cdata.h: New generated file.
31926 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
31929 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
31931 (input_varpool_node): Unpack dynamically_initialized bit.
31933 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
31935 PR rtl-optimization/79032
31936 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
31937 the alignment of the adjusted memory reference against that of MODE,
31938 instead of the alignment of the original memory reference.
31940 2017-01-11 Martin Jambor <mjambor@suse.cz>
31942 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
31944 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
31945 decorated functions.
31947 2017-01-11 Richard Biener <rguenther@suse.de>
31949 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
31950 set range/nonnull info for PHI results. Do not set it on
31951 stmts marked for removal.
31953 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
31955 * expr.c (store_field): In the bitfield case, fetch the return value
31956 from the registers before applying a single big-endian adjustment.
31957 Always do a final load for a BLKmode value not larger than a word.
31959 2017-01-10 David Malcolm <dmalcolm@redhat.com>
31962 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
31963 that we correctly handle column numbers greater than
31964 LINE_MAP_MAX_COLUMN_NUMBER.
31966 2017-01-10 Martin Sebor <msebor@redhat.com>
31968 PR middle-end/78245
31969 * gimple-ssa-sprintf.c (get_destination_size): Call
31970 {init,fini}object_sizes.
31971 * tree-object-size.c (addr_object_size): Adjust.
31972 (pass_through_call): Adjust.
31973 (pass_object_sizes::execute): Adjust.
31974 * tree-object-size.h (fini_object_sizes): Declare.
31976 2017-01-10 Martin Sebor <msebor@redhat.com>
31978 PR tree-optimization/78775
31979 * builtins.c (get_size_range): Move...
31980 * calls.c: ...to here.
31981 (alloc_max_size): Accept zero argument.
31982 (operand_signed_p): Remove.
31983 (maybe_warn_alloc_args_overflow): Call get_size_range.
31984 * calls.h (get_size_range): Declare.
31986 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
31988 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
31989 from TI's devices.csv file as of September 2016.
31990 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
31992 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
31994 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
31995 * doc/invoke.texi: Likewise.
31996 * doc/md.texi: Likewise.
31997 * doc/objc.texi: Likewise.
31999 2017-01-10 Joshua Conner <joshconner@google.com>
32001 * config/arm/fuchsia-elf.h: New file.
32002 * config/fuchsia.h: New file.
32003 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
32004 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
32006 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
32008 2016-01-10 Richard Biener <rguenther@suse.de>
32010 PR tree-optimization/79034
32011 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
32012 Propagate out degenerate PHIs in the joiner.
32014 2017-01-10 Martin Liska <mliska@suse.cz>
32016 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
32017 (sort_congruence_classes_by_decl_uid): Likewise.
32018 (sort_congruence_class_groups_by_decl_uid): Likewise.
32019 (sem_item_optimizer::merge_classes): Sort class, groups in these
32020 classes and members in the groups by DECL_UID of declarations.
32021 This would make merge operations stable.
32023 2017-01-10 Martin Liska <mliska@suse.cz>
32025 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
32026 usage of m_classes_vec.
32027 (sem_item_optimizer::~sem_item_optimizer): Likewise.
32028 (sem_item_optimizer::get_group_by_hash): Likewise.
32029 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
32030 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
32031 (sem_item_optimizer::verify_classes): Likewise.
32032 (sem_item_optimizer::process_cong_reduction): Likewise.
32033 (sem_item_optimizer::dump_cong_classes): Likewise.
32034 (sem_item_optimizer::merge_classes): Likewise.
32035 * ipa-icf.h (congruence_class_hash): Rename from
32036 congruence_class_group_hash. Remove declaration of m_classes_vec.
32038 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
32040 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
32041 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
32042 * config.gcc: Add avx512vpopcntdqintrin.h.
32043 * config/i386/avx512vpopcntdqintrin.h: New.
32044 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
32045 * config/i386/i386-builtin-types.def: Add new types.
32046 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
32047 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
32048 __builtin_ia32_vpopcountq_v8di_mask): New.
32049 * config/i386/i386-c.c (ix86_target_macros_internal): Define
32050 __AVX512VPOPCNTDQ__.
32051 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
32052 (PTA_AVX512VPOPCNTDQ): Define.
32053 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
32054 TARGET_AVX512VPOPCNTDQ_P): Define.
32055 * config/i386/i386.opt: Add mavx512vpopcntdq.
32056 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
32057 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
32059 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32061 PR middle-end/77484
32062 * predict.def (PRED_CALL): Set to 67.
32064 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
32066 * expr.c (store_field): In the bitfield case, if the value comes from
32067 a function call and is of an aggregate type returned in registers, do
32068 not modify the field mode; extract the value in all cases if the mode
32069 is BLKmode and the size is not larger than a word.
32071 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
32074 * config/i386/cpuid.h: Fix undefined behavior.
32076 2017-01-04 Jeff Law <law@redhat.com>
32078 PR tree-optimization/79007
32079 PR tree-optimization/67955
32080 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
32081 conservative for pt.null when flag_non_call_exceptions is on.
32083 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32085 PR translation/79019
32086 PR translation/79020
32087 * params.def (PARAM_INLINE_MIN_SPEEDUP,
32088 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
32089 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
32091 * config/avr/avr.opt (maccumulate-args): Likewise.
32092 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
32093 * common.opt (freport-bug): Likewise.
32094 * cif-code.def (CIF_FINAL_ERROR): Likewise.
32095 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
32096 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
32097 translatable string.
32098 * config/i386/i386.c (function_value_32): Likewise.
32099 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
32100 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
32102 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
32103 * common/config/msp430/msp430-common.c (msp430_handle_option):
32105 * symtab.c (symtab_node::verify_base): Likewise.
32106 * opts.c (set_debug_level): Likewise.
32107 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
32108 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
32109 missing whitespace to translatable strings.
32110 * config/avr/avr.md (bswapsi2): Fix typo in comment.
32111 * config/sh/superh.h: Likewise.
32112 * config/i386/xopintrin.h: Likewise.
32113 * config/i386/znver1.md: Likewise.
32114 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
32115 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
32116 * double-int.h (struct double_int): Likewise.
32117 * double-int.c (div_and_round_double): Likewise.
32118 * wide-int.cc: Likewise.
32119 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
32120 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
32121 * cfgcleanup.c (crossjumps_occured): Renamed to ...
32122 (crossjumps_occurred): ... this.
32123 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
32126 PR tree-optimization/78899
32127 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
32128 returning bool return struct loop *, NULL for failure and the new
32130 (versionable_outer_loop_p): Don't version outer loop if it has
32131 dont_vectorized bit set.
32132 (tree_if_conversion): When versioning outer loop, ensure
32133 tree_if_conversion is performed also on the inner loop of the
32134 non-vectorizable outer loop copy.
32135 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
32136 LOOP_VECTORIZED in inner loop of the scalar outer loop and
32137 prevent vectorization of it.
32138 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
32139 the outer loop vectorization of the non-scalar version is attempted
32140 before vectorization of the inner loop in scalar version. If
32141 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
32142 vectorization of its inner loop.
32143 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
32144 has 2 inner loops, rename also on edges from bb whose single pred
32145 is outer_loop->header. Fix typo in function comment.
32147 2017-01-09 Martin Sebor <msebor@redhat.com>
32150 * asan.c (asan_emit_stack_protection): Increase local buffer size
32151 to avoid snprintf truncation warning.
32153 2017-01-09 Andrew Pinski <apinski@cavium.com>
32155 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
32156 to reference thunderx2t99 for the tuning structure
32157 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
32159 (thunderx2t99_extra_costs): This.
32160 * config/aarch64/aarch64-tune.md: Regenerate.
32161 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
32162 (vulcan_addrcost_table): This.
32163 (vulcan_regmove_cost): Rename to ...
32164 (thunderx2t99_regmove_cost): This.
32165 (vulcan_vector_cost): Rename to ...
32166 (thunderx2t99_vector_cost): this.
32167 (vulcan_branch_cost): Rename to ...
32168 (thunderx2t99_branch_cost): This.
32169 (vulcan_tunings): Rename to ...
32170 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
32171 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
32173 2017-01-09 Martin Jambor <mjambor@suse.cz>
32177 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
32178 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
32179 (propagate_vr_accross_jump_function): Use the above function for all
32180 value range computations for pass-through jump functions and type
32181 converasion from explicit value range values.
32182 (ipcp_propagate_stage): Do not attempt to deduce types of formal
32183 parameters from TYPE_ARG_TYPES.
32184 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
32185 (ipa_write_node_info): Stream type of the actual argument.
32186 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
32188 2017-01-09 Martin Liska <mliska@suse.cz>
32191 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
32192 (lookup_compiler): Do not show error message with have_E.
32194 2017-01-09 Jakub Jelinek <jakub@redhat.com>
32196 PR tree-optimization/78938
32197 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
32198 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
32199 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
32200 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
32203 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
32205 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
32208 2017-01-09 Richard Biener <rguenther@suse.de>
32210 PR tree-optimization/78997
32211 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
32212 name condition properly.
32214 2017-01-09 Richard Biener <rguenther@suse.de>
32217 * dwarf2out.c (is_cxx): New overload with context.
32218 (is_naming_typedef_decl): Use it.
32220 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32222 * invoke.texi (Option Summary): Correct spacing in option lists
32223 and add line breaks to fix over-long lines.
32225 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32227 PR middle-end/17660
32229 * extend.texi (Common Variable Attributes): Add xref to GCC
32230 Internals manual to explain mode attribute keywords.
32232 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
32235 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
32236 and Preprocessor Options.
32237 (Options for Linking): Document -pthread here....
32238 (RS/6000 and PowerPC Options): ...not here.
32239 (Solaris 2 Options): ...or here.
32240 * doc/cppopts.texi: Document -pthread.
32242 2017-01-08 Martin Sebor <msebor@redhat.com>
32244 PR middle-end/77708
32245 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
32246 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
32247 New member functions.
32248 (format_directive): Used them.
32250 (pass_sprintf_length::handle_gimple_call): Same.
32251 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
32252 to avoid truncation for any argument.
32253 (extract_affine_mul): Same.
32254 * tree.c (get_file_function_name): Same.
32256 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32258 PR middle-end/77484
32259 * predict.def (PRED_INDIR_CALL): Set to 86.
32261 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32263 PR preprocessor/54124
32264 * doc/cppopts.texi: Reformat -d subtable to list the full name
32265 of the options. Add cross-reference to the docs for the general
32266 compiler -d options.
32267 * doc/invoke.texi (Developer Options): Add cross-reference to the
32268 preprocessor-specific -d option documentation.
32270 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32272 PR preprocessor/13498
32273 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
32274 redudant material, and reflect new command-line options.
32275 (System Headers): Likewise.
32277 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
32279 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
32280 -isystem, and -idirafter. Copy-edit.
32281 * doc/cppopts.texi: Copy-edit. Remove contradiction about
32282 default for -ftrack-macro-expansion. Delete obsolete and
32283 badly-formatted implementation details about -fdebug-cpp output.
32284 * doc/cppwarnopts.texi: Copy-edit.
32286 2017-01-07 David Malcolm <dmalcolm@redhat.com>
32289 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
32290 that the transition from a max line width >= 1<<10 to narrower
32291 lines works correctly.
32293 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
32295 * doc/options.texi (PerFunction): New.
32296 * opt-functions.awk (switch_flags): Map both Optimization and
32297 PerFunction to CL_OPTIMIZATION.
32298 * opth-gen.awk: Test for PerFunction flag along with
32300 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
32301 it only when the latter is present. Skip those that don't in
32302 the hash function generator.
32303 * common.opt (fvar-tracking): Mark as PerFunction instead of
32305 (fvar-tracking-assignments): Likewise.
32306 (fvar-tracking-assignments-toggle): Likewise.
32307 (fvar-tracking-uninit): Likewise.
32309 2017-01-07 Jakub Jelinek <jakub@redhat.com>
32311 PR translation/79018
32312 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
32315 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
32318 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
32319 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
32320 TARGET_LONG_JUMP_TABLE_OFFSETS.
32321 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
32322 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
32323 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
32324 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
32325 * config/m68k/m68k.md (tablejump expander): Likewise.
32326 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
32327 TARGET_LONG_JUMP_TABLE_OFFSETS.
32328 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
32329 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
32331 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
32332 David Holsgrove <david.holsgrove@xilinx.com>
32334 * common/config/microblaze/microblaze-common.c
32335 (TARGET_EXCEPT_UNWIND_INFO): Remove.
32336 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
32338 * config/microblaze/microblaze.c (microblaze_must_save_register)
32339 (microblaze_expand_epilogue, microblaze_return_addr): Handle
32341 (microblaze_eh_return): New function.
32342 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
32343 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
32344 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
32345 * config/microblaze/microblaze.md (eh_return): New pattern.
32347 2017-01-06 Jakub Jelinek <jakub@redhat.com>
32349 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
32350 GCC_DIAGNOSTIC_STRINGIFY): Define.
32352 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
32354 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32356 * config/arm/arm.md (<mcrr>): New.
32358 * config/arm/arm.c (arm_arch5te): New.
32359 (arm_option_override): Set arm_arch5te.
32360 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
32362 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
32363 (arm_mcrr_qualifiers): ... this. New.
32364 (MRRC_QUALIFIERS): Define to...
32365 (arm_mrrc_qualifiers): ... this. New.
32366 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
32368 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
32369 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
32370 (MRRCI, mrrc, MRRC): New.
32371 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
32372 VUNSPEC_MRRC2): New.
32374 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32376 * config/arm/arm.md (<mcr>): New.
32378 * config/arm/arm.c (arm_coproc_builtin_available): Add
32379 support for mcr, mrc, mcr2 and mrc2.
32380 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
32381 (arm_mcr_qualifiers): ... this. New.
32382 (MRC_QUALIFIERS): Define to ...
32383 (arm_mrc_qualifiers): ... this. New.
32384 (MCR_QUALIFIERS): Define to ...
32385 (arm_mcr_qualifiers): ... this. New.
32386 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
32388 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
32389 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
32390 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
32391 VUNSPEC_MRC2): New.
32393 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32395 * config/arm/arm.md (*ldc): New.
32399 * config/arm/arm.c (arm_coproc_builtin_available): Add
32400 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
32401 (arm_coproc_ldc_stc_legitimate_address): New.
32402 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
32403 'qualifier_const_pointer'.
32404 (LDC_QUALIFIERS): Define to...
32405 (arm_ldc_qualifiers): ... this. New.
32406 (STC_QUALIFIERS): Define to...
32407 (arm_stc_qualifiers): ... this. New.
32408 * config/arm/arm-protos.h
32409 (arm_coproc_ldc_stc_legitimate_address): New.
32410 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
32411 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
32412 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
32413 stc2, stcl, stc2l): New.
32414 * config/arm/constraints.md (Uz): New.
32415 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
32416 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
32417 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
32418 VUNSPEC_STC2L): New.
32420 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32422 * config/arm/arm.md (<cdp>): New.
32423 * config/arm/arm.c (neon_const_bounds): Rename this ...
32424 (arm_const_bounds): ... this.
32425 (arm_coproc_builtin_available): New.
32426 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
32427 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
32428 (CDP_QUALIFIERS): Define to...
32429 (arm_cdp_qualifiers): ... this. New.
32431 (arm_expand_builtin_args): Add case for 6 arguments.
32432 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
32433 (arm_const_bounds): ... this.
32434 (arm_coproc_builtin_available): New.
32435 * config/arm/arm_acle.h (__arm_cdp): New.
32437 * config/arm/arm_acle_builtins.def (cdp): New.
32439 * config/arm/iterators.md (CDPI,CDP,cdp): New.
32440 * config/arm/neon.md: Rename all 'neon_const_bounds' to
32441 'arm_const_bounds'.
32442 * config/arm/types.md (coproc): New.
32443 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
32444 * doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
32445 * doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
32446 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
32448 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32450 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
32451 (UBINOP_QUALIFIERS): New.
32453 (acle_builtin_data): New. Change comment.
32454 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
32455 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
32456 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
32457 arm_acle_builtins.def.
32458 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
32459 (arm_init_acle_builtins): New.
32460 (CRC32_BUILTIN): Remove.
32461 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
32462 crc32cb, crc32ch and crc32cw.
32463 (arm_init_crc32_builtins): Remove.
32464 (arm_init_builtins): Use arm_init_acle_builtins rather
32465 than arm_init_crc32_builtins.
32466 (arm_expand_acle_builtin): New.
32467 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
32468 * config/arm/arm_acle_builtins.def: New.
32470 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
32472 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
32473 (arm_builtin_datum): ... this.
32474 (arm_init_neon_builtin): Rename to ...
32475 (arm_init_builtin): ... this. Add a new parameters PREFIX
32476 and USE_SIG_IN_NAME.
32477 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
32478 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
32479 'arm_builtin_datum'.
32480 (arm_init_vfp_builtins): Likewise.
32481 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
32482 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
32483 (arm_expand_neon_args): Rename to ...
32484 (arm_expand_builtin_args): ... this. Rename builtin_arg
32485 enum values and differentiate between ARG_BUILTIN_MEMORY
32486 and ARG_BUILTIN_NEON_MEMORY.
32487 (arm_expand_neon_builtin_1): Rename to ...
32488 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
32489 values, arm_expand_builtin_args and add bool parameter NEON.
32490 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
32491 (arm_expand_vfp_builtin): Likewise.
32492 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
32494 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32496 PR middle-end/77484
32497 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
32498 * predict.c (tree_estimate_probability_bb): Reverse direction of
32499 polymorphic call predictor.
32501 2017-01-06 David Malcolm <dmalcolm@redhat.com>
32503 * passes.c (execute_one_pass): Split out pass-skipping logic into...
32504 (determine_pass_name_match): ...this new function and...
32505 (should_skip_pass_p): ...this new function.
32507 2017-01-06 Nathan Sidwell <nathan@acm.org>
32509 * ipa-visibility.c (function_and_variable_visibility): Reformat
32510 comments and long lines. Remove extrneous if.
32511 * symtab.c (symtab_node::make_decl_local): Fix code format.
32512 (symtab_node::set_section_for_node): Fix comment typo.
32514 2017-01-06 Martin Liska <mliska@suse.cz>
32517 * lra-constraints.c: Rename invariant to lra_invariant.
32518 * predict.c (set_even_probabilities): Initialize e to NULL.
32520 2017-01-05 Martin Sebor <msebor@redhat.com>
32522 PR tree-optimization/78910
32523 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
32524 (format_integer): Correct off-by-one error in the handling
32525 of precision with negative numbers in signed conversions..
32527 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
32529 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
32531 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32533 PR tree-optimization/71016
32534 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
32535 factor_out_conditional_conversion. Formatting fix.
32536 (factor_out_conditional_conversion): Add cond_stmt argument.
32537 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
32538 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
32541 2017-01-05 David Malcolm <dmalcolm@redhat.com>
32543 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
32544 read-rtl-function.o, and selftest-rtl.o.
32545 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
32546 (selftest::aarch64_test_loading_full_dump): New function.
32547 (selftest::aarch64_run_selftests): New function.
32548 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
32549 selftest::aarch64_run_selftests.
32550 * config/i386/i386.c
32551 (selftest::ix86_test_loading_dump_fragment_1): New function.
32552 (selftest::ix86_test_loading_call_insn): New function.
32553 (selftest::ix86_test_loading_full_dump): New function.
32554 (selftest::ix86_test_loading_unspec): New function.
32555 (selftest::ix86_run_selftests): Call the new functions.
32556 * emit-rtl.c (maybe_set_max_label_num): New function.
32557 * emit-rtl.h (maybe_set_max_label_num): New decl.
32558 * function.c (instantiate_decls): Guard call to
32559 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
32560 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
32562 * gensupport.c (gen_reader::gen_reader): Pass "false"
32563 for new "compact" param of rtx_reader.
32564 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
32565 rather than an empty string for NULL strings.
32566 * read-md.c: Potentially include config.h rather than bconfig.h.
32567 Wrap include of errors.h with #ifdef GENERATOR_FILE.
32568 (have_error): New global, copied from errors.c.
32569 (md_reader::read_name): Rename to...
32570 (md_reader::read_name_1): ...this, adding "out_loc" param,
32571 and converting "missing name or number" to returning false, rather
32573 (md_reader::read_name): Reimplement in terms of read_name_1.
32574 (md_reader::read_name_or_nil): New function.
32575 (md_reader::read_string): Handle "(nil)" by returning NULL.
32576 (md_reader::md_reader): Add new param "compact".
32577 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
32578 (md_reader::read_file): New method.
32579 * read-md.h (md_reader::md_reader): Add new param "compact".
32580 (md_reader::read_file): New method.
32581 (md_reader::is_compact): New accessor.
32582 (md_reader::read_name): Convert return type from void to file_location.
32583 (md_reader::read_name_or_nil): New decl.
32584 (md_reader::read_name_1): New decl.
32585 (md_reader::m_compact): New field.
32586 (noop_reader::noop_reader): Pass "false" for new "compact" param
32588 (rtx_reader::rtx_reader): Add new "compact" param.
32589 (rtx_reader::read_rtx_operand): Make virtual and convert return
32590 type from void to rtx.
32591 (rtx_reader::read_until): New decl.
32592 (rtx_reader::handle_any_trailing_information): New virtual function.
32593 (rtx_reader::postprocess): New virtual function.
32594 (rtx_reader::finalize_string): New virtual function.
32595 (rtx_reader::m_in_call_function_usage): New field.
32596 (rtx_reader::m_reuse_rtx_by_id): New field.
32597 * read-rtl-function.c: New file.
32598 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
32599 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
32600 (selftest::verify_three_block_rtl_cfg): New decl.
32601 * read-rtl-function.h: New file.
32602 * read-rtl.c: Potentially include config.h rather than bconfig.h.
32603 For host, include function.h, memmodel.h, and emit-rtl.h.
32604 (one_time_initialization): New function.
32605 (struct compact_insn_name): New struct.
32606 (compact_insn_names): New array.
32607 (find_code): Handle insn codes in compact dumps.
32608 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
32609 (bind_subst_iter_and_attr): Likewise.
32610 (add_condition_to_string): Likewise.
32611 (add_condition_to_rtx): Likewise.
32612 (apply_attribute_uses): Likewise.
32613 (add_current_iterators): Likewise.
32614 (apply_iterators): Likewise.
32615 (initialize_iterators): Guard usage of apply_subst_iterator with
32616 #ifdef GENERATOR_FILE.
32617 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
32618 (md_reader::read_mapping): Likewise.
32619 (add_define_attr_for_define_subst): Likewise.
32620 (add_define_subst_attr): Likewise.
32621 (read_subst_mapping): Likewise.
32622 (check_code_iterator): Likewise.
32623 (rtx_reader::read_rtx): Likewise. Move one-time initialization
32625 (one_time_initialization): New function.
32626 (rtx_reader::read_until): New method.
32627 (read_flags): New function.
32628 (parse_reg_note_name): New function.
32629 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
32630 Handle reuse_rtx ids.
32631 Wrap iterator lookup within #ifdef GENERATOR_FILE.
32632 Add parsing support for RTL dumps, mirroring the special-cases in
32633 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
32634 values, and calling handle_any_trailing_information.
32635 (rtx_reader::read_rtx_operand): Convert return type from void
32636 to rtx, returning return_rtx. Handle case 'e'. Call
32637 finalize_string on XSTR and XTMPL fields.
32638 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
32639 "(nil)" values were omitted. Call the postprocess vfunc on the
32641 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
32642 class ctor. Initialize m_in_call_function_usage. Call
32643 one_time_initialization.
32644 * rtl-tests.c (selftest::test_uncond_jump): Call
32645 set_new_first_and_last_insn.
32646 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
32647 * selftest-rtl.c: New file.
32648 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
32649 (selftest::get_insn_by_uid): New decl.
32650 * selftest-run-tests.c (selftest::run_tests): Call
32651 read_rtl_function_c_tests.
32652 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
32653 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
32656 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
32658 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
32659 operands in a special way. Assert that pos+len <= mode precision.
32661 2017-01-05 Jakub Jelinek <jakub@redhat.com>
32663 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
32664 3 argument Alias with unlimited for the negative form.
32665 (fno-vect-cost-model): Removed.
32667 2017-01-05 Martin Liska <mliska@suse.cz>
32669 * hsa-gen.c (gen_hsa_divmod): New function.
32670 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
32672 2017-01-05 Martin Liska <mliska@suse.cz>
32675 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
32678 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32680 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
32681 small constant length operands.
32683 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
32685 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
32686 between loop iterations.
32688 2017-01-05 Martin Liska <mliska@suse.cz>
32691 * gimplify.c (gimplify_decl_expr): Compare to
32692 asan_poisoned_variables instread of checking flags.
32693 (gimplify_target_expr): Likewise.
32694 (gimplify_expr): Likewise.
32695 (gimplify_function_tree): Conditionally initialize
32696 asan_poisoned_variables.
32698 2017-01-04 Jeff Law <law@redhat.com>
32700 PR tree-optimizatin/78812
32701 * rtl.h (contains_mem_rtx_p): Prototype.
32702 * ifcvt.c (containts_mem_rtx_p): Move from here to...
32703 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
32704 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
32705 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
32706 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
32708 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32710 * input.c (assert_char_at_range): Default-initialize actual_range.
32712 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32714 * df-scan.c (df_ref_create_structure): Make regno unsigned,
32715 to match the caller.
32717 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32719 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
32720 insns after final jump in test to emit dummy move.
32722 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32724 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
32725 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
32727 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
32729 * multiple_target.c (create_dispatcher_calls): Init e_next.
32730 * tree-ssa-loop-split.c (split_loop): Init border.
32731 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
32734 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
32739 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
32740 (altivec_register_operand): Do not return true if the operand
32741 contains a SUBREG mixing SImode and SFmode.
32742 (vsx_register_operand): Likewise.
32743 (vsx_reg_sfsubreg_ok): New predicate.
32744 (vfloat_operand): Do not return true if the operand contains a
32745 SUBREG mixing SImode and SFmode.
32746 (vint_operand): Likewise.
32747 (vlogical_operand): Likewise.
32748 (gpc_reg_operand): Likewise.
32749 (int_reg_operand): Likewise.
32750 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
32751 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
32752 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
32754 (rs6000_emit_move_si_sf_subreg): New helper function.
32755 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
32756 fixup SUBREGs involving SImode and SFmode.
32757 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
32758 numbers for the new peephole2 optimization.
32759 (peephole2 for SFmode unions): New peephole2 to optimize cases in
32760 the GLIBC math library that do AND/IOR/XOR operations on single
32761 precision floating point.
32762 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
32763 target macros to say whether we need to avoid SUBREGs mixing
32765 (TARGET_ALLOW_SF_SUBREG): Likewise.
32766 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
32767 (UNSPEC_SI_FROM_SF): Likewise.
32768 (iorxor): Change spacing.
32769 (and_ior_xor): New iterator for AND, IOR, and XOR.
32770 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
32771 (movdi_from_sf_zero_ext): Likewise.
32772 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
32773 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
32774 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
32775 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
32776 (fms<mode>4): Likewise.
32777 (fnma<mode>4): Likewise.
32778 (fnms<mode>4): Likewise.
32779 (nfma<mode>4): Likewise.
32780 (nfms<mode>4): Likewise.
32782 2017-01-04 Marek Polacek <polacek@redhat.com>
32785 * doc/invoke.texi: Document -Wpointer-compare.
32787 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32789 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
32792 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
32793 descriptions for -gdwarf-5 and emit them as uleb128 instead of
32796 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
32799 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
32800 documentation of the powerpc_popcntb_ok attribute.
32801 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
32802 code to issue warning messages if a requested CPU configuration is
32803 not supported by the binary (assembler and loader) toolchain.
32804 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
32805 made to define a built-in function that has been disabled.
32806 (paired_init_builtins): Add assertion to prevent ICE if attempt is
32807 made to define a built-in function that has been disabled.
32808 (altivec_init_builtins): Add comment explaining why definition
32809 of the DST built-in functions is not preceded by an assertion
32810 check. Add assertions to prevent ICE if attempts are made to
32811 define an altivec predicate or an abs* built-in function that has
32813 (htm_init_builtins): Add comment explaining why definition of the
32814 htm built-in functions is not preceded by an assertion check.
32816 2017-01-04 Jeff Law <law@redhat.com>
32818 PR tree-optimizatin/67955
32819 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
32820 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
32821 the points-to solution does not include pt_null. Use DECL_PT_UID
32824 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
32826 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
32827 Use gen_int_mode instead of gen_lopwart for const_int operands.
32829 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32831 PR tree-optimization/71563
32832 * match.pd: Simplify X << Y into X if Y is known to be 0 or
32833 out of range value - has low bits known to be zero.
32835 2017-01-04 Alan Modra <amodra@gmail.com>
32837 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
32838 * configure: Regenerate.
32839 * config.in: Regenerate.
32841 2017-01-04 Jakub Jelinek <jakub@redhat.com>
32844 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
32845 a substring of the message, but strcmp with the whole message. Ifdef
32846 ENABLE_NLS, translate the message first using dgettext.
32848 2017-01-03 Jeff Law <law@redhat.com>
32850 PR tree-optimizatin/78856
32851 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
32852 (mark_threaded_blocks): Remove code to truncate thread paths that
32853 cross multiple loop headers. Instead invalidate the cached loop
32854 iteration information and handle case of a thread path walking
32855 into an irreducible region.
32857 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
32860 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
32861 assertions. Add support for doing the signbit if the IEEE 128-bit
32862 floating point value is in a GPR.
32863 * config/rs6000/rs6000.md (Fsignbit): Delete.
32864 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
32865 Update the length attribute if the value is in a GPR.
32866 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
32867 the sign or zero extension instruction, since the value is always 0/1.
32868 (signbit<mode>2_dm2): Delete using <Fsignbit>.
32871 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
32872 extracting SImode to a GPR register so that we can generate a
32873 store, limit the vector to be in a traditional Altivec register
32874 for the vextuwrx instruction.
32876 2017-01-03 Ian Lance Taylor <iant@google.com>
32878 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
32880 2017-01-03 Martin Sebor <msebor@redhat.com>
32882 PR tree-optimization/78696
32883 * gimple-ssa-sprintf.c (format_floating): Correct handling of
32884 precision. Use MPFR for %f for greater fidelity. Correct handling
32886 (pass_sprintf_length::compute_format_length): Set width and precision
32887 specified by asrerisk to void_node for vararg functions.
32888 (try_substitute_return_value): Adjust dump output.
32890 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
32892 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
32894 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
32896 * doc/invoke.texi (SPARC options): Document -mlra as the default.
32897 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
32898 -mlra/-mno-lra was passed to the compiler.
32900 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
32902 PR rtl-optimization/65618
32903 * emit-rtl.c (try_split): Move initialization of "before" and
32904 "after" to just before the call to emit_insn_after_setloc.
32906 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
32908 * doc/md.texi (Standard Names): Remove reference to Java frontend.
32910 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
32912 * dwarf2out.c (gen_enumeration_type_die): When
32913 -gno-strict-dwarf, add a DW_AT_encoding attribute.
32915 2017-01-03 Jakub Jelinek <jakub@redhat.com>
32917 PR tree-optimization/78965
32918 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
32919 Change first argument from const call_info & to call_info &. For %n
32920 set info.nowrite to false.
32922 PR middle-end/78901
32923 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
32924 possibly throwing calls.
32926 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
32927 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
32928 and fns handling, rather than in a separate case SSA_NAME.
32930 2017-01-02 Jeff Law <law@redhat.com>
32932 * config/darwin-driver.c (darwin_driver_init): Const-correctness
32933 fixes for first_period and second_period variables.
32935 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
32938 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
32939 (*insvqi_1): New insn pattern.
32940 (*insvqi_1_mem_rex64): Ditto.
32941 (*insvqi_2): Ditto.
32942 (*insvqi_3): Rename from *insvqi.
32944 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
32946 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
32948 * doc/cfg.texi (Edges): Remove reference to Java.
32949 (Maintaining the CFG): Ditto.
32951 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32953 PR middle-end/77674
32954 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
32955 transparent aliases.
32957 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
32959 PR middle-end/77484
32960 * predict.def (PRED_CALL): Update hitrate.
32961 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
32962 * predict.c (tree_estimate_probability_bb): Split CALL predictor
32963 into direct/indirect/polymorphic variants.
32965 2017-01-01 Jakub Jelinek <jakub@redhat.com>
32967 Update copyright years.
32969 * gcc.c (process_command): Update copyright notice dates.
32970 * gcov-dump.c (print_version): Ditto.
32971 * gcov.c (print_version): Ditto.
32972 * gcov-tool.c (print_version): Ditto.
32973 * gengtype.c (create_file): Ditto.
32974 * doc/cpp.texi: Bump @copying's copyright year.
32975 * doc/cppinternals.texi: Ditto.
32976 * doc/gcc.texi: Ditto.
32977 * doc/gccint.texi: Ditto.
32978 * doc/gcov.texi: Ditto.
32979 * doc/install.texi: Ditto.
32980 * doc/invoke.texi: Ditto.
32982 Copyright (C) 2017 Free Software Foundation, Inc.
32984 Copying and distribution of this file, with or without modification,
32985 are permitted in any medium without royalty provided the copyright
32986 notice and this notice are preserved.