]> git.ipfire.org Git - thirdparty/gcc.git/blob - gcc/ChangeLog
PR middle-end/87188
[thirdparty/gcc.git] / gcc / ChangeLog
1 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
2
3 PR middle-end/87188
4 * dojump.c (do_compare_and_jump): Canonicalize function pointers
5 when one operand is a function pointer. Use POINTER_TYPE_P and
6 FUNC_OR_METHOD_TYPE_P.
7 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
8 * fold-const.c (build_range_check): Likewise.
9 * match.pd (simple_comparison): Likewise.
10
11 2018-09-14 David Malcolm <dmalcolm@redhat.com>
12
13 PR c/82967
14 * spellcheck.c (get_edit_distance_cutoff): New function.
15 (selftest::test_edit_distance_unit_test_oneway): Rename to...
16 (selftest::test_get_edit_distance_one_way): ...this.
17 (selftest::test_get_edit_distance_unit): Rename to...
18 (selftest::test_get_edit_distance_both_ways): ...this.
19 (selftest::test_edit_distances): Move tests to this new function,
20 and test some more pairs of strings. Update for above renaming.
21 (selftest::get_old_cutoff): New function.
22 (selftest::test_get_edit_distance_cutoff): New function.
23 (selftest::assert_suggested_for): New function.
24 (ASSERT_SUGGESTED_FOR): New macro.
25 (selftest::assert_not_suggested_for): New function.
26 (ASSERT_NOT_SUGGESTED_FOR): New macro.
27 (selftest::test_suggestions): New function.
28 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
29 tests to selftest::test_edit_distances and call it. Add calls to
30 selftest::test_get_edit_distance_cutoff and
31 selftest::test_suggestions.
32 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
33 (best_match::consider): Replace hard-coded cutoff calculation with
34 a call to...
35 (best_match::get_cutoff): New declaration.
36 (best_match::get_best_meaningful_candidate): Likewise.
37
38 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
39
40 * builtins.c (fold_builtin_strlen): Remove TODO comment.
41
42 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
43
44 revert:
45 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
46
47 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
48 terminated string literal.
49
50 2018-09-14 Martin Sebor <msebor@redhat.com>
51
52 * builtins.c (unterminated_array): Handle ARRAY_REF.
53 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
54 * builtins.h (unterminated_array): Declare extern.
55 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
56 arrays.
57 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
58 calls.
59
60 2018-09-14 Martin Sebor <msebor@redhat.com>
61 Jeff Law <law@redhat.com>
62
63 * builtins.c (unterminated_array): New.
64 (expand_builtin_strcpy): Adjust.
65 (expand_builtin_strcpy_args): Detect unterminated arrays.
66 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
67 unterminated arrays.
68 * gimple-fold.h (get_maxval_strlen): Add argument.
69 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
70
71 * gimple-fold.c (get_range_strlen): Add argument.
72 (get_maxval_strlen): Adjust.
73 * gimple-fold.h (get_range_strlen): Add argument.
74
75 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
76
77 * config/i386/movdirintrin.h: Fix copyright year.
78
79 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
80
81 * reg-stack.c: Include regs.h.
82 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
83 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
84 FIRST_STACK_REG, not DFmode.
85 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
86 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
87 (change stack): Default register mode to the reg_raw_mode of
88 FIRST_STACK_REG, not DFmode.
89 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
90 (*swapxf): Rename from swapxf.
91
92 2018-09-14 Carl Love <cel@us.ibm.com>
93
94 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
95 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
96
97 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
98
99 PR target/87224
100 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
101 alternatives.
102
103 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
104
105 PR target/85628
106 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
107
108 2018-09-14 Jason Merrill <jason@redhat.com>
109
110 Fix --enable-gather-detailed-mem-stats.
111 * hash-table.c (hash_table_usage): Change from variable to function.
112 * hash-table.h: Adjust.
113 * Makefile.in: Add missing dependencies on hash-table.h.
114
115 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
116
117 PR tree-optimization/87259
118 PR lto/87283
119 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
120 execute_cse_reciprocals_1 has tried transforming.
121
122 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
123
124 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
125 VR_VARYING for PLUS/MINUS_EXPR.
126
127 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
128
129 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
130 formatting.
131
132 2018-09-14 Richard Biener <rguenther@suse.de>
133
134 PR middle-end/63155
135 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
136 bits for the merged partition.
137
138 2018-09-13 Martin Sebor <msebor@redhat.com>
139 Bernd Edlinger <bernd.edlinger@hotmail.de>
140
141 * builtins.h (c_srlen): Add argument.
142 * builtins.c (warn_string_no_nul): New function.
143 (c_strlen): Add argument and use it. Update recursive calls.
144 Pass DECL argument to string_constant to get info on non
145 terminated strings. Update *NONSTR as needed.
146 (fold_builtin_strlen): Add argument to calls to c_strlen.
147 Warn for unterminated arrays.
148 (warn_string_no_null): Add prototype.
149 * expr.c (string_constant): Update arguments. Update recursive
150 calls appropriately. Detect missing NUL terminator and outermost
151 declaration its missing in.
152 Improve checks for arrays with nonzero lower bound or elements
153 that are not a single byte. Simplify offset computation.
154 Simplify checks for non-NUL terminated strings.
155 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
156 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
157
158 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
159
160 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
161 correctly.
162 * fold-const.c (c_getstr): Fix function comment. Remove unused third
163 argument. Fix range checks.
164 * fold-const.h (c_getstr): Adjust protoype.
165 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
166 string is constant but contains no NUL byte.
167
168 * expr.c (string_constant): Adjust function comment.
169 Remove bogus check for zero termination.
170
171 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
172
173 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
174 (get_constant_size): Don't make STRING_CSTs larger than they are.
175 (check_string_literal): New check function for STRING_CSTs.
176 (output_constant): Use it.
177
178 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
179
180 PR target/86812
181 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
182
183 2018-09-13 Richard Biener <rguenther@suse.de>
184
185 PR tree-optimization/87263
186 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
187 (struct unwind_state): Add max_rpo field.
188 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
189 Compute max_rpo, the max RPO number a block can be backwards reached
190 from. Re-write non-iterating mode to a RPO ordered worklist approach,
191 separating it from the iterating mode.
192
193 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
194
195 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
196 (rfs_decision): New scheduling decision.
197
198 2018-09-13 Richard Biener <rguenther@suse.de>
199
200 PR bootstrap/87134
201 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
202 (vn_nary_op_insert_pieces_predicated): Do not write useless
203 valid_dominated_by_p entry outside of the allocated storage.
204
205 2018-09-13 Omar Sandoval <osandov@osandov.com>
206 Tom de Vries <tdevries@suse.de>
207
208 PR debug/86985
209 * dwarf2out.c (is_c): New function.
210 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
211
212 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
213
214 PR target/85628
215 * config/aarch64/aarch64.md (*aarch64_bfxil):
216 Define.
217 * config/aarch64/constraints.md (Ulc): Define.
218 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
219 Define.
220 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
221 New function.
222
223 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
224
225 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
226 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
227 aarch64_layout_frame call.
228 (aarch64_expand_epilogue): Likewise.
229 (aarch64_initial_elimination_offset): Likewise.
230 (aarch64_get_separate_components): Likewise.
231 (aarch64_use_return_insn_p): Likewise.
232 (aarch64_layout_frame): Remove unneeded check.
233
234 2018-09-13 Jakub Jelinek <jakub@redhat.com>
235
236 * configure.ac: Only append
237 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
238 gcc_config_arguments if it was never reconfigured or last reconfigure
239 was with different arguments.
240 * configure: Regenerated.
241
242 2018-09-13 Jakub Jelinek <jakub@redhat.com>
243 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
244
245 PR middle-end/87290
246 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
247 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
248
249 2018-09-13 Jakub Jelinek <jakub@redhat.com>
250
251 PR tree-optimization/87287
252 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
253 X % C == 0 to X % (unsigned) C == 0 optimization to ...
254 * match.pd (X % C == 0): ... here. New optimization.
255
256 2018-09-12 Jakub Jelinek <jakub@redhat.com>
257
258 PR middle-end/82853
259 * expr.h (maybe_optimize_mod_cmp): Declare.
260 * expr.c (mod_inv): New function.
261 (maybe_optimize_mod_cmp): New function.
262 (do_store_flag): Use it.
263 * cfgexpand.c (expand_gimple_cond): Likewise.
264
265 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
266 Julian Brown <julian@codesourcery.com>
267
268 PR middle-end/86336
269 * gimplify.c (gimplify_scan_omp_clauses): Set
270 target_firstprivatize_array_bases in OpenACC parallel and kernels
271 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
272 OpenACC data regions.
273
274 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
275
276 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
277 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
278
279 2018-09-12 Richard Biener <rguenther@suse.de>
280
281 PR tree-optimization/87280
282 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
283 edge but unreachable target.
284 (do_rpo_vn): For conservatively handling a PHI only mark
285 the backedge executable but not the block reachable.
286
287 2018-09-12 Richard Biener <rguenther@suse.de>
288
289 PR tree-optimization/87266
290 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
291 visited blocks.
292
293 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
294
295 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
296 constants.
297 ("trunc<BFP:mode><DFP_ALL:mode>2")
298 ("trunc<DFP_ALL:mode><BFP:mode>2")
299 ("extend<BFP:mode><DFP_ALL:mode>2")
300 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
301 according to the target operand type.
302
303 2018-09-12 Jakub Jelinek <jakub@redhat.com>
304 Andreas Krebbel <krebbel@linux.ibm.com>
305
306 PR tree-optimization/86844
307 * gimple-ssa-store-merging.c
308 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
309 there are any overlapping stores in between them, make sure they are
310 also coalesced or we give up completely.
311
312 2018-09-12 Jakub Jelinek <jakub@redhat.com>
313
314 PR middle-end/87248
315 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
316 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
317
318 2018-09-12 Tom de Vries <tdevries@suse.de>
319
320 * common.opt (gdescribe-dies): Add option.
321 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
322 attribute for artifical and nameless decls.
323 (dwarf2out_register_external_die): Add description attribute to
324 external reference die.
325 (add_desc_attribute): New functions.
326 (gen_subprogram_die): Add description attribute to
327 DW_TAG_call_site_parameter.
328 * tree-pretty-print.c (print_generic_expr_to_str): New function.
329 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
330 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
331 -gno-describe-dies.
332 (@item -gdescribe-dies): Add.
333
334 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
335
336 * tree-vrp.c (vrp_shift_undefined_p): Remove.
337 (extract_range_from_binary_expr_1: Call
338 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
339 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
340 depend on sign.
341
342 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
343
344 * gimple-ssa-warn-alloca.c
345 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
346 field for ALLOCA_BOUND_*_LARGE.
347
348 2018-09-11 Nathan Sidwell <nathan@acm.org>
349
350 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
351
352 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
353
354 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
355 for clobbers. Remove obsolete comment.
356
357 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
358
359 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
360 mpxchk, mpxld and mpxst types.
361 (define_attr length_immediate): Remove all processing of mpx types.
362 (define_attr prefix_0f): Ditto.
363 (define_attr memory): Ditto.
364
365 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
366
367 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
368 (log<mode>2): Change operand 1 predicate to general_operand.
369 Extend operand 1 to XFmode and generate logxf3 insn.
370 (log10<mode>2): Change operand 1 predicate to general_operand.
371 Extend operand 1 to XFmode and generate log10xf3 insn.
372 (log2<mode>2): Change operand 1 predicate to general_operand.
373 Extend operand 1 to XFmode and generate log2xf3 insn.
374 (fyl2xp1_extend<mode>xf3_i387): Remove.
375 (log1p<mode>2): Change operand 1 predicate to general_operand.
376 Extend operand 1 to XFmode and generate log1pxf3 insn.
377 (fxtract_extend<mode>xf3_i387): Remove.
378 (logb<mode>2): Change operand 1 predicate to general_operand.
379 Extend operand 1 to XFmode and generate logbxf3 insn.
380 (ilogb<mode>2): Change operand 1 predicate to general_operand.
381 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
382 (significand<mode>2): Change operand 1 predicate to general_operand.
383 Extend operand 1 to XFmode and generate significandxf3 insn.
384
385 2018-09-11 Nathan Sidwell <nathan@acm.org>
386
387 * gcc.c (perror_with_name, pfatal_with_name): Delete.
388 (load_specs): Use fatal_error.
389 (DELETE_IF_ORDINARY, process_command): Use error.
390 (execute, run_attempt): Use fatal_error.
391
392 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
393
394 * diagnostic-core.h (sorry_at): New prototype.
395 * diagnostic.c (sorry_at): New function.
396
397 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
398
399 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
400 by zero as VR_UNDEFINED.
401
402 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
403
404 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
405 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
406 (<sincos>mode2): New expander.
407 (sincos_extend<mode>xf3_i387): Remove insn pattern.
408 (sincos -> sin, cos splitters): Remove splitter patterns.
409 (sincos<mode>3): Change operand 2 predicate to general_operand.
410 Extend operand 2 to XFmode and generate sincosxf3 insn.
411 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
412 Change operand 3 predicate to const1_operand.
413 (fptan_extend<mode>xf4_i387): Remove insn pattern.
414 (tanxf2): Update operands in the call to fptanxf4_i387.
415 (tan<mode>2): Change operand 1 predicate to general_operand.
416 Extend operand 1 to XFmode and generate tanxf3 insn.
417 (atan2xf3): Rename from *fpatanxf3_i387.
418 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
419 (atan2xf3): Remove expander.
420 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
421 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
422 (atan<mode>2): Change operand 1 predicate to general_operand.
423 Extend operand 1 to XFmode and generate atanxf3 insn.
424
425 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
426
427 * config/i386/i386.md (x87/SSE constant load splitter): Use
428 memory_operand instead of nonimmediate_operand for input operand
429 predicate.
430
431 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
432
433 * config/i386/i386.md (float partial SSE register stall splitter): Move
434 splitter near its instruction pattern.
435 (float_extend partial SSE register stall splitter): Ditto.
436 (float_truncate partial SSE register stall splitter): Ditto.
437
438 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
439
440 PR target/86794
441 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
442 to speculation_safe_value_not_needed.
443
444 PR target/85666
445 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
446 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
447 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
448 leaf_function_p, instead use has_hard_reg_initial_val.
449
450 2018-09-09 Nathan Sidwell <nathan@acm.org>
451
452 * gcc.h (pfatal_with_name): Don't declare here.
453 * gcc.c (pfatal_with_name): Make static.
454
455 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
456
457 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
458 earlyclobber.
459
460 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
461
462 PR rtl-optimization/85458
463 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
464 priority hook to reduce the priority of EXPR.
465
466 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
467
468 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
469 DImode for x87 on 32bit targets. Conditionally disable x87 modes
470 with X87_ENABLE_FLOAT. Remove preparation code.
471 (*float<SWI48:mode><MODEF:mode>2): Rename from
472 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
473 math using "enabled" attribute.
474 (*floatdi<MODEF:mode>2_i387): Rename from
475 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
476 enable for 32bit targets only.
477 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
478 splitter.
479 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
480 as operand 1 predicate. Rewrite as define_insn_and_split.
481 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
482
483 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
484
485 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
486 to fallthru to FLOAT case.
487
488 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
489
490 PR target/86731
491 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
492 around folding of vec_sl to handle out of range shift values.
493
494 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
495
496 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
497 Update callers to gen_fix_trunc<mode>_i387_fisttp
498 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
499 nonimmediate_operand.
500 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
501 and corresponding splitters.
502 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
503 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
504 (fix_truncdi_i387_with_temp): Remove insn pattern
505 and corresponding splitters.
506 (fix_trunc<mode>_i387): Change operand 0 predicate to
507 nonimmediate_operand.
508 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
509 and corresponding splitters.
510 (*fistdi2_1): Remove.
511 (fistdi2): Ditto.
512 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
513 (lrintxfdi2): Remove expander. Reimplement as define_insn.
514 (*fist<mode>2_1): Remove.
515 (fist<mode>2): Ditto.
516 (fist<mode>2_with_temp): Remove insn pattern and corresponding
517 splitters.
518 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
519 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
520 (fistdi2_<rounding>): Change operand 0 predicate to
521 nonimmediate_operand.
522 (fistdi2_<rounding>_with_temp): Remove insn pattern
523 and corresponding splitters.
524 (fist<mode>2_<rounding>): Change operand 0 predicate to
525 nonimmediate_operand.
526 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
527 and corresponding splitters.
528
529 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
530
531 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
532
533 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
534 the init value.
535
536 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
537
538 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
539 early gimple folding of vec_splat().
540 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
541 * gimple-fold.h: Add an extern define for tree_vec_extract().
542
543 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
544
545 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
546 wrappers around TREE_TYPE comparisons.
547
548 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
549
550 PR target/80080
551 * config/s390/predicates.md: Add nonsym_memory_operand.
552 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
553 contains a SYMBOL_REF, load it into an intermediate pseudo.
554 (s390_emit_compare_and_swap): Legitimize operand.
555 * config/s390/s390.md: Use the new nonsym_memory_operand
556 with UNSPECV_CAS patterns.
557
558 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
559
560 PR target/80080
561 * config/s390/s390-passes.def: New file.
562 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
563 declaration.
564 (make_pass_s390_early_mach): Add declaration.
565 * config/s390/s390.c (make_pass_s390_early_mach):
566 (s390_option_override): Remove dynamic registration.
567 * config/s390/t-s390: Add s390-passes.def.
568
569 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
570
571 * config/s390/s390.c (s390_decompose_constant_pool_ref):
572 Remove UNSPEC_LTREL_BASE check.
573 (annotate_constant_pool_refs): Likewise.
574 (find_constant_pool_ref): Likewise.
575 (find_ltrel_base): Removed.
576 (replace_ltrel_base): Removed.
577 (s390_mainpool_finish): Remove replace_ltrel_base call.
578 (s390_chunkify_start): Remove pending LTREL_BASE logic.
579 (s390_chunkify_finish): Remove replace_ltrel_base call.
580 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
581
582 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
583
584 PR target/86779
585 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
586 to speculation_safe_value_not_needed.
587
588 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
589 Bernd Schmidt <bernds_cb1@t-online.de>
590
591 * config/nvptx/nvptx-opts.h: New file.
592 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
593 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
594 (ASM_SPEC): Define.
595 (TARGET_SM35): New macro.
596 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
597 correct predicate.
598 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
599 values.
600 (misa=): New option.
601 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
602
603 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
604
605 * config/i386/i386.md (truncdfsf2): Remove expander.
606 (truncdfsf2_with_temp): Ditto.
607 (truncxf<mode>2): Ditto.
608 (*truncdfsf_fast_mixed): Remove insn pattern.
609 (*truncdfsf_fast_i387): Ditto.
610 (*truncdfsf_mixed): Ditto.
611 (*truncdfsf_i387): Ditto.
612 (*truncdfsf2_i387_1): Ditto.
613 (*truncxfsf2_mixed): Ditto.
614 (*truncxfdf2_mixed): Ditto.
615 (*truncxf<mode>2_i387_noop): Ditto. Update callers
616 to call gen_truncxf<mode>2 instead.
617 (*truncxf<mode>2_i387): Remove.
618 (reg->reg splitters): Remove splitter pattern.
619 (reg->mem splitters): Ditto.
620
621 (truncdfsf2): New insn pattern.
622 (truncxf<mode>2): Ditto.
623
624 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
625
626 * tree-ssa-math-opts.c (is_mult_by): New function.
627 (is_square_of): Use the above.
628 (optimize_recip_sqrt): New function.
629 (pass_cse_reciprocals::execute): Use the above.
630
631 2018-09-05 Richard Biener <rguenther@suse.de>
632
633 PR bootstrap/87134
634 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
635 to zero-init the emplaced vec.
636
637 2018-09-05 Martin Liska <mliska@suse.cz>
638
639 PR tree-optimization/87205
640 * tree-switch-conversion.c (pass_lower_switch::execute):
641 Group cases for switch statements.
642
643 2018-09-05 Richard Biener <rguenther@suse.de>
644
645 PR tree-optimization/87217
646 * tree-ssa-sccvn.c (vuse_valueize): New.
647 (vn_reference_lookup_pieces): Use it.
648 (vn_reference_lookup): Likewise.
649
650 2018-09-05 Nathan Sidwell <nathan@acm.org>
651
652 PR c++/87137
653 * stor-layout.c (place_field): Scan forwards to check last
654 bitfield when ms_bitfield_placement is in effect.
655
656 2018-09-05 Richard Biener <rguenther@suse.de>
657
658 PR bootstrap/87225
659 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
660 return.
661
662 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
663 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
664
665 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
666 * config.gcc (extra_objs): Build it.
667 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
668 Likewise.
669 * config/aarch64/aarch64-passes.def
670 (pass_tag_collision_avoidance): New pass.
671 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
672 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
673 (aarch64_classify_address): Remove static qualifier.
674 (aarch64_address_info, aarch64_address_type): Move to...
675 * config/aarch64/aarch64-protos.h: ... here.
676 (make_pass_tag_collision_avoidance): New function.
677 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
678 New tuning flag.
679
680 2018-09-05 Martin Liska <mliska@suse.cz>
681
682 * doc/gcov.texi: Update documentation of humar
683 readable mode.
684 * gcov.c (format_count): Print one decimal place, it provides
685 more fine number of situations like '1G' vs. '1.4G'.
686
687 2018-09-05 Martin Liska <mliska@suse.cz>
688
689 PR target/87164
690 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
691 * optc-gen.awk: Allow 'Var' for Deprecated options in order
692 to generate a MASK value.
693
694 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
695
696 PR debug/86593
697 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
698 if frame pointer isn't used.
699 (compute_frame_pointer_to_fb_displacement): Likewise.
700
701 2018-09-04 Jakub Jelinek <jakub@redhat.com>
702
703 PR target/87198
704 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
705 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
706 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
707 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
708 and OPTION_MASK_ISA_XSAVEC_UNSET.
709
710 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
711
712 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
713 XOR operations in NAND case.
714
715 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
716
717 * wide-int-range.cc (wide_int_range_convert): New.
718 * wide-int-range.h (wide_int_range_convert): New.
719 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
720 code into wide_int_range_convert.
721 (extract_range_into_wide_ints): Do not munge anti range constants
722 into the entire domain. Just return the range back.
723
724 2018-09-04 Martin Liska <mliska@suse.cz>
725
726 * genmatch.c (output_line_directive): Add new argument
727 fnargs.
728 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
729
730 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
731
732 * doc/invoke.texi (Option Summary): Add whitespace.
733
734 * doc/invoke.texi (Option Summary): Add -Waligned-new.
735
736 2018-09-04 Richard Biener <rguenther@suse.de>
737
738 PR tree-optimization/87211
739 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
740 backedge value we're supposed to treat as VARYING also number
741 the PHI to VARYING in case it got a different value-number already.
742
743 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
744
745 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
746 (extract_range_from_binary_expr_1): Do not call
747 vrp_can_optimize_bit_op.
748 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
749 static.
750 (wide_int_range_get_mask_and_bounds): New.
751 (wide_int_range_optimize_bit_op): New.
752 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
753 (wide_int_range_bit_and): Same.
754 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
755 (wide_int_range_optimize_bit_op): New.
756 (wide_int_range_get_mask_and_bounds): New.
757
758 2018-09-04 Richard Biener <rguenther@suse.de>
759
760 PR tree-optimization/87176
761 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
762 variable. When value-numbering a virtual PHI node make sure
763 to not value-number to the backedge value.
764
765 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
766
767 * doc/extend.texi (Long Long, Hex Floats): Document support for
768 long long and hex floats in more recent versions of ISO C++.
769
770 2018-09-03 Richard Biener <rguenther@suse.de>
771
772 PR tree-optimization/87177
773 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
774 cleanup.
775
776 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
777
778 * bb-reorder.c (edge_order): Convert to C-qsort-style
779 tri-state comparator.
780 (reorder_basic_blocks_simple): Change std::stable_sort to gcc_stablesort.
781
782 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
783
784 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
785 tri-state comparator.
786 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
787
788 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
789
790 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
791 (mergesort): ... here as maximum count for using netsort.
792 (gcc_qsort): Set nlim to 3 if stable sort is requested.
793 (gcc_stablesort): New.
794 * system.h (gcc_stablesort): Declare.
795
796 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
797
798 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
799 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
800 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
801
802 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
803
804 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
805 lxsdx and stxsdx alternatives.
806 (*mov<mode>_hardfloat64): Ditto.
807 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
808
809 2018-09-03 Richard Biener <rguenther@suse.de>
810
811 PR tree-optimization/87200
812 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
813 simplify result.
814
815 2018-09-03 Martin Liska <mliska@suse.cz>
816
817 PR tree-optimization/87201
818 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
819 Fix parenthesis in an expression.
820
821 2018-09-03 Richard Biener <rguenther@suse.de>
822
823 PR tree-optimization/87197
824 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
825 visited. CSE the VN_INFO hashtable lookup.
826
827 PR tree-optimization/87169
828 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
829 iterating make sure there's no extra backedges from irreducible
830 regions feeding the header. Mark the destination block
831 executable.
832
833 2018-09-03 Martin Liska <mliska@suse.cz>
834
835 PR driver/83193
836 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
837 * common/common-targhooks.c (default_get_valid_option_values):
838 New function.
839 * common/common-targhooks.h (default_get_valid_option_values):
840 Likewise.
841 * common/config/i386/i386-common.c: Move processor_target_table
842 from i386.c.
843 (ix86_get_valid_option_values): New function.
844 (TARGET_GET_VALID_OPTION_VALUES): New macro.
845 * config/i386/i386.c (struct ptt): Move to i386-common.c.
846 (PTA_*): Move all defined masks into i386-common.c.
847 (ix86_function_specific_restore): Use new processor_cost_table.
848 * config/i386/i386.h (struct ptt): Moved from i386.c.
849 (struct pta): Likewise.
850 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
851 * doc/tm.texi.in: Likewise.
852 * opt-suggestions.c (option_proposer::suggest_option):
853 Pass prefix to build_option_suggestions.
854 (option_proposer::get_completions): Likewise.
855 (option_proposer::build_option_suggestions): Use the new target
856 hook.
857 * opts.c (struct option_help_tuple): New struct.
858 (print_filtered_help): Use the new target hook.
859
860 2018-09-03 Martin Liska <mliska@suse.cz>
861
862 PR middle-end/59521
863 * predict.c (set_even_probabilities): Add likely_edges
864 argument and handle cases where we have precisely one
865 likely edge.
866 (combine_predictions_for_bb): Catch also likely_edges.
867 (tree_predict_by_opcode): Handle gswitch statements.
868 * tree-cfg.h (find_case_label_for_value): New declaration.
869 (find_taken_edge_switch_expr): Likewise.
870 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
871 Find pivot in decision tree based on probabily, not by number of
872 nodes.
873
874 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
875
876 * doc/standards.texi (Standards): Update Objective-C reference.
877
878 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
879
880 * doc/install.texi (Prerequisites): Update link for MPC.
881
882 2018-09-01 Michael Matz <matz@suse.de>
883
884 PR tree-optimization/87074
885 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
886 PHIs for outer-loop uses.
887
888 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
889
890 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
891 * doc/invoke.texi (C Dialect Options): Ditto.
892
893 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
894
895 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
896
897 2018-08-31 Richard Biener <rguenther@suse.de>
898
899 PR tree-optimization/87168
900 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
901 (rpo_elim::eliminate_avail): When OP was not visited it must
902 be available.
903
904 2018-08-31 David Malcolm <dmalcolm@redhat.com>
905
906 * tree-vrp.c (copy_value_range): Convert param "from" from
907 "value_range *" to "const value_range *".
908 (range_is_null): Likewise for param "vr".
909 (range_int_cst_p): Likewise.
910 (range_int_cst_singleton_p): Likewise.
911 (symbolic_range_p): Likewise.
912 (value_ranges_intersect_p): Likewise for both params.
913 (value_range_nonnegative_p): Likewise for param "vr".
914 (value_range_constant_singleton): Likewise.
915 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
916 (extract_range_into_wide_ints): Likewise for param "vr".
917 (extract_range_from_multiplicative_op): Likewise for params "vr0"
918 and "vr1".
919 (vrp_can_optimize_bit_op): Likewise.
920 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
921 "vr1_".
922 (extract_range_from_unary_expr): Likewise.
923 (debug_value_range): Likewise for param "vr".
924 (value_range::dump): Add "const" qualifier.
925 (vrp_prop::check_array_ref): Convert local "vr" from
926 "value_range *" to "const value_range *".
927 (vrp_prop::check_mem_ref): Likewise.
928 (vrp_prop::visit_stmt): Likewise for local "old_vr".
929 (vrp_intersect_ranges_1): Likewise for param "vr_1".
930 (vrp_intersect_ranges): Likewise.
931 (simplify_stmt_for_jump_threading): Likewise for local "vr".
932 (vrp_prop::vrp_finalize): Likewise.
933 * tree-vrp.h (value_range::dump): Add "const" qualifier.
934 (vrp_intersect_ranges): Add "const" qualifier to params as above.
935 (extract_range_from_unary_expr): Likewise.
936 (value_range_constant_singleton): Likewise.
937 (symbolic_range_p): Likewise.
938 (copy_value_range): Likewise.
939 (extract_range_from_binary_expr_1): Likewise.
940 (range_int_cst_p): Likewise.
941 (vrp_set_zero_nonzero_bits): Likewise.
942 (range_int_cst_singleton_p): Likewise.
943
944 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
945
946 * config/aarch64/arm_neon.h (vabsd_s64): New.
947 (vnegd_s64): Likewise.
948
949 2018-08-31 Martin Jambor <mjambor@suse.cz>
950
951 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
952
953 2018-08-31 Martin Liska <mliska@suse.cz>
954
955 * ipa-icf.c (sem_item::add_type): Use
956 sem_item::m_type_hash_cache.
957 * ipa-icf.h: Move the cache from sem_item_optimizer
958 to sem_item.
959
960 2018-08-31 Nathan Sidwell <nathan@acm.org>
961
962 * doc/extend.texi (Backwards Compatibility): Remove implicit
963 extern C leeway of () being (...).
964
965 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
966
967 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
968
969 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
970
971 PR target/86684
972 PR target/87149
973 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
974
975 2018-08-31 Jakub Jelinek <jakub@redhat.com>
976
977 PR middle-end/87138
978 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
979 gen_int_mode. Formatting fixes.
980
981 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
982
983 * target.def (custom_function_descriptors): Improve documentation.
984 * doc/tm.texi.in (Trampolines): Expand discussion of function
985 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
986 beginning of the section.
987 * doc/tm.texi: Regenerated.
988
989 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
990
991 * cfg.h (class auto_edge_flag): Spell out the template-id of the
992 base class in the initializer list. This is a workaround for
993 building with older GCC.
994 (class auto_bb_flag): Likewise.
995
996 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
997
998 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
999 (altivec_vcmpequ<VI_char>_p): Remove star.
1000 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
1001 vector load modes.
1002 (expand_strncmp_vec_sequence): New function.
1003 (emit_final_str_compare_vec): New function.
1004 (expand_strn_compare): Add support for vector strncmp.
1005 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
1006 length specification to bytes.
1007 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
1008 (vcmpnezb_p): New pattern.
1009 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
1010 for option -mstring-compare-inline-limit.
1011
1012 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
1013
1014 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
1015 (PTA_SKYLAKE): Add PTA_AES.
1016 (PTA_GOLDMONT): Likewise.
1017
1018 2018-08-29 Jan Hubicka <jh@suse.cz>
1019
1020 PR lto/86517
1021 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
1022 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
1023
1024 2018-08-29 Jan Hubicka <jh@suse.cz>
1025
1026 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
1027 TYPE_STUB_DECL.
1028 (hash_tree): Do not visit TYPE_STUB_DECL.
1029 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
1030 stream TYPE_STUB_DECL.
1031 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
1032 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
1033 after free_lang_data.
1034 (type_in_anonymous_namespace_p): Likewise.
1035
1036 2018-08-29 Jan Hubicka <jh@suse.cz>
1037
1038 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
1039 comment that it has to be even number.
1040 (class sreal): Change m_sig type to int32_t.
1041 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
1042 int64_t for temporary calculations.
1043 (sreal_verify_basics): Drop one bit from minimum and maximum.
1044
1045 2018-08-30 Richard Biener <rguenther@suse.de>
1046
1047 PR tree-optimization/87147
1048 * tree-ssa-sccvn.c (SSA_VISITED): New function.
1049 (visit_phi): When the degenerate result is from the backedge and
1050 we didn't visit its definition yet drop to VARYING.
1051 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
1052
1053 2018-08-29 Jan Hubicka <jh@suse.cz>
1054
1055 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
1056 DECL_VINDEX.
1057 (hash_tree): Likewise.
1058
1059 2018-08-29 Jan Hubicka <jh@suse.cz>
1060
1061 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
1062 and TYPE_NEXT_REF_TO.
1063
1064 2018-08-29 Jan Hubicka <jh@suse.cz>
1065
1066 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
1067 comment that it has to be even number.
1068 (class sreal): Change m_sig type to int32_t.
1069 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
1070 int64_t for temporary calculations.
1071 (sreal_verify_basics): Drop one bit from minimum and maximum.
1072
1073 2018-08-30 Tamar Christina <tamar.christina@arm.com>
1074
1075 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
1076
1077 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
1078
1079 PR middle-end/86995
1080 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
1081 if to_add is negative.
1082
1083 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
1084
1085 PR middle-end/87053
1086 * builtins.c (c_strlen): Improve range checks.
1087
1088 2018-08-29 Martin Sebor <msebor@redhat.com>
1089 Jeff Law <law@redhat.com>
1090
1091 PR tree-optimization/86714
1092 PR tree-optimization/86711
1093 * builtins.c (c_strlen): Add arguments to call to string_constant.
1094 * expr.c (string_constant): Add argument. Detect missing nul
1095 terminator and outermost declaration it's missing in.
1096 * expr.h (string_constant): Add argument.
1097 * fold-const.c (read_from_constant_string): Add arguments to call to
1098 string_constant.
1099 (c_getstr): Likewise.
1100 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
1101 to string_constant.
1102 * tree-ssa-strlen.c (get_stridx): Likewise.
1103
1104 2018-08-29 Jan Hubicka <jh@suse.cz>
1105
1106 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
1107 Do not stream DECL_VINDEX.
1108 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
1109 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
1110 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
1111
1112 2018-08-29 Richard Biener <rguenther@suse.de>
1113
1114 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
1115 virtual operands that are not default defs to honor region
1116 boundaries.
1117 (rpo_vn_valueize): Remove ineffective code here.
1118
1119 2018-08-29 Richard Biener <rguenther@suse.de>
1120
1121 PR tree-optimization/87132
1122 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
1123 when skipping defs reachable over backedges.
1124
1125 2018-08-29 Richard Biener <rguenther@suse.de>
1126
1127 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
1128 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
1129 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
1130 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
1131 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
1132 (vn_reference_lookup_3): Remove use of const_parms.
1133 (free_rpo_vn): Do not free const_parms.
1134 (do_rpo_vn): Do not call init_const_parms.
1135 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
1136 SSA_NAME_POINTS_TO_READONLY_MEMORY.
1137 (call_may_clobber_ref_p_1): Likewise.
1138
1139 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
1140
1141 PR other/86726
1142 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
1143 (-O): Ditto.
1144 (-ftree-scev-cprop): Document.
1145
1146 2018-08-29 Jan Hubicka <jh@suse.cz>
1147
1148 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
1149 parameters.
1150 (sreal constructor): Update.
1151 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
1152 sreal:operator/): Update.
1153
1154 2018-08-29 Martin Liska <mliska@suse.cz>
1155
1156 * tree-switch-conversion.c (switch_conversion::expand):
1157 Strenghten assumption about gswitch statements.
1158
1159 2018-08-29 Richard Biener <rguenther@suse.de>
1160
1161 PR tree-optimization/87117
1162 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
1163 re-value-number released SSA VDEFs.
1164
1165 2018-08-29 Richard Biener <rguenther@suse.de>
1166
1167 PR tree-optimization/87126
1168 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
1169
1170 2018-08-28 Jim Wilson <jimw@sifive.com>
1171
1172 * config/riscv/pic.md: Rewrite.
1173 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
1174 invalid address.
1175 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
1176 (SOFTF, default_load, softload, softstore): New.
1177
1178 2018-08-28 Jeff Law <law@redhat.com>
1179
1180 * fold-const.c (fold_binary_loc): Remove recently added assert.
1181
1182 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
1183
1184 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
1185 to OP parmeter of generated function.
1186
1187 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
1188
1189 PR tree-optimization/87009
1190 * match.pd: Add boolean optimizations.
1191
1192 2018-08-28 Martin Sebor <msebor@redhat.com>
1193
1194 PR middle-end/86631
1195 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
1196 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
1197 (pass_walloca::gate): Use it.
1198 (alloca_call_type): Same.
1199 (pass_walloca::execute): Same.
1200 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
1201
1202 2018-08-28 David Malcolm <dmalcolm@redhat.com>
1203
1204 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
1205 GCC_VERSION for usage of "__gcc_dump_printf__" format from
1206 >= 3005 to >= 9000.
1207
1208 2018-08-28 Richard Biener <rguenther@suse.de>
1209
1210 PR tree-optimization/87124
1211 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
1212 constants before looking up avail.
1213
1214 2018-08-28 Jakub Jelinek <jakub@redhat.com>
1215
1216 PR middle-end/87099
1217 * calls.c (maybe_warn_nonstring_arg): Punt early if
1218 warn_stringop_overflow is zero. Don't call get_range_strlen
1219 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
1220 Swap comparison operands to have constants on rhs. Only use
1221 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
1222 increment lenrng[0].
1223
1224 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
1225
1226 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
1227 use of tree_to_shwi. Remove duplicated test for the size being
1228 a whole number of bytes.
1229
1230 2018-08-28 Richard Biener <rguenther@suse.de>
1231
1232 PR tree-optimization/87117
1233 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
1234 Handle removed stmt without LHS (GIMPLE_NOP).
1235
1236 2018-08-28 Richard Biener <rguenther@suse.de>
1237
1238 PR tree-optimization/87117
1239 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
1240 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
1241
1242 2018-08-28 Richard Biener <rguenther@suse.de>
1243
1244 PR tree-optimization/87117
1245 * tree-ssa-pre.c (compute_avail): Do not make expressions
1246 with predicated values available.
1247 (get_expr_value_id): Assert we do not run into predicated value
1248 expressions.
1249
1250 2018-08-28 Richard Biener <rguenther@suse.de>
1251
1252 PR tree-optimization/87117
1253 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
1254 get virtual operands.
1255 (get_expr_operands): Handle STRING_CST like other decls.
1256
1257 2018-08-28 Martin Liska <mliska@suse.cz>
1258
1259 * tree.h: Update documentation of fndecl_built_in_p
1260 functions.
1261
1262
1263 2018-08-27 Jeff Law <law@redhat.com>
1264 PR tree-optimization/87110
1265 * tree-ssa-dse.c (compute_trims): Handle non-constant
1266 TYPE_SIZE_UNIT.
1267
1268 2018-08-27 Martin Sebor <msebor@redhat.com>
1269
1270 PR tree-optimization/86914
1271 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
1272
1273 2018-08-27 Martin Sebor <msebor@redhat.com>
1274
1275 PR tree-optimization/87112
1276 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
1277 the type of the bound argument.
1278
1279 2018-08-27 Jeff Law <law@redhat.com>
1280
1281 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
1282 type does not have a TYPE_SIZE_UNIT.
1283
1284 2018-08-27 Steve Ellcey <sellcey@cavium.com>
1285
1286 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
1287 with include of backend.h.
1288
1289 2018-08-27 Richard Biener <rguenther@suse.de>
1290
1291 PR tree-optimization/86927
1292 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
1293 use const cond reduction code.
1294
1295 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
1296
1297 PR tree-optimization/85758
1298 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
1299
1300 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1301
1302 PR c++/87091
1303 * diagnostic-show-locus.c (class layout_range): Update for
1304 conversion of show_caret_p to a tri-state.
1305 (layout_range::layout_range): Likewise.
1306 (make_range): Likewise.
1307 (layout::maybe_add_location_range): Likewise.
1308 (layout::should_print_annotation_line_p): Don't show annotation
1309 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
1310 (layout::get_state_at_point): Update for conversion of
1311 show_caret_p to a tri-state. Bail out early for
1312 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
1313 underlining or source colorization.
1314 (gcc_rich_location::add_location_if_nearby): Update for conversion
1315 of show_caret_p to a tri-state.
1316 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
1317 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
1318 Likewise.
1319 (selftest::test_one_liner_labels): Likewise.
1320 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
1321 conversion of show_caret_p to a tri-state.
1322 * pretty-print.c (text_info::set_location): Likewise.
1323 * pretty-print.h (text_info::set_location): Likewise.
1324 * substring-locations.c (format_warning_n_va): Likewise.
1325 * tree-diagnostic.c (default_tree_printer): Likewise.
1326 * tree-pretty-print.c (newline_and_indent): Likewise.
1327
1328 2018-08-27 David Malcolm <dmalcolm@redhat.com>
1329
1330 PR c++/87091
1331 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
1332 line above for line-insertion fix-it hints.
1333 (selftest::test_fixit_insert_containing_newline): Update the
1334 expected results, and add a test with line-numbering enabled.
1335
1336 2018-08-27 Martin Liska <mliska@suse.cz>
1337
1338 PR sanitizer/86962
1339 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
1340 params with DECL_HAS_VALUE_EXPR_P.
1341
1342 2018-08-27 Martin Liska <mliska@suse.cz>
1343
1344 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
1345 selected expansion strategy.
1346
1347 2018-08-27 Martin Liska <mliska@suse.cz>
1348
1349 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
1350 * builtins.c (is_builtin_fn): Likewise.
1351 * attribs.c (diag_attr_exclusions): Use new function
1352 fndecl_built_in_p and remove check for FUNCTION_DECL if
1353 possible.
1354 (builtin_mathfn_code): Likewise.
1355 (fold_builtin_expect): Likewise.
1356 (fold_call_expr): Likewise.
1357 (fold_builtin_call_array): Likewise.
1358 (fold_call_stmt): Likewise.
1359 (set_builtin_user_assembler_name): Likewise.
1360 (is_simple_builtin): Likewise.
1361 * calls.c (gimple_alloca_call_p): Likewise.
1362 (maybe_warn_nonstring_arg): Likewise.
1363 * cfgexpand.c (expand_call_stmt): Likewise.
1364 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
1365 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
1366 (cgraph_node::verify_node): Likewise.
1367 * cgraphclones.c (build_function_decl_skip_args): Likewise.
1368 (cgraph_node::create_clone): Likewise.
1369 * config/arm/arm.c (arm_insert_attributes): Likewise.
1370 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
1371 * dse.c (scan_insn): Likewise.
1372 * expr.c (expand_expr_real_1): Likewise.
1373 * fold-const.c (operand_equal_p): Likewise.
1374 (fold_binary_loc): Likewise.
1375 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
1376 * gimple-low.c (lower_stmt): Likewise.
1377 * gimple-pretty-print.c (dump_gimple_call): Likewise.
1378 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Likewise.
1379 * gimple.c (gimple_build_call_from_tree): Likewise.
1380 (gimple_call_builtin_p): Likewise.
1381 (gimple_call_combined_fn): Likewise.
1382 * gimplify.c (gimplify_call_expr): Likewise.
1383 (gimple_boolify): Likewise.
1384 (gimplify_modify_expr): Likewise.
1385 (gimplify_addr_expr): Likewise.
1386 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
1387 * ipa-cp.c (determine_versionability): Likewise.
1388 * ipa-fnsummary.c (compute_fn_summary): Likewise.
1389 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
1390 * ipa-split.c (visit_bb): Likewise.
1391 (split_function): Likewise.
1392 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
1393 * lto-cgraph.c (input_node): Likewise.
1394 * lto-streamer-out.c (write_symbol): Likewise.
1395 * omp-low.c (setjmp_or_longjmp_p): Likewise.
1396 (lower_omp_1): Likewise.
1397 * predict.c (strip_predict_hints): Likewise.
1398 * print-tree.c (print_node): Likewise.
1399 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
1400 * trans-mem.c (is_tm_irrevocable): Likewise.
1401 (is_tm_load): Likewise.
1402 (is_tm_simple_load): Likewise.
1403 (is_tm_store): Likewise.
1404 (is_tm_simple_store): Likewise.
1405 (is_tm_abort): Likewise.
1406 (tm_region_init_1): Likewise.
1407 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
1408 * tree-cfg.c (verify_gimple_call): Likewise.
1409 (move_stmt_r): Likewise.
1410 (stmt_can_terminate_bb_p): Likewise.
1411 * tree-eh.c (lower_eh_constructs_2): Likewise.
1412 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
1413 * tree-inline.c (remap_gimple_stmt): Likewise.
1414 (copy_bb): Likewise.
1415 (estimate_num_insns): Likewise.
1416 (fold_marked_statements): Likewise.
1417 * tree-sra.c (scan_function): Likewise.
1418 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
1419 (optimize_stack_restore): Likewise.
1420 (pass_fold_builtins::execute): Likewise.
1421 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
1422 (mark_all_reaching_defs_necessary_1): Likewise.
1423 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
1424 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
1425 (pass_forwprop::execute): Likewise.
1426 * tree-ssa-loop-im.c (stmt_cost): Likewise.
1427 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
1428 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
1429 * tree-ssa-strlen.c (get_string_length): Likewise.
1430 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
1431 (find_func_aliases_for_call): Likewise.
1432 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
1433 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
1434 * tree-tailcall.c (find_tail_calls): Likewise.
1435 * tree.c (need_assembler_name_p): Likewise.
1436 (free_lang_data_in_decl): Likewise.
1437 (get_call_combined_fn): Likewise.
1438 * ubsan.c (is_ubsan_builtin_p): Likewise.
1439 * varasm.c (incorporeal_function_p): Likewise.
1440 * tree.h (DECL_BUILT_IN): Remove and replace with
1441 fndecl_built_in_p.
1442 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
1443 (fndecl_built_in_p): New.
1444
1445 2018-08-27 Martin Liska <mliska@suse.cz>
1446
1447 PR tree-optimization/86847
1448 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
1449 Dump also subtree probability.
1450 (switch_decision_tree::do_jump_if_equal): New function.
1451 (switch_decision_tree::emit_case_nodes): Handle special
1452 situations in balanced tree that can be emitted much simpler.
1453 Fix calculation of probabilities that happen in tree expansion.
1454 * tree-switch-conversion.h (struct cluster): Add
1455 is_single_value_p.
1456 (struct simple_cluster): Likewise.
1457 (struct case_tree_node): Add new function has_child.
1458 (do_jump_if_equal): New.
1459
1460 2018-08-27 Martin Liska <mliska@suse.cz>
1461
1462 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
1463 Add new argument to bit_test_cluster constructor.
1464 (bit_test_cluster::emit): Set bits really number of values
1465 handlel by a test.
1466 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
1467 probability argument.
1468 * tree-switch-conversion.h (struct bit_test_cluster):
1469 Add m_handles_entire_switch.
1470
1471 2018-08-27 Martin Liska <mliska@suse.cz>
1472
1473 PR tree-optimization/86702
1474 * tree-switch-conversion.c (jump_table_cluster::emit):
1475 Make probabilities even for values in jump table
1476 according to number of cases handled.
1477 (switch_decision_tree::compute_cases_per_edge): Pass
1478 argument to reset_out_edges_aux function.
1479 (switch_decision_tree::analyze_switch_statement): Likewise.
1480 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
1481 Make it static.
1482
1483 2018-08-27 Martin Liska <mliska@suse.cz>
1484
1485 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
1486 cfun argument explicitly.
1487 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
1488 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
1489 function gimple_switch_default_bb.
1490 (convert_switch_statements):
1491 (expand_builtins):
1492 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
1493 * stmt.c (label_to_block_fn): Use label_to_block and pass
1494 cfun argument explicitly and use gimple_switch_label_bb.
1495 (expand_case): Likewise.
1496 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
1497 cfun argument explicitly. Likewise.
1498 (make_edges_bb): Likewise.
1499 (make_cond_expr_edges): Likewise.
1500 (get_cases_for_edge): Likewise.
1501 (make_gimple_switch_edges): Likewise.
1502 (label_to_block_fn): Likewise.
1503 (label_to_block): Likewise.
1504 (make_goto_expr_edges): Likewise.
1505 (make_gimple_asm_edges): Likewise.
1506 (main_block_label): Likewise.
1507 (group_case_labels_stmt): Likewise.
1508 (find_taken_edge_computed_goto): Likewise.
1509 (find_taken_edge_switch_expr): Likewise.
1510 (gimple_verify_flow_info): Likewise.
1511 (gimple_redirect_edge_and_branch): Likewise.
1512 (gimple_switch_label_bb): New function.
1513 (gimple_switch_default_bb): Likewise.
1514 (gimple_switch_edge): Likewise.
1515 (gimple_switch_default_edge): Likewise.
1516 * tree-cfg.h (label_to_block_fn): Remove and replace ...
1517 (label_to_block): ... with this.
1518 (gimple_switch_label_bb): New.
1519 (gimple_switch_default_bb): Likewise.
1520 (gimple_switch_edge): Likewise.
1521 (gimple_switch_default_edge): Likewise.
1522 * tree-cfgcleanup.c (convert_single_case_switch): Use
1523 new gimple functions and pass new argument to label_to_block.
1524 (cleanup_control_flow_bb):
1525 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
1526 cfun argument explicitly.
1527 (make_eh_edges): Likewise.
1528 (redirect_eh_dispatch_edge): Likewise.
1529 (lower_resx): Likewise.
1530 (lower_eh_dispatch): Likewise.
1531 (maybe_remove_unreachable_handlers): Likewise.
1532 (unsplit_eh): Likewise.
1533 (cleanup_empty_eh): Likewise.
1534 (verify_eh_edges): Likewise.
1535 (verify_eh_dispatch_edge): Likewise.
1536 * tree-ssa-dom.c (record_edge_info): Likewise.
1537 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
1538 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
1539 (thread_through_normal_block): Likewise.
1540 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
1541 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
1542 * tree-switch-conversion.c (switch_conversion::collect): Use new
1543 gimple functions.
1544 (switch_conversion::check_final_bb): Likewise.
1545 (switch_conversion::gather_default_values): Pass new argument
1546 to label_to_block.
1547 (switch_conversion::build_constructors): Likewise.
1548 (switch_decision_tree::compute_cases_per_edge): Use new
1549 gimple_switch_edge function.
1550 (switch_decision_tree::analyze_switch_statement): Pass new argument
1551 to label_to_block.
1552 (switch_decision_tree::try_switch_expansion): Use
1553 gimple_switch_default_edge.
1554 * tree-vrp.c (find_switch_asserts): Pass new argument
1555 to label_to_block.
1556 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
1557 (vr_values::simplify_switch_using_ranges): Likewise.
1558
1559 2018-08-27 Richard Biener <rguenther@suse.de>
1560
1561 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
1562 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
1563
1564 * tree-ssa-sccvn.h (struct vn_pval): New structure.
1565 (struct vn_nary_op_s): Add unwind_to member. Add
1566 predicated_values flag and put result into a union together
1567 with a linked list of vn_pval.
1568 (struct vn_ssa_aux): Add name member to make maintaining
1569 a map of SSA name to vn_ssa_aux possible. Remove no longer
1570 needed info, dfsnum, low, visited, on_sccstack, use_processed
1571 and range_info_anti_range_p members.
1572 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
1573 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
1574 New functions.
1575 (vn_valueize): New global.
1576 (vn_context_bb): Likewise.
1577 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
1578 VN_INFO_PTR_INFO): Remove.
1579 * tree-ssa-sccvn.c: ... (rewrite)
1580 (pass_fre::execute): For -O2+ initialize loops and run
1581 RPO VN in optimistic mode (iterating). For -O1 and -Og
1582 run RPO VN in non-optimistic mode.
1583 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
1584 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
1585 * doc/invoke.texi (sccvn-max-scc-size): Remove.
1586 (rpo-vn-max-loop-depth): Document.
1587 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
1588 when valuezing the VUSE signals we walked out of the region.
1589 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
1590 (phi_translate): Set VN context block to use for availability
1591 lookup.
1592 (compute_avail): Likewise.
1593 (pre_valueize): New function.
1594 (pass_pre::execute): Adjust to the RPO VN API.
1595
1596 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
1597 (propagate_constants_for_unrolling): Remove.
1598 (tree_unroll_loops_completely): Perform value-numbering
1599 on the unrolled bodies loop parent.
1600
1601 2018-08-27 Richard Biener <rguenther@suse.de>
1602
1603 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
1604 for partial antic compute.
1605
1606 2018-08-27 Jakub Jelinek <jakub@redhat.com>
1607
1608 PR rtl-optimization/87065
1609 * combine.c (simplify_if_then_else): Formatting fix.
1610 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
1611 check.
1612 (known_cond): Don't return const_true_rtx for vector modes. Use
1613 CONST0_RTX instead of const0_rtx. Formatting fixes.
1614
1615 2018-08-27 Martin Liska <mliska@suse.cz>
1616
1617 PR gcov-profile/87069
1618 * gcov.c (process_file): Record files already processed
1619 and warn about a file being processed multiple times.
1620
1621 2018-08-27 Martin Liska <mliska@suse.cz>
1622
1623 PR driver/83193
1624 * config/aarch64/aarch64.c (aarch64_override_options_internal):
1625 Set default values for x_aarch64_*_string strings.
1626 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
1627 prefix. For -mabi do not print '=ABI' in help and use
1628 <option_value> format for -msve-vector-bits and -moverride
1629 options.
1630
1631 2018-08-26 Jeff Law <law@redhat.com>
1632
1633 * config/mips/frame-header-opt.c: Include "backend.h" rather than
1634 "cfg.h"
1635
1636 2018-08-26 Marek Polacek <polacek@redhat.com>
1637
1638 PR c++/87029, Implement -Wredundant-move.
1639 * doc/invoke.texi: Document -Wredundant-move.
1640
1641 2018-08-25 Martin Sebor <msebor@redhat.com>
1642
1643 PR tree-optimization/87059
1644 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
1645 to the same type as the other.
1646 * fold-const.c (fold_binary_loc): Assert expectation.
1647
1648 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
1649
1650 * config/darwin.c (machopic_legitimize_pic_address): Clean up
1651 extraneous parentheses, dead code section and formatting.
1652
1653 2018-08-24 David Malcolm <dmalcolm@redhat.com>
1654
1655 PR c++/87091
1656 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
1657 wide enough for jumps in the line-numbering to be visible.
1658 (layout::print_gap_in_line_numbering): New member function.
1659 (layout::calculate_line_spans): When using line numbering, merge
1660 line spans that are only 1 line apart.
1661 (diagnostic_show_locus): When printing line numbers, show gaps in
1662 line numbering directly, rather than printing headers.
1663 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
1664 line-numbering with multiple line spans.
1665 (selftest::test_fixit_insert_containing_newline_2): Add test of
1666 line-numbering, in which the spans are close enough to be merged.
1667
1668 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
1669
1670 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
1671 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
1672 * tree-vrp.c (range_is_nonnull): Remove.
1673 (range_includes_zero_p): Accept value_range instead of min/max.
1674 (extract_range_from_binary_expr_1): Do not early bail on
1675 POINTER_PLUS_EXPR.
1676 Use range_includes_zero_p instead of range_is_nonnull.
1677 (extract_range_from_unary_expr): Use range_includes_zero_p instead
1678 of range_is_nonnull.
1679 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
1680 special case VR_ANTI_RANGE.
1681 (vrp_finalize): Same.
1682 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
1683 instead of min/max.
1684 (range_is_nonnull): Remove.
1685 * vr-values.c (vrp_stmt_computes_nonzero): Use
1686 range_includes_zero_p instead of range_is_nonnull.
1687 (extract_range_basic): Pass value_range to range_includes_zero_p
1688 instead of range_is_nonnull.
1689
1690 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
1691
1692 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
1693 * emit-rtl.h (rtl_data): Remove return_bnd.
1694 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
1695 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
1696 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
1697 (POINTER_BOUNDS_MODE): Remove definition.
1698 (make_pointer_bounds_mode): Remove.
1699 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
1700 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
1701 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
1702 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
1703 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
1704 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
1705 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
1706
1707 * config/i386/i386-modes.def (BND32, BND64): Remove.
1708 * config/i386/i386.c (dbx_register_map): Remove bound registers.
1709 (dbx64_register_map): Ditto.
1710 (svr4_dbx_register_map): Ditto.
1711 (indirect_thunk_bnd_needed): Remove.
1712 (indirect_thunks_bnd_used): Ditto.
1713 (indirect_return_bnd_needed): Ditto.
1714 (indirect_return_via_cx_bnd): Ditto.
1715 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
1716 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
1717 (output_indirect_thunk): Ditto. Remove need_prefix argument.
1718 (output_indirect_thunk_function): Remove handling of
1719 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
1720 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
1721 (ix86_save_reg): Remove handling of crtl->return_bnd.
1722 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
1723 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
1724 and UNSPEC_BNDLX_ADDR.
1725 (ix86_output_indirect_branch_via_reg): Remove handling of
1726 indirect_thunk_prefix_bnd.
1727 (ix86_output_indirect_branch_via_push): Ditto.
1728 (ix86_output_function_return): Ditto.
1729 (ix86_output_indirect_function_return): Ditto.
1730 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
1731 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
1732 (CALL_USED_REGISTERS): Ditto.
1733 (REG_ALLOC_ORDER): Update for removal of bound registers.
1734 (HI_REGISTER_NAMES): Ditto.
1735 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
1736 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
1737 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
1738 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
1739 (FIRST_PSEUDO_REG): Update.
1740 (BND): Remove mode iterator.
1741 * config/i386/predicates.md (bnd_mem_operator): Remove.
1742
1743 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
1744
1745 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
1746 vectors.
1747
1748 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
1749
1750 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
1751 the case in which the permute needs only a single element and
1752 repeats for every vector of the result. Extend that case to
1753 handle variable-length vectors.
1754 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
1755
1756 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
1757
1758 PR debug/79342
1759 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
1760 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
1761
1762 2018-08-24 Richard Biener <rguenther@suse.de>
1763
1764 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
1765 bb_flags_allocated members.
1766 (auto_flag): New RAII class for allocating flags.
1767 (auto_edge_flag): New RAII class for allocating edge flags.
1768 (auto_bb_flag): New RAII class for allocating bb flags.
1769 * cfgloop.c (verify_loop_structure): Allocate temporary edge
1770 flag dynamically.
1771 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
1772 in favor of temporarily allocated BB flag.
1773 * hsa-brig.c: Re-order includes.
1774 * hsa-dump.c: Likewise.
1775 * hsa-regalloc.c: Likewise.
1776 * print-rtl.c: Likewise.
1777 * profile-count.c: Likewise.
1778
1779 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
1780
1781 PR target/86989
1782 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
1783 the TOC register.
1784
1785 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
1786
1787 PR 87073/bootstrap
1788 * wide-int-range.cc (wide_int_range_div): Do not ignore result
1789 from wide_int_range_multiplicative_op.
1790
1791 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
1792
1793 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
1794 "permutaion".
1795
1796 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
1797
1798 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
1799 to 'expanded'.
1800
1801 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
1802
1803 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
1804 full GENERIC expression used for replacement.
1805
1806 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
1807
1808 * tree-vrp.c (abs_extent_range): Remove.
1809 (extract_range_into_wide_ints): Pass wide ints by reference.
1810 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
1811 Pass wide ints by reference in all calls to
1812 extract_range_into_wide_ints.
1813 * wide-int-range.cc (wide_int_range_div): New.
1814 * wide-int-range.h (wide_int_range_div): New.
1815 (wide_int_range_includes_zero_p): New.
1816 (wide_int_range_zero_p): New.
1817
1818 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
1819
1820 * config/aarch64/aarch64.md (arches): New enum.
1821 (arch): New enum attr.
1822 (arch_enabled): New attr.
1823 (enabled): Now uses arch_enabled only.
1824 (simd, sve, fp16): Removed attribute.
1825 (fp): Attr now defined in terms of 'arch'.
1826 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
1827 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
1828 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
1829 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
1830 attributes into 'arch'.
1831 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
1832 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
1833 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
1834 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
1835 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
1836 'simd' attribute into 'arch'.
1837 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
1838 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
1839 Convert use of 'fp' attribute to 'arch'.
1840 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
1841 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
1842 into 'arch'.
1843 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
1844 (different modes) Merge 'fp' and 'simd' into 'arch'.
1845 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
1846 'simd' into 'arch'.
1847
1848 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
1849
1850 PR rtl-optimization/87026
1851 * expmed.c (canonicalize_comparison): If we can no longer create
1852 pseudoregisters, don't.
1853
1854 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
1855
1856 PR target/86951
1857 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
1858 prototype.
1859 * config/arm/arm.c (speculation_barrier_libfunc): New static
1860 variable.
1861 (arm_init_libfuncs): Initialize it.
1862 (arm_emit_speculation_barrier): New function.
1863 * config/arm/arm.md (speculation_barrier): Call
1864 arm_emit_speculation_barrier for architectures that do not have
1865 DSB or ISB.
1866 (speculation_barrier_insn): Only match on Armv7 or later.
1867
1868 2018-08-23 Richard Biener <rguenther@suse.de>
1869
1870 PR middle-end/87024
1871 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
1872 calls.
1873
1874 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
1875
1876 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
1877 of single-vector TBLs.
1878 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
1879 one input is given.
1880
1881 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
1882
1883 PR target/85910
1884 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
1885 aarch64_evpc_tbl guard.
1886
1887 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
1888
1889 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
1890 behaviour.
1891
1892 2018-08-22 Martin Sebor <msebor@redhat.com>
1893
1894 PR middle-end/87052
1895 * tree-pretty-print.c (pretty_print_string): Add argument.
1896 (dump_generic_node): Call to pretty_print_string with string size.
1897
1898 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
1899
1900 PR rtl-optimization/86771
1901 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
1902 of two SETs into those two SETs, one to be placed at i2, if that SETs
1903 destination is modified between i2 and i3.
1904
1905 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1906
1907 PR tree-optimization/86725
1908 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
1909 function.
1910 (vect_analyze_scalar_cycles_1): Check it.
1911
1912 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1913
1914 PR tree-optimization/86725
1915 * tree-vect-loop.c (vect_is_simple_reduction): When treating
1916 an outer loop phi as a double reduction, make sure that the
1917 single user of the phi result is an inner loop phi.
1918
1919 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1920
1921 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
1922 grouped stores with gaps to a strided group.
1923
1924 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
1925
1926 * tree-vect-stmts.c (get_group_load_store_type)
1927 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
1928 first statement in a group.
1929
1930 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1931
1932 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
1933 the sequence used in gcc/gcc.c.
1934
1935 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1936
1937 PR other/704
1938 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
1939 building it.
1940
1941 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1942
1943 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
1944 Darwin10-specific unwinder-shim.
1945 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
1946 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
1947 New to cater for Darwin10 Rosetta.
1948
1949 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1950
1951 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
1952 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
1953
1954 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
1955
1956 PR bootstrap/81033
1957 PR target/81733
1958 PR target/52795
1959 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
1960 (dwarf2out_switch_text_section): Generate a local label for the
1961 second function sub-section and apply it as the second FDE start
1962 label.
1963 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
1964 second sub-section start.
1965
1966 2018-08-22 Richard Biener <rguenther@suse.de>
1967
1968 PR tree-optimization/86988
1969 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
1970
1971 2018-08-22 Richard Biener <rguenther@suse.de>
1972
1973 PR tree-optimization/86945
1974 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
1975
1976 2018-08-22 Alexandre Oliva <oliva@adacore.com>
1977
1978 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
1979 a comment about how uses of r2 for .sdata2 come about.
1980
1981 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
1982
1983 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
1984
1985 2018-08-21 Marek Polacek <polacek@redhat.com>
1986
1987 PR c++/86981, Implement -Wpessimizing-move.
1988 * doc/invoke.texi: Document -Wpessimizing-move.
1989
1990 2018-08-21 Jan Hubicka <jh@suse.cz>
1991
1992 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
1993 * tree.h (is_redundant_typedef): Remove.
1994 * dwarf2out.c (is_redundant_typedef): Turn into static function.
1995
1996 2018-08-21 Jan Hubicka <jh@suse.cz>
1997
1998 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
1999 when possible.
2000
2001 2018-08-21 Tamar Christina <tamar.christina@arm.com>
2002
2003 * expmed.c (extract_low_bits): Reject invalid subregs early.
2004
2005 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
2006
2007 PR middle-end/86121
2008 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
2009 behaviour.
2010
2011 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2012
2013 * config/vxworks.h: Guard vxworks_asm_out_constructor and
2014 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
2015 * config/vxworks.c: Likewise.
2016
2017 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2018
2019 * config/vxworks.c: Set targetm.have_ctors_dtors
2020 if HAVE_INITFINI_ARRAY_SUPPORT.
2021 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
2022 if HAVE_INITFINI_ARRAY_SUPPORT.
2023
2024 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
2025
2026 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
2027 default search path for VxWorks < 7.
2028
2029 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
2030
2031 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
2032 (get_format_string): Refer to c_getstr.
2033
2034 2018-08-21 Tom de Vries <tdevries@suse.de>
2035
2036 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
2037 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
2038 (debuginfo_early_stop): Declare.
2039 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
2040 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
2041 (debuginfo_early_stop): New function.
2042 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
2043 and debuginfo_early_stop.
2044 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
2045 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
2046 (general_init): Call debuginfo_early_init.
2047 (finalize): Call debuginfo_fini.
2048 (do_compile): Call debuginfo_init.
2049 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
2050 -fdump-early-debug.
2051 (@item -fdump-debug, @item -fdump-earlydebug): Add.
2052
2053 2018-08-21 Tom de Vries <tdevries@suse.de>
2054
2055 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
2056 flag_dump_noaddr and flag_dump_unnumbered.
2057
2058 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
2059
2060 * wide-int-range.cc (wide_int_range_abs): New.
2061 (wide_int_range_order_set): Rename from wide_int_range_min_max.
2062 * wide-int-range.h (wide_int_range_abs): New.
2063 (wide_int_range_min_max): New.
2064 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
2065 case to call wide_int_range_abs.
2066 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
2067 (extract_range_from_abs_expr): Delete.
2068
2069 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
2070
2071 PR target/87033
2072 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
2073 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
2074 for indexed loads.
2075
2076 2018-08-20 Nathan Sidwell <nathan@acm.org>
2077 Jeff Law <law@redhat.com>
2078
2079 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
2080 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
2081
2082 2018-08-20 David Malcolm <dmalcolm@redhat.com>
2083
2084 PR other/84889
2085 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
2086 (decl_attributes): Likewise.
2087 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
2088 instance.
2089 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
2090 * diagnostic-core.h (class auto_diagnostic_group): New class.
2091 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
2092 (diagnostic_report_diagnostic): Handle the first diagnostics within
2093 a group.
2094 (emit_diagnostic): Add auto_diagnostic_group instance.
2095 (inform): Likewise.
2096 (inform_n): Likewise.
2097 (warning): Likewise.
2098 (warning_at): Likewise.
2099 (warning_n): Likewise.
2100 (pedwarn): Likewise.
2101 (permerror): Likewise.
2102 (error): Likewise.
2103 (error_n): Likewise.
2104 (error_at): Likewise.
2105 (sorry): Likewise.
2106 (fatal_error): Likewise.
2107 (internal_error): Likewise.
2108 (internal_error_no_backtrace): Likewise.
2109 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
2110 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
2111 * diagnostic.h (struct diagnostic_context): Add fields
2112 "diagnostic_group_nesting_depth",
2113 "diagnostic_group_emission_count", "begin_group_cb",
2114 "end_group_cb".
2115 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
2116 Add auto_diagnostic_group instance(s).
2117 (find_explicit_erroneous_behavior): Likewise.
2118 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
2119 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
2120 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
2121 (gimplify_va_arg_expr): Likewise.
2122 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
2123 (HSA_SORRY_AT): Likewise.
2124 * ipa-devirt.c (compare_virtual_tables): Likewise.
2125 (warn_odr): Likewise.
2126 * multiple_target.c (expand_target_clones): Likewise.
2127 * opts-common.c (cmdline_handle_error): Likewise.
2128 * reginfo.c (globalize_reg): Likewise.
2129 * substring-locations.c (format_warning_n_va): Likewise.
2130 * tree-inline.c (expand_call_inline): Likewise.
2131 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
2132 * tree-ssa-loop-niter.c
2133 (do_warn_aggressive_loop_optimizations): Likewise.
2134 * tree-ssa-uninit.c (warn_uninit): Likewise.
2135 * tree.c (warn_deprecated_use): Likewise.
2136
2137 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
2138
2139 PR target/87014
2140 * config/i386/i386.md (eh_return): Always update EH return
2141 address in word_mode.
2142
2143 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
2144
2145 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
2146 TARGET_SPLIT_COMPLEX_ARG is defined.
2147
2148 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
2149
2150 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
2151
2152 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
2153
2154 PR target/86984
2155 * expr.c (expand_assignment): Assert that bitpos is positive.
2156 (store_field): Likewise
2157 (expand_expr_real_1): Make sure that bitpos is positive.
2158 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
2159 integer overflow.
2160
2161 2018-08-20 Nathan Sidwell <nathan@acm.org>
2162
2163 * Makefile.in (CPP_ID_DATA_H): Delete.
2164 (CPP_INTERNAL_H): Don't add it.
2165 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
2166 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
2167
2168 2018-08-20 Richard Biener <rguenther@suse.de>
2169
2170 PR tree-optimization/78655
2171 * tree-vrp.c (extract_range_from_binary_expr_1): Make
2172 pointer + offset nonnull if either operand is nonnull work.
2173
2174 2018-08-20 Tom de Vries <tdevries@suse.de>
2175
2176 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
2177 unless the referenced die describes the added property using
2178 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
2179 Otherwise, add a DW_AT_location to the referenced die.
2180
2181 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
2182
2183 PR target/86994
2184 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
2185 register_operand when calling ix86_set_reg_reg_cost.
2186 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
2187 Set *total to 0 for operands that satisfy x86_64_immediate_operand
2188 predicate and to 1 otherwise.
2189
2190 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
2191
2192 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
2193 emit a diagnostic that it is not supported and reset the option.
2194 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
2195 supported and consume the option. (ASM_FINAL_SPEC): New.
2196
2197 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2198
2199 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
2200 a sentence.
2201
2202 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
2203
2204 C-SKY port: Documentation
2205
2206 * doc/extend.texi (C-SKY Function Attributes): New section.
2207 * doc/invoke.texi (Option Summary): Add C-SKY options.
2208 (C-SKY Options): New section.
2209 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
2210
2211 2018-08-17 Jojo <jijie_rong@c-sky.com>
2212 Huibin Wang <huibin_wang@c-sky.com>
2213 Sandra Loosemore <sandra@codesourcery.com>
2214 Chung-Lin Tang <cltang@codesourcery.com>
2215
2216 C-SKY port: Backend implementation
2217
2218 * config/csky/*: New.
2219 * common/config/csky/*: New.
2220
2221 2018-08-17 Jojo <jijie_rong@c-sky.com>
2222 Huibin Wang <huibin_wang@c-sky.com>
2223 Sandra Loosemore <sandra@codesourcery.com>
2224 Chung-Lin Tang <cltang@codesourcery.com>
2225 Andrew Jenner <andrew@codesourcery.com>
2226
2227 C-SKY port: Configury
2228
2229 * config.gcc (csky-*-*): New.
2230 * configure.ac: Add csky to targets for dwarf2 debug_line support.
2231 * configure: Regenerated.
2232
2233 2018-08-17 David Malcolm <dmalcolm@redhat.com>
2234
2235 * dump-context.h: Include "dumpfile.h".
2236 (dump_context::dump_printf_va): Convert final param from va_list
2237 to va_list *. Convert from ATTRIBUTE_PRINTF to
2238 ATTRIBUTE_GCC_DUMP_PRINTF.
2239 (dump_context::dump_printf_loc_va): Likewise.
2240 * dumpfile.c: Include "stringpool.h".
2241 (make_item_for_dump_printf_va): Delete.
2242 (make_item_for_dump_printf): Delete.
2243 (class dump_pretty_printer): New class.
2244 (dump_pretty_printer::dump_pretty_printer): New ctor.
2245 (dump_pretty_printer::emit_items): New member function.
2246 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
2247 function.
2248 (dump_pretty_printer::emit_item): New member function.
2249 (dump_pretty_printer::stash_item): New member function.
2250 (dump_pretty_printer::format_decoder_cb): New member function.
2251 (dump_pretty_printer::decode_format): New member function.
2252 (dump_context::dump_printf_va): Reimplement in terms of
2253 dump_pretty_printer.
2254 (dump_context::dump_printf_loc_va): Convert final param from va_list
2255 to va_list *.
2256 (dump_context::begin_scope): Reimplement call to
2257 make_item_for_dump_printf.
2258 (dump_printf): Update for change to dump_printf_va.
2259 (dump_printf_loc): Likewise.
2260 (selftest::test_capture_of_dump_calls): Convert "stmt" from
2261 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
2262 with %T, %E, and %G.
2263 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
2264 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
2265 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
2266 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
2267 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
2268 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
2269 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
2270 within a dump_printf_loc call to "%wu".
2271 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
2272 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
2273 missing space after "=".
2274 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
2275 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
2276 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
2277 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
2278 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
2279 duplicate "vectorized" from message.
2280
2281 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
2282
2283 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
2284 polyNxK_t element's TYPE_STRING_FLAG.
2285
2286 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
2287
2288 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
2289 (they were unnamed before). Fix comments.
2290
2291 2018-08-17 Nathan Sidwell <nathan@acm.org>
2292
2293 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
2294
2295 2018-08-17 Richard Biener <rguenther@suse.de>
2296
2297 PR tree-optimization/86841
2298 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
2299
2300 2018-08-17 Martin Liska <mliska@suse.cz>
2301
2302 * common.opt: Remove Warn, Init and Report for options with
2303 Ignore/Deprecated flag. Warning is done automatically for
2304 Deprecated flags.
2305 * config/i386/i386.opt: Likewise.
2306 * config/ia64/ia64.opt: Likewise.
2307 * config/rs6000/rs6000.opt: Likewise.
2308 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
2309 Remove usage of flag_check_pointer_bounds.
2310 * lto-wrapper.c (merge_and_complain): Do not handle
2311 OPT_fcheck_pointer_bounds.
2312 (append_compiler_options): Likewise.
2313 * opt-functions.awk: Do not handle Deprecated.
2314 * optc-gen.awk: Check that Var, Report and Init are not
2315 used for an option with Ignore/Deprecated flag.
2316 * opts-common.c (decode_cmdline_option): Do not report
2317 CL_ERR_DEPRECATED.
2318 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
2319 options.
2320 * opts.h (struct cl_option): Remove cl_deprecated flag.
2321 (CL_ERR_DEPRECATED): Remove error enum value.
2322
2323 2018-08-17 Richard Biener <rguenther@suse.de>
2324
2325 PR middle-end/86505
2326 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
2327 across a va-arg-pack using call adjust its return value accordingly.
2328
2329 2018-08-16 Martin Sebor <msebor@redhat.com>
2330
2331 PR tree-optimization/86853
2332 * gimple-ssa-sprintf.c (struct format_result): Rename member.
2333 (struct fmtresult): Add member and initialize it in ctors.
2334 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
2335 (format_string): Handle %S the same as %ls. Set MAYFAIL.
2336 (format_directive): Set POSUNDER4K when MAYFAIL is set.
2337 (parse_directive): Handle %C same as %c.
2338 (sprintf_dom_walker::compute_format_length): Adjust.
2339 (is_call_safe): Adjust.
2340
2341 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
2342
2343 * builtins.c (c_strlen): Add new parameter eltsize. Use it
2344 for determining how to count the elements.
2345 * builtins.h (c_strlen): Adjust prototype.
2346 * expr.c (string_constant): Add new parameter mem_size.
2347 Set *mem_size appropriately.
2348 * expr.h (string_constant): Adjust protoype.
2349 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
2350 * gimple-fold.h (get_range_strlen): Adjust prototype.
2351 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
2352 (format_string): Call get_string_length with eltsize.
2353
2354 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2355
2356 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
2357 to emit the span, rather than setting it as the prefix.
2358
2359 2018-08-16 David Malcolm <dmalcolm@redhat.com>
2360
2361 * diagnostic-show-locus.c (layout::start_annotation_line): Add
2362 "margin_char" parameter, defaulting to space. Use it in place
2363 of pp_space for the initial part of the margin.
2364 (layout::print_leading_fixits): Use '+' when filling the margin
2365 of line-insertion fix-it hints.
2366
2367 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2368
2369 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
2370 Delete.
2371
2372 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2373
2374 * config/rs6000/altivec.md: Don't set length attribute to the default
2375 value.
2376 * config/rs6000/darwin.md: Ditto.
2377 * config/rs6000/dfp.md: Ditto.
2378 * config/rs6000/htm.md: Ditto.
2379 * config/rs6000/rs6000.md: Ditto.
2380 * config/rs6000/sync.md: Ditto.
2381 * config/rs6000/vsx.md: Ditto.
2382
2383 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2384
2385 * config/rs6000/altivec.md: Don't set length attribute to the default
2386 value, for branch instructions.
2387 * config/rs6000/darwin.md: Ditto.
2388 * config/rs6000/rs6000.md: Ditto.
2389
2390 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
2391
2392 * config/rs6000/rs6000.md (length): Always define as const_int 4.
2393 (unnamed conditional branch define_insn): Set length to 4 or 8
2394 depending on offset.
2395 (<bd>_<mode>): Similar, for alternative 0.
2396 (<bd>tf_<mode>): Ditto.
2397
2398 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2399
2400 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
2401
2402 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
2403
2404 * doc/rtl.texi: Replace old RTX class names with new names.
2405
2406
2407 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
2408
2409 * expmed.h (canonicalize_comparison): New declaration.
2410 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
2411 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
2412 * optabs.c (prepare_cmp_insn): Likewise.
2413 * rtl.h (unsigned_condition_p): New function which checks if a
2414 comparison operator is unsigned.
2415
2416 2018-08-16 Nathan Sidwell <nathan@acm.org>
2417
2418 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
2419 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
2420
2421 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2422
2423 PR target/84711
2424 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
2425 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
2426 (mov<mov>): ..this and enable unconditionally.
2427
2428 2018-08-16 Tamar Christina <tamar.christina@arm.com>
2429
2430 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
2431
2432 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
2433
2434 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
2435 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
2436 "Common" with "Target".
2437
2438 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2439
2440 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
2441 * doc/invoke.texi (mmitigate-rop): Remove.
2442 * config/i386/i386.c: Do not include "regrename.h".
2443 (ix86_rop_should_change_byte_p, reg_encoded_number)
2444 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
2445 Remove.
2446 (ix86_reorg): Remove call to ix86_mitigate_rop.
2447 * config/i386/i386.md (attr "modrm_class"): Remove.
2448 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
2449 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
2450 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
2451
2452 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
2453
2454 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
2455 allow folding of mergeh() and mergel() for the float and double types.
2456 (fold_mergehl_helper): Rework to handle building a permute tree
2457 for float vectors.
2458
2459 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
2460
2461 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
2462 for TARGET_SSE.
2463
2464 2018-08-15 David Malcolm <dmalcolm@redhat.com>
2465
2466 * common.opt (fdiagnostics-show-labels): New option.
2467 * diagnostic-show-locus.c (class layout_range): Add field
2468 "m_label".
2469 (class layout): Add field "m_show_labels_p".
2470 (layout_range::layout_range): Add param "label" and use it to
2471 initialize m_label.
2472 (make_range): Pass in NULL for new "label" param of layout_range's
2473 ctor.
2474 (layout::layout): Initialize m_show_labels_p.
2475 (layout::maybe_add_location_range): Pass in loc_range->m_label
2476 when constructing layout_range instances.
2477 (struct line_label): New struct.
2478 (layout::print_any_labels): New member function.
2479 (layout::print_line): Call it if label-printing is enabled.
2480 (selftest::test_one_liner_labels): New test.
2481 (selftest::test_diagnostic_show_locus_one_liner): Call it.
2482 * diagnostic.c (diagnostic_initialize): Initialize
2483 context->show_labels_p.
2484 * diagnostic.h (struct diagnostic_context): Add field
2485 "show_labels_p".
2486 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
2487 -fno-diagnostics-show-labels.
2488 * dwarf2out.c (gen_producer_string): Add
2489 OPT_fdiagnostics_show_labels to the ignored options.
2490 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
2491 param.
2492 (gcc_rich_location::maybe_add_expr): Likewise.
2493 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
2494 label" param, defaulting to NULL.
2495 (gcc_rich_location::add_expr): Add "label" param.
2496 (gcc_rich_location::maybe_add_expr): Likewise.
2497 (class text_range_label): New class.
2498 (class range_label_for_type_mismatch): New class.
2499 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
2500 of format_warning_va.
2501 (fmtwarn_n): Likewise for new params of format_warning_n_va.
2502 * lto-wrapper.c (merge_and_complain): Add
2503 OPT_fdiagnostics_show_labels to the "pick one setting" options.
2504 (append_compiler_options): Likewise to the dropped options.
2505 (append_diag_options): Likewise to the passed-on options.
2506 * opts.c (common_handle_option): Handle the new option.
2507 * selftest-diagnostic.c
2508 (test_diagnostic_context::test_diagnostic_context): Enable
2509 show_labels_p.
2510 * substring-locations.c: Include "gcc-rich-location.h".
2511 (format_warning_n_va): Add "fmt_label" and "param_label" params
2512 and use them as appropriate.
2513 (format_warning_va): Add "fmt_label" and "param_label" params,
2514 passing them on to format_warning_n_va.
2515 (format_warning_at_substring): Likewise.
2516 (format_warning_at_substring_n): Likewise.
2517 * substring-locations.h (format_warning_va): Add "fmt_label" and
2518 "param_label" params.
2519 (format_warning_n_va): Likewise.
2520 (format_warning_at_substring): Likewise.
2521 (format_warning_at_substring_n): Likewise.
2522 * toplev.c (general_init): Initialize global_dc->show_labels_p.
2523
2524 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
2525
2526 PR testsuite/86519
2527 * builtins.c (expand_builtin_memcmp): Do not expand the call
2528 when overflow is detected.
2529
2530 2018-08-15 Martin Sebor <msebor@redhat.com>
2531
2532 PR tree-optimization/71625
2533 * config/aarch64/aarch64-builtins.c
2534 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
2535
2536 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
2537
2538 * config/s390/s390.c (s390_reorg): Remove loop.
2539
2540 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
2541
2542 * config/darwin.c
2543 (darwin_function_switched_text_sections): Delete.
2544 * gcc/config/darwin.h
2545 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
2546
2547 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
2548
2549 PR target/81685
2550 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
2551 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
2552 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
2553
2554 2018-08-15 Martin Liska <mliska@suse.cz>
2555
2556 PR tree-optimization/86925
2557 * predict.c (expr_expected_value_1): When taking
2558 later predictor, assign also probability.
2559 Use fold_build2_initializer_loc in order to fold
2560 the expression in -frounding-math.
2561
2562 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
2563
2564 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
2565 patterns.
2566 (expand_vec_perm_1): Try the new method.
2567
2568 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
2569
2570 PR target/86547
2571 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
2572 Check whether lra_live_max_point is 0 before dividing.
2573
2574 2018-08-14 Martin Sebor <msebor@redhat.com>
2575
2576 PR tree-optimization/86650
2577 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
2578 (vrp_prop::check_mem_ref): Same.
2579
2580 2018-08-13 Liu Hao <lh_mouse@126.com>
2581
2582 * pretty-print.c (eat_esc_sequence): Swap the foreground and
2583 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
2584 and clear it thereafter, as it only works for DBCS.
2585
2586 2018-08-13 Liu Hao <lh_mouse@126.com>
2587
2588 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
2589 handle returned by _get_osf_handle().
2590
2591 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2592
2593 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
2594 for folding vec_perm.
2595
2596 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2597
2598 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
2599 Add support for gimple-folding of vec_pack() and vec_unpack()
2600 intrinsics.
2601
2602 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
2603
2604 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
2605 vec_xst variants to the list.
2606 (rs6000_gimple_fold_builtin): Add support for folding unaligned
2607 vector loads and stores.
2608
2609 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
2610
2611 * config.gcc (rs6000-ibm-aix4.x): Delete.
2612 (rs6000-ibm-aix5.1): Delete.
2613 (rs6000-ibm-aix5.2): Delete.
2614 (rs6000-ibm-aix5.3): Delete.
2615 * config/rs6000/aix43.h: Delete.
2616 * config/rs6000/aix51.h: Delete.
2617 * config/rs6000/aix52.h: Delete.
2618 * config/rs6000/t-aix43: Delete.
2619
2620 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
2621
2622 * config/s390/s390.c (s390_decompose_constant_pool_ref):
2623 New function.
2624 (s390_decompose_address): Factor out constant pool ref
2625 decomposition.
2626
2627 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2628
2629 * config/nds32/nds32-predicates.c
2630 (nds32_can_use_bclr_p): Change return type as bool.
2631 (nds32_can_use_bset_p): Ditto.
2632 (nds32_can_use_btgl_p): Ditto.
2633 (nds32_can_use_bitci_p): Ditto.
2634 * config/nds32/nds32-protos.h
2635 (nds32_can_use_bclr_p): Change declaration.
2636 (nds32_can_use_bset_p): Ditto.
2637 (nds32_can_use_btgl_p): Ditto.
2638 (nds32_can_use_bitci_p): Ditto.
2639
2640 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2641
2642 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
2643 Support -msched-prolog-epilog option.
2644 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
2645
2646 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2647
2648 * common/config/nds32/nds32-common.c
2649 (nds32_option_optimization_table): Enalbe -malways-align.
2650
2651 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
2652
2653 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
2654 extra_headers.
2655 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
2656 OPT_misr_secure_ case.
2657 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
2658 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
2659 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
2660 secure attribute.
2661 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
2662 (nds32_isr_info): New field security_level.
2663 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
2664 * config/nds32/nds32.md (return_internal): Consider critical attribute.
2665 * config/nds32/nds32.opt (misr-secure): New option.
2666 * config/nds32/nds32_init.inc: New file.
2667 * config/nds32/nds32_isr.h: New file.
2668
2669 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
2670
2671 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
2672 Update comment for atomic instructions.
2673 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
2674 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
2675 Remove.
2676 (atomic_loaddi): Revise fence expansion to only emit fence prior to
2677 load for __ATOMIC_SEQ_CST model.
2678 (atomic_loaddi_1): Remove float register target.
2679 (atomic_storedi): Handle CONST_INT values.
2680 (atomic_storedi_1): Remove float register source. Add special case
2681 for zero value.
2682 (memory_barrier): New expander and insn.
2683
2684 2018-08-11 Jakub Jelinek <jakub@redhat.com>
2685
2686 PR tree-optimization/86835
2687 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
2688 new_stmt after def_gsi, make sure to insert new_square_stmt after
2689 that stmt, not 2 stmts before it.
2690
2691 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
2692
2693 PR target/82418
2694 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
2695 instead of SWI48.
2696
2697 2018-08-10 Martin Liska <mliska@suse.cz>
2698
2699 PR target/83610
2700 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
2701 function type.
2702 * builtins.c (expand_builtin_expect_with_probability):
2703 New function.
2704 (expand_builtin_expect_with_probability): New function.
2705 (build_builtin_expect_predicate): Add new argumnet probability
2706 for BUILT_IN_EXPECT_WITH_PROBABILITY.
2707 (fold_builtin_expect):
2708 (fold_builtin_2):
2709 (fold_builtin_3):
2710 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
2711 * builtins.h (fold_builtin_expect): Set new argument.
2712 * doc/extend.texi: Document __builtin_expect_with_probability.
2713 * doc/invoke.texi: Likewise.
2714 * gimple-fold.c (gimple_fold_call): Pass new argument.
2715 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
2716 also BUILT_IN_EXPECT_WITH_PROBABILITY.
2717 * predict.c (get_predictor_value): New function.
2718 (expr_expected_value): Add new argument probability. Assume
2719 that predictor and probability are always non-null.
2720 (expr_expected_value_1): Likewise. For __builtin_expect and
2721 __builtin_expect_with_probability set probability. Handle
2722 combination in binary expressions.
2723 (tree_predict_by_opcode): Simplify code by simply calling
2724 get_predictor_value.
2725 (pass_strip_predict_hints::execute): Add handling of
2726 BUILT_IN_EXPECT_WITH_PROBABILITY.
2727 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
2728 new predictor.
2729 * tree.h (DECL_BUILT_IN_P): New function.
2730
2731 2018-08-10 Martin Liska <mliska@suse.cz>
2732
2733 PR tree-optimization/85799
2734 * passes.def: Add argument for pass_strip_predict_hints.
2735 * predict.c (class pass_strip_predict_hints): Add new argument
2736 early_p.
2737 (strip_predictor_early): New function.
2738 (pass_strip_predict_hints::execute): Call the function to
2739 strip predictors.
2740 (strip_predict_hints): New function.
2741 * predict.def: Fix comment.
2742
2743 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
2744
2745 * Makefile.in: Clarify which tm.texi to copy over to assert the
2746 right to grant a GFDL license for all.
2747
2748 2018-08-09 Jeff Law <law@redhat.com>
2749
2750 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
2751 unused variable.
2752
2753 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
2754
2755 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
2756 prototype.
2757
2758 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2759
2760 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
2761 reductions for variable-length vectors.
2762
2763 2018-08-09 David Malcolm <dmalcolm@redhat.com>
2764
2765 PR other/84889
2766 * common.opt (fdiagnostics-show-line-numbers): New option.
2767 * diagnostic-show-locus.c (class layout): Add fields
2768 "m_show_line_numbers_p" and "m_linenum_width";
2769 (num_digits): New function.
2770 (test_num_digits): New function.
2771 (layout::layout): Initialize new fields. Update m_x_offset
2772 logic to handle any left margin.
2773 (layout::print_source_line): Print line number when requested.
2774 (layout::start_annotation_line): New member function.
2775 (layout::print_annotation_line): Call it.
2776 (layout::print_leading_fixits): Likewise.
2777 (layout::print_trailing_fixits): Likewise. Update calls to
2778 move_to_column for new parameter.
2779 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
2780 it to potentially call start_annotation_line.
2781 (layout::show_ruler): Call start_annotation_line.
2782 (selftest::test_line_numbers_multiline_range): New selftest.
2783 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
2784 and selftest::test_line_numbers_multiline_range.
2785 * diagnostic.c (diagnostic_initialize): Initialize
2786 show_line_numbers_p.
2787 * diagnostic.h (struct diagnostic_context): Add field
2788 "show_line_numbers_p".
2789 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
2790 -fno-diagnostics-show-line-numbers.
2791 * dwarf2out.c (gen_producer_string): Add
2792 OPT_fdiagnostics_show_line_numbers to the ignored options.
2793 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
2794 one setting" options.
2795 (append_compiler_options): Likewise to the dropped options.
2796 (append_diag_options): Likewise to the passed-on options.
2797 * opts.c (common_handle_option): Handle the new option.
2798 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
2799
2800 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
2801
2802 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
2803 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
2804 third argument of type "const signed char" to descriptions of
2805 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
2806 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
2807 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
2808 __builtin_bcdsub_ov functions.
2809
2810 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2811
2812 PR tree-optimization/86858
2813 * tree-vect-loop.c (vect_is_simple_reduction): Restore
2814 flow_bb_inside_loop_p calls.
2815
2816 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
2817
2818 PR tree-optimization/86871
2819 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
2820 instead of gimple_assign_lhs.
2821
2822 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
2823
2824 PR target/86887
2825 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
2826 register constraint to operand 0.
2827 (add<mode>3_carryinC): Likewise.
2828 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
2829
2830 2018-08-09 Martin Liska <mliska@suse.cz>
2831
2832 PR c/86895
2833 * common.opt: Remove extra line.
2834
2835 2018-08-09 Martin Liska <mliska@suse.cz>
2836
2837 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
2838 at the end of a line, make first letter capital and end up
2839 a sentence with a dot.
2840 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
2841 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
2842 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
2843 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
2844 (PARAM_MAX_ISL_OPERATIONS): Likewise.
2845 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
2846 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
2847 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
2848 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
2849 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
2850 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
2851 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
2852 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
2853 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
2854 (PARAM_TREE_REASSOC_WIDTH): Likewise.
2855 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
2856 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
2857 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
2858
2859 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
2860
2861 PR target/84332
2862 * config/s390/s390.c (s390_option_override_internal): Reduce the
2863 stack-clash-protection-probe-interval param if it would be too big
2864 for z900.
2865
2866 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
2867
2868 PR target/46179
2869 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
2870 * config/m68k/m68k.c (handle_move_double): Don't call
2871 m68k_final_prescan_insn.
2872 (m68k_adjust_decorated_operand): Renamed from
2873 m68k_final_prescan_insn, remove first and third operand and
2874 simplify.
2875 (print_operand): Call it.
2876 (print_operand_address): Call it.
2877
2878 2018-08-08 Nathan Sidwell <nathan@acm.org>
2879
2880 * diagnostic.c (diagnostic_report_current_module): Use
2881 linemap_included_from & linemap_included_from_linemap.
2882
2883 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
2884
2885 * config/aarch64/aarch64-cores.def: Add phecda core.
2886 * config/aarch64/aarch64-tune.md: Regenerate.
2887 * doc/invoke.texi: Add phecda core.
2888
2889 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
2890
2891 PR target/85295
2892 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
2893 definitions.
2894 * config/s390/s390.md ("movti"): Add more alternatives for
2895 constant to GPR copies.
2896
2897 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
2898
2899 * config/s390/s390.c: Fix whitespace damage throughout the file.
2900 * config/s390/s390.h: Likewise.
2901 * config/s390/tpf.h: Likewise.
2902
2903 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
2904
2905 * config/s390/s390.c (s390_loadrelative_operand_p):
2906 Remove TARGET_CPU_ZARCH usages.
2907 (s390_rtx_costs): Likewise.
2908 (s390_legitimate_constant_p): Likewise.
2909 (s390_cannot_force_const_mem): Likewise.
2910 (legitimate_reload_constant_p): Likewise.
2911 (s390_preferred_reload_class): Likewise.
2912 (legitimize_pic_address): Likewise.
2913 (legitimize_tls_address): Likewise.
2914 (s390_split_branches): Removed.
2915 (s390_add_execute): Removed.
2916 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
2917 (s390_mainpool_start): Likewise.
2918 (s390_mainpool_finish): Likewise.
2919 (s390_mainpool_cancel): Removed.
2920 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
2921 (s390_chunkify_cancel): Likewise.
2922 (s390_return_addr_rtx): Likewise.
2923 (s390_register_info): Remove split_branches_pending_p uages.
2924 (s390_optimize_register_info): Likewise.
2925 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
2926 split_branches_pending_p usages.
2927 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
2928 (s390_load_got): Likewise.
2929 (s390_expand_split_stack_prologue): Likewise.
2930 (output_asm_nops): Likewise.
2931 (s390_function_profiler): Likewise.
2932 (s390_emit_call): Likewise.
2933 (s390_conditional_register_usage): Likewise.
2934 (s390_optimize_prologue): Likewise.
2935 (s390_reorg): Remove TARGET_CPU_ZARCH and
2936 split_branches_pending_p usages.
2937 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
2938 usages.
2939 (s390_output_indirect_thunk_function): Likewise.
2940 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
2941 (TARGET_CPU_ZARCH_P): Removed.
2942 (struct machine_function): Remove split_branches_pending_p.
2943 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
2944
2945 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
2946
2947 * common/config/s390/s390-common.c (processor_flags_table):
2948 Remove flags.
2949 * config.gcc: Remove with_arch/with_tune support.
2950 * config/s390/2064.md: Remove cpu attribute comparisons.
2951 * config/s390/driver-native.c (s390_host_detect_local_cpu):
2952 Remove MTN.
2953 * config/s390/linux.h (ASM_SPEC):
2954 Remove -march support.
2955 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
2956 Use a table to get an arch level.
2957 * config/s390/s390-opts.h (enum processor_type):
2958 Remove enum values.
2959 * config/s390/s390.c
2960 (processor_table): Remove entries, add arch_level values.
2961 (s390_issue_rate): Remove cases.
2962 (s390_option_override): Adjust
2963 s390_option_override_internal() call.
2964 (s390_option_override_internal): Remove deprecation warning.
2965 (s390_valid_target_attribute_tree): Adjust
2966 s390_option_override_internal() call.
2967 * config/s390/s390.h (struct s390_processor):
2968 Share with s390-c.c, add arch_level field.
2969 * config/s390/s390.md:
2970 Remove occurrences in cpu attribute.
2971 * config/s390/s390.opt: Remove -march/-mtune support.
2972 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
2973 * doc/invoke.texi: Remove deprecation warning.
2974
2975 2018-08-08 Luis Machado <luis.machado@linaro.org>
2976
2977 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
2978 global.
2979 (qdf24xx_tunings): Set vector cost structure to
2980 qdf24xx_vector_cost.
2981
2982 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
2983 <register_sextend>: Set to 3.
2984
2985 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
2986
2987 PR target/86838
2988 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
2989 * config/aarch64/aarch64-simd.md
2990 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
2991 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
2992 (aarch64_frecpx<mode>): ...this new pattern.
2993 * config/aarch64/aarch64-simd-builtins.def: Remove comment
2994 about aarch64_frecp<FRECP:frecp_suffix><mode>.
2995
2996 2018-08-07 Martin Liska <mliska@suse.cz>
2997
2998 PR middle-end/83023
2999 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
3000 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
3001 * predict.def (PRED_MALLOC_NONNULL): New predictor.
3002 * doc/extend.texi: Document that malloc attribute adds
3003 hit to compiler.
3004
3005 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
3006
3007 PR target/86785
3008 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3009 Define to speculation_safe_value_not_needed.
3010
3011 2018-08-06 Jeff Law <law@redhat.com>
3012
3013 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
3014 the vr_values instance to cprop_into_stmt.
3015 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
3016 (cprop_operand): Also query EVRP to determine if OP is a constant.
3017
3018 2018-08-06 Nathan Sidwell <nathan@acm.org>
3019
3020 * diagnostic.c (diagnostic_report_current_module): Reroll
3021 included-at loop. Translate text.
3022
3023 2018-08-06 David Malcolm <dmalcolm@redhat.com>
3024
3025 * function-tests.c (selftest::test_expansion_to_rtl): Call
3026 free_after_compilation.
3027
3028 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3029
3030 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
3031
3032 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
3033
3034 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
3035 loops with memory block operations from getting unrolled.
3036
3037 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
3038
3039 PR target/86807
3040 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3041 Define to speculation_safe_value_not_needed.
3042
3043 2018-08-06 Jeff Law <law@redhat.com>
3044
3045 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
3046 assert.
3047
3048 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3049
3050 PR target/86662
3051 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
3052 with all enabled __intN types.
3053
3054 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
3055
3056 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3057
3058 * alias.c (record_set): Check for clobber high.
3059 * cfgexpand.c (expand_gimple_stmt): Likewise.
3060 * combine-stack-adj.c (single_set_for_csa): Likewise.
3061 * combine.c (find_single_use_1): Likewise.
3062 (set_nonzero_bits_and_sign_copies): Likewise.
3063 (get_combine_src_dest): Likewise.
3064 (is_parallel_of_n_reg_sets): Likewise.
3065 (try_combine): Likewise.
3066 (record_dead_and_set_regs_1): Likewise.
3067 (reg_dead_at_p_1): Likewise.
3068 (reg_dead_at_p): Likewise.
3069 * dce.c (deletable_insn_p): Likewise.
3070 (mark_nonreg_stores_1): Likewise.
3071 (mark_nonreg_stores_2): Likewise.
3072 * df-scan.c (df_find_hard_reg_defs): Likewise.
3073 (df_uses_record): Likewise.
3074 (df_get_call_refs): Likewise.
3075 * dwarf2out.c (mem_loc_descriptor): Likewise.
3076 * haifa-sched.c (haifa_classify_rtx): Likewise.
3077 * ira-build.c (create_insn_allocnos): Likewise.
3078 * ira-costs.c (scan_one_insn): Likewise.
3079 * ira.c (equiv_init_movable_p): Likewise.
3080 (rtx_moveable_p): Likewise.
3081 (interesting_dest_for_shprep): Likewise.
3082 * jump.c (mark_jump_label_1): Likewise.
3083 * postreload-gcse.c (record_opr_changes): Likewise.
3084 * postreload.c (reload_cse_simplify): Likewise.
3085 (struct reg_use): Add source expr.
3086 (reload_combine): Check for clobber high.
3087 (reload_combine_note_use): Likewise.
3088 (reload_cse_move2add): Likewise.
3089 (move2add_note_store): Likewise.
3090 * print-rtl.c (print_pattern): Likewise.
3091 * recog.c (decode_asm_operands): Likewise.
3092 (store_data_bypass_p): Likewise.
3093 (if_test_bypass_p): Likewise.
3094 * regcprop.c (kill_clobbered_value): Likewise.
3095 (kill_set_value): Likewise.
3096 * reginfo.c (reg_scan_mark_refs): Likewise.
3097 * reload1.c (maybe_fix_stack_asms): Likewise.
3098 (eliminate_regs_1): Likewise.
3099 (elimination_effects): Likewise.
3100 (mark_not_eliminable): Likewise.
3101 (scan_paradoxical_subregs): Likewise.
3102 (forget_old_reloads_1): Likewise.
3103 * reorg.c (find_end_label): Likewise.
3104 (try_merge_delay_insns): Likewise.
3105 (redundant_insn): Likewise.
3106 (own_thread_p): Likewise.
3107 (fill_simple_delay_slots): Likewise.
3108 (fill_slots_from_thread): Likewise.
3109 (dbr_schedule): Likewise.
3110 * resource.c (update_live_status): Likewise.
3111 (mark_referenced_resources): Likewise.
3112 (mark_set_resources): Likewise.
3113 * rtl.c (copy_rtx): Likewise.
3114 * rtlanal.c (reg_referenced_p): Likewise.
3115 (single_set_2): Likewise.
3116 (noop_move_p): Likewise.
3117 (note_stores): Likewise.
3118 * sched-deps.c (sched_analyze_reg): Likewise.
3119 (sched_analyze_insn): Likewise.
3120
3121 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3122
3123 * cse.c (invalidate_reg): New function extracted from...
3124 (invalidate): ...here.
3125 (canonicalize_insn): Check for clobber high.
3126 (invalidate_from_clobbers): invalidate clobber highs.
3127 (invalidate_from_sets_and_clobbers): Likewise.
3128 (count_reg_usage): Check for clobber high.
3129 (insn_live_p): Likewise.
3130 * cselib.c (cselib_expand_value_rtx_1):Likewise.
3131 (cselib_invalidate_regno): Check for clobber in setter.
3132 (cselib_invalidate_rtx): Pass through setter.
3133 (cselib_invalidate_rtx_note_stores):
3134 (cselib_process_insn): Check for clobber high.
3135 * cselib.h (cselib_invalidate_rtx): Add operand.
3136
3137 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3138
3139 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
3140 (mark_not_eliminable): Likewise.
3141 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
3142 * lra-lives.c (process_bb_lives): Check for clobber high.
3143 * lra.c (new_insn_reg): Remember clobber highs.
3144 (collect_non_operand_hard_regs): Check for clobber high.
3145 (lra_set_insn_recog_data): Likewise.
3146 (add_regs_to_insn_regno_info): Likewise.
3147 (lra_update_insn_regno_info): Likewise.
3148
3149 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3150
3151 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
3152 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
3153
3154 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3155
3156 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
3157 (copy_insn_1): Likewise.
3158 (gen_hard_reg_clobber_high): New gen function.
3159 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
3160 * genemit.c (gen_exp): Likewise.
3161 (gen_emit_seq): Pass through info.
3162 (gen_insn): Check for CLOBBER_HIGH.
3163 (gen_expand): Pass through info.
3164 (gen_split): Likewise.
3165 (output_add_clobbers): Likewise.
3166 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
3167 (remove_clobbers): Likewise.
3168 * rtl.h (gen_hard_reg_clobber_high): New declaration.
3169
3170 2018-08-06 Alan Hayward <alan.hayward@arm.com>
3171
3172 * doc/rtl.texi (clobber_high): Add.
3173 (parallel): Add in clobber high
3174 * rtl.c (rtl_check_failed_code3): Add function.
3175 * rtl.def (CLOBBER_HIGH): Add expression.
3176 * rtl.h (RTL_CHECKC3): Add macro.
3177 (rtl_check_failed_code3): Add declaration.
3178 (XC3EXP): Add macro.
3179
3180 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
3181
3182 PR target/86386
3183 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
3184 cfun->machine->max_used_stack_alignment if needed.
3185
3186 2018-08-04 Martin Sebor <msebor@redhat.com>
3187
3188 PR tree-optimization/86571
3189 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
3190 NaN output to 4.
3191
3192 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
3193
3194 PR target/86799
3195 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3196 Define.
3197
3198 2018-08-03 Jeff Law <law@redhat.com>
3199
3200 PR target/86795
3201 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3202 Define to speculation_safe_value_not_needed.
3203
3204 2018-08-03 David Malcolm <dmalcolm@redhat.com>
3205
3206 * doc/gcov.texi (-x): Remove duplicate "to".
3207 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
3208 (-Wif-not-aligned): Remove duplicate "is".
3209 (-flto): Remove duplicate "the".
3210 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
3211 duplicate "v5.00.b".
3212 (MSP430 Options): Remove duplicate "and" from the description
3213 of "-mgprel-sec=regexp".
3214 (x86 Options): Remove duplicate copies of "vmldLog102" and
3215 vmlsLog104 from description of "-mveclibabi=type".
3216
3217 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
3218
3219 * internal-fn.h (first_commutative_argument): Declare.
3220 * internal-fn.c (first_commutative_argument): New function.
3221 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
3222 restrictions for pattern statements. Use first_commutative_argument
3223 to look for commutative operands in calls to internal functions.
3224
3225 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
3226
3227 * Makefile.in (wide-int-range.o): New.
3228 * tree-vrp.c: Move all the wide_int_* functions to...
3229 * wide-int-range.cc: ...here.
3230 * tree-vrp.h: Move all the wide_int_* prototypes to...
3231 * wide-int-range.h: ...here.
3232
3233 2018-08-03 Tom de Vries <tdevries@suse.de>
3234
3235 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
3236 UI_NONE.
3237 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
3238 * except.c (output_function_exception_table): Do early exit if
3239 targetm_common.except_unwind_info (&global_options) == UI_NONE.
3240
3241 2018-08-03 Martin Liska <mliska@suse.cz>
3242
3243 * predict.c (dump_prediction): Change to 2 digits
3244 in fraction part.
3245
3246 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
3247
3248 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
3249 neon_dup_q to...
3250 (falkor_am_1_gtov_gtov): ... a new insn reservation.
3251
3252 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
3253
3254 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
3255 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
3256 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
3257 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
3258 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
3259
3260 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3261
3262 * diagnostic-show-locus.c (diagnostic_show_locus): Use
3263 pp_take_prefix when saving the existing prefix.
3264 * diagnostic.c (diagnostic_append_note): Likewise.
3265 * langhooks.c (lhd_print_error_function): Likewise.
3266 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
3267 param's type. Free the existing prefix.
3268 (pp_take_prefix): New function.
3269 (pretty_printer::pretty_printer): Drop the prefix parameter.
3270 Rename the length parameter to match the comment.
3271 (pretty_printer::~pretty_printer): Free the prefix.
3272 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
3273 parameter.
3274 (struct pretty_printer): Drop the "const" from "prefix" field's
3275 type and clarify memory management.
3276 (pp_set_prefix): Drop the "const" from the 2nd param.
3277 (pp_take_prefix): New decl.
3278
3279 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
3280
3281 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
3282 for word_mode_ok here instead of passing as argument.
3283 (expand_block_compare): Change select_block_compare_mode() call.
3284 (expand_strncmp_gpr_sequence): New function.
3285 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
3286
3287 2018-08-02 Jeff Law <law@redhat.com>
3288
3289 PR target/86790
3290 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3291 Define to speculation_safe_value_not_needed.
3292
3293 PR target/86784
3294 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3295 Define to speculation_safe_value_not_needed.
3296
3297 2018-08-02 Tom de Vries <tdevries@suse.de>
3298
3299 PR target/86660
3300 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
3301 function. Return UI_TARGET unconditionally.
3302 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
3303 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
3304
3305 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3306
3307 * genemit.c (print_overload_test): Fix typo.
3308
3309 2018-08-02 Richard Biener <rguenther@suse.de>
3310
3311 PR tree-optimization/86816
3312 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
3313 which checks for value availability before querying it.
3314 (gvn_uses_equal): Use it.
3315 (same_succ_hash): Likewise.
3316 (gimple_equal_p): Likewise.
3317
3318 2018-08-02 Nick Clifton <nickc@redhat.com>
3319
3320 PR target/86813
3321 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3322 Define to speculation_safe_value_not_needed.
3323
3324 PR target/86810
3325 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3326 Define to speculation_safe_value_not_needed.
3327
3328 PR target/86810
3329 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3330 Define to speculation_safe_value_not_needed.
3331
3332 PR target/86803
3333 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3334 Define to speculation_safe_value_not_needed.
3335
3336 PR target/86797
3337 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3338 Define to speculation_safe_value_not_needed.
3339
3340 PR target/86791
3341 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3342 Define to speculation_safe_value_not_needed.
3343
3344 PR target/86789
3345 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3346 Define to speculation_safe_value_not_needed.
3347
3348 PR target/86787
3349 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
3350 Define to speculation_safe_value_not_needed.
3351
3352 PR target/86782
3353 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
3354 speculation_safe_value_not_needed.
3355
3356 PR target/86781
3357 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
3358 to speculation_safe_value_not_needed.
3359
3360 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3361
3362 * doc/md.texi: Expand the documentation of instruction names
3363 to mention port-local uses. Document '@' in pattern names.
3364 * read-md.h (overloaded_instance, overloaded_name): New structs.
3365 (mapping): Declare.
3366 (md_reader::handle_overloaded_name): New member function.
3367 (md_reader::get_overloads): Likewise.
3368 (md_reader::m_first_overload): New member variable.
3369 (md_reader::m_next_overload_ptr): Likewise.
3370 (md_reader::m_overloads_htab): Likewise.
3371 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
3372 m_next_overload_ptr and m_overloads_htab.
3373 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
3374 (get_mode_token, get_code_token, get_int_token): New functions.
3375 (map_attr_string): Add an optional argument that passes back
3376 the associated iterator.
3377 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
3378 (md_reader::handle_overloaded_name, add_overload_instance): New
3379 functions.
3380 (apply_iterators): Handle '@' names. Report an error if '@'
3381 is used without iterators.
3382 (initialize_iterators): Initialize the new iterator_group fields.
3383 * genopinit.c (handle_overloaded_code_for)
3384 (handle_overloaded_gen): New functions.
3385 (main): Use them to print declarations of maybe_code_for_* and
3386 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
3387 * genemit.c (print_overload_arguments, print_overload_test)
3388 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
3389 (main): Use it to print definitions of maybe_code_for_* and
3390 maybe_gen_* functions.
3391 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
3392 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
3393 instead of explicit mode checks.
3394 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
3395 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
3396 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
3397 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
3398 (aarch64_expand_compare_and_swap): Likewise
3399 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
3400 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
3401 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
3402 (aarch64_constant_pool_reload_icode): Delete.
3403 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
3404 instead of aarch64_constant_pool_reload_icode. Use
3405 code_for_aarch64_reload_mov instead of explicit mode checks.
3406 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
3407 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
3408 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
3409 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
3410 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
3411 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
3412 (aarch64_atomic_load_op_code): Delete.
3413 (aarch64_emit_atomic_load_op): Likewise.
3414 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
3415 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
3416 instead of aarch64_emit_atomic_load_op.
3417 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
3418 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
3419 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
3420 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
3421 character before the pattern name.
3422 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
3423 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
3424 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
3425 (aarch64_frecps<mode>): Likewise.
3426 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
3427 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
3428 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
3429 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
3430 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
3431
3432 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
3433
3434 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
3435 Allow HFmode constants if TARGET_FP_F16INST.
3436
3437 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
3438
3439 PR target/86014
3440 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
3441 No longer check last store for clobber of address register.
3442
3443 2018-08-02 Martin Liska <mliska@suse.cz>
3444
3445 PR gcov-profile/86817
3446 * gcov.c (process_all_functions): New function.
3447 (main): Call it.
3448 (process_file): Move functions processing to
3449 process_all_functions.
3450
3451 2018-08-02 David Malcolm <dmalcolm@redhat.com>
3452
3453 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
3454 "const" to the "gimple *" and "rtx_insn *" parameters.
3455 * dumpfile.h (dump_user_location_t::dump_user_location_t):
3456 Likewise.
3457 (dump_location_t::dump_location_t): Likewise.
3458
3459 2018-08-01 Martin Sebor <msebor@redhat.com>
3460
3461 PR tree-optimization/86650
3462 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
3463 rather than a "gcall *". Directly pass the data of interest
3464 to percent_K_format, rather than building a temporary CALL_EXPR
3465 to hold it.
3466 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
3467 (gimple_fold_builtin_strncat): Adjust.
3468 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
3469 gcall* argument with gimple*.
3470 * gimple-ssa-warn-restrict.c (check_call): Same.
3471 (wrestrict_dom_walker::before_dom_children): Same.
3472 (builtin_access::builtin_access): Same.
3473 (check_bounds_or_overlap): Same
3474 (maybe_diag_overlap): Same.
3475 (maybe_diag_offset_bounds): Same.
3476 * tree-diagnostic.c (default_tree_printer): Move usage of
3477 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
3478 to this callsite.
3479 * tree-pretty-print.c (percent_K_format): Add argument.
3480 * tree-pretty-print.h: Add argument.
3481 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
3482 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
3483 (maybe_diag_stxncpy_trunc): Same.
3484 (handle_builtin_stxncpy): Same.
3485 (handle_builtin_strcat): Same.
3486
3487 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3488
3489 * match.pd: Optimise pointer range checks.
3490
3491 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3492
3493 PR tree-optimization/86758
3494 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
3495 to remove pattern statements.
3496
3497 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3498
3499 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
3500 result of dfs_enumerate_from when constructing stmt_vec_infos,
3501 instead of additionally calling get_loop_body.
3502
3503 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3504
3505 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
3506 parameter.
3507 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
3508 When creating an iv, assert that the step is not known to be zero.
3509 (vect_setup_realignment): Update call accordingly.
3510 * tree-vect-stmts.c (vectorizable_store): Likewise.
3511 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
3512
3513 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3514
3515 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
3516 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
3517 (vectorizable_reduction): Likewise.
3518 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
3519 (vect_detect_hybrid_slp_stmts): Likewise.
3520 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
3521
3522 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
3523
3524 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
3525 (wide_int_set_zero_nonzero_bits): ...this.
3526 (zero_nonzero_bits_from_vr): Rename to...
3527 (vrp_set_zero_nonzero_bits): ...this.
3528 (extract_range_from_multiplicative_op_1): Abstract wide int
3529 code...
3530 (wide_int_range_multiplicative_op): ...here.
3531 (extract_range_from_binary_expr_1): Extract wide int binary
3532 operations into their own functions.
3533 (wide_int_range_lshift): New.
3534 (wide_int_range_can_optimize_bit_op): New.
3535 (wide_int_range_shift_undefined_p): New.
3536 (wide_int_range_bit_xor): New.
3537 (wide_int_range_bit_ior): New.
3538 (wide_int_range_bit_and): New.
3539 (wide_int_range_trunc_mod): New.
3540 (extract_range_into_wide_ints): New.
3541 (vrp_shift_undefined_p): New.
3542 (extract_range_from_multiplicative_op): New.
3543 (vrp_can_optimize_bit_op): New.
3544 * tree-vrp.h (value_range::dump): New.
3545 (wide_int_range_multiplicative_op): New.
3546 (wide_int_range_lshift):New.
3547 (wide_int_range_shift_undefined_p): New.
3548 (wide_int_range_bit_xor): New.
3549 (wide_int_range_bit_ior): New.
3550 (wide_int_range_bit_and): New.
3551 (wide_int_range_trunc_mod): New.
3552 (zero_nonzero_bits_from_bounds): Rename to...
3553 (wide_int_set_zero_nonzero_bits): ...this.
3554 (zero_nonzero_bits_from_vr): Rename to...
3555 (vrp_set_zero_nonzero_bits): ...this.
3556 (range_easy_mask_min_max): Rename to...
3557 (wide_int_range_can_optimize_bit_op): this.
3558 * vr-values.c (simplify_bit_ops_using_ranges): Rename
3559 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
3560
3561 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3562
3563 * tree-vectorizer.h (vect_orig_stmt): New function.
3564 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
3565 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
3566 (vect_create_epilog_for_reduction): Likewise.
3567 (vectorizable_live_operation): Likewise.
3568 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
3569 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
3570 * tree-vect-stmts.c (vectorizable_call): Likewise.
3571 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
3572
3573 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3574
3575 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
3576 argument.
3577 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
3578 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
3579 (vect_transform_loop): Likewise.
3580 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3581
3582 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3583
3584 * tree-vectorizer.h (vect_schedule_slp): Return void.
3585 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
3586 (vect_schedule_slp): Likewise.
3587
3588 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3589
3590 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
3591 argument.
3592 (vect_transform_loop): Update calls accordingly. Schedule SLP
3593 instances before the main loop, if any exist.
3594
3595 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
3596
3597 PR tree-optimization/86749
3598 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
3599 If the lhs is used in a COND_EXPR, check that it is being used
3600 as the "then" or "else" value.
3601
3602 2018-08-01 Tom de Vries <tdevries@suse.de>
3603
3604 PR target/86800
3605 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
3606 speculation_safe_value_not_needed.
3607
3608 2018-08-01 Richard Biener <rguenther@suse.de>
3609
3610 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
3611 as base and offset.
3612
3613 2018-08-01 Martin Liska <mliska@suse.cz>
3614
3615 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
3616 format how successful transformation is dumped.
3617 (gimple_mod_pow2_value_transform): Likewise.
3618 (gimple_mod_subtract_transform): Likewise.
3619 (gimple_stringops_transform): Likewise.
3620
3621 2018-08-01 Martin Liska <mliska@suse.cz>
3622
3623 PR value-prof/35543
3624 * value-prof.c (interesting_stringop_to_profile_p):
3625 Simplify the code and add BUILT_IN_MEMMOVE.
3626 (gimple_stringops_transform): Likewise.
3627
3628 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
3629
3630 * config/aarch64/aarch64-simd.md
3631 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
3632 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
3633 use GPI iterator instead of SI mode.
3634
3635 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
3636
3637 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
3638 rs6000_speculation_barrier.
3639 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
3640 new barrier pattern name.
3641
3642 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
3643
3644 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
3645 (speculation_barrier): New insn.
3646
3647 2018-08-01 Richard Biener <rguenther@suse.de>
3648
3649 PR bootstrap/86724
3650 * graphite.h: Include isl/id.h and isl/space.h to allow build
3651 with ISL 0.20.
3652
3653 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
3654
3655 PR target/86651
3656 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
3657 mode for COFF targets.
3658 * defaults.h (TARGET_COFF): Define.
3659 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
3660 TARGET_COFF): Define.
3661 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
3662 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
3663 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
3664
3665 2018-07-31 Alexandre Oliva <oliva@adacore.com>
3666
3667 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
3668 * gimple-streamer-out.c (output_bb): Save it.
3669 * lto-streamer-in.c (input_struct_function_base): Restore
3670 instance discriminator if available. Create map on demand.
3671 * lto-streamer-out.c (output_struct_function_base): Save it if
3672 available.
3673 * final.c (decl_to_instance_map): Document LTO strategy.
3674
3675 2018-07-31 Alexandre Oliva <oliva@adacore.com>
3676 Olivier Hainque <hainque@adacore.com>
3677
3678 * debug.h (decl_to_instance_map_t): New type.
3679 (decl_to_instance_map): Declare.
3680 (maybe_create_decl_to_instance_map): New inline function.
3681 * final.c (bb_discriminator, last_bb_discriminator): New statics,
3682 to track basic block discriminators.
3683 (final_start_function_1): Initialize them.
3684 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
3685 bb_discriminator.
3686 (decl_to_instance_map): New variable.
3687 (map_decl_to_instance, maybe_set_discriminator): New functions.
3688 (notice_source_line): Set discriminator.
3689
3690 2018-07-31 Ian Lance Taylor <iant@golang.org>
3691
3692 * targhooks.c (default_have_speculation_safe_value): Add
3693 ATTRIBUTE_UNUSED.
3694
3695 2018-07-31 David Malcolm <dmalcolm@redhat.com>
3696
3697 * dump-context.h: Include "pretty-print.h".
3698 (dump_context::refresh_dumps_are_enabled): New decl.
3699 (dump_context::emit_item): New decl.
3700 (class dump_context): Add fields "m_test_pp" and
3701 "m_test_pp_flags".
3702 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
3703 (temp_dump_context::get_dumped_text): New decl.
3704 (class temp_dump_context): Add field "m_pp".
3705 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
3706 (dump_context::refresh_dumps_are_enabled): ...and add a test for
3707 m_test_pp.
3708 (set_dump_file): Update for above change.
3709 (set_alt_dump_file): Likewise.
3710 (dump_loc): New overload, taking a pretty_printer *.
3711 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
3712 to any test pretty-printer.
3713 (make_item_for_dump_gimple_stmt): New function, adapted from
3714 optinfo::add_gimple_stmt.
3715 (dump_context::dump_gimple_stmt): Call it, and use the result,
3716 eliminating the direct usage of dump_file and alt_dump_file in
3717 favor of indirectly using them via emit_item.
3718 (make_item_for_dump_gimple_expr): New function, adapted from
3719 optinfo::add_gimple_expr.
3720 (dump_context::dump_gimple_expr): Call it, and use the result,
3721 eliminating the direct usage of dump_file and alt_dump_file in
3722 favor of indirectly using them via emit_item.
3723 (make_item_for_dump_generic_expr): New function, adapted from
3724 optinfo::add_tree.
3725 (dump_context::dump_generic_expr): Call it, and use the result,
3726 eliminating the direct usage of dump_file and alt_dump_file in
3727 favor of indirectly using them via emit_item.
3728 (make_item_for_dump_printf_va): New function, adapted from
3729 optinfo::add_printf_va.
3730 (make_item_for_dump_printf): New function.
3731 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
3732 and use the result, eliminating the direct usage of dump_file and
3733 alt_dump_file in favor of indirectly using them via emit_item.
3734 (make_item_for_dump_dec): New function.
3735 (dump_context::dump_dec): Call it, and use the result,
3736 eliminating the direct usage of dump_file and alt_dump_file in
3737 favor of indirectly using them via emit_item.
3738 (make_item_for_dump_symtab_node): New function, adapted from
3739 optinfo::add_symtab_node.
3740 (dump_context::dump_symtab_node): Call it, and use the result,
3741 eliminating the direct usage of dump_file and alt_dump_file in
3742 favor of indirectly using them via emit_item.
3743 (dump_context::begin_scope): Reimplement, avoiding direct usage
3744 of dump_file and alt_dump_file in favor of indirectly using them
3745 via emit_item.
3746 (dump_context::emit_item): New member function.
3747 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
3748 Set up test pretty-printer on the underlying context. Call
3749 refresh_dumps_are_enabled.
3750 (temp_dump_context::~temp_dump_context): Call
3751 refresh_dumps_are_enabled.
3752 (temp_dump_context::get_dumped_text): New member function.
3753 (selftest::verify_dumped_text): New function.
3754 (ASSERT_DUMPED_TEXT_EQ): New macro.
3755 (selftest::test_capture_of_dump_calls): Run all tests twice, with
3756 and then without optinfo enabled. Add uses of
3757 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
3758 * dumpfile.h: Update comment for the dump_* API.
3759 * optinfo-emit-json.cc
3760 (selftest::test_building_json_from_dump_calls): Update for new
3761 param for temp_dump_context ctor.
3762 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
3763 and "m_owned" field.
3764 (optinfo_item::~optinfo_item): Likewise.
3765 (optinfo::add_item): New member function.
3766 (optinfo::emit): Update comment.
3767 (optinfo::add_string): Delete.
3768 (optinfo::add_printf): Delete.
3769 (optinfo::add_printf_va): Delete.
3770 (optinfo::add_gimple_stmt): Delete.
3771 (optinfo::add_gimple_expr): Delete.
3772 (optinfo::add_tree): Delete.
3773 (optinfo::add_symtab_node): Delete.
3774 (optinfo::add_dec): Delete.
3775 * optinfo.h (class dump_context): New forward decl.
3776 (optinfo::add_item): New decl.
3777 (optinfo::add_string): Delete.
3778 (optinfo::add_printf): Delete.
3779 (optinfo::add_printf_va): Delete.
3780 (optinfo::add_gimple_stmt): Delete.
3781 (optinfo::add_gimple_expr): Delete.
3782 (optinfo::add_tree): Delete.
3783 (optinfo::add_symtab_node): Delete.
3784 (optinfo::add_dec): Delete.
3785 (optinfo::add_poly_int): Delete.
3786 (optinfo_item::optinfo_item): Remove "owned" param.
3787 (class optinfo_item): Remove field "m_owned".
3788
3789 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3790
3791 PR middle-end/86705
3792 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
3793 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
3794 requested variable alignment.
3795 (expand_one_ssa_partition): Likewise.
3796 (expand_one_var): Likewise.
3797
3798 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3799
3800 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
3801 to speculation_safe_value_not_needed.
3802
3803 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3804
3805 * targhooks.h (speculation_safe_value_not_needed): New prototype.
3806 * targhooks.c (speculation_safe_value_not_needed): New function.
3807 * target.def (have_speculation_safe_value): Update documentation.
3808 * doc/tm.texi: Regenerated.
3809
3810 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3811
3812 * config/aarch64/iterators.md (ALLI_TI): New iterator.
3813 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
3814 expand.
3815 (despeculate_copy<ALLI:mode>_insn): New insn.
3816 (despeculate_copyti_insn): New insn.
3817 (despeculate_simple<ALLI:mode>): New insn
3818 (despeculate_simpleti): New insn.
3819 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
3820 function.
3821 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
3822 aarch64_speculation_safe_value.
3823 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
3824
3825 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3826
3827 * config/aarch64/aarch64-speculation.cc: New file.
3828 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
3829 before pass_reorder_blocks.
3830 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
3831 prototype.
3832 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
3833 X14 and X15 when tracking speculation.
3834 * config/aarch64/aarch64.md (register name constants): Add
3835 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
3836 (unspec): Add UNSPEC_SPECULATION_TRACKER.
3837 (speculation_barrier): New insn attribute.
3838 (cmp<mode>): Allow SP in comparisons.
3839 (speculation_tracker): New insn.
3840 (speculation_barrier): Add speculation_barrier attribute.
3841 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
3842 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
3843 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
3844
3845 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3846
3847 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
3848 aarch64_track_speculation is true.
3849 (tb<optab><mode>1): Likewise.
3850 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
3851 generate CB[N]Z when tracking speculation.
3852 (aarch64_split_compare_and_swap): Likewise.
3853 (aarch64_split_atomic_op): Likewise.
3854
3855 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3856
3857 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
3858
3859 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3860
3861 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
3862 (speculation_barrier): New insn.
3863
3864 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3865
3866 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
3867 * config/arm/arm.md (speculation_barrier): New expand.
3868 (speculation_barrier_insn): New pattern.
3869
3870 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
3871
3872 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
3873 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
3874 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
3875 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
3876 list.
3877 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
3878 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
3879 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
3880 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
3881 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
3882 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
3883 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
3884 * builtins.c (expand_speculation_safe_value): New function.
3885 (expand_builtin): Call it.
3886 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
3887 * doc/extend.texi: Document __builtin_speculation_safe_value.
3888 * doc/md.texi: Document "speculation_barrier" pattern.
3889 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
3890 TARGET_HAVE_SPECULATION_SAFE_VALUE.
3891 * doc/tm.texi: Regenerated.
3892 * target.def (have_speculation_safe_value, speculation_safe_value): New
3893 hooks.
3894 * targhooks.c (default_have_speculation_safe_value): New function.
3895 (default_speculation_safe_value): New function.
3896 * targhooks.h (default_have_speculation_safe_value): Add prototype.
3897 (default_speculation_safe_value): Add prototype.
3898
3899 2018-07-31 David Malcolm <dmalcolm@redhat.com>
3900
3901 * dump-context.h (dump_context::dump_loc): New decl.
3902 * dumpfile.c (dump_context::dump_loc): New member function.
3903 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
3904 and dump_gimple_stmt.
3905 (dump_context::dump_gimple_expr_loc): Likewise, using
3906 dump_gimple_expr.
3907 (dump_context::dump_generic_expr_loc): Likewise, using
3908 dump_generic_expr.
3909 (dump_context::dump_printf_loc_va): Likewise, using
3910 dump_printf_va.
3911 (dump_context::begin_scope): Explicitly using the global function
3912 "dump_loc", rather than the member function.
3913
3914 2018-07-31 Martin Sebor <msebor@redhat.com>
3915
3916 PR tree-optimization/86741
3917 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
3918
3919 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
3920
3921 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
3922 depend on whether prefetch instructions will be emitted or not.
3923 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
3924 will be emitted or not.
3925 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
3926 (TARGET_SETMEM_PFD): New macros.
3927
3928 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3929
3930 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
3931 (NULL_STMT_VEC_INFO): Delete.
3932 (stmt_vec_info::operator*): Likewise.
3933 (stmt_vec_info::operator gimple *): Likewise.
3934 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
3935 of NULL_STMT_VEC_INFO.
3936 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
3937 (vect_reassociating_reduction_p): Likewise.
3938 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
3939 (vectorizable_store): Likewise.
3940 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
3941 (vec_info::free_stmt_vec_infos): Likewise.
3942
3943 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3944
3945 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
3946 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
3947
3948 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3949
3950 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
3951 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
3952 (vec_info::free_stmt_vec_info): New private member functions.
3953 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
3954 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
3955 * tree-parloops.c (gather_scalar_reductions): Remove calls to
3956 set_stmt_vec_info_vec and free_stmt_vec_infos.
3957 * tree-vect-loop.c (_loop_vec_info): Remove call to
3958 set_stmt_vec_info_vec.
3959 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
3960 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
3961 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
3962 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
3963 (vec_info::free_stmt_vec_info): ...these new functions. Remove
3964 assignments in {vec_info::,}new_stmt_vec_info that are redundant
3965 with the clearing in the xcalloc.
3966 (stmt_vec_info_vec): Delete.
3967 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
3968 (vectorize_loops): Likewise.
3969 (vec_info::~vec_info): Remove argument from call to
3970 free_stmt_vec_infos.
3971 (vec_info::add_stmt): Remove vinfo argument from call to
3972 new_stmt_vec_info.
3973
3974 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3975
3976 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
3977 rather than a gimple stmt.
3978 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
3979 information for pattern statements when passed the original
3980 statement; instead wait to be passed the pattern statement itself.
3981 Don't call set_vinfo_for_stmt here.
3982 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
3983 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
3984 stmt_vec_infos here.
3985 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
3986 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
3987 stmt_vec_infos entry.
3988
3989 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3990
3991 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
3992 * tree-vectorizer.c (vec_info::replace_stmt): New function.
3993 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
3994 * tree-vect-stmts.c (vectorizable_call): Likewise.
3995 (vectorizable_simd_clone_call): Likewise.
3996
3997 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
3998
3999 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
4000 * tree-vectorizer.c (vec_info::remove_stmt): New function.
4001 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
4002 * tree-vect-loop.c (vect_transform_loop): Likewise.
4003 * tree-vect-slp.c (vect_schedule_slp): Likewise.
4004 * tree-vect-stmts.c (vect_remove_stores): Likewise.
4005
4006 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4007
4008 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
4009 (vect_dr_stmt): Delete.
4010 * tree-vectorizer.c (vec_info::lookup_dr): New function.
4011 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
4012 of DR_VECT_AUX.
4013 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
4014 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
4015 (vect_verify_datarefs_alignment, vect_peeling_supportable)
4016 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
4017 (vect_analyze_data_refs): Likewise.
4018 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
4019 argument.
4020 (vect_find_same_alignment_drs): Likewise.
4021 (vect_slp_analyze_node_dependences): Update calls accordingly.
4022 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
4023 instead of DR_VECT_AUX.
4024 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
4025 of a vector data references. Use vec_info::lookup_dr instead of
4026 DR_VECT_AUX.
4027 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
4028 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
4029 instead of DR_VECT_AUX.
4030
4031 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4032
4033 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
4034 dr_vec_info.
4035 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
4036 accordingly.
4037 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4038 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
4039 (vect_gen_prolog_loop_niters): Likewise.
4040
4041 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4042
4043 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
4044 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
4045 (vect_known_alignment_in_bytes, vect_dr_behavior)
4046 (vect_get_scalar_dr_size): Take references as dr_vec_infos
4047 instead of data_references. Update calls to other routines for
4048 which the same change has been made.
4049 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
4050 dr_vec_infos instead of stmt_vec_infos.
4051 (vect_analyze_data_ref_dependence): Update call accordingly.
4052 (vect_slp_analyze_data_ref_dependence)
4053 (vect_record_base_alignments): Use DR_VECT_AUX.
4054 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
4055 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
4056 (vector_alignment_reachable_p, vect_get_data_access_cost)
4057 (vect_peeling_supportable, vect_analyze_group_access_1)
4058 (vect_analyze_group_access, vect_analyze_data_ref_access)
4059 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
4060 (vect_compile_time_alias, vect_small_gap_p)
4061 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
4062 (vect_supportable_dr_alignment): Take references as dr_vec_infos
4063 instead of data_references. Update calls to other routines for
4064 which the same change has been made.
4065 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
4066 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
4067 (vect_slp_analyze_and_verify_node_alignment)
4068 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
4069 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
4070 (vect_setup_realignment): Use dr_vec_infos. Update calls after
4071 above changes.
4072 (_vect_peel_info::dr): Replace with...
4073 (_vect_peel_info::dr_info): ...this new field.
4074 (vect_peeling_hash_get_most_frequent)
4075 (vect_peeling_hash_choose_best_peeling): Update accordingly.
4076 (vect_peeling_hash_get_lowest_cost):
4077 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
4078 routines for which the same change has been made.
4079 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
4080 data_reference.
4081 * tree-vect-loop-manip.c (get_misalign_in_elems)
4082 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
4083 above changes.
4084 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4085 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
4086 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
4087 (get_group_load_store_type, get_negative_load_store_type)
4088 (vect_get_data_ptr_increment, vectorizable_store)
4089 (vectorizable_load): Likewise.
4090 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
4091 Update calls to other routines for which the same change has been made.
4092
4093 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4094
4095 * tree-vectorizer.h (vec_info::move_dr): New member function.
4096 (dataref_aux): Rename to...
4097 (dr_vec_info): ...this and add "dr" and "stmt" fields.
4098 (_stmt_vec_info::dr_aux): Update accordingly.
4099 (_stmt_vec_info::data_ref_info): Delete.
4100 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
4101 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
4102 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
4103 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
4104 of data_ref.
4105 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
4106 (STMT_VINFO_DR_INFO): New macro.
4107 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
4108 (set_dr_misalignment): Update after rename of dataref_aux.
4109 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
4110 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
4111 initialization of STMT_VINFO_DATA_REF.
4112 * tree-vectorizer.c (vec_info::move_dr): New function.
4113 * tree-vect-patterns.c (vect_recog_bool_pattern)
4114 (vect_recog_mask_conversion_pattern)
4115 (vect_recog_gather_scatter_pattern): Use it.
4116 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
4117 the "dr" and "stmt" fields of dr_vec_info instead of
4118 STMT_VINFO_DATA_REF.
4119
4120 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4121
4122 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
4123 (is_pattern_stmt_p): Use it.
4124 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
4125 on pattern statements.
4126
4127 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4128
4129 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
4130 original stmt as a stmt_vec_info rather than a gimple stmt.
4131 (vect_pattern_recog_1): Take the statement directly as a
4132 stmt_vec_info, rather than via a gimple_stmt_iterator.
4133 Update call to vect_mark_pattern_stmts.
4134 (vect_pattern_recog): Update calls accordingly.
4135
4136 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4137
4138 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
4139 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
4140 a vect_def_type for the first argument.
4141 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
4142 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
4143 operand if it isn't defined by a vectorized statement.
4144 (vect_build_gather_load_calls): Remove the mask_dt argument and
4145 update calls to vect_get_vec_def_for_stmt_copy.
4146 (vectorizable_bswap): Likewise the dt argument.
4147 (vectorizable_call): Update calls to vectorizable_bswap and
4148 vect_get_vec_def_for_stmt_copy.
4149 (vectorizable_simd_clone_call, vectorizable_assignment)
4150 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
4151 (vectorizable_comparison): Update calls to
4152 vect_get_vec_def_for_stmt_copy.
4153 (vectorizable_store): Likewise. Remove now-unnecessary calls to
4154 vect_is_simple_use.
4155 (vect_get_loop_based_defs): Remove dt argument and update call
4156 to vect_get_vec_def_for_stmt_copy.
4157 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
4158 and vect_get_vec_def_for_stmt_copy.
4159 (vectorizable_load): Update calls to vect_build_gather_load_calls
4160 and vect_get_vec_def_for_stmt_copy.
4161 * tree-vect-loop.c (vect_create_epilog_for_reduction)
4162 (vectorizable_reduction, vectorizable_live_operation): Update calls
4163 to vect_get_vec_def_for_stmt_copy.
4164
4165 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4166
4167 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
4168 and gimple stmt arguments with a stmt_vec_info.
4169 (vect_record_base_alignments): Update calls accordingly.
4170 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
4171 and gimple stmt arguments with a stmt_vec_info.
4172 (vect_build_slp_tree_1): Remove vinfo argument and update call
4173 to vect_record_max_nunits.
4174 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
4175 and vect_record_max_nunits.
4176
4177 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4178
4179 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
4180 file and take a stmt_vec_info instead of a gimple stmt.
4181 (supportable_widening_operation, vect_finish_replace_stmt)
4182 (vect_finish_stmt_generation, vect_get_store_rhs)
4183 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
4184 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
4185 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
4186 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
4187 (vect_create_data_ref_ptr, bump_vector_ptr)
4188 (vect_permute_store_chain, vect_setup_realignment)
4189 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
4190 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
4191 (vectorizable_reduction, vectorizable_induction)
4192 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
4193 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
4194 than gimple stmts as arguments.
4195 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
4196 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
4197 (can_group_stmts_p, vect_check_gather_scatter)
4198 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
4199 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
4200 (vect_permute_load_chain, vect_shift_permute_load_chain)
4201 (vect_transform_grouped_load)
4202 (vect_record_grouped_load_vectors): Likewise.
4203 * tree-vect-loop.c (vect_fixup_reduc_chain)
4204 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
4205 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
4206 (vectorizable_reduction, vectorizable_induction)
4207 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
4208 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
4209 (vect_get_load_store_mask): Likewise.
4210 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
4211 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
4212 * tree-vect-stmts.c (vect_mark_relevant)
4213 (is_simple_and_all_uses_invariant)
4214 (exist_non_indexing_operands_for_use_p, process_use)
4215 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
4216 (vect_get_vec_def_for_operand, vect_get_vec_defs)
4217 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
4218 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
4219 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
4220 (get_negative_load_store_type, get_load_store_type)
4221 (vect_check_load_store_mask, vect_check_store_rhs)
4222 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
4223 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4224 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4225 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4226 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
4227 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
4228 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
4229 (supportable_widening_operation): Likewise.
4230
4231 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4232
4233 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
4234 a stmt_vec_info instead of a gcall.
4235 (vect_check_gather_scatter): Update call accordingly.
4236 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
4237 of a gphi.
4238 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
4239 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
4240 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
4241 instead of a gimple stmt.
4242 (vect_transform_loop): Update calls accordingly.
4243 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
4244 stmt_vec_infos instead of gimple stmts.
4245 (vect_analyze_slp_instance): Update use accordingly.
4246 * tree-vect-stmts.c (read_vector_array, write_vector_array)
4247 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
4248 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
4249 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
4250 (vect_gen_widened_results_half, vect_get_loop_based_defs)
4251 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
4252 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
4253 down to subroutines.
4254
4255 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4256
4257 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
4258 of the worklist from a vector of gimple stmts to a vector of
4259 stmt_vec_infos.
4260 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4261 (vect_mark_stmts_to_be_vectorized): Likewise
4262
4263 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4264
4265 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
4266 statement before passing it to vect_analyze_stmt.
4267 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
4268 the chain of phi vector definitions. Track the exit phi via its
4269 stmt_vec_info.
4270 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
4271 STMT_VINFO_REDUC_DEF.
4272 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
4273 stmt_vec_infos to handle the statement chains.
4274 (vect_get_slp_defs): Record the first statement in the node
4275 using a stmt_vec_info.
4276 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
4277 statements here and pass their stmt_vec_info down to subroutines.
4278 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
4279 down to vect_finish_stmt_generation.
4280 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
4281 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
4282 stmt_vec_infos to subroutines.
4283 (vect_remove_stores): Use stmt_vec_infos to handle the statement
4284 chains.
4285
4286 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4287
4288 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
4289 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
4290 (vect_permute_store_chain, vect_setup_realignment)
4291 (vect_permute_load_chain, vect_shift_permute_load_chain)
4292 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
4293 stmts internally, and when passing values to other vectorizer routines.
4294 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4295 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
4296 (vect_analyze_loop_operations, get_initial_def_for_reduction)
4297 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
4298 (vectorizable_reduction, vectorizable_induction)
4299 (vectorizable_live_operation, vect_transform_loop_stmt)
4300 (vect_transform_loop): Likewise.
4301 * tree-vect-patterns.c (vect_reassociating_reduction_p)
4302 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
4303 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
4304 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4305 (vect_slp_analyze_node_operations_1): Likewise.
4306 * tree-vect-stmts.c (vect_mark_relevant, process_use)
4307 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
4308 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
4309 (vect_finish_stmt_generation_1, get_group_load_store_type)
4310 (get_load_store_type, vect_build_gather_load_calls)
4311 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4312 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4313 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4314 (vectorizable_store, vectorizable_load, vectorizable_condition)
4315 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
4316 (supportable_widening_operation): Likewise.
4317 (vect_get_vector_types_for_stmt): Likewise.
4318 * tree-vectorizer.h (vect_dr_behavior): Likewise.
4319
4320 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4321
4322 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4323 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
4324 (vect_permute_store_chain, vect_permute_load_chain)
4325 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
4326 repeated stmt_vec_info lookups.
4327 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
4328 (vect_update_ivs_after_vectorizer): Likewise.
4329 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
4330 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
4331 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
4332 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4333 (vect_bb_slp_scalar_cost): Likewise.
4334 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
4335
4336 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4337
4338 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
4339 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
4340 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
4341 to get gassigns and gcalls, rather than operating on generc gimple
4342 stmts.
4343 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
4344 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
4345 (vectorizable_load, vect_analyze_stmt): Likewise.
4346 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
4347
4348 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4349
4350 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
4351 return stmt_vec_infos rather than gimple stmts. Do not accept
4352 null arguments.
4353 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
4354 of a gimple stmt.
4355 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
4356 Update use of get_later_stmt.
4357 (vect_get_constant_vectors): Update call accordingly.
4358 (vect_schedule_slp_instance): Likewise
4359 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
4360 (vect_slp_analyze_instance_dependence): Likewise.
4361 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
4362
4363 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4364
4365 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
4366 (stmt_info_for_cost::stmt_info): ...this new field.
4367 (add_stmt_costs): Update accordingly.
4368 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
4369 (vect_get_known_peeling_cost): Likewise.
4370 (vect_estimate_min_profitable_iters): Likewise.
4371 * tree-vect-stmts.c (record_stmt_cost): Likewise.
4372
4373 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4374
4375 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
4376 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4377 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
4378 accordingly.
4379 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
4380
4381 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4382
4383 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
4384 a gimple stmt to a stmt_vec_info.
4385 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
4386
4387 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4388
4389 * tree-vectorizer.h (vec_info::grouped_stores): Change from
4390 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4391 (_loop_vec_info::reduction_chains): Likewise.
4392 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
4393 accordingly.
4394 * tree-vect-slp.c (vect_analyze_slp): Likewise.
4395
4396 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4397
4398 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
4399 a gimple stmt to a stmt_vec_info.
4400 (_stmt_vec_info::next_element): Likewise.
4401 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
4402 (vect_slp_analyze_and_verify_node_alignment)
4403 (vect_analyze_group_access_1, vect_analyze_group_access)
4404 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
4405 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
4406 (vect_supportable_dr_alignment): Update accordingly.
4407 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
4408 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
4409 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
4410 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
4411 * tree-vect-slp.c (vect_build_slp_tree_1)
4412 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
4413 (vect_split_slp_store_group, vect_analyze_slp_instance)
4414 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
4415 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
4416 (get_group_load_store_type, get_load_store_type)
4417 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
4418 (vect_transform_stmt, vect_remove_stores): Likewise.
4419
4420 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4421
4422 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
4423 than a gimple stmt.
4424 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
4425 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
4426 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
4427 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
4428 (vector_alignment_reachable_p, vect_get_data_access_cost)
4429 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
4430 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
4431 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
4432 (vect_analyze_group_access_1, vect_analyze_group_access)
4433 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
4434 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
4435 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
4436 result of vect_dr_stmt and use the stmt_vec_info instead of
4437 the associated gimple stmt.
4438 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
4439 (vect_gen_prolog_loop_niters): Likewise.
4440 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
4441
4442 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4443
4444 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
4445 to a vec<stmt_vec_info>.
4446 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
4447 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
4448 vec<stmt_vec_info>.
4449 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
4450 to a vec<stmt_vec_info>.
4451 (bst_traits::value_type, bst_traits::value_type): Likewise.
4452 (bst_traits::hash): Update accordingly.
4453 (vect_get_and_check_slp_defs): Change the stmts parameter from
4454 a vec<gimple *> to a vec<stmt_vec_info>.
4455 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
4456 (vect_build_slp_tree): Likewise.
4457 (vect_build_slp_tree_2): Likewise. Update uses of
4458 SLP_TREE_SCALAR_STMTS.
4459 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
4460 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
4461 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
4462 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
4463 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
4464 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
4465 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
4466 (vect_get_constant_vectors, vect_get_slp_defs)
4467 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
4468 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
4469 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
4470 instead of gimple stmts.
4471 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
4472 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
4473 (vect_slp_analyze_instance_dependence): Update uses of
4474 SLP_TREE_SCALAR_STMTS.
4475 (vect_slp_analyze_and_verify_node_alignment): Likewise.
4476 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
4477 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
4478 (get_initial_defs_for_reduction): Likewise.
4479 (vect_create_epilog_for_reduction): Likewise.
4480 (vectorize_fold_left_reduction): Likewise.
4481 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
4482 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
4483 (can_vectorize_live_stmts): Likewise.
4484
4485 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4486
4487 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
4488 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
4489 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
4490 than gimple stmts.
4491 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
4492 of a gimple stmt.
4493 (gather_scalar_reductions): Update after above interface changes.
4494 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
4495 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
4496 than gimple stmts.
4497 (vect_force_simple_reduction): Likewise.
4498 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
4499 LOOP_VINFO_REDUCTIONS.
4500 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
4501
4502 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4503
4504 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
4505 a gimple stmt to a stmt_vec_info.
4506 * tree-vect-loop.c (vect_active_double_reduction_p)
4507 (vect_force_simple_reduction, vectorizable_reduction): Update
4508 accordingly.
4509
4510 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4511
4512 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
4513 vec<gimple *> to a vec<stmt_vec_info>.
4514 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
4515 the reduction_phis argument from a vec<gimple *> to a
4516 vec<stmt_vec_info>.
4517 (vectorizable_reduction): Likewise the phis local variable that
4518 is passed to vect_create_epilog_for_reduction. Update for new type
4519 of SLP_TREE_VEC_STMTS.
4520 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
4521 (vectorizable_live_operation): Likewise.
4522 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
4523 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
4524
4525 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4526
4527 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
4528 a gimple stmt to a stmt_vec_info.
4529 (vectorizable_condition, vectorizable_live_operation)
4530 (vectorizable_reduction, vectorizable_induction): Pass back the
4531 vectorized statement as a stmt_vec_info.
4532 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
4533 use of STMT_VINFO_VEC_STMT.
4534 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
4535 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
4536 as stmt_vec_infos rather than gimple stmts.
4537 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
4538 to a stmt_vec_info.
4539 (vectorizable_live_operation): Likewise.
4540 (vectorizable_reduction, vectorizable_induction): Likewise,
4541 updating use of STMT_VINFO_VEC_STMT.
4542 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
4543 of STMT_VINFO_VEC_STMT.
4544 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
4545 (vectorizable_simd_clone_call, vectorizable_conversion)
4546 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4547 (vectorizable_store, vectorizable_load, vectorizable_condition)
4548 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
4549 from a gimple stmt to a stmt_vec_info.
4550 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
4551 pointer to a stmt_vec_info to the vectorizable_* routines.
4552
4553 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4554
4555 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
4556 a gimple stmt to a stmt_vec_info.
4557 (is_pattern_stmt_p): Update accordingly.
4558 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
4559 (vect_record_grouped_load_vectors): Likewise.
4560 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
4561 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
4562 (vect_model_reduction_cost): Likewise.
4563 (vect_create_epilog_for_reduction): Likewise.
4564 (vectorizable_reduction, vectorizable_induction): Likewise.
4565 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4566 Return the stmt_vec_info for the pattern statement.
4567 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
4568 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
4569 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
4570 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
4571 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
4572 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
4573 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
4574 (free_stmt_vec_info, vect_is_simple_use): Likewise.
4575
4576 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4577
4578 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
4579 (vect_finish_stmt_generation): Likewise.
4580 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
4581 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
4582 (vect_build_gather_load_calls): Use the return value of the above
4583 functions instead of a separate call to vinfo_for_stmt. Use narrow
4584 scopes for the input gimple stmt and wider scopes for the associated
4585 stmt_vec_info. Use vec_info::lookup_def when setting these
4586 stmt_vec_infos from an SSA_NAME definition.
4587 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
4588 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
4589 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
4590 (vectorizable_store, vectorizable_load, vectorizable_condition)
4591 (vectorizable_comparison): Likewise.
4592 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
4593 (vectorizable_reduction): Likewise.
4594
4595 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4596
4597 * tree-vectorizer.h (vect_is_simple_use): Add an optional
4598 stmt_vec_info * parameter before the optional gimple **.
4599 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4600 (process_use, vect_get_vec_def_for_operand_1): Update callers.
4601 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
4602 * tree-vect-loop.c (vectorizable_reduction): Likewise.
4603 (vectorizable_live_operation): Likewise.
4604 * tree-vect-patterns.c (type_conversion_p): Likewise.
4605 (vect_look_through_possible_promotion): Likewise.
4606 (vect_recog_rotate_pattern): Likewise.
4607 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
4608
4609 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4610
4611 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
4612 a typedef to a wrapper class.
4613 (NULL_STMT_VEC_INFO): New macro.
4614 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
4615 (stmt_vec_info::operator*): New function.
4616 (stmt_vec_info::operator gimple *): Likewise.
4617 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
4618 (add_stmt_costs): Likewise.
4619 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
4620 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
4621 (vect_get_known_peeling_cost): Likewise.
4622 (vect_estimate_min_profitable_iters): Likewise.
4623 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4624 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
4625 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
4626 (vectorizable_store, free_stmt_vec_infos): Likewise.
4627 (new_stmt_vec_info): Change return type of xcalloc to
4628 _stmt_vec_info *.
4629
4630 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4631
4632 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
4633 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
4634 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
4635 a single_imm_use-based sequence.
4636 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
4637
4638 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4639
4640 * tree-vectorizer.h (vec_info::lookup_def): Declare.
4641 * tree-vectorizer.c (vec_info::lookup_def): New function.
4642 * tree-vect-patterns.c (vect_get_internal_def): Use it.
4643 (vect_widened_op_tree): Likewise.
4644 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
4645 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
4646 (vectorizable_reduction): Likewise.
4647 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
4648 of a gimple *.
4649 (vect_is_slp_reduction): Update calls accordingly. Use
4650 vec_info::lookup_def.
4651 (vect_is_simple_reduction): Likewise
4652 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
4653
4654 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4655
4656 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
4657 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
4658 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
4659 of vinfo_for_stmt.
4660 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
4661 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
4662 (vect_update_vf_for_slp, vect_analyze_loop_operations)
4663 (vect_is_slp_reduction, vectorizable_induction)
4664 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
4665 * tree-vect-patterns.c (vect_init_pattern_stmt):
4666 (vect_determine_min_output_precision_1, vect_determine_precisions)
4667 (vect_pattern_recog): Likewise.
4668 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
4669 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
4670 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
4671 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
4672 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
4673 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
4674 info field from a loop to a loop_vec_info.
4675
4676 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4677
4678 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
4679 (vec_info::add_stmt): Declare.
4680 * tree-vectorizer.c (vec_info::add_stmt): New function.
4681 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
4682 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
4683 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
4684 (vectorizable_induction): Likewise.
4685 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
4686 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
4687 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
4688 (vectorizable_load): Likewise.
4689 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
4690 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
4691 (vect_recog_gather_scatter_pattern): Likewise.
4692 (append_pattern_def_seq): Likewise. Remove a check that is
4693 performed by add_stmt itself.
4694
4695 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4696
4697 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
4698 which make_ssa_name was called with new_stmt before new_stmt
4699 had been created.
4700
4701 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4702
4703 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
4704 split out from...
4705 (vect_is_slp_reduction): ...here...
4706 (vect_is_simple_reduction): ...and here. Remove repetition of tests
4707 that are already known to be false.
4708
4709 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4710
4711 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
4712 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
4713 STMT_VINFO_NUM_SLP_USES when it's true.
4714 (vect_free_slp_instance): Add a final_p parameter and pass it to
4715 vect_free_slp_tree.
4716 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
4717 (vect_analyze_slp_instance): Likewise.
4718 (vect_slp_analyze_operations): Likewise.
4719 (vect_slp_analyze_bb_1): Likewise.
4720 * tree-vectorizer.c (vec_info): Likewise.
4721 * tree-vect-loop.c (vect_transform_loop): Likewise.
4722
4723 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4724
4725 * tree-vect-loop.c (vectorizable_reduction): Assert that the
4726 function is not called for second and subsequent members of
4727 a reduction group.
4728
4729 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
4730
4731 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
4732 cases for nested loops from here to ...
4733 (vect_create_epilog_for_reduction): ...here. Only call
4734 vect_is_simple_use for inner-loop reductions.
4735
4736 2018-07-31 Martin Liska <mliska@suse.cz>
4737
4738 PR gcov-profile/85338
4739 PR gcov-profile/85350
4740 PR gcov-profile/85372
4741 * profile.c (struct location_triplet): New.
4742 (struct location_triplet_hash): Likewise.
4743 (output_location): Do not output a BB that
4744 is already recorded for a line.
4745 (branch_prob): Use streamed_locations.
4746
4747 2018-07-31 Martin Liska <mliska@suse.cz>
4748
4749 PR gcov-profile/85370
4750 * coverage.c (coverage_begin_function): Do not mark target
4751 clones as artificial functions.
4752
4753 2018-07-31 Martin Liska <mliska@suse.cz>
4754
4755 PR gcov-profile/83813
4756 PR gcov-profile/84758
4757 PR gcov-profile/85217
4758 PR gcov-profile/85332
4759 * profile.c (branch_prob): Do not record GOTO expressions
4760 for GIMPLE statements which locations are already streamed.
4761
4762 2018-07-31 Olivier Hainque <hainque@adacore.com>
4763
4764 * gcc.c (handle_spec_function): Accept a soft_matched_part
4765 argument, as do_spec_1. Pass it down to ...
4766 (eval_spec_function): Accept a soft_matched_part argument,
4767 and pass it down to ...
4768 (do_spec_2): Accept a soft_matched_part argument, and pass
4769 it down to do_spec_1.
4770 (do_spec_1): Pass soft_matched_part to handle_spec_function.
4771 (handle_braces): Update call to handle_spec_function.
4772 (driver::set_up_specs): Update calls to do_spec_2.
4773 (compare_debug_dump_opt_spec_function): Likewise.
4774 (compare_debug_self_opt_spec_function): Likewise.
4775
4776 2018-07-31 Olivier Hainque <hainque@adacore.com>
4777
4778 * common.opt (nolibc): New option.
4779 * doc/invoke.texi (Link Options): Document it.
4780 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
4781 * config/alpha/linux.h: Likewise.
4782 * config/arc/elf.h: Likewise.
4783 * config/arm/uclinux-elf.h: Likewise.
4784 * config/arm/unknown-elf.h: Likewise.
4785 * config/avr/avrlibc.h: Likewise.
4786 * config/bfin/bfin.h: Likewise.
4787 * config/bfin/linux.h: Likewise.
4788 * config/bfin/uclinux.h: Likewise.
4789 * config/darwin.h: Likewise.
4790 * config/darwin10.h: Likewise.
4791 * config/darwin12.h: Likewise.
4792 * config/gnu-user.h: Likewise.
4793 * config/lm32/uclinux-elf.h: Likewise.
4794 * config/pa/pa-hpux11.h: Likewise.
4795 * config/pa/pa64-hpux.h: Likewise.
4796 * config/sparc/sparc.h: Likewise.
4797
4798 2018-07-31 Olivier Hainque <hainque@adacore.com>
4799
4800 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
4801 undefined variables.
4802
4803 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
4804
4805 PR target/86640
4806 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
4807 instead of GEN_INT.
4808
4809 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
4810
4811 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
4812 terminated string literal.
4813
4814 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
4815
4816 PR rtl-optimization/85160
4817 * combine.c (is_just_move): New function.
4818 (try_combine): Allow combining two instructions into two if neither of
4819 the original instructions was a move.
4820
4821 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
4822
4823 PR target/86673
4824 * doc/extend.texi (Global Register Variables): Discourage use of type
4825 qualifiers.
4826 (Local Register Variables): Likewise.
4827
4828 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
4829
4830 PR tree-optimization/86506
4831 * hwint.h (ceil_log2): Resync with hwint.c implementation.
4832
4833 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
4834
4835 PR target/86547
4836 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
4837 hard_regno, make sure no insn between `from` and `to` clobbers it.
4838
4839 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
4840 Tom de Vries <tdevries@suse.de>
4841
4842 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
4843 (PTX_DEFAULT_RUNTIME_DIM): ... this.
4844 (nvptx_goacc_validate_dims): Set default worker and gang dims to
4845 PTX_DEFAULT_RUNTIME_DIM.
4846 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
4847
4848 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
4849
4850 * config/pa/pa.c (pa_output_addr_vec): Align address table.
4851 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
4852 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
4853
4854 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
4855
4856 * config/rs6000/constraints.md (wG constraint): Delete, no longer
4857 used.
4858 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
4859 predicate to reflect toc fusion has been deleted.
4860 (toc_fusion_mem_raw): Delete, no longer used.
4861 (toc_fusion_mem_wrapped): Likewise.
4862 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
4863 fusion mask bit.
4864 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
4865 Delete, no longer used.
4866 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
4867 meant to be used for toc fusion.
4868 (rs6000_debug_print_mode): Delete toc fusion debugging.
4869 (rs6000_debug_reg_global): Likewise.
4870 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
4871 fusion and secondary reload support that were never used.
4872 (rs6000_option_override_internal): Delete TOC fusion, that was only
4873 partially defined, and it did not work unless you also used the
4874 -mcmodel= switch.
4875 (rs6000_legitimate_address_p): Delete TOC fusion support.
4876 (rs6000_opt_masks): Likewise.
4877 (fusion_wrap_memory_address): Delete function, no longer used.
4878 (fusion_split_address); Delete TOC fusion support.
4879 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
4880 longer used with toc fusion being deleted.
4881 (TARGET_TOC_FUSION_FP): Likewise.
4882 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
4883 UNSPEC.
4884 (toc fusion spliter): Delete TOC fusion support.
4885 (toc_fusionload_<mode>): Likewise.
4886 (toc_fusionload_di): Likewise.
4887 (fusion_gpr_load_<mode>): Delete generator function, this insn no
4888 longer needs to be named. Rename predicate to delete TOC fusion.
4889 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
4890 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
4891 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
4892 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
4893 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
4894
4895 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
4896
4897 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
4898 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
4899 __int128 in built-in function prototypes.
4900 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
4901 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
4902
4903 2018-07-27 Martin Sebor <msebor@redhat.com>
4904
4905 PR tree-optimization/86696
4906 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
4907 types, including enums.
4908 (handle_char_store): Be prepared for the above function to fail.
4909
4910 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
4911
4912 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
4913 when optimization level is lower than 2 or optimize for size.
4914
4915 2018-07-26 Martin Sebor <msebor@redhat.com>
4916
4917 PR tree-optimization/86043
4918 PR tree-optimization/86042
4919 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
4920 (get_string_cst_length): Rename...
4921 (get_min_string_length): ...to this. Add argument.
4922 (handle_char_store): Extend to handle multi-character stores by
4923 MEM_REF.
4924 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
4925 * tree.h (initializer_zerop): Add argument.
4926
4927 2018-07-26 Jakub Jelinek <jakub@redhat.com>
4928
4929 PR middle-end/86660
4930 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
4931 declare target to variables if they have always,{to,from,tofrom} map
4932 kinds.
4933
4934 2018-07-26 Martin Liska <mliska@suse.cz>
4935
4936 PR lto/86548
4937 * lto-wrapper.c: Add linker_output as prefix
4938 for ltrans_output_file.
4939
4940 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
4941
4942 PR rtl-optimization/85805
4943 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
4944 value for hard registers if that was written in the same mode.
4945
4946 2018-07-26 Martin Liska <mliska@suse.cz>
4947
4948 PR gcov-profile/86536
4949 * gcov.c (format_gcov): Use printf format %.*f directly
4950 and do not handle special values.
4951
4952 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4953
4954 * common/config/arc/arc-common.c (arc_option_optimization_table):
4955 Update default optimizations for size.
4956
4957 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4958
4959 * config/arc/arc.md (movsf_insn): Add short instruction selection.
4960 * config/arc/constraints.md (CfZ): New constraint.
4961 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
4962 (subsf3_fpu): Likewise.
4963 (cmpsf_fpu): Likewise.
4964 (cmpsf_fpu_uneq): Likewise.
4965
4966 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4967
4968 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
4969 accesses as well.
4970 (arc_is_uncached_mem_p): uncached applies to both the variable and
4971 the pointer.
4972
4973 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
4974
4975 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
4976 register names.
4977
4978 2018-07-25 David Malcolm <dmalcolm@redhat.com>
4979
4980 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
4981 field "m_scopes" from vec to auto_vec.
4982
4983 2018-07-25 Martin Liska <mliska@suse.cz>
4984
4985 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
4986 return type.
4987
4988 2018-07-25 Richard Biener <rguenther@suse.de>
4989
4990 PR debug/86654
4991 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
4992 special wrt context_die late.
4993 (gen_subprogram_die): Re-use DIEs in local scope.
4994
4995 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
4996
4997 PR tree-optimization/86644
4998 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
4999
5000 2018-07-25 Martin Liska <mliska@suse.cz>
5001
5002 PR middle-end/86645
5003 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
5004 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
5005
5006 2018-07-25 Martin Liska <mliska@suse.cz>
5007
5008 PR sanitizer/79635
5009 * params.def: Explain ASan abbreviation and provide
5010 a documentation link.
5011
5012 2018-07-24 Martin Sebor <msebor@redhat.com>
5013
5014 PR tree-optimization/86622
5015 PR tree-optimization/86532
5016 * builtins.h (string_length): Declare.
5017 * builtins.c (c_strlen): Correct handling of non-constant offsets.
5018 (check_access): Be prepared for non-constant length ranges.
5019 (string_length): Make extern.
5020 * expr.c (string_constant): Only handle the minor non-constant
5021 array index. Use string_constant to compute the length of
5022 a generic string constant.
5023
5024 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
5025
5026 PR tree-optimization/86618
5027 * tree-vect-stmts.c (vectorizable_call): Don't take the address
5028 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
5029
5030 2018-07-24 David Malcolm <dmalcolm@redhat.com>
5031
5032 PR tree-optimization/86636
5033 * json.cc (json::object::set): Fix comment. Add assertions.
5034 (json::array::append): Move here from json.h. Add comment and an
5035 assertion.
5036 (json::string::string): Likewise.
5037 * json.h (json::array::append): Move to json.cc.
5038 (json::string::string): Likewise.
5039 * optinfo-emit-json.cc
5040 (optrecord_json_writer::impl_location_to_json): Assert that we
5041 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
5042 wrapper around it. Expand the location once, rather than three
5043 times.
5044 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
5045 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
5046 wrappers.
5047 (optrecord_json_writer::optinfo_to_json): Likewise, in four
5048 places. Fix some overlong lines.
5049
5050 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
5051
5052 * config/aarch64/aarch64-simd.md
5053 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
5054 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
5055 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
5056 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
5057 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
5058 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
5059 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
5060 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
5061 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
5062
5063 2018-07-24 Jakub Jelinek <jakub@redhat.com>
5064
5065 PR middle-end/86627
5066 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
5067 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
5068 and abs_d == d, do the power of two handling if profitable.
5069
5070 2018-07-24 Richard Biener <rguenther@suse.de>
5071
5072 * match.pd: Add BIT_FIELD_REF canonicalizations.
5073
5074 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
5075
5076 PR c/86617
5077 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
5078
5079 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
5080
5081 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
5082 terminated STRING_CST object.
5083
5084 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
5085
5086 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
5087
5088 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
5089
5090 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
5091 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
5092 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
5093 the elements into a register.
5094 (rs6000_split_v4si_init_di_reg): Delete.
5095 (rs6000_split_v4si_init): Delete.
5096 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
5097 (vsx_init_v4si): Rewrite as a define_expand.
5098
5099 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
5100
5101 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
5102 zero_extend argument from memory): New.
5103
5104 2018-07-22 Martin Sebor <msebor@redhat.com>
5105
5106 PR bootstrap/86621
5107 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
5108 diagnosing calls with unknown arguments unless -Walloca-larger-than
5109 is restricted to less than PTRDIFF_MAX bytes.
5110
5111 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
5112
5113 * doc/gcov.texi (Invoking Gcov): Editorial changes.
5114
5115 2018-07-20 David Malcolm <dmalcolm@redhat.com>
5116
5117 * pretty-print.c (text_info::set_location): Remove redundant
5118 "line_table" parameter from call to rich_location::set_range.
5119
5120 2018-07-20 Martin Sebor <msebor@redhat.com>
5121
5122 PR middle-end/82063
5123 * builtins.c (expand_builtin_alloca): Adjust.
5124 * calls.c (alloc_max_size): Simplify.
5125 * cgraphunit.c (cgraph_node::expand): Adjust.
5126 * common.opt (larger_than_size, warn_frame_larger_than): Remove
5127 variables.
5128 (frame_larger_than_size): Same.
5129 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
5130 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
5131 Initialize.
5132 * doc/invoke.texi (GCC Command Options): Document option arguments.
5133 Explain byte-size arguments and suffixes.
5134 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
5135 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
5136 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
5137 * doc/options.texi (UInteger): Expand.
5138 (Host_Wide_Int, ByteSize): Document new properties.
5139 * final.c (final_start_function_1): Include sizes in an error message.
5140 * function.c (frame_offset_overflow): Same.
5141 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
5142 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
5143 Diagnose unbounded alloca calls only for limits of less than
5144 PTRDIFF_MAX.
5145 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
5146 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
5147 for alloca(0).
5148 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
5149 only for limits of less than PTRDIFF_MAX.
5150 * langhooks-def.h (lhd_handle_option): Change function argument
5151 to HOST_WIDE_INT.
5152 * langhooks.c (lhd_handle_option): Same.
5153 * langhooks.h (handle_option): Same.
5154 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
5155 ByteSize flags.
5156 (var_type, var_type_struct): Same.
5157 (var_set): Handle ByteSize flag.
5158 * optc-gen.awk: Add comments to output to ease debugging. Make
5159 use of HOST_WIDE_INT where appropriate.
5160 * opts-gen-save.awk: Use %lx to format unsigned long.
5161 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
5162 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
5163 arguments. Parse bytes-size suffixes.
5164 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
5165 (enum_value_to_arg): Same.
5166 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
5167 (handle_option): Adjust.
5168 (generate_option): Change function argument to HOST_WIDE_INT.
5169 (cmdline_handle_error): Adjust.
5170 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
5171 (set_option): Change function argument to HOST_WIDE_INT.
5172 (option_enabled): Handle cl_host_wide_int.
5173 (get_option_state): Handle CLVC_SIZE.
5174 (control_warning_option): Same.
5175 * opts.c (common_handle_option): Change function argument to
5176 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
5177 OPT_Wvla_larger_than_.
5178 * opts.h (enum cl_var_type): Add an enumerator.
5179 * stor-layout.c (layout_decl): Print a more meaningful warning.
5180 * toplev.c (output_stack_usage): Adjust.
5181
5182 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
5183
5184 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
5185 call to inline_expand_builtin_string_cmp.
5186 (expand_builtin_strcmp): Likewise.
5187 (expand_builtin_strncmp): Likewise.
5188 (inline_string_cmp): Delete the last parameter, change char_type_node
5189 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
5190 two operands.
5191 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
5192 the inlining expansion on target where the type of the call has same or
5193 narrower precision than unsigned char.
5194
5195 2018-07-20 David Malcolm <dmalcolm@redhat.com>
5196
5197 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
5198 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
5199 * common.opt (fsave-optimization-record): New option.
5200 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
5201 * doc/invoke.texi (-fsave-optimization-record): New option.
5202 * dumpfile.c: Include "optinfo-emit-json.h".
5203 (struct kv_pair): Move to coretypes.h.
5204 (optgroup_options): Make non-static.
5205 (dump_context::end_scope): Call
5206 optimization_records_maybe_pop_dump_scope.
5207 * dumpfile.h (optgroup_options): New decl.
5208 * json.cc: New file.
5209 * json.h: New file.
5210 * optinfo-emit-json.cc: New file.
5211 * optinfo-emit-json.h: New file.
5212 * optinfo.cc: Include "optinfo-emit-json.h".
5213 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
5214 (optinfo_enabled_p): Check optimization_records_enabled_p.
5215 (optinfo_wants_inlining_info_p): Likewise.
5216 * optinfo.h: Update comment.
5217 * profile-count.c (profile_quality_as_string): New function.
5218 * profile-count.h (profile_quality_as_string): New decl.
5219 (profile_count::quality): New accessor.
5220 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
5221 and optinfo_emit_json_cc_tests.
5222 * selftest.h (selftest::json_cc_tests): New decl.
5223 (selftest::optinfo_emit_json_cc_tests): New decl.
5224 * toplev.c: Include "optinfo-emit-json.h".
5225 (compile_file): Call optimization_records_finish.
5226 (do_compile): Call optimization_records_start.
5227 * tree-ssa-live.c: Include optinfo.h.
5228 (remove_unused_scope_block_p): Retain inlining information if
5229 optinfo_wants_inlining_info_p returns true.
5230
5231 2018-07-20 Richard Biener <rguenther@suse.de>
5232
5233 PR debug/86585
5234 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
5235 to cover -flto-partition=none.
5236
5237 2018-07-20 Martin Liska <mliska@suse.cz>
5238
5239 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
5240 (get_decl_source_range): Remove unused function.
5241
5242 2018-07-20 Richard Biener <rguenther@suse.de>
5243
5244 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
5245 (struct vn_phi_s): Likewise.
5246 (struct vn_reference_s): Likewise.
5247 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
5248 for searching the slot of an entry known to be in the hash itself.
5249 (vn_phi_hasher::equal): Likewise.
5250 (vn_reference_hasher::equal): Likewise.
5251 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
5252 globals.
5253 (optimistic_info, current_info): Remove, keeping only valid_info.
5254 (vn_reference_lookup_1): Remove fallback lookup.
5255 (vn_reference_lookup_2): Likewise.
5256 (vn_nary_op_lookup_1): Likewise.
5257 (vn_phi_lookup): Likewise.
5258 (vn_nary_build_or_lookup_1): Make sure to not chain the built
5259 hash element.
5260 (vn_reference_insert): Adjust, chain the inserted hash element
5261 at last_inserted_ref.
5262 (vn_reference_insert_pieces): Likewise.
5263 (visit_reference_op_call): Likewise.
5264 (vn_nary_op_insert_into): Chain the inserted hash element at
5265 last_inserted_nary.
5266 (vn_nary_op_insert_pieces): Adjust.
5267 (vn_nary_op_insert): Likewise.
5268 (vn_nary_op_insert_stmt): Likewise.
5269 (vn_phi_insert): Adjust, chain the inserted hash element at
5270 last_inserted_phi.
5271 (process_scc): Remove clearing and copying the optimistic
5272 table. Instead remove elements inserted during an optimistic
5273 iteration from the single table we maintain.
5274 (init_scc_vn): Adjust.
5275 (free_scc_vn): Likewise.
5276 (sccvn_dom_walker::record_cond): Likewise.
5277 (sccvn_dom_walker::after_dom_children): Likewise.
5278
5279 2018-07-19 Martin Sebor <msebor@redhat.com>
5280
5281 PR tree-optimization/84047
5282 PR tree-optimization/83776
5283 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
5284 (check_array_bounds): Call it.
5285
5286 2018-07-19 Martin Sebor <msebor@redhat.com>
5287
5288 * align.h (align_flags): Use member initialization.
5289
5290 2018-07-19 David Malcolm <dmalcolm@redhat.com>
5291
5292 * Makefile.in (OBJS): Add optinfo.o.
5293 * coretypes.h (class symtab_node): New forward decl.
5294 (struct cgraph_node): New forward decl.
5295 (class varpool_node): New forward decl.
5296 * dump-context.h: New file.
5297 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
5298 "tree-pass.h".
5299 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
5300 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
5301 (set_alt_dump_file): Likewise.
5302 (dump_context::~dump_context): New dtor.
5303 (dump_gimple_stmt): Move implementation to...
5304 (dump_context::dump_gimple_stmt): ...this new member function.
5305 Add the stmt to any pending optinfo, creating one if need be.
5306 (dump_gimple_stmt_loc): Move implementation to...
5307 (dump_context::dump_gimple_stmt_loc): ...this new member function.
5308 Start a new optinfo and add the stmt to it.
5309 (dump_gimple_expr): Move implementation to...
5310 (dump_context::dump_gimple_expr): ...this new member function.
5311 Add the stmt to any pending optinfo, creating one if need be.
5312 (dump_gimple_expr_loc): Move implementation to...
5313 (dump_context::dump_gimple_expr_loc): ...this new member function.
5314 Start a new optinfo and add the stmt to it.
5315 (dump_generic_expr): Move implementation to...
5316 (dump_context::dump_generic_expr): ...this new member function.
5317 Add the tree to any pending optinfo, creating one if need be.
5318 (dump_generic_expr_loc): Move implementation to...
5319 (dump_context::dump_generic_expr_loc): ...this new member
5320 function. Add the tree to any pending optinfo, creating one if
5321 need be.
5322 (dump_printf): Move implementation to...
5323 (dump_context::dump_printf_va): ...this new member function. Add
5324 the text to any pending optinfo, creating one if need be.
5325 (dump_printf_loc): Move implementation to...
5326 (dump_context::dump_printf_loc_va): ...this new member function.
5327 Start a new optinfo and add the stmt to it.
5328 (dump_dec): Move implementation to...
5329 (dump_context::dump_dec): ...this new member function. Add the
5330 value to any pending optinfo, creating one if need be.
5331 (dump_context::dump_symtab_node): New member function.
5332 (dump_context::get_scope_depth): New member function.
5333 (dump_context::begin_scope): New member function.
5334 (dump_context::end_scope): New member function.
5335 (dump_context::ensure_pending_optinfo): New member function.
5336 (dump_context::begin_next_optinfo): New member function.
5337 (dump_context::end_any_optinfo): New member function.
5338 (dump_context::s_current): New global.
5339 (dump_context::s_default): New global.
5340 (dump_scope_depth): Delete global.
5341 (dumpfile_ensure_any_optinfo_are_flushed): New function.
5342 (dump_symtab_node): New function.
5343 (get_dump_scope_depth): Reimplement in terms of dump_context.
5344 (dump_begin_scope): Likewise.
5345 (dump_end_scope): Likewise.
5346 (selftest::temp_dump_context::temp_dump_context): New ctor.
5347 (selftest::temp_dump_context::~temp_dump_context): New dtor.
5348 (selftest::verify_item): New function.
5349 (ASSERT_IS_TEXT): New macro.
5350 (ASSERT_IS_TREE): New macro.
5351 (ASSERT_IS_GIMPLE): New macro.
5352 (selftest::test_capture_of_dump_calls): New test.
5353 (selftest::dumpfile_c_tests): Call it.
5354 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
5355 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
5356 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
5357 descriptive comment.
5358 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
5359 (dump_node, dump_bb): Move these unrelated decls.
5360 (class dump_manager): Add leading comment.
5361 * optinfo.cc: New file.
5362 * optinfo.h: New file.
5363
5364 2018-07-19 Michael Collison <michael.collison@arm.com>
5365 Richard Henderson <rth@redhat.com>
5366
5367 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
5368 (subti): Handle op1 zero.
5369 (subvti4, usub4ti4): New.
5370 (*sub<GPI>3_compare1_imm): New.
5371 (sub<GPI>3_carryinCV): New.
5372 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
5373 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
5374
5375 2018-07-19 Michael Collison <michael.collison@arm.com>
5376 Richard Henderson <rth@redhat.com>
5377
5378 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
5379 (addti3): Create simpler code if low part is already known to be 0.
5380 (addvti4, uaddvti4): New.
5381 (*add<GPI>3_compareC_cconly_imm): New.
5382 (*add<GPI>3_compareC_cconly): New.
5383 (*add<GPI>3_compareC_imm): New.
5384 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
5385 handle constants within this pattern..
5386 (*add<GPI>3_compareV_cconly_imm): New.
5387 (*add<GPI>3_compareV_cconly): New.
5388 (*add<GPI>3_compareV_imm): New.
5389 (add<GPI>3_compareV): New.
5390 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
5391 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
5392 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
5393 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
5394 with 'comparison' operator.
5395 (*add<GPI>3_compareV_cconly_imm): Ditto.
5396 (*add<GPI>3_compareV_cconly): Ditto.
5397 (*add<GPI>3_compareV_imm): Ditto.
5398 (add<GPI>3_compareV): Ditto.
5399 (add<mode>3_carryinC): Ditto.
5400 (*add<mode>3_carryinC_zero): Ditto.
5401 (*add<mode>3_carryinC): Ditto.
5402 (add<mode>3_carryinV): Ditto.
5403 (*add<mode>3_carryinV_zero): Ditto.
5404 (*add<mode>3_carryinV): Ditto.
5405
5406 2018-07-19 Michael Collison <michael.collison@arm.com>
5407 Richard Henderson <rth@redhat.com>
5408
5409 * config/aarch64/aarch64-modes.def (CC_V): New.
5410 * config/aarch64/aarch64-protos.h
5411 (aarch64_addti_scratch_regs): Declare
5412 (aarch64_subvti_scratch_regs): Declare.
5413 (aarch64_expand_subvti): Declare.
5414 (aarch64_gen_unlikely_cbranch): Declare
5415 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
5416 for signed overflow using CC_Vmode.
5417 (aarch64_get_condition_code_1): Handle CC_Vmode.
5418 (aarch64_gen_unlikely_cbranch): New function.
5419 (aarch64_addti_scratch_regs): New function.
5420 (aarch64_subvti_scratch_regs): New function.
5421 (aarch64_expand_subvti): New function.
5422
5423 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5424
5425 * config/aarch64/aarch64-option-extensions.def: New entry for profile
5426 extension.
5427 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
5428 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
5429 extension.
5430
5431 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5432
5433 PR target/83009
5434 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
5435 address check not strict.
5436
5437 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
5438
5439 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
5440 Umq with Umn.
5441 (store_pair_lanes<mode>): Likewise.
5442 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
5443 enum value 'ADDR_QUERY_LDP_STP_N'.
5444 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
5445 (aarch64_print_address_internal): Add declaration.
5446 (aarch64_print_ldpstp_address): Remove.
5447 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
5448 (aarch64_print_operand): Change printing of 'y'.
5449 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
5450 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
5451 'true' rather than '1'.
5452 * config/aarch64/constraints.md (Uml): Likewise.
5453 (Uml): Rename to Umn.
5454 (Umq): Remove.
5455
5456 2018-07-19 Richard Biener <rguenther@suse.de>
5457
5458 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
5459 a trailing array.
5460 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
5461 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
5462 (vn_reference_hasher): Likewise.
5463 (struct vn_tables_s): Remove obstack and alloc-pool members.
5464 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
5465 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
5466 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
5467 (vn_reference_insert_pieces): Likewise.
5468 (alloc_vn_nary_op_noinit): Adjust.
5469 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
5470 (vn_phi_eq): Adjust.
5471 (shared_lookup_phiargs): Remove.
5472 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
5473 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
5474 (visit_reference_op_call): Likewise.
5475 (copy_nary, copy_phi, copy_reference): Remove.
5476 (process_scc): Rewind the obstack when iterating. Do not
5477 copy the elements to valid_info but just move them from one
5478 hashtable to the other.
5479 (allocate_vn_table): Adjust.
5480 (free_vn_table): Likewise.
5481 (init_scc_vn): Likewise.
5482 (free_scc_vn): Likewise.
5483
5484 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
5485
5486 PR target/86560
5487 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
5488 indirect_return as function type attribute.
5489 (ix86_attribute_table): Change indirect_return to function
5490 type attribute.
5491 * doc/extend.texi: Update indirect_return attribute.
5492
5493 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
5494
5495 * wide-int.h (widest2_int): New.
5496 * gimple-fold.c (arith_overflowed_p): Use it.
5497 * tree.h (widest2_int_cst): New.
5498 * tree-vrp.c (wide_int_binop_overflow): Rename from
5499 vrp_int_const_binop.
5500 Rewrite to work on trees.
5501 (extract_range_from_multiplicative_op_1): Abstract code to...
5502 (wide_int_range_min_max): ...here.
5503 (wide_int_range_cross_product): ...and here.
5504 (extract_range_from_binary_expr_1): Abstract overflow code to...
5505 (wide_int_range_mult_wrapping): ...here.
5506 * tree-vrp.h (wide_int_range_cross_product): New.
5507 (wide_int_range_mult_wrapping): New.
5508
5509 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
5510 Julia Koval <julia.koval@intel.com>
5511
5512 * config/i386/x86-tune-costs.h (skylake_memcpy,
5513 skylake_memset): Replace rep_prefix with unrolling for size 512.
5514
5515 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
5516
5517 PR middle-end/86544
5518 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
5519 comparision with EQ_EXPR in last stmt.
5520
5521 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
5522
5523 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
5524 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
5525 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
5526 previously known as "PowerPC AltiVec Built-in Functions". Move
5527 some material to new subsubsections "PowerPC AltiVec Built-in
5528 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
5529 ISA 2.07".
5530 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
5531 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
5532 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
5533 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
5534
5535 2018-07-18 Richard Biener <rguenther@suse.de>
5536
5537 PR tree-optimization/86557
5538 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
5539 EXACT_DIV_EXPR.
5540
5541 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
5542
5543 * config/s390/s390.c (s390_function_profiler): Generate CFI.
5544
5545 2018-07-17 Jeff Law <law@redhat.com>
5546
5547 * config/arm/arm.c (get_label_padding): Update for recent
5548 changes to label_to_alignment.
5549
5550 PR tree-optimization/86010
5551 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
5552
5553 * config/mips/mips.c (vr4130_align_insns): Update for recent
5554 changes to label_to_alignment.
5555
5556 * config/frv/frv.c (frv_label_align): Update for recent changes
5557 to label_to_alignment.
5558
5559 * config/nios2/nios2.c (nios2_label_align): Update for recent
5560 changes which dropped ALIGN_LABELS_LOG.
5561
5562 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
5563
5564 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
5565 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
5566
5567 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
5568
5569 * config/arc/arc.c (arc_label_align): Use align_labels instead of
5570 deprecated align_labels_log.
5571
5572 2018-07-17 Richard Biener <rguenther@suse.de>
5573
5574 PR lto/86456
5575 * dwarf2out.c (init_sections_and_labels): Always generate
5576 a debug_line_str_section for early LTO debug.
5577 (dwarf2out_finish): Reset debug_line_str_hash output early.
5578 Bump counter for extra dwarf5 .debug_loc labels to not conflict
5579 with fat LTO part.
5580 (dwarf2out_early_finish): Output debug_line_str.
5581
5582 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
5583
5584 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
5585 index register on z196 or later.
5586
5587 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
5588
5589 * config/s390/s390.c (s390_default_align): Set default function
5590 alignment to 16.
5591 (s390_override_options_after_change): Call s390_default align.
5592 (s390_option_override_internal): Call s390_default_align.
5593 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
5594
5595 2018-07-17 Jakub Jelinek <jakub@redhat.com>
5596
5597 PR middle-end/86542
5598 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
5599 to _looptemp_ clauses, other than the first two.
5600
5601 2018-07-17 Martin Liska <mliska@suse.cz>
5602
5603 * opts.c: Do not enable OPT_falign_* for -Os.
5604
5605 2018-07-17 Martin Liska <mliska@suse.cz>
5606
5607 * align.h (MAX_CODE_ALIGN): New.
5608 (MAX_CODE_ALIGN_VALUE): New.
5609 * common/config/i386/i386-common.c (ix86_handle_option):
5610 (MAX_CODE_ALIGN): Moved to align.h.
5611 * final.c (MAX_CODE_ALIGN): Likewise.
5612 * opts.c (parse_and_check_align_values):
5613 (MAX_CODE_ALIGN): Likewise.
5614 (MAX_CODE_ALIGN_VALUE): Likewise.
5615
5616 2018-07-17 Martin Liska <mliska@suse.cz>
5617
5618 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
5619 in order to fulfil coding style.
5620 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
5621 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5622 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
5623 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
5624 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
5625 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
5626 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
5627 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
5628 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
5629
5630 2018-07-17 Martin Liska <mliska@suse.cz>
5631
5632 * align.h: New file.
5633 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
5634 directly.
5635 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
5636 align_flags of label_to_alignment.
5637 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
5638 align_flags class.
5639 * config/m68k/m68k.c: Do not use removed align_labels_value and
5640 align_loops_value.
5641 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
5642 (LOOP_ALIGN): Likewise.
5643 (LABEL_ALIGN): Likewise.
5644 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
5645 Remove not used macro.
5646 (rs6000_loop_align): Change return type to align_flags.
5647 (rs6000_loop_align_max_skip): Remove.
5648 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
5649 Change return type to align_flags.
5650 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
5651 Remove not used macro.
5652 (rs6000_loop_align): Change return type to align_flags.
5653 (rs6000_loop_align_max_skip): Remove.
5654 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
5655 * config/rx/rx-protos.h (rx_align_for_label): Make it
5656 static function.
5657 * config/rx/rx.c (rx_align_for_label): Change return type
5658 to align_flags.
5659 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
5660 macro definitions.
5661 into align_flags class.
5662 (LABEL_ALIGN): Likewise.
5663 (LOOP_ALIGN): Likewise.
5664 * config/s390/s390.c (s390_label_align): Use align_flags
5665 class member.
5666 (s390_asm_output_function_label): Likewise.
5667 * config/sh/sh.c (sh_override_options_after_change):
5668 Use align_flags class directly without macros.
5669 (find_barrier): Likewise.
5670 (barrier_align): Likewise.
5671 (sh_loop_align): Likewise.
5672 * config/spu/spu.c (spu_option_override):
5673 Use align_flags_tuple::get_value instead of removed macros.
5674 (spu_sched_init): Likewise.
5675 * config/spu/spu.h (GTY): Likewise.
5676 * config/visium/visium.c (visium_option_override):
5677 Set "8" as default secondary alignment.
5678 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
5679 in order to guarantee secondary alignment of 8.
5680 * coretypes.h: Include align.h header file.
5681 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
5682 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
5683 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
5684 * doc/tm.texi.in: Likewise.
5685 * final.c (struct label_alignment): Remove not used structure.
5686 (LABEL_ALIGN): Change type to align_flags.
5687 (LOOP_ALIGN): Likewise.
5688 (JUMP_ALIGN): Likewise.
5689 (default_loop_align_max_skip): Remove.
5690 (default_label_align_max_skip): Likewise.
5691 (default_jump_align_max_skip): Likewise.
5692 (default_label_align_after_barrier_max_skip):
5693 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
5694 (LABEL_TO_MAX_SKIP): Remove.
5695 (label_to_alignment): Return align_flags type instead of integer.
5696 (label_to_max_skip): Remove.
5697 (align_fuzz): Use align_flags type.
5698 (compute_alignments): Use align_flags type and use align_flags::max
5699 to combine multiple alignments.
5700 (grow_label_align): Grow vec instead of C array.
5701 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
5702 (shorten_branches): Use align_flags type and use align_flags::max
5703 to combine multiple alignments.
5704 (final_scan_insn_1): Remove usage of secondary alignment that comes
5705 from label alignment, but instead use proper secondary alignment
5706 which is computed in grow_label_align.
5707 * flags.h (struct align_flags_tuple): Move to align.h.
5708 (struct align_flags): Likewise.
5709 (state_align_loops): Rename to align_loops.
5710 (state_align_jumps): Rename to align_jumps.
5711 (state_align_labels): Rename to align_labels.
5712 (state_align_functions): Rename to align_functions.
5713 (align_loops_log): Remove.
5714 (align_jumps_log): Remove.
5715 (align_labels_log): Remove.
5716 (align_functions_log): Remove.
5717 (align_loops_max_skip): Remove.
5718 (align_jumps_max_skip): Remove.
5719 (align_labels_max_skip): Remove.
5720 (align_functions_max_skip): Remove.
5721 (align_loops_value): Remove.
5722 (align_jumps_value): Remove.
5723 (align_labels_value): Remove.
5724 (align_functions_value): Remove.
5725 * output.h (label_to_alignment): Change return type to align_flags.
5726 (label_to_max_skip): Remove.
5727 * target.def: Remove loop_align_max_skip, label_align_max_skip,
5728 jump_align_max_skip macros.
5729 * targhooks.h (default_loop_align_max_skip): Remove.
5730 (default_label_align_max_skip): Likewise.
5731 (default_jump_align_max_skip): Likewise.
5732 (default_label_align_after_barrier_max_skip): Remove.
5733 * toplev.c (read_log_maxskip): Use ::normalize function.
5734 (parse_N_M): Remove not used argument and also call ::normalize.
5735 (parse_alignment_opts): Do not pass unused arguments.
5736 * varasm.c (assemble_start_function): Use directly align_functions
5737 instead of removed macros.
5738 * system.h: Do not poison removed macros.
5739
5740 2018-07-17 Jakub Jelinek <jakub@redhat.com>
5741
5742 PR middle-end/86539
5743 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
5744 and cond temporaries don't have reference type if iterator has
5745 pointer type. For init use &for_pre_body instead of pre_p if
5746 for_pre_body is non-empty.
5747
5748 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5749
5750 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
5751 double-double modes to SFmode directly directly.
5752 (trunc<mode>sf2_fprs): Delete.
5753
5754 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5755
5756 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
5757 for conversions between IFmode and the decimal floating point modes.
5758 (init_float128_ieee): Use the correct names for conversions between
5759 KFmode and the decimal floating point modes.
5760
5761 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
5762
5763 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
5764 for the conversions between TDmode and IFmode.
5765 (init_float128_ieee): Use more correct names for the conversions
5766 between TDmode and KFmode.
5767
5768 2018-07-16 Jakub Jelinek <jakub@redhat.com>
5769
5770 PR tree-optimization/86526
5771 * builtins.c (expand_builtin_memcmp): Formatting fixes.
5772 (inline_expand_builtin_string_cmp): Likewise.
5773 (inline_string_cmp): Likewise. Use c_readstr instead of
5774 builtin_memcpy_read_str. Add unit_mode temporary.
5775
5776 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
5777
5778 PR middle-end/86528
5779 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
5780 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
5781
5782 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
5783
5784 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
5785 Alphabetize prototypes of built-in functions, separating out
5786 built-in functions that are listed in this section but should be
5787 described elsewhere.
5788
5789 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
5790
5791 PR target/86511
5792 * expmed.c (emit_store_flag): Do not emit setcc followed by a
5793 conditional move when trapping comparison was split to a
5794 non-trapping one (and vice versa).
5795
5796 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5797
5798 * config/s390/s390.c (s390_function_profiler): Generate nops
5799 instead of profiler call sequences.
5800 * config/s390/s390.opt: Add the new option.
5801
5802 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5803
5804 * config/s390/s390.c (s390_function_profiler): Generate
5805 __mcount_loc section.
5806 * config/s390/s390.opt: Add the new option.
5807
5808 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
5809
5810 * common.opt: Add the new warning.
5811 * config/s390/s390.c (s390_function_profiler): Emit "brasl
5812 %r0,__fentry__" when -mfentry is specified.
5813 (s390_option_override_internal): Disallow -mfentry for 31-bit
5814 CPUs.
5815 * config/s390/s390.opt: Add the new option.
5816
5817 2018-07-16 Richard Biener <rguenther@suse.de>
5818
5819 PR lto/86523
5820 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
5821 for function-local FUNCTION_DECL and RESULT_DECL immediately.
5822
5823 2018-07-16 Martin Liska <mliska@suse.cz>
5824
5825 PR ipa/86529
5826 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
5827 to ::get_create.
5828
5829 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
5830
5831 * config/arc/arcHS.md: Update ARCHS scheduling rules.
5832
5833 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
5834
5835 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
5836 for ARCHS4x.
5837 * config/arc/arc-cpus.def (hs4x): New cpu.
5838 (hs4xd): Likewise.
5839 * config/arc/arc-tables.opt: Regenerate.
5840 * config/arc/arc.c (arc_sched_issue_rate): New function.
5841 (TARGET_SCHED_ISSUE_RATE): Define.
5842 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
5843 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
5844 fpu_cvt.
5845 (attr tune): Add ARCHS4x tune values.
5846 (attr tune_dspmpy): Define.
5847 (*tst): Correct instruction type.
5848 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
5849 * config/arc/arcHS4x.md: New file.
5850 * config/arc/fpu.md: Update instruction type attributes.
5851 * config/arc/t-multilib: Regenerate.
5852
5853 2018-07-16 Tom de Vries <tdevries@suse.de>
5854
5855 PR debug/86455
5856 * var-tracking.c (vt_initialize): Fix pre_dec handling.
5857
5858 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
5859
5860 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
5861 early clobber.
5862
5863 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
5864
5865 PR tree-optimization/86514
5866 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
5867 conversion to a boolean type from a type with greater precision.
5868
5869 2018-07-16 Tom de Vries <tdevries@suse.de>
5870
5871 * var-tracking.c (vt_initialize): Print adjusted insn slim if
5872 dump_flags request TDF_SLIM.
5873
5874 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
5875
5876 * fold-const.c (int_const_binop_1): Abstract...
5877 (wide_int_binop): ...wide int code here.
5878 (poly_int_binop): ...poly int code here.
5879 Abstract the rest of int_const_binop_1 into int_const_binop.
5880 * fold-const.h (wide_int_binop): New.
5881 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
5882 Remove useless PLUS/MINUS_EXPR case.
5883 (zero_nonzero_bits_from_vr): Move wide int code...
5884 (zero_nonzero_bits_from_bounds): ...here.
5885 (extract_range_from_binary_expr_1): Move mask optimization code...
5886 (range_easy_mask_min_max): ...here.
5887 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
5888 (range_easy_mask_min_max): New.
5889
5890 2018-07-15 Jeff Law <law@redhat.com>
5891
5892 PR target/85993
5893 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
5894 block.
5895
5896 2018-07-14 Jim Wilson <jimw@sifive.com>
5897
5898 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
5899
5900 2018-07-14 Paul Koning <ni1d@arrl.net>
5901
5902 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
5903
5904 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
5905
5906 * lto-streamer-out.c (copy_function_or_variable): Dump info about
5907 copying section.
5908
5909 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
5910 Steve Munroe <munroesj52@gmail.com>
5911
5912 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
5913 (_mm_andnot_si128): Likewise.
5914 (_mm_or_si128): Likewise.
5915 (_mm_xor_si128): Likewise.
5916
5917 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
5918
5919 PR middle-end/78809
5920 * builtins.c (expand_builtin_memcmp): Inline the calls first
5921 when result_eq is false.
5922 (expand_builtin_strcmp): Inline the calls first.
5923 (expand_builtin_strncmp): Likewise.
5924 (inline_string_cmp): New routine. Expand a string compare
5925 call by using a sequence of char comparison.
5926 (inline_expand_builtin_string_cmp): New routine. Inline expansion
5927 a call to str(n)cmp/memcmp.
5928 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
5929 New option.
5930 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
5931
5932 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
5933
5934 * config/arm/driver-arm.c: Include arm-native.h.
5935 (host_detect_local_cpu): Use auto-generated data tables.
5936 (vendors, arm_cpu_table): Delete. Move part information to ...
5937 * config/arm/arm-cpus.in: ... here.
5938 * config/arm/parsecpu.awk (gen_native): New function.
5939 (vendor, part): New CPU fields.
5940 (END): Add support for building the native CPU detection tables.
5941 * config/arm/t-arm (arm-native.h): Add build rule.
5942 (driver-arm.o): Add dependency on arm-native.h.
5943
5944 2018-07-13 Richard Biener <rguenther@suse.de>
5945
5946 PR middle-end/85974
5947 * match.pd (addr1 - addr2): Allow either of the operand to
5948 have a conversion.
5949
5950 2018-07-13 Tom de Vries <tdevries@suse.de>
5951
5952 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
5953 in remap_ssa_name.
5954
5955 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
5956
5957 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
5958 arrays instead of numbered variables.
5959
5960 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
5961
5962 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
5963 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
5964
5965 2018-07-13 Richard Biener <rguenther@suse.de>
5966
5967 PR debug/86452
5968 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
5969 instead of get_context_die.
5970
5971 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
5972 Richard Biener <rguenther@suse.de>
5973
5974 PR middle-end/86489
5975 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
5976 that the loop latch destination where phi is defined.
5977
5978 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
5979
5980 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
5981 (riscv_expand_epilogue): Add assertion to check interrupt mode.
5982 (riscv_set_current_function): Extract getting interrupt type to new
5983 function.
5984 (riscv_get_interrupt_type): New function.
5985 (riscv_merge_decl_attributes): New function, checking interrupt type is
5986 same.
5987 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
5988
5989 2018-07-12 Paul Koning <ni1d@arrl.net>
5990
5991 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
5992 directive.
5993
5994 2018-07-12 Paul Koning <ni1d@arrl.net>
5995
5996 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
5997 zero reference, add doloop_end instead.
5998 * doc/md.texi (decrement_and_branch_until_zero): Remove.
5999 (Looping patterns): Remove decrement_and_branch_until_zero. Add
6000 detail for doloop_end.
6001
6002 2018-07-12 Martin Sebor <msebor@redhat.com>
6003
6004 PR c/86453
6005 * attribs.c (decl_attributes): Reject conflicting attributes before
6006 calling attribute handlers.
6007
6008 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
6009
6010 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
6011 parameter.
6012 (gcc::dump_manager::get_dump_file_name): likewise.
6013 (dump_begin): Likewise.
6014 * dumpfile.h (dump_begin): Update prototype.
6015 (gcc::dump_manager::get_dump_file_name,
6016 gcc::dump_manager::get_dump_file_name): Update prototype.
6017
6018 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6019
6020 * internal-fn.h (vectorizable_internal_fn_p): New function.
6021 * tree-vect-slp.c (compatible_calls_p): Likewise.
6022 (vect_build_slp_tree_1): Remove nops argument. Handle calls
6023 to internal functions.
6024 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
6025
6026 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6027
6028 * fold-const.h (inverse_conditions_p): Declare.
6029 * fold-const.c (inverse_conditions_p): New function.
6030 * match.pd: Use inverse_conditions_p. Add folds of view_converts
6031 that test the inverse condition of a conditional internal function.
6032 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
6033 * internal-fn.c (internal_fn_mask_index): Handle conditional
6034 internal functions.
6035 (vectorized_internal_fn_supported_p): New function.
6036 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
6037 (any_pred_load_store): Replace with...
6038 (need_to_predicate): ...this new variable.
6039 (redundant_ssa_names): New variable.
6040 (ifcvt_can_use_mask_load_store): Move initial checks to...
6041 (ifcvt_can_predicate): ...this new function. Handle tree codes
6042 for which a conditional internal function exists.
6043 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
6044 instead of ifcvt_can_use_mask_load_store. Update after variable
6045 name change.
6046 (predicate_load_or_store): New function, split out from
6047 predicate_mem_writes.
6048 (check_redundant_cond_expr): New function.
6049 (value_available_p): Likewise.
6050 (predicate_rhs_code): Likewise.
6051 (predicate_mem_writes): Rename to...
6052 (predicate_statements): ...this. Use predicate_load_or_store
6053 and predicate_rhs_code.
6054 (combine_blocks, tree_if_conversion): Update after above name changes.
6055 (ifcvt_local_dce): Handle redundant_ssa_names.
6056 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
6057 general conditional functions.
6058 * tree-vect-stmts.c (vectorizable_call): Likewise.
6059
6060 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6061 Alan Hayward <alan.hayward@arm.com>
6062 David Sherwood <david.sherwood@arm.com>
6063
6064 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
6065 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
6066 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
6067 plus and minus and convert them into IFN_COND_FMA-based sequences.
6068 (convert_mult_to_fma): Handle conditional plus and minus.
6069
6070 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6071
6072 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
6073 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
6074 (cond_fnms_optab): New optabs.
6075 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
6076 internal functions.
6077 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
6078 * internal-fn.h (get_conditional_internal_fn): Declare.
6079 (get_unconditional_internal_fn): Likewise.
6080 * internal-fn.c (cond_ternary_direct): New macro.
6081 (expand_cond_ternary_optab_fn): Likewise.
6082 (direct_cond_ternary_optab_supported_p): Likewise.
6083 (FOR_EACH_COND_FN_PAIR): Likewise.
6084 (get_conditional_internal_fn): New function.
6085 (get_unconditional_internal_fn): Likewise.
6086 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
6087 (gimple_match_op::gimple_match_op): Add a new overload for 5
6088 operands.
6089 (gimple_match_op::set_op): Likewise.
6090 (gimple_resimplify5): Declare.
6091 * genmatch.c (decision_tree::gen): Generate simplifications for
6092 5 operands.
6093 * gimple-match-head.c (gimple_simplify): Define an overload for
6094 5 operands. Handle calls with 5 arguments in the top-level overload.
6095 (convert_conditional_op): Handle conversions from unconditional
6096 internal functions to conditional ones.
6097 (gimple_resimplify5): New function.
6098 (build_call_internal): Pass a fifth operand.
6099 (maybe_push_res_to_seq): Likewise.
6100 (try_conditional_simplification): Try converting conditional
6101 internal functions to unconditional internal functions.
6102 Handle 3-operand unconditional forms.
6103 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
6104 Define ternary equivalents of the current rules for binary conditional
6105 internal functions.
6106 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
6107 ternary operations.
6108 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
6109 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
6110 (optab): Handle them.
6111 (SVE_COND_FP_TERNARY): New int iterator.
6112 (sve_fmla_op, sve_fmad_op): New int attributes.
6113 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
6114 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
6115 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
6116
6117 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
6118
6119 * target.def (preferred_else_value): New target hook.
6120 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
6121 * doc/tm.texi: Regenerate.
6122 * targhooks.h (default_preferred_else_value): Declare.
6123 * targhooks.c (default_preferred_else_value): New function.
6124 * internal-fn.h (conditional_internal_fn_code): Declare.
6125 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
6126 (get_conditional_internal_fn): Use it.
6127 (conditional_internal_fn_code): New function.
6128 * gimple-match.h (gimple_match_cond): New struct.
6129 (gimple_match_op): Add a cond member function.
6130 (gimple_match_op::gimple_match_op): Update all forms to take a
6131 gimple_match_cond.
6132 * genmatch.c (expr::gen_transform): Use the same condition as res_op
6133 for the suboperation, but don't specify a particular else_value.
6134 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
6135 (visit_nary_op, visit_reference_op_load): Pass
6136 gimple_match_cond::UNCOND to the gimple_match_op constructor.
6137 * gimple-match-head.c: Include tree-eh.h
6138 (convert_conditional_op): New function.
6139 (maybe_resimplify_conditional_op): Likewise.
6140 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
6141 (gimple_resimplify2): Likewise.
6142 (gimple_resimplify3): Likewise.
6143 (gimple_resimplify4): Likewise.
6144 (maybe_push_res_to_seq): Return null for conditional operations.
6145 (try_conditional_simplification): New function.
6146 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
6147 constructor.
6148 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
6149 IFN_COND_* call.
6150 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
6151 function.
6152 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
6153
6154 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
6155
6156 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
6157 DECL_FCONTEXT
6158 (hash_tree): Do not hash DECL_FCONTEXT
6159 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
6160 Do not stream DECL_FCONTEXT.
6161 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
6162 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
6163
6164 2018-07-12 Richard Biener <rguenther@suse.de>
6165
6166 PR debug/86462
6167 * dwarf2out.c (gen_block_die): Only output blocks when they have
6168 at least one !DECL_IGNORED_P variable.
6169
6170 2018-07-12 Richard Biener <rguenther@suse.de>
6171
6172 PR target/84829
6173 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
6174 Remove -mieee-fp handling.
6175
6176 2018-07-12 Richard Biener <rguenther@suse.de>
6177
6178 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
6179 left-over from last patch.
6180
6181 2018-07-12 Jakub Jelinek <jakub@redhat.com>
6182
6183 PR tree-optimization/86492
6184 * gimple-ssa-store-merging.c
6185 (imm_store_chain_info::coalesce_immediate_stores): Call
6186 check_no_overlap even for the merge_overlapping case. Formatting fix.
6187
6188 2018-07-12 Richard Biener <rguenther@suse.de>
6189
6190 PR middle-end/86479
6191 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
6192 move possibly trapping operations into the conditional.
6193
6194 2018-07-12 Richard Biener <rguenther@suse.de>
6195
6196 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
6197 (vn_lookup_simplify_result): Remove recursion limit applied
6198 here.
6199 (vn_nary_build_or_lookup_1): Adjust.
6200 (try_to_simplify): Likewise.
6201 * gimple-match-head.c (gimple_resimplify1): Instead apply one
6202 here.
6203 (gimple_resimplify2): Likewise.
6204 (gimple_resimplify3): Likewise.
6205 (gimple_resimplify4): Likewise.
6206
6207 2018-07-11 Jakub Jelinek <jakub@redhat.com>
6208
6209 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
6210 Use __mmask64 type instead of __mmask8 for __M argument.
6211 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
6212 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
6213 __U argument.
6214 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
6215 __mmask16 for __M argument.
6216 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
6217 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
6218 to __mmask16 instead of __mmask8.
6219 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
6220 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
6221 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
6222 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
6223 instead of __mmask16 for __U argument.
6224 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
6225 __mmask16 instead of __mmask8 for __U argument.
6226 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
6227 __U argument.
6228 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
6229 __mmask16.
6230 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
6231 argument.
6232 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
6233 __U argument.
6234 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
6235 __mmask16.
6236 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
6237 of __mmask16.
6238 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
6239 __U argument.
6240 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
6241 __U argument.
6242 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
6243 __U argument.
6244 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
6245 __U argument.
6246 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
6247 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
6248 return type as well as __M argument type and all casts from __mmask8
6249 to __mmask32.
6250 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
6251 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
6252 return type as well as __M argument type and all casts from __mmask8
6253 to __mmask16.
6254 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
6255 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
6256 return type as well as __M argument type and all casts from __mmask8
6257 to __mmask32.
6258 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
6259 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
6260 return type as well as __M argument type and all casts from __mmask8
6261 to __mmask16.
6262 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
6263 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
6264 __mmask16.
6265
6266 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
6267
6268 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
6269 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
6270 for __U argument.
6271
6272 2018-07-11 Paul Koning <ni1d@arrl.net>
6273
6274 * doc/md.texi (define_subst): Document how multiple occurrences of
6275 the same argument in the replacement pattern are handled.
6276
6277 2018-07-11 Paul Koning <ni1d@arrl.net>
6278
6279 * doc/extend.texi (Common Variable Attributes): Move "mode" into
6280 alphabetical order.
6281 (Common Type Attributes): Add "mode" attribute.
6282
6283 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
6284
6285 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
6286 stream DECL_ORIGINAL_TYPE.
6287 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
6288 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
6289 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
6290 Do not walk original type.
6291 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
6292 external decls.
6293 (write_ts_decl_non_common_tree_pointers): Do not stream
6294 DECL_ORIGINAL_TYPE
6295 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
6296 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
6297
6298 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6299
6300 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
6301 thread twice from the same starting edge.
6302
6303 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
6304
6305 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
6306 * gimple.c (gimple_call_nonnull_result_p): ...here...
6307 (gimple_call_nonnull_arg): ...and here.
6308 * gimple.h (gimple_call_nonnull_result_p): New.
6309 (gimple_call_nonnull_arg): New.
6310
6311 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
6312
6313 * config/arm/arm-cpus.in: Move information from fpu field of each
6314 cpu definition to the isa field.
6315 * config/arm/parsecpu.awk (fpu): Delete match rule.
6316 (gen_comm_data): Don't add bits from the CPU's FPU entry.
6317
6318 2018-07-11 Richard Biener <rguenther@suse.de>
6319
6320 PR debug/86457
6321 * dwarf2out.c (init_sections_and_labels): Use
6322 output_asm_line_debug_info consistently.
6323 (dwarf2out_early_finish): Likewise.
6324 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
6325 type units.
6326
6327 2018-07-11 Richard Biener <rguenther@suse.de>
6328
6329 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
6330 Rework father_bb setting in a way to avoid propagating constants
6331 multiple times on a loop body.
6332
6333 2018-07-10 Mark Wielaard <mark@klomp.org>
6334
6335 PR debug/86459
6336 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
6337
6338 2018-07-10 Richard Biener <rguenther@suse.de>
6339
6340 * hash-map.h (hash_map::iterator::operator*): Return
6341 references to key and value.
6342
6343 2018-07-10 Jakub Jelinek <jakub@redhat.com>
6344
6345 PR c++/86443
6346 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
6347 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
6348 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
6349 (gimplify_omp_for): For composite loops, move outer
6350 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
6351 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
6352 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
6353 TREE_LIST for both the original class iterator and the "last" helper
6354 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
6355 loop, remember has_decl_expr from outer composite loops for the
6356 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
6357
6358 2018-07-09 Martin Sebor <msebor@redhat.com>
6359
6360 PR middle-end/77357
6361 PR middle-end/86428
6362 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
6363 accessing implicitly initialized array elements.
6364 * expr.c (string_constant): Handle string initializers of
6365 character arrays within aggregates.
6366 * gimple-fold.c (fold_array_ctor_reference): Add argument.
6367 Store element offset. As a special case, handle zero size.
6368 (fold_nonarray_ctor_reference): Same.
6369 (fold_ctor_reference): Add argument. Store subobject offset.
6370 * gimple-fold.h (fold_ctor_reference): Add argument.
6371
6372 2018-07-09 Paul Koning <ni1d@arrl.net>
6373
6374 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
6375 (pdp11_insn_cost): New function.
6376 (pdp11_md_asm_adjust): New function.
6377 (TARGET_INVALID_WITHIN_DOLOOP): Define.
6378 (pdp11_rtx_costs): Update to match machine better.
6379 (output_addr_const_pdp11): Correct format mismatch warnings.
6380 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
6381 * config/pdp11/pdp11.md: General change to add base_cost and/or
6382 length attributes for use by new pdp11_insn_cost function.
6383 (MIN_BRANCH): Correct definition.
6384 (MIN_SOB): Ditto.
6385 (doloop_end): Use standard pattern name for looping pattern.
6386 (doloop_end_nocc): New.
6387 (movsf): Add another constraint alternative.
6388 (zero_extendqihi2): Add constraint alternatives for not in place
6389 extend.
6390 (zero_extendhisi2): Remove.
6391 (shift patterns): Add CC handling variants.
6392 (bswaphi2): New.
6393 (bswapsi2): New.
6394 (rothi3): New.
6395 (define_peephole2): New peephole to recognize mov that sets CC for
6396 subsequent test.
6397
6398 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
6399
6400 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
6401 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
6402 wi::add.
6403
6404 2018-07-09 Jakub Jelinek <jakub@redhat.com>
6405
6406 PR c/86420
6407 * real.c (real_nextafter): Return true if result is denormal.
6408
6409 2018-07-09 Martin Liska <mliska@suse.cz>
6410
6411 * common.opt: Add back wrongly removed attribute.
6412
6413 2018-07-09 Richard Biener <rguenther@suse.de>
6414
6415 PR debug/86413
6416 * dwarf2out.c (gen_block_die): For an early generated DIE
6417 always output high/low PC attributes.
6418
6419 2018-07-09 Tom de Vries <tdevries@suse.de>
6420
6421 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
6422 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
6423 onto VAR_DECL with abstract origin.
6424
6425 2018-07-07 Jim Wilson <jimw@sifive.com>
6426
6427 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
6428
6429 2018-07-07 Tom de Vries <tdevries@suse.de>
6430
6431 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
6432
6433 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
6434
6435 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
6436 overflow_type.
6437 (combine_bound): Use wide-int overflow calculation instead of
6438 rolling our own.
6439 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
6440 overflow_type.
6441 * fold-const.c (int_const_binop_2): Same.
6442 (extract_muldiv_1): Same.
6443 (fold_div_compare): Same.
6444 (fold_abs_const): Same.
6445 * match.pd: Same.
6446 * poly-int.h (add): Same.
6447 (sub): Same.
6448 (neg): Same.
6449 (mul): Same.
6450 * predict.c (predict_iv_comparison): Same.
6451 * profile-count.c (slow_safe_scale_64bit): Same.
6452 * simplify-rtx.c (simplify_const_binary_operation): Same.
6453 * tree-chrec.c (tree_fold_binomial): Same.
6454 * tree-data-ref.c (split_constant_offset_1): Same.
6455 * tree-if-conv.c (idx_within_array_bound): Same.
6456 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
6457 * tree-ssa-phiopt.c (minmax_replacement): Same.
6458 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
6459 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
6460 * vr-values.c (vr_values::adjust_range_with_scev): Same.
6461 * wide-int.cc (wi::add_large): Same.
6462 (wi::mul_internal): Same.
6463 (wi::sub_large): Same.
6464 (wi::divmod_internal): Same.
6465 * wide-int.h: Change overflow type to overflow_type for neg, add,
6466 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
6467 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
6468 mul_internal, divmod_internal.
6469 (overflow_type): New enum.
6470 (accumulate_overflow): New.
6471
6472 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
6473
6474 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
6475 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
6476
6477 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6478
6479 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
6480 argument is checked for zero before entering loop, avoid checking again.
6481
6482 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
6483
6484 * gimplify.h (generic_expr_could_trap_p): Set as global function.
6485 * gimplify.h (generic_expr_could_trap_p): Likwise.
6486 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
6487
6488 2018-07-06 Jakub Jelinek <jakub@redhat.com>
6489
6490 PR tree-optimization/86401
6491 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
6492 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
6493 (fold_bit_and_mask): ... here. New helper function for match.pd.
6494 * fold-const.h (fold_bit_and_mask): Declare.
6495 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
6496
6497 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
6498
6499 PR target/86324
6500 * target.def (translate_mode_attribute): New hook.
6501 * targhooks.h (default_translate_mode_attribute): Declare.
6502 * targhooks.c (default_translate_mode_attribute): New function.
6503 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
6504 * doc/tm.texi: Regenerate.
6505 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
6506 (rs6000_translate_mode_attribute): New function.
6507
6508 2018-07-06 Paul Koning <ni1d@arrl.net>
6509
6510 * doc/md.texi (define_split): Document DONE and FAIL.
6511 (define_peephole2): Ditto.
6512
6513 2018-07-05 Jeff Law <law@redhat.com>
6514
6515 PR tree-optimization/86010
6516 * tree-ssa-dse.c (compute_trims): More aggressively trim at
6517 both the head and tail of mem* and str* calls.
6518
6519 2018-07-05 Jim Wilson <jimw@sifive.com>
6520
6521 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
6522
6523 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
6524
6525 * config/aarch64/aarch64-simd.md: correct flags text for
6526 MIN_EXPR replacement.
6527
6528 2018-07-05 James Clarke <jrtc27@jrtc27.com>
6529
6530 * configure: Regenerated.
6531
6532 2018-07-05 Carl Love <cel@us.ibm.com>
6533
6534 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
6535 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
6536 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
6537 VSX_BUILTIN_DOUBLEL_V4SF.
6538
6539 2018-07-05 Martin Sebor <msebor@redhat.com>
6540
6541 PR c++/86400
6542 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
6543 than its domain to compute its the upper bound of a char array.
6544
6545 2018-07-05 Nathan Sidwell <nathan@acm.org>
6546
6547 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
6548 * doc/cpp.texi: Update comment.
6549 * doc/tm.texi: Rebuilt.
6550 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
6551 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
6552 * doc/extend.texi (Backwards Compatibility): Clarify it is system
6553 headers affected by extern "C".
6554 * system.h: Poison NO_IMPLICIT_EXTERN_C.
6555 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
6556 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
6557 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
6558 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
6559 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
6560 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
6561 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
6562 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
6563 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
6564 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
6565 config/sparc/sp64-elf.h, config/spu/spu.h,
6566 config/stormy16/stormy16.h, config/v850/v850.h,
6567 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
6568 define NO_IMPLICIT_EXTERN_C.
6569 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
6570
6571 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6572
6573 PR target/84711
6574 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
6575 instead of GET_MODE_SIZE when comparing Units.
6576
6577 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6578
6579 PR target/84711
6580 * rtlanal.c (set_noop_p): Constrain on mode change,
6581 include hard-reg-set.h
6582
6583 2018-07-05 Tamar Christina <tamar.christina@arm.com>
6584
6585 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
6586
6587 2018-07-05 Jakub Jelinek <jakub@redhat.com>
6588
6589 Revert
6590 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
6591
6592 PR sanitizer/84250
6593 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
6594 libasan.
6595 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
6596
6597 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
6598
6599 PR sanitizer/84250
6600 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
6601 libasan.
6602 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
6603
6604 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
6605
6606 PR middle-end/86380
6607 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
6608
6609 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
6610
6611 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
6612 neg_*_op* variables.
6613
6614 2018-07-04 Martin Liska <mliska@suse.cz>
6615
6616 * tree-switch-conversion.c: Define
6617 max_ratio_for_speed and max_ratio_for_size constants.
6618
6619 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
6620 Martin Liska <mliska@suse.cz>
6621
6622 PR middle-end/66240
6623 PR target/45996
6624 PR c/84100
6625 * common.opt: Rename align options with 'str_' prefix.
6626 * common/config/i386/i386-common.c (set_malign_value): New
6627 function.
6628 (ix86_handle_option): Use it to set -falign-* options/
6629 * config/aarch64/aarch64-protos.h (struct tune_params): Change
6630 type from int to string.
6631 * config/aarch64/aarch64.c: Update default values from int
6632 to string.
6633 * config/alpha/alpha.c (alpha_override_options_after_change):
6634 Likewise.
6635 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
6636 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6637 max skip conditionally.
6638 * config/i386/freebsd.h (SUBALIGN_LOG): New.
6639 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6640 max skip conditionally.
6641 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6642 max skip conditionally.
6643 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
6644 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6645 max skip conditionally.
6646 * config/i386/i386.c (struct ptt): Change type from int to
6647 string.
6648 (ix86_default_align): Set default values.
6649 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
6650 max skip conditionally.
6651 * config/i386/iamcu.h (SUBALIGN_LOG): New.
6652 (ASM_OUTPUT_MAX_SKIP_ALIGN):
6653 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
6654 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6655 max skip conditionally.
6656 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
6657 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
6658 * config/i386/x86-64.h (SUBALIGN_LOG): New.
6659 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
6660 max skip conditionally.
6661 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
6662 * config/ia64/ia64.c (ia64_option_override): Set default values
6663 for alignment options.
6664 * config/m68k/m68k.c: Handle new str_align_* options.
6665 * config/mips/mips.c (mips_set_compression_mode): Change
6666 type of constants.
6667 (mips_option_override): Set default values for options.
6668 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
6669 Likewise.
6670 * config/rs6000/rs6000.c (rs6000_option_override_internal):
6671 Likewise.
6672 * config/rx/rx.c (rx_option_override): Likewise.
6673 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
6674 (LABEL_ALIGN): Use align_labels_log.
6675 (LOOP_ALIGN): Use align_loops_align.
6676 * config/s390/s390.c (s390_asm_output_function_label): Use new
6677 macros.
6678 * config/sh/sh.c (sh_override_options_after_change):
6679 Change type of constants.
6680 * config/spu/spu.c (spu_sched_init): Likewise.
6681 * config/sparc/sparc.c (sparc_option_override): Set default
6682 values for options.
6683 * config/visium/visium.c (visium_option_override): Likewise.
6684 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
6685 emit p2align format with last argument if it's not needed.
6686 * doc/invoke.texi: Document extended format of -falign-*.
6687 * final.c: Use align_labels alignment.
6688 * flags.h (struct target_flag_state): Change type to use
6689 align_flags.
6690 (struct align_flags_tuple): New.
6691 (struct align_flags): Likewise.
6692 (align_loops_log): Redefine macro to use new types.
6693 (align_loops_max_skip): Redefine macro to use new types.
6694 (align_jumps_log): Redefine macro to use new types.
6695 (align_jumps_max_skip): Redefine macro to use new types.
6696 (align_labels_log): Redefine macro to use new types.
6697 (align_labels_max_skip): Redefine macro to use new types.
6698 (align_functions_log): Redefine macro to use new types.
6699 (align_loops): Redefine macro to use new types.
6700 (align_jumps): Redefine macro to use new types.
6701 (align_labels): Redefine macro to use new types.
6702 (align_functions): Redefine macro to use new types.
6703 (align_functions_max_skip): Redefine macro to use new types.
6704 (align_loops_value): New macro.
6705 (align_jumps_value): New macro.
6706 (align_labels_value): New macro.
6707 (align_functions_value): New macro.
6708 * function.c (invoke_set_current_function_hook): Propagate
6709 alignment values from flags to global variables default in
6710 topleev.h.
6711 * ipa-icf.c (sem_function::equals_wpa): Use
6712 cl_optimization_option_eq instead of memcmp.
6713 * lto-streamer.h (cl_optimization_stream_out): Support streaming
6714 of string types.
6715 (cl_optimization_stream_in): Likewise.
6716 * optc-save-gen.awk: Support strings in cl_optimization.
6717 * opth-gen.awk: Likewise.
6718 * opts.c (finish_options): Remove error checking of invalid
6719 value ranges.
6720 (MAX_CODE_ALIGN): Remove.
6721 (MAX_CODE_ALIGN_VALUE): Likewise.
6722 (parse_and_check_align_values): New function.
6723 (check_alignment_argument): Likewise.
6724 (common_handle_option): Use check_alignment_argument.
6725 * opts.h (parse_and_check_align_values): Declare.
6726 * toplev.c (init_alignments): Remove.
6727 (read_log_maxskip): New.
6728 (parse_N_M): Likewise.
6729 (parse_alignment_opts): Likewise.
6730 (backend_init_target): Remove usage of init_alignments.
6731 * toplev.h (parse_alignment_opts): Declare.
6732 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
6733 argument.
6734 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
6735 * tree.c (cl_option_hasher::equal): New.
6736 * varasm.c: Use new global macros.
6737
6738 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
6739
6740 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
6741 Use a simpler align directive also if MAXSKIP = ALIGN-1.
6742 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6743 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6744 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6745 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
6746 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
6747 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
6748 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
6749 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6750 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6751 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6752 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
6753
6754 2018-07-04 Martin Liska <mliska@suse.cz>
6755 Jonathan Wakely <jwakely@redhat.com>
6756
6757 * coverage.c: Use correct type.
6758 * doc/invoke.texi: Language correction.
6759
6760 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
6761
6762 PR target/85620
6763 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
6764 ENDBRANCH for non-tail call which may return via indirect branch.
6765 * doc/extend.texi: Document indirect_return attribute.
6766
6767 2018-07-03 Martin Sebor <msebor@redhat.com>
6768
6769 PR tree-optimization/86274
6770 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
6771 precondition.
6772 (format_floating): Correct handling of infinities and NaNs.
6773
6774 2018-07-03 Martin Sebor <msebor@redhat.com>
6775
6776 * print-tree.c (print_real_cst): New function.
6777 (print_node_brief): Call it.
6778 (print_node): Ditto.
6779
6780 2018-07-03 Jeff Law <law@redhat.com>
6781
6782 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
6783 into a single pattern.
6784
6785 * config/h8300/h8300.md (ors code_iterator): New.
6786 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
6787 a single pattern and single splitter.
6788 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
6789 (iorqi3_1, xorqi3_1): Likewise.
6790 (iorqi3, xorqi3 expanders): Similarly.
6791
6792 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
6793 (movmd_internal) into a single pattern using the P mode iterator.
6794 (movmd splitters): Similarly.
6795 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
6796 (movsd splitters): Similarly.
6797
6798 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
6799 ADDB, ADDW and ADDL into a single ADD attribute which selects the
6800 right table based on the size of the operand.
6801 * config/h8300/h8300.md (length_table): Corresponding changes. All
6802 references to "addb", "addw" and "addl" changed to "add".
6803 (btst patterns): Merge two variants into a single pattern.
6804 (tstqi, tsthi): Likewise.
6805 (addhi3_incdec, addsi3_incdec): Likewise.
6806 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
6807 (mulhi3, mulsi3): Likewise.
6808 (udivhi3, udivsi3): Likewise.
6809 (divhi3, divsi3): Likewise.
6810 (andorqi3, andorhi3, andorsi3): Likewise.
6811
6812 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
6813
6814 PR target/85694
6815 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
6816 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
6817
6818 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6819
6820 PR tree-optimization/85694
6821 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
6822 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
6823 UNSPEC_URHADD.
6824 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
6825 (<u>avg<mode>3_ceil): New patterns.
6826
6827 2018-07-03 David Malcolm <dmalcolm@redhat.com>
6828
6829 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
6830 scan-tree-dump directive.
6831 * gcc.dg/vect/slp-perm-2.c: Likewise.
6832 * gcc.dg/vect/slp-perm-3.c: Likewise.
6833 * gcc.dg/vect/slp-perm-5.c: Likewise.
6834 * gcc.dg/vect/slp-perm-6.c: Likewise.
6835 * gcc.dg/vect/slp-perm-7.c: Likewise.
6836 * gcc.dg/vect/slp-perm-8.c: Likewise.
6837
6838 2018-07-03 Marek Polacek <polacek@redhat.com>
6839
6840 PR middle-end/86202
6841 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
6842
6843 2018-07-03 Richard Biener <rguenther@suse.de>
6844
6845 PR ipa/86389
6846 * tree-ssa-structalias.c (find_func_clobbers): Properly
6847 handle indirect calls.
6848
6849 2018-07-03 Jeff Law <law@redhat.com>
6850
6851 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
6852 (shifts): New code iterator.
6853 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
6854 expander. Fix HImode handling on H8/SX.
6855 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
6856 (subqi3, subhi3, subsi3 expanders): Likewise.
6857 (andqi3, andhi3, andsi3 expanders): Likewise.
6858 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
6859 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
6860 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
6861 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
6862 (zero_extendqihi2, zero_extendqisi2): Likewise.
6863 (extendqihi2, extendqisi2): Likewise.
6864 (rotlqi3, rotlhi3, rotlsi3): Likewise.
6865 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
6866 (rotlqi3_1, rotlhi3_1): Likewise.
6867 (logicalhi3_sn, logicalsi3_sn): Likewise.
6868 (logicalhi3, logicalsi3): Likewise.
6869
6870 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6871
6872 * tree-vect-patterns.c (vect_recog_rotate_pattern)
6873 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
6874 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
6875 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
6876 type to append_pattern_def_seq instead of creating a stmt_vec_info
6877 directly.
6878 (build_mask_conversion): Likewise. Remove vinfo argument.
6879 (vect_add_conversion_to_patterm): Likewise, renaming to...
6880 (vect_add_conversion_to_pattern): ...this.
6881 (vect_recog_mask_conversion_pattern): Update call to
6882 build_mask_conversion. Pass the vector type to
6883 append_pattern_def_seq here too.
6884 (vect_recog_gather_scatter_pattern): Update call to
6885 vect_add_conversion_to_pattern.
6886
6887 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6888
6889 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
6890 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
6891 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
6892 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
6893 STMT_VINFO_PATTERN_DEF_SEQ to null here.
6894 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
6895 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
6896 append_pattern_def_seq instead of new_pattern_def_seq.
6897 (vect_recog_divmod_pattern): Do both of the above.
6898 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
6899 is null.
6900
6901 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6902
6903 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
6904 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
6905 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
6906 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
6907 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
6908 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
6909 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
6910 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
6911 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
6912 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
6913 parameter with a single stmt_vec_info.
6914 (vect_recog_func_ptr): Likewise.
6915 (vect_recog_gather_scatter_pattern): Likewise, folding in...
6916 (vect_try_gather_scatter_pattern): ...this.
6917 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
6918 the stmt_vec_info of the statement to be matched. Don't clear
6919 STMT_VINFO_RELATED_STMT.
6920 (vect_pattern_recog): Update call accordingly.
6921
6922 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6923
6924 PR tree-optimization/85694
6925 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
6926 (uavgM3_ceil): Document new optabs.
6927 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
6928 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
6929 functions.
6930 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
6931 (savg_ceil_optab): New optabs.
6932 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
6933 (vect_vect_recog_func_ptrs): Add it.
6934 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
6935 constant directly from the associated lhs.
6936
6937 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6938
6939 * tree-vect-patterns.c (vect_split_statement): New function.
6940 (vect_convert_input): Use it to try to split an existing cast.
6941
6942 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6943
6944 * poly-int.h (print_hex): New function.
6945 * dumpfile.h (dump_dec, dump_hex): Declare.
6946 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
6947 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
6948 min_input_precision, operation_precision and operation_sign.
6949 * tree-vect-patterns.c (vect_get_range_info): New function.
6950 (vect_same_loop_or_bb_p, vect_single_imm_use)
6951 (vect_operation_fits_smaller_type): Delete.
6952 (vect_look_through_possible_promotion): Add an optional
6953 single_use_p parameter.
6954 (vect_recog_over_widening_pattern): Rewrite to use new
6955 stmt_vec_info infomration. Handle one operation at a time.
6956 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
6957 (vect_truncatable_operation_p, vect_set_operation_type)
6958 (vect_set_min_input_precision): New functions.
6959 (vect_determine_min_output_precision_1): Likewise.
6960 (vect_determine_min_output_precision): Likewise.
6961 (vect_determine_precisions_from_range): Likewise.
6962 (vect_determine_precisions_from_users): Likewise.
6963 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
6964 (vect_vect_recog_func_ptrs): Put over_widening first.
6965 Add cast_forwprop.
6966 (vect_pattern_recog): Call vect_determine_precisions.
6967
6968 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
6969
6970 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
6971 statements that have been replaced by further pattern statements.
6972 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
6973
6974 2018-07-03 Richard Biener <rguenther@suse.de>
6975
6976 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
6977 always set *dt. Dump vectype in vectype overload.
6978 * dumpfile.h (dump_gimple_expr): New function.
6979 (dump_gimple_expr_loc): Likewise.
6980 * dumpfile.c (dump_gimple_expr): New function.
6981 (dump_gimple_expr_loc): Likewise.
6982
6983 2018-07-02 Jeff Law <law@redhat.com>
6984
6985 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
6986 the H8/300, H8/300H and H8/S variants into a single pattern.
6987 (movhi_h8300, movqi_h8300hs): Similarly.
6988 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
6989 (QHI mode iterator): New.
6990
6991 * config/h8300/h8300.md: Remove trailing whitespace.
6992
6993 2018-07-02 Jim Wilson <jimw@sifive.com>
6994
6995 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
6996 instead of emit_insn for interrupt returns.
6997 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
6998 (riscv_sret, riscv_uret): Likewise.
6999
7000 2018-07-02 David Malcolm <dmalcolm@redhat.com>
7001
7002 * pretty-print.c (selftest::test_pp_format): Move save and restore
7003 of quotes to class auto_fix_quotes, and add an instance.
7004 * selftest.c: Include "intl.h".
7005 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
7006 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
7007 * selftest.h (selftest::auto_fix_quotes): New class.
7008
7009 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
7010
7011 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
7012 (aarch64_sve_prepare_conditional_op): Remove.
7013 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
7014 Allow aarch64_simd_reg_or_zero as select operand; remove
7015 the aarch64_sve_prepare_conditional_op call.
7016 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
7017 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
7018 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
7019 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
7020 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
7021 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
7022 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
7023 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
7024 and a splitters to match all of the *_any patterns.
7025 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
7026
7027 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
7028 (SVE_COND_FP_BINARY_REV): Remove.
7029 (sve_int_op_rev, sve_fp_op_rev): New.
7030 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
7031 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
7032 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
7033 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
7034 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
7035 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
7036 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
7037 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
7038 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
7039
7040 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
7041 Remove match_dup 1 from the inner unspec.
7042 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
7043
7044 * config/aarch64/aarch64.md (movprfx): New attr.
7045 (length): Default movprfx to 8.
7046 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
7047 (*madd<SVE_I>, *msub<SVE_I): Likewise.
7048 (*<su>mul<SVE_I>3_highpart): Likewise.
7049 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
7050 (*v<ASHIFT><SVE_I>3): Likewise.
7051 (*<su><MAXMIN><SVE_I>3): Likewise.
7052 (*<su><MAXMIN><SVE_F>3): Likewise.
7053 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
7054 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
7055 (*div<SVE_F>4): Likewise.
7056
7057 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
7058
7059 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
7060 in dump string.
7061
7062 2018-07-02 Richard Biener <rguenther@suse.de>
7063
7064 PR tree-optimization/86363
7065 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
7066 memset argument refers to a non-variable address.
7067
7068 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
7069
7070 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
7071 {PLUS,MINUS}_EXPR code to...
7072 (adjust_symbolic_bound): ...here,
7073 (combine_bound): ...here,
7074 (set_value_range_with_overflow): ...and here.
7075
7076 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
7077
7078 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
7079 code...
7080 (extract_range_from_abs_expr): ...here.
7081
7082 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
7083
7084 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
7085 -fno-omit-frame-pointer when not optimizing.
7086
7087 2018-07-02 Martin Liska <mliska@suse.cz>
7088
7089 PR ipa/86279
7090 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
7091 (propagate_nothrow): Likewise.
7092
7093 2018-07-02 Martin Liska <mliska@suse.cz>
7094
7095 PR ipa/86323
7096 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
7097
7098 2018-07-02 David Malcolm <dmalcolm@redhat.com>
7099
7100 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
7101 function in r262149, changing "loc" param from source_location to
7102 const dump_location_t &.
7103 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
7104 declaration, as above.
7105
7106 2018-07-01 Paul Koning <ni1d@arrl.net>
7107
7108 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
7109 -munit-asm, -mgnu-asm, -mdec-asm.
7110 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
7111 (pdp11_output_labelref): New.
7112 (pdp11_output_def): New.
7113 (pdp11_output_addr_vec_elt): New.
7114 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
7115 %# and %@ format codes.
7116 (pdp11_option_override): New.
7117 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
7118 (pdp11_output_ident): New.
7119 (pdp11_asm_named_section): New.
7120 (pdp11_asm_init_sections): New.
7121 (pdp11_file_start): New.
7122 (pdp11_file_end): New.
7123 (output_ascii): Use .ascii/.asciz for -mdec-asm.
7124 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
7125 %o, like %c but octal.
7126 (pdp11_option_override): New.
7127 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
7128 -mdec-asm.
7129 (DATA_SECTION_ASM_OP): Ditto.
7130 (READONLY_DATA_SECTION_ASM_OP): New.
7131 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
7132 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
7133 (ASM_OUTPUT_LABELREF): Ditto.
7134 (ASM_OUTPUT_DEF): Ditto.
7135 (ASM_OUTPUT_EXTERNAL): New.
7136 (ASM_OUTPUT_SOURCE_FILENAME): New.
7137 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
7138 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
7139 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
7140 %# and %@ format codes.
7141 * config/pdp11/pdp11.opt (mgnu-asm): New.
7142 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
7143 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
7144 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
7145
7146 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
7147
7148 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
7149 dereferencing path[] beyond its length.
7150 (debug_path): New.
7151 (debug_all_paths): New.
7152 (rewire_first_differing_edge): New.
7153 (adjust_paths_after_duplication): New.
7154 (duplicate_thread_path): Call adjust_paths_after_duplication.
7155 Add new argument.
7156 (thread_through_all_blocks): Add new argument to
7157 duplicate_thread_path.
7158
7159 2018-06-30 Jim Wilson <jimw@sifive.com>
7160
7161 * config/riscv/predicates.md (p2m1_shift_operand): New.
7162 (high_mask_shift_operand): New.
7163 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
7164 pattern using p2m1_shift_operand.
7165 (lshsi3_zero_extend_3+2): New combiner pattern using
7166 high_mask_shift_operand.
7167
7168 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7169
7170 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
7171 split out from...
7172 (vect_recog_rotate_pattern): ...here.
7173 (vect_convert_input): Try to insert casts of invariants in the
7174 preheader.
7175 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
7176 preheader to be empty.
7177
7178 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7179
7180 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
7181 vector type. If given, install it in the new statement's
7182 STMT_VINFO_VECTYPE.
7183 (vect_element_precision): New function.
7184 (vect_unpromoted_value): New struct.
7185 (vect_unpromoted_value::vect_unpromoted_value): New function.
7186 (vect_unpromoted_value::set_op): Likewise.
7187 (vect_look_through_possible_promotion): Likewise.
7188 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
7189 (vect_widened_op_tree, vect_convert_input): Likewise.
7190 (vect_convert_inputs, vect_convert_output): Likewise.
7191 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
7192 to handle the optional cast of the multiplication result and
7193 vect_widened_op_tree to detect the widened multiplication itself.
7194 Do not require the input and output of promotion casts to have
7195 the same sign, but base the signedness of the operation on the
7196 input rather than the result. If the pattern includes two
7197 promotions, check that those promotions have the same sign.
7198 Do not restrict the MULT_EXPR handling to a double-width result;
7199 handle quadruple-width results and wider. Use vect_convert_inputs
7200 to convert the inputs to the common type.
7201 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
7202 to handle the optional cast of the ABS result. Also allow a sign
7203 change or a sign extension between the ABS and MINUS.
7204 Use vect_widened_op_tree to detect the widened subtraction and use
7205 vect_convert_inputs to convert the inputs to the common type.
7206 (vect_handle_widen_op_by_const): Delete.
7207 (vect_recog_widen_op_pattern): New function.
7208 (vect_recog_widen_mult_pattern): Use it.
7209 (vect_recog_widen_shift_pattern): Likewise.
7210 (vect_recog_widen_sum_pattern): Use
7211 vect_look_through_possible_promotion to handle the promoted
7212 PLUS_EXPR operand.
7213
7214 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7215
7216 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
7217 the containing gimple_seq *.
7218 * gimple-iterator.h (gsi_for_stmt): Declare it.
7219 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
7220 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
7221 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
7222 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
7223 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
7224 checks.
7225 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
7226 split out from...
7227 (vect_mark_pattern_stmts): ...here. Handle cases in which the
7228 statement being replaced is part of an existing pattern
7229 definition sequence, inserting the new pattern statements before
7230 the original one.
7231 (vect_pattern_recog_1): Don't return a bool. If the statement
7232 is already part of a pattern, instead apply pattern matching
7233 to the pattern definition statements. Don't clear the
7234 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
7235 (vect_pattern_recog): Don't break after the first match;
7236 continue processing the pattern definition statements instead.
7237 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
7238
7239 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7240
7241 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
7242 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
7243 (vect_recog_widen_sum_pattern): Use it.
7244
7245 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7246
7247 * tree-vect-loop.c (vectorizable_reduction): Assert that the
7248 phi is not a pattern statement and has not been replaced by
7249 a pattern statement.
7250 * tree-vect-patterns.c (type_conversion_p): Don't check
7251 STMT_VINFO_IN_PATTERN_P.
7252 (vect_recog_vector_vector_shift_pattern): Likewise.
7253 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
7254 the pattern statement rather than the original statement; check
7255 directly for a WIDEN_MULT_EXPR here.
7256 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
7257 vect_is_simple_use to return the pattern statement rather
7258 than the original statement; use is_pattern_stmt_p to check
7259 for such a pattern statement.
7260 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
7261 to return the pattern statement rather than the original statement;
7262 don't do the same transformation here.
7263 (vect_is_simple_use): If the defining statement has been replaced
7264 by a pattern statement, return the pattern statement instead.
7265 Remove the corresponding (local) transformation from the vectype
7266 overload.
7267
7268 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
7269
7270 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
7271 end and default to null.
7272 * tree-vect-loop.c (vect_create_epilog_for_reduction)
7273 (vectorizable_reduction): Update calls accordingly, dropping the
7274 gimple ** argument if the passed-back statement isn't needed.
7275 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
7276 (vect_recog_rotate_pattern): Likewise.
7277 (vect_recog_mask_conversion_pattern): Likewise.
7278 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
7279 (vect_mask_constant_operand_p): Likewise.
7280 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
7281 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
7282 (get_group_load_store_type, get_load_store_type): Likewise.
7283 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
7284 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
7285 (vectorizable_conversion, vectorizable_assignment): Likewise.
7286 (vectorizable_shift, vectorizable_operation): Likewise.
7287 (vectorizable_store, vect_is_simple_cond): Likewise.
7288 (vectorizable_condition, vectorizable_comparison): Likewise.
7289 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
7290 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
7291 and move it to the end. Cope with null def_stmt_outs.
7292
7293 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
7294
7295 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
7296
7297 2018-06-29 Jeff Law <law@redhat.com>
7298
7299 * config/v850/v850.c (v850_legitimate_address_p): Handle large
7300 displacements for TARGET_V850E2V3 and newer.
7301 (TARGET_LRA_P): Remove. Defaults to LRA now.
7302 * config/v850/v850.md (sign23byte_load): Remove.
7303 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
7304 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
7305
7306 2018-06-29 Martin Liska <mliska@suse.cz>
7307
7308 PR lto/85759
7309 * coverage.c (coverage_init): Mangle full path name.
7310 * doc/invoke.texi: Document the change.
7311 * gcov-io.c (mangle_path): New.
7312 * gcov-io.h (mangle_path): Likewise.
7313 * gcov.c (mangle_name): Use mangle_path for path mangling.
7314
7315 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7316
7317 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
7318 if starting source register is not even.
7319
7320 2018-06-29 Martin Liska <mliska@suse.cz>
7321
7322 PR tree-optimization/86263
7323 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
7324 Make edge redirection.
7325
7326 2018-06-29 David Malcolm <dmalcolm@redhat.com>
7327
7328 * dumpfile.c (dump_loc): Add indentation based on scope depth.
7329 (dump_scope_depth): New variable.
7330 (get_dump_scope_depth): New function.
7331 (dump_begin_scope): New function.
7332 (dump_end_scope): New function.
7333 * dumpfile.h (get_dump_scope_depth): New declaration.
7334 (dump_begin_scope): New declaration.
7335 (dump_end_scope): New declaration.
7336 (class auto_dump_scope): New class.
7337 (AUTO_DUMP_SCOPE): New macro.
7338 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
7339 AUTO_DUMP_SCOPE.
7340
7341 2018-06-29 Richard Biener <rguenther@suse.de>
7342
7343 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
7344 compute_all_dependences succeeds.
7345 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
7346 exceed --param loop-max-datarefs-for-datadeps.
7347
7348 2018-06-29 Jakub Jelinek <jakub@redhat.com>
7349
7350 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
7351
7352 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
7353
7354 PR target/86348
7355 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
7356 alternative 0 in preferred_for_speed attribute.
7357
7358 2018-06-28 Paul Koning <ni1d@arrl.net>
7359
7360 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
7361 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
7362 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
7363 * config/pdp11/pdp11.md: Correct "length" attribute calculation
7364 for shift insn patterns.
7365
7366 2018-06-28 David Malcolm <dmalcolm@redhat.com>
7367
7368 * cgraph.c (cgraph_node::get_body): Replace assignments to
7369 "dump_file" with calls to set_dump_file.
7370 * dumpfile.c (alt_dump_file): Make static, and group with...
7371 (alt_flags): ...this definition.
7372 (dumps_are_enabled): New variable.
7373 (refresh_dumps_are_enabled): New function.
7374 (set_dump_file): New function.
7375 (set_alt_dump_file): New function.
7376 (gcc::dump_manager::dump_start): Replace assignments to
7377 "dump_file" and "alt_dump_file" with calls to set_dump_file and
7378 set_alt_dump_file.
7379 (gcc::dump_manager::dump_finish): Likewise.
7380 * dumpfile.h (alt_dump_file): Delete decl.
7381 (dumps_are_enabled): New variable decl.
7382 (set_dump_file): New function decl.
7383 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
7384 global.
7385 * tree-nested.c (lower_nested_functions): Replace assignments to
7386 "dump_file" with calls to set_dump_file.
7387
7388 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7389
7390 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
7391 goto_locus of each outgoing edge of each basic block.
7392
7393 2018-06-28 Richard Biener <rguenther@suse.de>
7394
7395 * dwarf2out.c (decl_scope_table): Remove.
7396 (push_decl_scope): Likewise.
7397 (pop_decl_scope): Likewise.
7398 (gen_type_die_for_member): Do not call push/pop_decl_scope.
7399 (gen_struct_or_union_type_die): Likewise.
7400 (gen_tagged_type_die): Likewise.
7401 (dwarf2out_init): Do not initialize decl_scope_table.
7402 (dwarf2out_c_finalize): Do not free it.
7403
7404 2018-06-28 Richard Biener <rguenther@suse.de>
7405
7406 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
7407 deciding whether to not re-use a DIE.
7408
7409 2018-06-28 Richard Biener <rguenther@suse.de>
7410
7411 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
7412 DW_AT_abstract_origin attribute.
7413
7414 2018-06-28 Martin Liska <mliska@suse.cz>
7415
7416 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
7417 Use newly introduced constants.
7418 * tree-switch-conversion.h (struct jump_table_cluster):
7419 Define max_ratio_for_size and max_ratio_for_speed.
7420
7421 2018-06-28 Martin Liska <mliska@suse.cz>
7422
7423 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
7424 Add new checking assert to catch invalid state.
7425 (jump_table_cluster::can_be_handled): Handle single case
7426 clusters.
7427 (jump_table_cluster::is_beneficial): Bail out for such case.
7428 (bit_test_cluster::find_bit_tests):
7429 Add new checking assert to catch invalid state.
7430 (bit_test_cluster::can_be_handled): Handle single case
7431 clusters.
7432 (bit_test_cluster::is_beneficial): Bail out for such case.
7433 (switch_decision_tree::analyze_switch_statement):
7434 Fix comment.
7435
7436 2018-06-28 Martin Liska <mliska@suse.cz>
7437
7438 * common.opt: Introduce -completion option.
7439 * gcc.c (driver_handle_option): Handle it.
7440 (driver::main): Print completions if completion
7441 is set.
7442 * opt-suggestions.c (option_proposer::get_completions):
7443 New function.
7444 (option_proposer::suggest_completion): Likewise.
7445 (option_proposer::find_param_completions): Likewise.
7446 (verify_autocompletions): Likewise.
7447 (test_completion_valid_options): Likewise.
7448 (test_completion_valid_params): Likewise.
7449 (in_completion_p): Likewise.
7450 (empty_completion_p): Likewise.
7451 (test_completion_partial_match): Likewise.
7452 (test_completion_garbage): Likewise.
7453 (opt_proposer_c_tests): Likewise.
7454 * opt-suggestions.h: Declare new functions.
7455 * opts.c (common_handle_option): Handle OPT__completion_.
7456 * selftest-run-tests.c (selftest::run_tests): Add
7457 opt_proposer_c_tests.
7458 * selftest.c (assert_str_startswith): New.
7459 * selftest.h (assert_str_startswith): Likewise.
7460 (opt_proposer_c_tests): New.
7461 (ASSERT_STR_STARTSWITH): Likewise.
7462
7463 2018-06-28 Martin Liska <mliska@suse.cz>
7464
7465 * Makefile.in: Add opt-suggestions.o.
7466 * gcc-main.c: Include opt-suggestions.h.
7467 * gcc.c (driver::driver): Likewise.
7468 (driver::~driver): Remove m_option_suggestions.
7469 (driver::build_option_suggestions): Moved to option_proposer.
7470 (driver::suggest_option): Likewise.
7471 (driver::handle_unrecognized_options): Use option_proposer.
7472 * gcc.h (class driver): Add new memver m_option_proposer.
7473 * opt-suggestions.c: New file.
7474 * opt-suggestions.h: New file.
7475
7476 2018-06-28 Martin Liska <mliska@suse.cz>
7477
7478 * vec.h (class auto_string_vec): New (moved from auto_argvec).
7479 (auto_string_vec::~auto_string_vec): Likewise.
7480
7481 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
7482
7483 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
7484 prevent_decl_creation_for_types fields up and add reset_location field.
7485 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
7486 statement if id->reset_location is true.
7487 (copy_edges_for_bb): Do not set goto_locus on the new edges if
7488 id->reset_location is true.
7489 (copy_phis_for_bb): Force input_location on the arguments if
7490 id->reset_location is true.
7491 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
7492 is set on the function to be inlined.
7493
7494 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
7495
7496 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
7497
7498 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
7499
7500 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
7501 registers for Pmode.
7502 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
7503 hard registers for the clobbered pseudo.
7504
7505 2018-06-27 Paul Koning <ni1d@arrl.net>
7506
7507 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
7508 mutually exclusive options.
7509 * config/pdp11/constraints.md (h): New constraint.
7510 (O): Update definition to match shift code generation.
7511 (D): New constraint.
7512 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
7513 (CCFP): Remove.
7514 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
7515 function.
7516 (output_jump): Change arguments.
7517 (pdp11_fixed_cc_regs): New function.
7518 (pdp11_cc_mode): Ditto.
7519 (pdp11_expand_shift): Ditto.
7520 (pdp11_assemble_shift): Ditto.
7521 (pdp11_small_shift): Ditto.
7522 (pdp11_branch_cost): Remove.
7523 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
7524 from output.
7525 (pdp11_register_move_cost): Update for CC registers.
7526 (pdp11_rtx_costs): Add case for LSHIFTRT.
7527 (pdp11_output_jump): Add CCNZ mode conditional branches.
7528 (notice_update_cc_on_set): Remove.
7529 (pdp11_cc_mode): New function.
7530 (simple_memory_operand): Correct pre/post decrement case.
7531 (no_side_effect_operand): New function.
7532 (pdp11_regno_reg_class): Add CC_REGS class.
7533 (pdp11_fixed_cc_regs): New function.
7534 (pdp11_small_shift): New function.
7535 (pdp11_expand_shift): New function to expand shift insns.
7536 (pdp11_assemble_shift): New function to output shifts.
7537 (pdp11_branch_cost): Remove.
7538 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
7539 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
7540 (WCHAR_TYPE): Ditto.
7541 (PTRDIFF_TYPE): Ditto.
7542 (ADJUST_INSN_LENGTH): New macro.
7543 (FIXED_REGISTERS): Add CC registers.
7544 (CALL_USED_REGISTERS): Ditto.
7545 (reg_class): Ditto.
7546 (REG_CLASS_NAMES): Ditto.
7547 (REG_CLASS_CONTENTS): Ditto.
7548 (SELECT_CC_MODE): Use new function.
7549 (TARGET_FLAGS_REGNUM): New macro.
7550 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
7551 (cc0_reg_rtx): Remove.
7552 (CC_STATUS_MDEP): Remove.
7553 (CC_STATUS_MDEFP_INIT): Remove.
7554 (CC_IN_FPU): Remove.
7555 (NOTICE_UPDATE_CC): Remove.
7556 (REGISTER_NAMES): Add CC registers.
7557 (BRANCH_COST): Change to constant 1.
7558 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
7559 handling.
7560 * config/pdp11/pdp11.opt (mbcopy): Remove.
7561 (mbcopy-builtin): Remove.
7562 (mbranch-cheap): Remove.
7563 (mbranch-expensive): Remove.
7564 * config/pdp11/predicates.md (expand_shift_operand): Update to
7565 match shift code generation.
7566 (ccnz_operator): New predicate.
7567 * doc/invoke.texi (PDP-11 Options): Remove deleted options
7568 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
7569 Remove non-existent option -mabshi, -mno-abshi. Document mutually
7570 exclusive options.
7571 * doc/md.texi (PDP-11): Document new D and h constraints. Update
7572 description of O constraint.
7573
7574 2018-06-27 Jeff Law <law@redhat.com>
7575 Austin Law <austinklaw@gmail.com>
7576
7577 * config/v850/v850.md (addsi3_set_flags): New pattern.
7578 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
7579 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
7580 (zero_extendhisi2_v850_set_flags): Likewise.
7581 (zero_extendqisi2_v850_set_flags): Likewise.
7582 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
7583 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
7584 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
7585
7586 * config/v850/v850-protos.h (notice_update_cc): Remove.
7587 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
7588 (v850_print_operand): Handle 'D' and "d".
7589 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
7590 Add handling of arithmetic/logical operations compared against zero.
7591 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
7592 Do not look at v850_compare_op, instead get mode from last argument.
7593 (v850_gen_compare): Remove
7594 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
7595 after reload for prologue insns.
7596 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
7597 patterns.
7598 (construct_save_jarl): Likewise.
7599 (TARGET_FLAGS_REGNUM): Define.
7600 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
7601 (NOTICE_UPDATE_CC): Remove.
7602 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
7603 than cc0. Conditionalize on reload_completed.
7604 (cmpsi_insn, setfcc_insn): Likewise.
7605 (tst1 splitter): Turn into define_and_split which sets the flags
7606 after reload.
7607 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
7608 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
7609 (cstoresf4, cstoredf4): Clobber the flags.
7610 (cmpsi, cmpsf, cmpdf): Remove expanders.
7611 (setf_insn): Remove pattern.
7612 (addsi3): Turn into define_and_split which clobbers the flags after
7613 reload and a suitable pattern (addsi3_clobber_flags) for use after
7614 reload.
7615 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
7616 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
7617 (ashrsi3, ashrsi3_v850e2): Likewise.
7618 (bins): Clobber the flags.
7619 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
7620 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
7621 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
7622 (call_value_internal_short, call_value_internal_long): Likewise.
7623 (callt_save_interrupt, callt_return_interrupt): Likewise.
7624 (save_interrupt, return_interrupt): Likewise.
7625 (callt_save_all_interrupt, save_all_interrupt): Likewise.
7626 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
7627 (restore_all_interrupt, _restore_all_interrupt): Likewise.
7628 (All FP comparisons): Only allow after reload has completed.
7629 (trfsr): Likewise.
7630 (divh, divhu): Tweak output template.
7631 (branch_z_normal, branch_z_invert): Remove
7632 (branch_nz_normal, branch_nz_invert): Likewise.
7633 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
7634
7635 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
7636 * config/v850/v850.c (notice_update_cc): Remove.
7637 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
7638 (CC_NO_CARRY): Likewise.
7639 (NOTICE_UPDATE_CC): Define to nothing.
7640 * config/v850/v850.md: Remove block comment on cc0 handling
7641 Remove "cc" attribute from all patterns. Remove cc_status handling
7642 from all patterns. Minor formatting fixes.
7643
7644 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7645
7646 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
7647 (cortex-a76.cortex-a55): Likewise.
7648 * config/aarch64/aarch64-tune.md: Regenerate.
7649 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
7650 cortex-a76.cortex-a55.
7651
7652 2018-06-27 Jeff Law <law@redhat.com>
7653
7654 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
7655 (MULTILIB_DIRNAMES): Similarly.
7656
7657 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
7658
7659 * gimple.h (gimple_return_retbnd): Delete.
7660 (gimple_return_set_retbnd): Likewise.
7661 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
7662 gimple_return_set_retbnd.
7663 * gimple-pretty-print.c (dump_gimple_return): Remove call to
7664 gimple_return_retbnd and adjust.
7665 * tree-inline.h (struct copy_body_data): Remove retbnd field.
7666 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
7667 Explicitly return NULL in a couple more cases. Move assertion
7668 on debug statements and remove unreachable code.
7669 (reset_debug_binding): Do not test id->retbnd.
7670 (expand_call_inline): Do not set it.
7671
7672 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
7673
7674 * configure.ac: Add --disable-gcov option.
7675 * configure: Regenerate.
7676 * Makefile.in: Honour @enable_gcov@.
7677 * doc/install.texi: Document --disable-gcov.
7678
7679 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7680
7681 * config/arm/arm-cpus.in (cortex-a76): New entry.
7682 (cortex-a76.cortex-a55): Likewise.
7683 * config/arm/arm-tables.opt: Regenerate.
7684 * config/arm/arm-tune.md: Likewise.
7685 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
7686 * doc/invoke.texi (ARM Options): Document cortex-a76 and
7687 cortex-a76.cortex-a55.
7688
7689 2018-06-27 Tamar Christina <tamar.christina@arm.com>
7690
7691 PR target/85769
7692 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
7693
7694 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
7695
7696 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
7697 comment.
7698 (EPILOGUE_USES): Likewise.
7699
7700 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
7701
7702 * tree-inline.c (remap_location): New function extracted from...
7703 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
7704 (copy_phis_for_bb): ...here. Call remap_location.
7705 (copy_cfg_body): Adjust call to copy_edges_for_bb.
7706
7707 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
7708
7709 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
7710 unaligned vsx for 16B memset.
7711
7712 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
7713
7714 PR target/86285
7715 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
7716 ieee128_float_type_node to long_double_type_node unless
7717 TARGET_LONG_DOUBLE_128 is set.
7718
7719 2018-06-26 David Malcolm <dmalcolm@redhat.com>
7720
7721 * cfgloop.c (get_loop_location): Convert return type from
7722 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
7723 by implicit construction from rtx_insn *, and using
7724 dump_user_location_t::from_function_decl for the fallback case.
7725 * cfgloop.h (get_loop_location): Convert return type from
7726 location_t to dump_user_location_t.
7727 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
7728 dump_printf_loc to pass in a dump_location_t rather than a
7729 location_t, via the gimple stmt.
7730 * coverage.c (get_coverage_counts): Update calls to
7731 dump_printf_loc to pass in dump_location_t rather than a
7732 location_t.
7733 * doc/optinfo.texi (Dump types): Convert example of
7734 dump_printf_loc from taking "locus" to taking "insn". Update
7735 description of the "_loc" calls to cover dump_location_t.
7736 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
7737 "selftest.h".
7738 (dump_user_location_t::dump_user_location_t): New constructors,
7739 from gimple *stmt and rtx_insn *.
7740 (dump_user_location_t::from_function_decl): New function.
7741 (dump_loc): Make static.
7742 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
7743 const dump_location_t &.
7744 (dump_generic_expr_loc): Delete.
7745 (dump_printf_loc): Convert param "loc" from location_t to
7746 const dump_location_t &.
7747 (selftest::test_impl_location): New function.
7748 (selftest::dumpfile_c_tests): New function.
7749 * dumpfile.h: Include "profile-count.h".
7750 (class dump_user_location_t): New class.
7751 (struct dump_impl_location_t): New struct.
7752 (class dump_location_t): New class.
7753 (dump_printf_loc): Convert 2nd param from source_location to
7754 const dump_location_t &.
7755 (dump_generic_expr_loc): Delete.
7756 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
7757 const dump_location_t &.
7758 * gimple-fold.c (fold_gimple_assign): Update call to
7759 dump_printf_loc to pass in a dump_location_t rather than a
7760 location_t, via the gimple stmt.
7761 (gimple_fold_call): Likewise.
7762 * gimple-loop-interchange.cc
7763 (loop_cand::analyze_iloop_reduction_var): Update for change to
7764 check_reduction_path.
7765 (tree_loop_interchange::interchange): Update for change to
7766 find_loop_location.
7767 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
7768 change in return-type of find_loop_location.
7769 (graphite_regenerate_ast_isl): Likewise.
7770 * graphite-optimize-isl.c (optimize_isl): Likewise.
7771 * graphite.c (graphite_transform_loops): Likewise.
7772 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
7773 pass in a dump_location_t rather than a location_t, via the
7774 gimple stmt.
7775 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
7776 * ipa.c (walk_polymorphic_call_targets): Likewise.
7777 * loop-unroll.c (report_unroll): Convert "locus" param from
7778 location_t to dump_location_t.
7779 (decide_unrolling): Update for change to get_loop_location's
7780 return type.
7781 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
7782 location_t to dump_user_location_t.
7783 (grid_find_single_omp_among_assignments_1): Updates calls to
7784 dump_printf_loc to pass in a dump_location_t rather than a
7785 location_t, via the gimple stmt.
7786 (grid_parallel_clauses_gridifiable): Convert "tloc" from
7787 location_t to dump_location_t. Updates calls to dump_printf_loc
7788 to pass in a dump_location_t rather than a location_t, via the
7789 gimple stmt.
7790 (grid_inner_loop_gridifiable_p): Likewise.
7791 (grid_dist_follows_simple_pattern): Likewise.
7792 (grid_gfor_follows_tiling_pattern): Likewise.
7793 (grid_target_follows_gridifiable_pattern): Likewise.
7794 (grid_attempt_target_gridification): Convert initialization
7795 of local "grid" from memset to zero-initialization; FIXME: does
7796 this require C++11? Update call to dump_printf_loc to pass in a
7797 optinfo_location rather than a location_t, via the gimple stmt.
7798 * profile.c (read_profile_edge_counts): Updates call to
7799 dump_printf_loc to pass in a dump_location_t rather than a
7800 location_t
7801 (compute_branch_probabilities): Likewise.
7802 * selftest-run-tests.c (selftest::run_tests): Call
7803 dumpfile_c_tests.
7804 * selftest.h (dumpfile_c_tests): New decl.
7805 * tree-loop-distribution.c (pass_loop_distribution::execute):
7806 Update for change in return type of find_loop_location.
7807 * tree-parloops.c (parallelize_loops): Likewise.
7808 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
7809 "locus" from location_t to dump_user_location_t.
7810 (canonicalize_loop_induction_variables): Likewise.
7811 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
7812 for change in return type of find_loop_location.
7813 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
7814 to dump_printf_loc to pass in a dump_location_t rather than a
7815 location_t, via the stmt.
7816 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
7817 Likewise.
7818 * tree-vect-loop-manip.c (find_loop_location): Convert return
7819 type from source_location to dump_user_location_t.
7820 (vect_do_peeling): Update for above change.
7821 (vect_loop_versioning): Update for change in type of
7822 vect_location.
7823 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
7824 from location_t to dump_user_location_t.
7825 (vect_estimate_min_profitable_iters): Update for change in type
7826 of vect_location.
7827 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
7828 location_t to dump_location_t.
7829 (vect_slp_bb): Update for change in type of vect_location.
7830 * tree-vectorizer.c (vect_location): Convert from source_location
7831 to dump_user_location_t.
7832 (try_vectorize_loop_1): Update for change in vect_location's type.
7833 (vectorize_loops): Likewise.
7834 (increase_alignment): Likewise.
7835 * tree-vectorizer.h (vect_location): Convert from source_location
7836 to dump_user_location_t.
7837 (find_loop_location): Convert return type from source_location to
7838 dump_user_location_t.
7839 (check_reduction_path): Convert 1st param from location_t to
7840 dump_user_location_t.
7841 * value-prof.c (check_counter): Update call to dump_printf_loc to
7842 pass in a dump_user_location_t rather than a location_t; update
7843 call to error_at for change in type of "locus".
7844 (check_ic_target): Update call to dump_printf_loc to
7845 pass in a dump_user_location_t rather than a location_t, via the
7846 call_stmt.
7847
7848 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
7849
7850 * config/s390/s390.h (enum processor_flags): Do not use
7851 default tune parameter when -march was specified.
7852
7853 2018-06-26 Jakub Jelinek <jakub@redhat.com>
7854
7855 PR target/86314
7856 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
7857 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
7858 operands.
7859
7860 2018-06-26 Richard Biener <rguenther@suse.de>
7861
7862 PR tree-optimization/86287
7863 PR bootstrap/86316
7864 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
7865 (vect_analyze_loop): Initialize n_stmts.
7866
7867 2018-06-26 Richard Biener <rguenther@suse.de>
7868
7869 PR middle-end/86271
7870 * fold-const.c (fold_convertible_p): Pointer extension
7871 isn't valid.
7872
7873 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
7874
7875 PR debug/86064
7876 * dwarf2out.c (loc_list_has_views): Adjust comments.
7877 (dw_loc_list): Split single cross-partition range with
7878 nonzero locview.
7879
7880 2018-06-25 Jeff Law <law@redhat.com>
7881
7882 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
7883 on -mbig-switch by default.
7884
7885 * config/v850/predicates.md (const_float_1_operand): Fix match_code
7886 test.
7887 (const_float_0_operand): Remove unused predicate.
7888 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
7889 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
7890 (recipsf2): New expander. Original pattern now called
7891 (recipsf2_insn).
7892 (recipdf2, recipdf2_insn): Similarly.
7893 (rsqrtsf2, rsqrtsf2_insn): Similarly
7894 (rsqrtdf2, rsqrtdf2_insn): Similarly
7895
7896 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
7897
7898 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
7899 Simplify logic for FreeBSD (twice).
7900
7901 2018-06-25 Martin Sebor <msebor@redhat.com>
7902
7903 PR tree-optimization/86204
7904 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
7905 a strnlen result if it's less than the length of the string.
7906
7907 2018-06-25 Martin Sebor <msebor@redhat.com>
7908
7909 PR tree-optimization/85700
7910 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
7911 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
7912 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
7913
7914 2018-06-25 Martin Sebor <msebor@redhat.com>
7915
7916 * doc/extend.texi (Zero-length arrays): Update and clarify.
7917
7918 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
7919
7920 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
7921 added IEEE/IBM long double multilib support on PowerPC little
7922 endian Linux systems.
7923 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
7924 (MULTILIB_DEFAULTS): Likewise.
7925 * config/rs6000/rs6000.c (rs6000_option_override_internal):
7926 Likewise.
7927 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
7928 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
7929 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
7930
7931 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
7932
7933 PR middle-end/86311
7934 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
7935 (REORDER_45): Likewise.
7936
7937 2018-06-25 Jeff Law <law@redhat.com>
7938
7939 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
7940 dividend to 32 bits. Adjust length.
7941 (udivmodhi4): Cleanup output template. Fix length.
7942
7943 2018-06-25 Carl Love <cel@us.ibm.com>
7944
7945 * config/rs6000/vsx.md: Change word selector to prefered location.
7946
7947 2018-06-25 Richard Biener <rguenther@suse.de>
7948
7949 PR tree-optimization/86304
7950 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
7951 epilogue-if-converted loops as well.
7952
7953 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
7954
7955 * lto-section-out.c (lto_begin_section): Do not print section
7956 name for noaddr and unnumbered dumps.
7957
7958 2018-06-25 Richard Biener <rguenther@suse.de>
7959
7960 * tree-vectorizer.h (struct vec_info_shared): New structure
7961 with parts split out from struct vec_info and loop_nest from
7962 struct _loop_vec_info.
7963 (struct vec_info): Adjust accordingly.
7964 (struct _loop_vec_info): Likewise.
7965 (LOOP_VINFO_LOOP_NEST): Adjust.
7966 (LOOP_VINFO_DATAREFS): Likewise.
7967 (LOOP_VINFO_DDRS): Likewise.
7968 (struct _bb_vec_info): Likewise.
7969 (BB_VINFO_DATAREFS): Likewise.
7970 (BB_VINFO_DDRS): Likewise.
7971 (struct _stmt_vec_info): Add dr_aux member.
7972 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
7973 (DR_MISALIGNMENT_UNINITIALIZED): New.
7974 (set_dr_misalignment): Adjust.
7975 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
7976 (vect_analyze_loop): Adjust prototype.
7977 (vect_analyze_loop_form): Likewise.
7978 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
7979 Compute dependences lazily.
7980 (vect_record_base_alignments): Use shared datarefs/ddrs.
7981 (vect_verify_datarefs_alignment): Likewise.
7982 (vect_analyze_data_refs_alignment): Likewise.
7983 (vect_analyze_data_ref_accesses): Likewise.
7984 (vect_analyze_data_refs): Likewise.
7985 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
7986 constructor parameter for shared part.
7987 (vect_analyze_loop_form): Pass in shared part and adjust.
7988 (vect_analyze_loop_2): Pass in storage for the number of
7989 stmts. Move loop nest finding to the caller. Compute
7990 datarefs lazily.
7991 (vect_analyze_loop): Pass in shared part.
7992 (vect_transform_loop): Verify shared datarefs are unchanged.
7993 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
7994 constructor parameter for shared part.
7995 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
7996 (vect_slp_bb): Verify shared datarefs are unchanged before
7997 transform.
7998 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
7999 change.
8000 (new_stmt_vec_info): Initialize DR_AUX misalignment to
8001 DR_MISALIGNMENT_UNINITIALIZED.
8002 * tree-vectorizer.c (vec_info::vec_info): Add constructor
8003 parameter for shared part.
8004 (vec_info::~vec_info): Adjust.
8005 (vec_info_shared::vec_info_shared): New.
8006 (vec_info_shared::~vec_info_shared): Likewise.
8007 (vec_info_shared::save_datarefs): Likewise.
8008 (vec_info_shared::check_datarefs): Likewise.
8009 (try_vectorize_loop_1): Construct shared part live for analyses
8010 of a single loop for multiple vector sizes.
8011 * tree-parloops.c (gather_scalar_reductions): Adjust.
8012
8013 2018-06-25 Richard Biener <rguenther@suse.de>
8014
8015 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
8016 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
8017 (vect_analyze_data_refs): Remove similar code from here and
8018 simplify accordingly.
8019
8020 2018-06-25 Richard Biener <rguenther@suse.de>
8021
8022 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
8023 for reverse storage order accesses rather than asserting
8024 they cannot happen here.
8025
8026 2018-06-25 Tom de Vries <tdevries@suse.de>
8027
8028 PR debug/86257
8029 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
8030 Use data16 instead of .byte for insn prefix.
8031
8032 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
8033
8034 PR C++/86082
8035 * parser.c (make_char_string_pack): Pass this literal chars
8036 through cpp_interpret_string.
8037 (cp_parser_userdef_numeric_literal): Check the result of
8038 make_char_string_pack.
8039
8040 2018-06-24 Maya Rashish <coypu@sdf.org>
8041
8042 * ginclude/stddef.h: Simplify conditions around avoiding
8043 re-definition of __size_t.
8044
8045 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
8046
8047 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
8048 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
8049
8050 2018-06-22 Maya Rashish <coypu@sdf.org>
8051
8052 * doc/invoke.texi (mno-fancy-math-387): Update for changes
8053 made to OpenBSD and NetBSD through the years.
8054
8055 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
8056
8057 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
8058 behavior of vec_pack (vector double, vector double) to match
8059 behavior of vec_float2 (vector double, vector double).
8060
8061 2018-06-22 Olivier Hainque <hainque@adacore.com>
8062
8063 * gimplify.c (gimplify_function_tree): Prevent creation
8064 of a trampoline for the address of the current function
8065 passed to entry/exit instrumentation hooks.
8066
8067 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
8068
8069 PR target/86222
8070 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
8071 correctly.
8072
8073 2018-06-22 Martin Liska <mliska@suse.cz>
8074
8075 PR tree-optimization/86263
8076 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
8077 Bail out if is_enabled is false.
8078 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
8079 New declaration.
8080 (jump_table_cluster::is_enabled): New function.
8081
8082 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
8083
8084 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
8085 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
8086 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
8087 (lto_input_ts_binfo_tree_pointers): Likewise.
8088 * tree-streamer-out.c (streamer_write_tree_bitfields,
8089 write_ts_binfo_tree_pointers): Likewise.
8090 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
8091
8092 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
8093
8094 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
8095
8096 2018-06-22 Martin Liska <mliska@suse.cz>
8097
8098 * symbol-summary.h (get): Make it pure and inline move
8099 functionality from ::get function.
8100 (get): Remove and inline into ::get and ::get_create.
8101 (get_create): Move code from ::get function.
8102
8103 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8104
8105 PR target/85994
8106 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
8107 -x assembler-with-cpp.
8108
8109 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8110
8111 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
8112 _FILE_OFFSET_BITS=64 for C++.
8113
8114 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
8115
8116 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
8117 conversion insn that shows up when pr85657-3.c is compiled using
8118 IEEE 128-bit long double.
8119 (neg<mode>2_internal): Use the correct mode to check whether the
8120 mode is IBM extended.
8121 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
8122 multiply and divide external functions from being created more
8123 than once.
8124
8125 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8126
8127 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
8128 functions.
8129 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
8130 the edge can be forwarded.
8131 (cfg_layout_merge_blocks): Likewise.
8132
8133 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8134
8135 * except.c (finish_eh_generation): Commit edge insertions only after
8136 the EH edges have been redirected from post-landing to landing pads.
8137
8138 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8139
8140 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
8141 create_tmp_var_for to create the FRAME decl.
8142 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
8143
8144 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
8145
8146 * tree-inline.c (copy_edges_for_bb): Minor tweak.
8147 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
8148 debug statement when resetting its value.
8149 (expand_call_inline): Copy the locus of the call onto the assignment
8150 of the return value, if any. Use local variable in more cases.
8151
8152 2018-06-21 Martin Liska <mliska@suse.cz>
8153
8154 * ipa-pure-const.c (propagate_nothrow): Use
8155 funct_state_summaries->get.
8156 (dump_malloc_lattice): Likewise.
8157 (propagate_malloc): Likewise.
8158
8159 2018-06-21 Richard Biener <rguenther@suse.de>
8160
8161 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
8162 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
8163 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
8164 comment.
8165 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
8166 BLOCK_ABSTRACT_ORIGIN unconditionally.
8167
8168 2018-06-21 David Malcolm <dmalcolm@redhat.com>
8169
8170 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
8171 deleting it.
8172 * ipa-reference.c (ipa_reference_c_finalize): Delete
8173 ipa_ref_opt_sum_summaries and set it to NULL.
8174
8175 2018-06-21 Tom de Vries <tdevries@suse.de>
8176
8177 PR tree-optimization/85859
8178 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
8179 test with comment from bb_no_side_effects_p.
8180
8181 2018-06-21 Richard Biener <rguenther@suse.de>
8182
8183 PR tree-optimization/86232
8184 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
8185 max for constant niter.
8186
8187 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8188
8189 * config/aarch64/aarch64-simd.md
8190 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
8191
8192 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
8193
8194 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
8195 Make opernads of the unspec commutative.
8196
8197 2018-06-21 Richard Biener <rguenther@suse.de>
8198
8199 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
8200 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
8201 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
8202 (vect_analyze_data_ref_dependence): Re-order checks to deal with
8203 NULL DR_STEP.
8204 (vect_record_base_alignments): Do not record base alignment
8205 for gathers or scatters.
8206 (vect_compute_data_ref_alignment): Drop return value that is always
8207 true. Bail out early for gathers or scatters.
8208 (vect_enhance_data_refs_alignment): Bail out early for gathers
8209 or scatters.
8210 (vect_find_same_alignment_drs): Likewise.
8211 (vect_analyze_data_refs_alignment): Remove dead code.
8212 (vect_slp_analyze_and_verify_node_alignment): Likewise.
8213 (vect_analyze_data_refs): For possible gathers or scatters do
8214 not create an alternate DR, just check their possible validity
8215 and mark them. Adjust DECL_NONALIASED handling to not rely
8216 on DR_BASE_ADDRESS.
8217 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
8218 update inits of gathers or scatters.
8219 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
8220 Also copy gather/scatter flag to pattern vinfo.
8221
8222 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
8223
8224 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
8225 behavior of vec_packsu (vector unsigned long long, vector unsigned
8226 long long) to match behavior of vec_packs with same signature.
8227
8228 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
8229 Thomas Schwinge <thomas@codesourcery.com>
8230 Cesar Philippidis <cesar@codesourcery.com>
8231
8232 * gimplify.c (gimplify_scan_omp_clauses): Add support for
8233 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8234 (gimplify_adjust_omp_clauses): Likewise.
8235 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
8236 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
8237 (gimplify_omp_target_update): Update handling of acc update and
8238 enter/exit data.
8239 * omp-low.c (install_var_field): Remove unused parameter
8240 base_pointers_restrict.
8241 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
8242 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
8243 FINALIZE}
8244 (omp_target_base_pointers_restrict_p): Delete.
8245 (scan_omp_target): Update call to scan_sharing_clauses.
8246 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
8247 FINALIZE}.
8248 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
8249 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
8250 (convert_local_omp_clauses): Likewise.
8251 * tree-pretty-print.c (dump_omp_clause): Likewise.
8252 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
8253 FINALIZE}.
8254 (omp_clause_code_name): Likewise.
8255
8256 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8257
8258 PR debug/86194
8259 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
8260 be narrowed.
8261
8262 PR tree-optimization/86231
8263 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
8264 anti-range don't overwrite *vr0min before using it to compute *vr0max.
8265
8266 2018-06-20 Tom de Vries <tdevries@suse.de>
8267
8268 PR tree-optimization/86097
8269 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
8270 iv type if signedness of iv type is not the same as that of *nit.
8271
8272 2018-06-20 Jakub Jelinek <jakub@redhat.com>
8273
8274 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
8275 EDGE_EH edges, verify they are all EDGE_EH.
8276
8277 2018-06-20 Maya Rashish <coypu@sdf.org>
8278
8279 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
8280
8281 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8282
8283 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
8284 * config/aarch64/aarch64.c (xgene1_tunings): Add
8285 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
8286 (aarch64_mode_valid_for_sched_fusion_p):
8287 Allow 16-byte modes.
8288 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
8289 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
8290 128-bit modes.
8291 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
8292 New pattern.
8293 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
8294 * config/aarch64/iterators.md (VQ2): New mode iterator.
8295
8296 2018-06-20 Martin Liska <mliska@suse.cz>
8297
8298 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
8299 Change default ratio from 10 to 8.
8300
8301 2018-06-20 Martin Liska <mliska@suse.cz>
8302
8303 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
8304 New.
8305 (bit_test_cluster::find_bit_tests): Likewise.
8306 (switch_decision_tree::analyze_switch_statement): Find clusters.
8307 * tree-switch-conversion.h (struct jump_table_cluster): Document
8308 hierarchy.
8309
8310 2018-06-20 Martin Liska <mliska@suse.cz>
8311
8312 * tree-switch-conversion.c (switch_conversion::collect):
8313 Record m_uniq property.
8314 (switch_conversion::expand): Bail out for special conditions.
8315 (group_cluster::~group_cluster): New.
8316 (group_cluster::group_cluster): Likewise.
8317 (group_cluster::dump): Likewise.
8318 (jump_table_cluster::emit): New.
8319 (switch_decision_tree::fix_phi_operands_for_edges): New.
8320 (struct case_node): Remove struct.
8321 (jump_table_cluster::can_be_handled): New.
8322 (case_values_threshold): Moved to header.
8323 (reset_out_edges_aux): Likewise.
8324 (jump_table_cluster::is_beneficial): New.
8325 (bit_test_cluster::can_be_handled): Likewise.
8326 (add_case_node): Remove.
8327 (bit_test_cluster::is_beneficial): New.
8328 (case_bit_test::cmp): New.
8329 (bit_test_cluster::emit): New.
8330 (expand_switch_as_decision_tree_p): Remove.
8331 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
8332 (fix_phi_operands_for_edge): Likewise.
8333 (switch_decision_tree::analyze_switch_statement): New.
8334 (compute_cases_per_edge): Move ...
8335 (switch_decision_tree::compute_cases_per_edge): ... here.
8336 (try_switch_expansion): Likewise.
8337 (switch_decision_tree::try_switch_expansion): Likewise.
8338 (record_phi_operand_mapping): Likewise.
8339 (switch_decision_tree::record_phi_operand_mapping): Likewise.
8340 (emit_case_decision_tree): Likewise.
8341 (switch_decision_tree::emit): Likewise.
8342 (balance_case_nodes): Likewise.
8343 (switch_decision_tree::balance_case_nodes): Likewise.
8344 (dump_case_nodes): Likewise.
8345 (switch_decision_tree::dump_case_nodes): Likewise.
8346 (emit_jump): Likewise.
8347 (switch_decision_tree::emit_jump): Likewise.
8348 (emit_cmp_and_jump_insns): Likewise.
8349 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
8350 (emit_case_nodes): Likewise.
8351 (switch_decision_tree::emit_case_nodes): Likewise.
8352 (conditional_probability): Remove.
8353 * tree-switch-conversion.h (enum cluster_type): New.
8354 (PRINT_CASE): New.
8355 (struct cluster): Likewise.
8356 (cluster::cluster): Likewise.
8357 (struct simple_cluster): Likewise.
8358 (simple_cluster::simple_cluster): Likewise.
8359 (struct group_cluster): Likewise.
8360 (struct jump_table_cluster): Likewise.
8361 (struct bit_test_cluster): Likewise.
8362 (struct min_cluster_item): Likewise.
8363 (struct case_tree_node): Likewise.
8364 (case_tree_node::case_tree_node): Likewise.
8365 (jump_table_cluster::case_values_threshold): Likewise.
8366 (struct case_bit_test): Likewise.
8367 (struct switch_decision_tree): Likewise.
8368 (struct switch_conversion): Likewise.
8369 (switch_decision_tree::reset_out_edges_aux): Likewise.
8370
8371 2018-06-20 Martin Liska <mliska@suse.cz>
8372
8373 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
8374 (hoist_edge_and_branch_if_true): Likewise.
8375 (expand_switch_using_bit_tests_p): Likewise.
8376 (struct case_bit_test): Likewise.
8377 (case_bit_test_cmp): Likewise.
8378 (emit_case_bit_tests): Likewise.
8379 (switch_conversion::switch_conversion): New class.
8380 (struct switch_conv_info): Remove old struct.
8381 (collect_switch_conv_info): More to ...
8382 (switch_conversion::collect): ... this.
8383 (check_range): Likewise.
8384 (switch_conversion::check_range): Likewise.
8385 (check_all_empty_except_final): Likewise.
8386 (switch_conversion::check_all_empty_except_final): Likewise.
8387 (check_final_bb): Likewise.
8388 (switch_conversion::check_final_bb): Likewise.
8389 (create_temp_arrays): Likewise.
8390 (switch_conversion::create_temp_arrays): Likewise.
8391 (free_temp_arrays): Likewise.
8392 (gather_default_values): Likewise.
8393 (switch_conversion::gather_default_values): Likewise.
8394 (build_constructors): Likewise.
8395 (switch_conversion::build_constructors): Likewise.
8396 (constructor_contains_same_values_p): Likewise.
8397 (switch_conversion::contains_same_values_p): Likewise.
8398 (array_value_type): Likewise.
8399 (switch_conversion::array_value_type): Likewise.
8400 (build_one_array): Likewise.
8401 (switch_conversion::build_one_array): Likewise.
8402 (build_arrays): Likewise.
8403 (switch_conversion::build_arrays): Likewise.
8404 (gen_def_assigns): Likewise.
8405 (switch_conversion::gen_def_assigns): Likewise.
8406 (prune_bbs): Likewise.
8407 (switch_conversion::prune_bbs): Likewise.
8408 (fix_phi_nodes): Likewise.
8409 (switch_conversion::fix_phi_nodes): Likewise.
8410 (gen_inbound_check): Likewise.
8411 (switch_conversion::gen_inbound_check): Likewise.
8412 (process_switch): Use the newly created class.
8413 (switch_conversion::expand): New.
8414 (switch_conversion::~switch_conversion): New.
8415 * tree-switch-conversion.h: New file.
8416
8417 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8418
8419 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
8420 tree-vect-patterns.c.
8421 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
8422 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
8423 (vect_recog_sad_pattern): Likewise.
8424 (vect_recog_widen_sum_pattern): Likewise.
8425 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
8426 (vect_recog_widen_shift_pattern): Remove the type_in argument.
8427 (vect_recog_rotate_pattern): Likewise.
8428 (vect_recog_mult_pattern): Likewise.
8429 (vect_recog_vector_vector_shift_pattern): Likewise.
8430 (vect_recog_divmod_pattern): Likewise.
8431 (vect_recog_mixed_size_cond_pattern): Likewise.
8432 (vect_recog_bool_pattern): Likewise.
8433 (vect_recog_mask_conversion_pattern): Likewise.
8434 (vect_try_gather_scatter_pattern): Likewise.
8435 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
8436 (vect_recog_over_widening_pattern): Likewise.
8437 (vect_recog_gather_scatter_pattern): Likewise.
8438 (vect_recog_func_ptr): Move from tree-vectorizer.h
8439 (vect_vect_recog_func_ptrs): Move further down the file.
8440 (vect_recog_func): Likewise. Remove the third argument.
8441 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
8442 (vect_pattern_recog_1): Expect the pattern function to do any
8443 necessary target tests. Also expect it to provide a vector type.
8444 Remove the type_in handling.
8445
8446 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8447
8448 * tree-vect-patterns.c (vect_pattern_detected): New function.
8449 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
8450 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
8451 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
8452 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
8453 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
8454 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
8455 (vect_recog_mask_conversion_pattern)
8456 (vect_try_gather_scatter_pattern): Likewise.
8457
8458 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8459
8460 * tree-vect-patterns.c (vect_get_internal_def): New function.
8461 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
8462 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
8463 (search_type_for_mask_1): Use it.
8464
8465 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8466
8467 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8468 redundant WIDEN_SUM_EXPR handling.
8469 (vect_recog_sad_pattern): Likewise.
8470
8471 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8472
8473 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
8474 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
8475 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
8476 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
8477 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
8478
8479 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8480
8481 * tree-vect-stmts.c (vectorizable_call): Make sure that we
8482 use the stmt_vec_info of the original bb statement for the
8483 new zero assignment, even if the call is part of a pattern.
8484
8485 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8486
8487 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
8488 that the sequence is attached to the original statement rather
8489 than the pattern statement.
8490 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
8491 PATTERN_DEF_SEQ from the original statement rather than
8492 the main pattern statement.
8493 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
8494 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
8495 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
8496
8497 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
8498
8499 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
8500 definition statements before the early exit for statements that aren't
8501 live or relevant.
8502 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
8503 split out from...
8504 (vect_transform_loop): ...here. Process pattern definition
8505 statements without first checking whether the main pattern
8506 statement is live or relevant.
8507
8508 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
8509
8510 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
8511 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
8512
8513 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
8514
8515 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
8516 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
8517 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
8518 (expand_block_compare): Change select_block_compare_mode call.
8519 (expand_strncmp_align_check): Use new functions, fix comment.
8520 (emit_final_str_compare_gpr): New function.
8521 (expand_strn_compare): Refactor and clean up code.
8522 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
8523
8524 2018-06-19 Tony Reix <tony.reix@atos.com>
8525 Damien Bergamini <damien.bergamini@atos.com>
8526 David Edelsohn <dje.gcc@gmail.com>
8527
8528 * collect2.c (static_obj): New variable.
8529 (static_libs): New variable.
8530 (is_in_list): Uncomment declaration.
8531 (main): Track AIX libraries linked statically.
8532 (is_in_list): Uncomment definition.
8533 (scan_prog_file): Don't add AIX shared libraries initializer
8534 to constructor list if linking statically.
8535
8536 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
8537
8538 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
8539 constant.
8540 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
8541
8542 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
8543
8544 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
8545 blocks.
8546
8547 2018-06-19 Martin Liska <mliska@suse.cz>
8548
8549 * config/i386/i386.c (ix86_can_inline_p): Do not use
8550 ipa_fn_summaries::get_create.
8551 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
8552 get.
8553 (devirtualization_time_bonus): Likewise.
8554 (ipcp_propagate_stage): Likewise.
8555 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
8556 (edge_set_predicate): Likewise.
8557 (evaluate_conditions_for_known_args): Likewise.
8558 (evaluate_properties_for_edge): Likewise.
8559 (ipa_call_summary::reset): Tranform to ...
8560 (ipa_call_summary::~ipa_call_summary): ... this.
8561 (ipa_fn_summary::reset): Transform to ...
8562 (ipa_fn_summary::~ipa_fn_summary): ... this.
8563 (ipa_fn_summary_t::remove): Rename to ...
8564 (ipa_fn_summary_t::remove_callees): ... this.
8565 (ipa_fn_summary_t::duplicate): Use placement new
8566 instead of memory copy.
8567 (ipa_call_summary_t::duplicate): Likewise.
8568 (ipa_call_summary_t::remove): Remove.
8569 (dump_ipa_call_summary): Change get_create to get.
8570 (ipa_dump_fn_summary): Dump only when summary exists.
8571 (analyze_function_body): Use symbol_summary::get instead
8572 of get_create.
8573 (compute_fn_summary): Likewise.
8574 (estimate_edge_devirt_benefit): Likewise.
8575 (estimate_edge_size_and_time): Likewise.
8576 (inline_update_callee_summaries): Likewise.
8577 (remap_edge_change_prob): Likewise.
8578 (remap_edge_summaries): Likewise.
8579 (ipa_merge_fn_summary_after_inlining): Likewise.
8580 (write_ipa_call_summary): Likewise.
8581 (ipa_fn_summary_write): Likewise.
8582 (ipa_free_fn_summary): Likewise.
8583 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
8584 (struct ipa_call_summary): Likewise.
8585 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
8586 of get_create.
8587 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
8588 (estimate_size_after_inlining): Likewise.
8589 (estimate_growth): Likewise.
8590 (growth_likely_positive): Likewise.
8591 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
8592 (inline_call): Likewise.
8593 * ipa-inline.c (caller_growth_limits): Likewise.
8594 (can_inline_edge_p): Likewise.
8595 (can_inline_edge_by_limits_p): Likewise.
8596 (compute_uninlined_call_time): Likewise.
8597 (compute_inlined_call_time): Likewise.
8598 (want_inline_small_function_p): Likewise.
8599 (edge_badness): Likewise.
8600 (update_caller_keys): Likewise.
8601 (update_callee_keys): Likewise.
8602 (inline_small_functions): Likewise.
8603 (inline_to_all_callers_1): Likewise.
8604 (dump_overall_stats): Likewise.
8605 (early_inline_small_functions): Likewise.
8606 (early_inliner): Likewise.
8607 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
8608 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
8609 * ipa-pure-const.c (malloc_candidate_p): Likewise.
8610 * ipa-split.c (execute_split_functions): Likewise.
8611 * symbol-summary.h: Likewise.
8612 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
8613
8614 2018-06-19 Richard Biener <rguenther@suse.de>
8615
8616 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
8617 (vectorize_loops): ... here. Fix dbgcnt handling.
8618 (try_vectorize_loop): Wrap try_vectorize_loop_1.
8619
8620 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
8621
8622 PR target/86197
8623 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
8624 ieee128 argument takes up only one (vector) register, not two (floating
8625 point) registers.
8626
8627 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
8628
8629 * gimplify.c (gimplify_init_constructor): Really never clear for an
8630 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
8631
8632 2018-06-19 Richard Biener <rguenther@suse.de>
8633
8634 PR tree-optimization/86179
8635 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
8636 after failed recognition.
8637
8638 2018-06-18 Martin Sebor <msebor@redhat.com>
8639
8640 PR middle-end/85602
8641 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
8642 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
8643 Handle integer subtraction.
8644 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
8645 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
8646
8647 2018-06-18 David Malcolm <dmalcolm@redhat.com>
8648
8649 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
8650 param from rtx to rtx_insn *.
8651 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
8652 param.
8653 (frv_ifcvt_modify_insn): Likwise.
8654 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
8655 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
8656 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
8657 as_a <rtx_insn *> cast to local "unprotected_region" once
8658 it's been established that it's not NULL or pc_rtx.
8659 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
8660 param "sethi" from rtx to rtx_insn *.
8661 (nds32_group_float_insns): Likewise for param "insn".
8662 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
8663 param.
8664 (vax_output_int_subtract): Likewise.
8665 * config/vax/vax.c (vax_output_int_add): Likewise for param
8666 "insn".
8667 (vax_output_int_subtract): Likewise.
8668 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
8669 (emit_pattern_after): Likewise for param "after".
8670 (emit_insn_after): Likewise.
8671 (emit_jump_insn_after): Likewise.
8672 (emit_call_insn_after): Likewise.
8673 (emit_debug_insn_after): Likewise.
8674 (emit_pattern_before): Likewise for param "before".
8675 (emit_insn_before): Likewise.
8676 (emit_jump_insn_before): Likewise.
8677 * final.c (get_insn_template): Likewise for param "insn", removing
8678 a cast.
8679 * output.h (get_insn_template): Likewise for 2nd param.
8680 * rtl.h (emit_insn_before): Likewise.
8681 (emit_jump_insn_before): Likewise.
8682 (emit_debug_insn_before_noloc): Likewise.
8683 (emit_insn_after): Likewise.
8684 (emit_jump_insn_after): Likewise.
8685 (emit_call_insn_after): Likewise.
8686 (emit_debug_insn_after): Likewise.
8687 (set_insn_deleted): Likewise for param.
8688
8689 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
8690
8691 PR target/85358
8692 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
8693 floating point modes, so that IFmode is numerically greater than
8694 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
8695 to declare the ordering. This prevents IFmode from being
8696 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
8697 machine. Include rs6000-modes.h to share the fractional values
8698 between genmodes* and the rest of the compiler.
8699 (IFmode): Likewise.
8700 (KFmode): Likewise.
8701 (TFmode): Likewise.
8702 * config/rs6000/rs6000-modes.h: New file.
8703 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
8704 meaning of rs6000_long_double_size so that 126..128 selects an
8705 appropriate 128-bit floating point type.
8706 (rs6000_option_override_internal): Likewise.
8707 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
8708 (TARGET_LONG_DOUBLE_128): Change the meaning of
8709 rs6000_long_double_size so that 126..128 selects an appropriate
8710 128-bit floating point type.
8711 (LONG_DOUBLE_TYPE_SIZE): Update comment.
8712 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
8713 source and destination to match the standard usage.
8714 (truncifkf2): Likewise.
8715 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
8716 ISA 2.07 to use an explicit clobber, instead of passing in a
8717 temporary.
8718 (copysign<mode>3_soft): Likewise.
8719
8720 2018-06-18 David Malcolm <dmalcolm@redhat.com>
8721
8722 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
8723 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
8724 (vect_slp_analyze_instance_dependence): Likewise.
8725 (vect_enhance_data_refs_alignment): Likewise.
8726 (vect_analyze_data_refs_alignment): Likewise.
8727 (vect_slp_analyze_and_verify_instance_alignment
8728 (vect_analyze_data_ref_accesses): Likewise.
8729 (vect_prune_runtime_alias_test_list): Likewise.
8730 (vect_analyze_data_refs): Likewise.
8731 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
8732 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
8733 (vect_analyze_scalar_cycles_1): Likewise.
8734 (vect_get_loop_niters): Likewise.
8735 (vect_analyze_loop_form_1): Likewise.
8736 (vect_update_vf_for_slp): Likewise.
8737 (vect_analyze_loop_operations): Likewise.
8738 (vect_analyze_loop): Likewise.
8739 (vectorizable_induction): Likewise.
8740 (vect_transform_loop): Likewise.
8741 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
8742 * tree-vect-slp.c (vect_analyze_slp): Likewise.
8743 (vect_make_slp_decision): Likewise.
8744 (vect_detect_hybrid_slp): Likewise.
8745 (vect_slp_analyze_operations): Likewise.
8746 (vect_slp_bb): Likewise.
8747 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
8748 (vectorizable_bswap): Likewise.
8749 (vectorizable_call): Likewise.
8750 (vectorizable_simd_clone_call): Likewise.
8751 (vectorizable_conversion): Likewise.
8752 (vectorizable_assignment): Likewise.
8753 (vectorizable_shift): Likewise.
8754 (vectorizable_operation): Likewise.
8755 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
8756
8757 2018-06-18 Martin Sebor <msebor@redhat.com>
8758
8759 PR tree-optimization/81384
8760 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
8761 * builtins.c (expand_builtin_strnlen): New function.
8762 (expand_builtin): Call it.
8763 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
8764 * builtins.def (BUILT_IN_STRNLEN): New.
8765 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
8766 Warn for bounds in excess of maximum object size.
8767 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
8768 single-value ranges. Handle strnlen.
8769 (handle_builtin_strlen): Handle strnlen.
8770 (strlen_check_and_optimize_stmt): Same.
8771 * doc/extend.texi (Other Builtins): Document strnlen.
8772
8773 2018-06-18 Maya Rashish <coypu@sdf.org>
8774
8775 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
8776 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
8777 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
8778
8779 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
8780 here to ...
8781 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
8782
8783 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8784
8785 * tree.c (escaped_string::escape): Replace cast to char * by
8786 const_cast<char *> (unescaped).
8787
8788 2018-06-18 Nick Clifton <nickc@redhat.com>
8789
8790 PR 84195
8791 * tree.c (escaped_string): New class. Converts an unescaped
8792 string into its escaped equivalent.
8793 (warn_deprecated_use): Use the new class to convert the
8794 deprecation message, if present.
8795 (test_escaped_strings): New self test.
8796 (test_c_tests): Add test_escaped_strings.
8797 * doc/extend.texi (deprecated): Add a note that the
8798 deprecation message is affected by the -fmessage-length
8799 option, and that control characters will be escaped.
8800 (#pragma GCC error): Document this pragma.
8801 (#pragma GCC warning): Likewise.
8802 * doc/invoke.texi (-fmessage-length): Document this option's
8803 effect on the #warning and #error preprocessor directives and
8804 the deprecated attribute.
8805
8806 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
8807
8808 * tree.c (decl_value_expr_lookup): Revert latest change.
8809 (decl_value_expr_insert): Likewise.
8810
8811 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
8812
8813 * gimplify.c (nonlocal_vlas): Delete.
8814 (nonlocal_vla_vars): Likewise.
8815 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
8816 referenced VLAs.
8817 (gimplify_body): Do not create and destroy nonlocal_vlas.
8818 * tree-nested.c: Include diagnostic.h.
8819 (use_pointer_in_frame): Tweak.
8820 (lookup_field_for_decl): Add assertion and declare the transformation.
8821 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
8822 internal error when the reference is in a wrong context. Do not
8823 create a debug decl by default.
8824 (note_nonlocal_block_vlas): Delete.
8825 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
8826 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
8827 create a debug decl by default.
8828 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
8829 call is in a wrong context.
8830 (fixup_vla_decls): New function.
8831 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
8832 debug variables were created.
8833 * tree.c (decl_value_expr_lookup): Add checking assertion.
8834 (decl_value_expr_insert): Likewise.
8835
8836 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
8837
8838 PR middle-end/82479
8839 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
8840 * tree-scalar-evolution.c (interpret_expr): Likewise.
8841 (expression_expensive_p): Likewise.
8842 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
8843 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
8844 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
8845 (ssa_defined_by_minus_one_stmt_p): New.
8846
8847 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
8848
8849 PR middle-end/64946
8850 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
8851 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
8852 * dojump.c (do_jump): Likewise.
8853 * expr.c (expand_expr_real_2): Check operand type's sign.
8854 * fold-const.c (const_unop): Handle ABSU_EXPR.
8855 (fold_abs_const): Likewise.
8856 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
8857 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
8858 (strip_sign_op_1): Likesise.
8859 * match.pd: Add new pattern to generate ABSU_EXPR.
8860 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
8861 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
8862 * tree-eh.c (operation_could_trap_helper_p): Likewise.
8863 * tree-inline.c (estimate_operator_cost): Likewise.
8864 * tree-pretty-print.c (dump_generic_node): Likewise.
8865 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
8866 * tree.def (ABSU_EXPR): New.
8867
8868 2018-06-16 Jakub Jelinek <jakub@redhat.com>
8869
8870 PR middle-end/86095
8871 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
8872 documented as preserved for backward compatibility only.
8873 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
8874
8875 PR rtl-optimization/86108
8876 * bb-reorder.c (create_forwarder_block): Renamed to ...
8877 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
8878 jump from new landing pad to the second part.
8879 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
8880 Adjust callers.
8881
8882 2018-06-15 Jakub Jelinek <jakub@redhat.com>
8883
8884 PR middle-end/85878
8885 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
8886 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
8887 Only call store_expr for halves if the mode is the same.
8888
8889 PR middle-end/86123
8890 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
8891 Fix up comment formatting.
8892
8893 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
8894
8895 * typed-splay-tree.h (typed_splay_tree::remove): New function.
8896 (typed_splay_tree::closure,
8897 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
8898 (typed_splay_tree::typed_splay_tree,
8899 typed_splay_tree::operator =): Declared private.
8900 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
8901 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
8902 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
8903 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
8904 typed_splay_tree::splay_tree_splay,
8905 typed_splay_tree::splay_tree_foreach_helper,
8906 typed_splay_tree::splay_tree_insert,
8907 typed_splay_tree::splay_tree_remove,
8908 typed_splay_tree::splay_tree_lookup,
8909 typed_splay_tree::splay_tree_predecessor,
8910 typed_splay_tree::splay_tree_successor,
8911 typed_splay_tree::splay_tree_min,
8912 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
8913 (typed_splay_tree::root, typed_splay_tree::comp,
8914 typed_splay_tree::delete_key,
8915 typed_splay_tree::delete_value): New data members.
8916 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
8917 typed_splay_tree::remove.
8918
8919 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
8920
8921 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
8922 -mginv and -mno-ginv to the assembler.
8923 * config/mips/mips.opt (-mcrc): New option.
8924 (-mginv): Likewise.
8925 * doc/invoke.text (-mcrc): Document.
8926 (-mginv): Likewise.
8927
8928 2018-06-15 Nick Clifton <nickc@redhat.com>
8929
8930 PR 84195
8931 * tree.c (escaped_string): New class. Converts an unescaped
8932 string into its escaped equivalent.
8933 (warn_deprecated_use): Use the new class to convert the
8934 deprecation message, if present.
8935 (test_escaped_strings): New self test.
8936 (test_c_tests): Add test_escaped_strings.
8937 * doc/extend.texi (deprecated): Add a note that the
8938 deprecation message is affected by the -fmessage-length
8939 option, and that control characters will be escaped.
8940 (#pragma GCC error): Document this pragma.
8941 (#pragma GCC warning): Likewise.
8942 * doc/invoke.texi (-fmessage-length): Document this option's
8943 effect on the #warning and #error preprocessor directives and
8944 the deprecated attribute.
8945
8946 2018-06-15 Richard Biener <rguenther@suse.de>
8947
8948 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
8949 here, also noting vector size used.
8950 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
8951 size used in MSG_OPTIMIZED_LOCATIONS dump.
8952 (pass_slp_vectorize::execute): Adjust.
8953
8954 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
8955
8956 PR target/85968
8957 * config/arc/arc.c (arc_return_address_register): Fix
8958 if-condition.
8959
8960 2018-06-15 Richard Biener <rguenther@suse.de>
8961
8962 PR middle-end/86159
8963 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
8964 leave useless conversion stripping to force_gimple_operand_gsi.
8965 (gimplify_build2): Likewise.
8966 (gimplify_build1): Likewise.
8967
8968 2018-06-15 Richard Biener <rguenther@suse.de>
8969
8970 PR middle-end/86076
8971 * tree-cfg.c (move_stmt_op): unshare invariant addresses
8972 before adjusting their block.
8973
8974 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
8975
8976 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
8977 multilibs for *-*-rtems*.
8978 * config/riscv/t-rtems: New file.
8979
8980 2018-06-14 Jakub Jelinek <jakub@redhat.com>
8981
8982 PR middle-end/86122
8983 * match.pd ((A +- CST1) +- CST2): Punt if last resort
8984 unsigned_type_for returns NULL.
8985
8986 PR target/85945
8987 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
8988 subregs of multi-word pseudos unless the float mode has word size.
8989
8990 2018-06-14 Richard Biener <rguenther@suse.de>
8991
8992 PR middle-end/86139
8993 * tree-vect-generic.c (build_word_mode_vector_type): Remove
8994 duplicate and harmful type_hash_canon.
8995 * tree.c (type_hash_canon): Assert we didn't find ourselves.
8996
8997 2018-06-14 Richard Biener <rguenther@suse.de>
8998
8999 PR ipa/86124
9000 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
9001 NULL cgraph_node.
9002
9003 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
9004
9005 * config/rtems.h (STDINT_LONG32): Define.
9006
9007 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
9008 Prachi Godbole <prachi.godbole@imgtec.com>
9009
9010 * config/mips/mips-cpus.def: Define P6600.
9011 * config/mips/mips-tables.opt: Regenerate.
9012 * config/mips/mips.c (mips_ucbranch_type): New enum.
9013 (mips_rtx_cost_data): Add support for P6600.
9014 (mips_issue_rate): Likewise.
9015 (mips_multipass_dfa_lookahead): Likewise.
9016 (mips_avoid_hazard): Likewise.
9017 (mips_reorg_process_insns): Likewise.
9018 (mips_classify_branch_p6600): New function.
9019 * config/mips/mips.h (TUNE_P6600): New define.
9020 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
9021 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
9022 * config/mips/mips.md: Include p6600.md.
9023 (processor): Add p6600.
9024 * config/mips/p6600.md: New file.
9025 * doc/invoke.texi: Add p6600 to supported architectures.
9026
9027 2018-06-13 Martin Sebor <msebor@redhat.com>
9028
9029 PR tree-optimization/86114
9030 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
9031 of integer types.
9032 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
9033
9034 2018-06-13 Richard Biener <rguenther@suse.de>
9035
9036 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
9037 Properly set vector type of the intermediate stmt.
9038 * tree-vect-stmts.c (vectorizable_operation): The destination
9039 var always has vectype_out type.
9040
9041 2018-06-13 Jeff Law <law@redhat.com>
9042
9043 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
9044 integer 0 for argument to print_rtl_with_bb.
9045 (rl78_reorg): Likewise.
9046
9047 2018-06-13 David Malcolm <dmalcolm@redhat.com>
9048
9049 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
9050 from rtx to rtx_insn *.
9051 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
9052 "label".
9053 (add_sched_insns_for_speculation): Likewise for local "target",
9054 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
9055 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
9056 from rtx_insn ** to rtx_code_label **.
9057 (reorg_emit_nops): Likewise.
9058 (c6x_reorg): Likewise for local "call_labels".
9059 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
9060 rtx to rtx_insn *.
9061 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
9062 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
9063 the loops over LABEL_REFS.
9064 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
9065 braf_label.
9066 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
9067 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
9068 (split_branches): Strengthen local "olabel" from rtx to
9069 rtx_insn *, adding a safe_as_a cast.
9070 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
9071 to "rtx_insn *".
9072 (add_insn_after): Likewise for first two params.
9073 (add_insn_before): Likewise.
9074 (remove_insn): Likewise for param.
9075 (emit_pattern_before_noloc): Likewise for second and third params.
9076 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
9077 (emit_call_insn_before_noloc): Likewise.
9078 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
9079 to "rtx_insn *".
9080 (emit_barrier_before): Likewise.
9081 (emit_label_before): Strengthen "label" param from "rtx" to
9082 "rtx_code_label *". Strengthen "before" param from "rtx" to
9083 "rtx_insn *".
9084 (emit_insn_after_1): Strengthen "after" param from "rtx" to
9085 "rtx_insn *".
9086 (emit_pattern_after_noloc): Likewise.
9087 (emit_insn_after_noloc): Likewise.
9088 (emit_jump_insn_after_noloc): Likewise.
9089 (emit_call_insn_after_noloc): Likewise.
9090 (emit_debug_insn_after_noloc): Likewise.
9091 (emit_barrier_after): Likewise.
9092 (emit_label_after): Likewise for both params.
9093 (emit_pattern_after_setloc): Likewise for "after" param. Convert
9094 "loc" param from "int" to "location_t".
9095 (emit_insn_after_setloc): Likewise.
9096 (emit_jump_insn_after_setloc): Likewise.
9097 (emit_call_insn_after_setloc): Likewise.
9098 (emit_debug_insn_after_setloc): Likewise.
9099 (emit_pattern_before_setloc): Likewise for "before" param. Convert
9100 "loc" param from "int" to "location_t".
9101 (emit_pattern_before): Convert NULL_RTX to NULL.
9102 (emit_insn_before_setloc): Convert "loc" param from "int" to
9103 "location_t".
9104 (emit_jump_insn_before_setloc): Likewise.
9105 (emit_call_insn_before_setloc): Likewise.
9106 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
9107 rtx_insn *. Convert "loc" param from "int" to "location_t".
9108 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
9109 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
9110 Convert 3rd param from "int" to "location_t".
9111 (emit_barrier_before, emit_barrier_after, next_real_insn):
9112 Strengthen param from rtx to rtx_insn *.
9113 (emit_label_before): Strengthen 1st param from "rtx" to
9114 "rtx_code_label *". Strengthen 2nd param from "rtx" to
9115 "rtx_insn *".
9116 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
9117 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
9118 Strengthen 2nd param from "rtx" to "rtx_insn *".
9119 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
9120 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
9121 Likewise. Convert 3rd param from "int" to "location_t".
9122 (emit_label_after): Strengthen 1st param from "rtx" to
9123 "rtx_code_label *".
9124 (next_real_insn, remove_insn): Strengthen param from "rtx" to
9125 "rtx_insn *".
9126 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
9127 from "rtx" to "rtx_insn *".
9128
9129 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
9130
9131 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
9132 bodies streamed in with -Q.
9133 * dumpfile.c (dump_files): Add lto-stream-out dump file.
9134 * dumpfile.h (tree_dump_index): Add lto_stream_out.
9135 * gimple-streamer-out.c: Include gimple-pretty-print.h
9136 (output_bb): Dump stmts streamed.
9137 * lto-section-out.c: Include print-tree.h
9138 (lto_begin_section): Dump sections created.
9139 (lto_output_decl_index): Dump decl encoded.
9140 * lto-streamer-out.c: Include print-tree.h
9141 (create_output_block): Dump output block created.
9142 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
9143 (output_function): Dump function output.
9144 (output_constructor): Dump constructor streamed.
9145 (write_global_stream): Output indexes encoded.
9146 (produce_asm_for_decls): Dump streams encoded.
9147 * lto-streamer.c (streamer_dump_file): New global var.
9148 * lto-streamer.h (streamer_dump_file): Declare.
9149 * passes.c (ipa_write_summaries): Initialize streamer dump.
9150 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
9151 in.
9152
9153 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
9154
9155 PR target/86048
9156 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
9157 offsets for register save directives. Emit a second batch of save
9158 directives, if need be, when the function accesses prior frames.
9159
9160 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9161
9162 * config/arc/fpu.md (fmasf4): Force operand to register.
9163 (fnmasf4): Likewise.
9164
9165 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9166
9167 * config/arc/arc-protos.h (arc_pad_return): Remove.
9168 * config/arc/arc.c (machine_function): Remove force_short_suffix
9169 and size_reason.
9170 (arc_print_operand): Adjust printing of '&'.
9171 (arc_verify_short): Remove conditional printing of short suffix.
9172 (arc_final_prescan_insn): Remove reference to size_reason.
9173 (pad_return): New function.
9174 (arc_reorg): Call pad_return.
9175 (arc_pad_return): Remove.
9176 (arc_init_machine_status): Remove reference to force_short_suffix.
9177 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
9178 (attr length): When attribute iscompact is true force to 2
9179 regardless; in the case of maybe check if we want to force the
9180 instruction to have 4 bytes length.
9181 (nopv): Change it to generate 4 byte long nop as well.
9182 (blockage): New pattern.
9183 (simple_return): Remove call to arc_pad_return.
9184 (p_return_i): Likewise.
9185
9186 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9187
9188 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
9189
9190 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9191
9192 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
9193 ARC cores.
9194
9195 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9196
9197 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
9198 for ARC700 and ARCv2.
9199
9200 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
9201
9202 PR target/86076
9203 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
9204 operands[2] instead of operands[1].
9205
9206
9207 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9208
9209 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
9210 case, check whether the outer register overlaps an unallocatable
9211 register, not just whether it fits the required class.
9212
9213 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
9214
9215 * poly-int.h (can_div_trunc_p): Add new overload in which all values
9216 are poly_ints.
9217 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
9218 (memrefs_conflict_p): Likewise.
9219 (init_alias_analysis): Likewise.
9220 * cfgexpand.c (expand_debug_expr): Likewise.
9221 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
9222 * cse.c (fold_rtx): Likewise.
9223 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
9224 * expr.c (emit_block_move_hints): Likewise.
9225 (clear_storage_hints, push_block, emit_push_insn): Likewise.
9226 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
9227 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
9228 (emit_group_store): Likewise.
9229 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
9230 to read the PRE/POST_MODIFY increment.
9231 * calls.c (store_one_arg): Use strip_offset.
9232 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
9233 poly_int_rtx_p.
9234 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
9235 by a VEC_SELECT.
9236 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
9237 (simplify_binary_operation_1): Extend CONST_INT handling to
9238 poly_int_rtx_p.
9239 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
9240 than a HOST_WIDE_INT.
9241 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
9242 poly_int64.
9243 (adjust_mems, add_stores): Update accodingly.
9244 (vt_canonicalize_addr): Track polynomial offsets.
9245 (emit_note_insn_var_location): Likewise.
9246 (vt_add_function_parameter): Likewise.
9247 (vt_initialize): Likewise.
9248
9249 2018-06-12 Jeff Law <law@redhat.com>
9250
9251 * config.gcc (alpha*-*-freebsd*): Remove.
9252 * config/alpha/freebsd.h: Remove.
9253
9254 2018-06-12 David Malcolm <dmalcolm@redhat.com>
9255
9256 PR other/69968
9257 * spellcheck-tree.c (levenshtein_distance): Rename to...
9258 (get_edit_distance): ...this, and update for underlying renaming.
9259 * spellcheck-tree.h (levenshtein_distance): Rename to...
9260 (get_edit_distance): ...this.
9261 * spellcheck.c (levenshtein_distance): Rename to...
9262 (get_edit_distance): ...this. Convert from Levenshtein distance
9263 to Damerau-Levenshtein distance by supporting transpositions of
9264 adjacent characters. Rename "v1" to "v_next" and "v0" to
9265 "v_one_ago".
9266 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
9267 (selftest::test_edit_distance_unit_test_oneway): ...this, and
9268 update for underlying renaming.
9269 (selftest::levenshtein_distance_unit_test): Rename to...
9270 (selftest::test_get_edit_distance_unit): ...this, and update for
9271 underlying renaming.
9272 (selftest::test_find_closest_string): Add example from PR 69968
9273 where transposition helps
9274 (selftest::test_metric_conditions): Update for renaming.
9275 (selftest::test_metric_conditions): Likewise.
9276 (selftest::spellcheck_c_tests): Likewise.
9277 * spellcheck.h (levenshtein_distance): Rename both overloads to...
9278 (get_edit_distance): ...this.
9279 (best_match::consider): Update for renaming.
9280
9281 2018-06-12 Martin Sebor <msebor@redhat.com>
9282
9283 PR tree-optimization/85259
9284 * builtins.c (compute_objsize): Handle constant offsets.
9285 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
9286 true iff a warning has been issued.
9287 * gimple.h (gimple_nonartificial_location): New function.
9288 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
9289 gimple_nonartificial_location and handle -Wno-system-headers.
9290 (handle_builtin_stxncpy): Same.
9291
9292 2018-06-12 Martin Sebor <msebor@redhat.com>
9293
9294 PR c/85931
9295 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
9296
9297 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
9298
9299 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
9300 BUILTIN_VEC_XST entries for pointer to double and long long.
9301
9302 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
9303
9304 PR target/85990
9305 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9306 Update comments.
9307 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
9308 Likewise.
9309
9310 2018-06-12 Martin Liska <mliska@suse.cz>
9311
9312 * doc/options.texi: Document IntegerRange.
9313
9314 2018-06-12 Martin Liska <mliska@suse.cz>
9315
9316 * config/i386/i386.opt: Make MPX-related options as Deprecated.
9317 * opt-functions.awk: Handle Deprecated flag.
9318 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
9319 and report error.
9320 (read_cmdline_option): Report warning for a deprecated option.
9321 * opts.h (struct cl_option): Add new field cl_deprecated.
9322 (CL_ERR_DEPRECATED): New.
9323
9324 2018-06-12 Martin Liska <mliska@suse.cz>
9325
9326 * doc/options.texi: Document Deprecated option flag.
9327
9328 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9329
9330 * config/arc/arc-arch.h (arc_extras): New enum.
9331 (arc_cpu_t):Add field extra.
9332 (arc_cpu_types): Consider the extras.
9333 * config/arc/arc-cpus.def: Add extras info.
9334 * config/arc/arc-opts.h (processor_type): Consider extra field.
9335 * config/arc/arc.c (arc_override_options): Handle extra field.
9336
9337 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
9338
9339 * config/arc/arc-arch.h: Update ARC_OPTX macro.
9340 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
9341 field.
9342 * config/arc/arc.c (arc_init): Update pic warning.
9343 (irq_range): Update irq range parsing warnings.
9344 (arc_override_options): Update various warning messages.
9345 (arc_handle_aux_attribute): Likewise.
9346
9347 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
9348
9349 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
9350
9351 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
9352
9353 * doc/sourcebuild.texi: Document usage of line number 0 in verify
9354 compiler messages directives.
9355
9356 2018-06-12 Matthew Fortune <mfortune@gmail.com>
9357
9358 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
9359 * config/mips/mips-tables.opt: Regenerate.
9360 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
9361 mips64r6.
9362 * doc/invoke.texi: Document -march=i6500.
9363
9364 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
9365
9366 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
9367 (i6400_gpmul): Add cpu_unit.
9368 (i6400_gpdiv): Likewise.
9369 (i6400_msa_add_d): Update reservations.
9370 (i6400_msa_int_add) Likewise.
9371 (i6400_msa_short_logic3) Likewise.
9372 (i6400_msa_short_logic2) Likewise.
9373 (i6400_msa_short_logic) Likewise.
9374 (i6400_msa_move) Likewise.
9375 (i6400_msa_cmp) Likewise.
9376 (i6400_msa_short_float2) Likewise.
9377 (i6400_msa_div_d) Likewise.
9378 (i6400_msa_long_logic1) Likewise.
9379 (i6400_msa_long_logic2) Likewise.
9380 (i6400_msa_mult) Likewise.
9381 (i6400_msa_long_float2) Likewise.
9382 (i6400_msa_long_float4) Likewise.
9383 (i6400_msa_long_float5) Likewise.
9384 (i6400_msa_long_float8) Likewise.
9385 (i6400_fpu_fadd): Include frint type.
9386 (i6400_fpu_store): New define_insn_reservation.
9387 (i6400_fpu_load): Likewise.
9388 (i6400_fpu_move): Likewise.
9389 (i6400_fpu_fcmp): Likewise.
9390 (i6400_fpu_fmadd): Likewise.
9391 (i6400_int_mult): Include imul3nc type and update reservation.
9392 (i6400_int_div): Include idiv3 type and update reservation.
9393 (i6400_int_load): Update to check type not move_type.
9394 (i6400_int_store): Likewise.
9395 (i6400_int_prefetch): Set zero latency.
9396
9397 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
9398
9399 * gcc.c: Document new %@{...} sequence.
9400 (LINK_COMMAND_SPEC): Use it for the -L switches.
9401 (cpp_unique_options): Use it for the -I switches.
9402 (at_file_argbuf): New global variable.
9403 (in_at_file): Likewise.
9404 (alloc_args): Create at_file_argbuf.
9405 (clear_args): Truncate at_file_argbuf.
9406 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
9407 (open_at_file): New function.
9408 (close_at_file): Likewise.
9409 (create_at_file): Delete.
9410 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
9411 <'o'>: Likewise.
9412 <'@'>: New case.
9413 (validate_switches_from_spec): Deal with %@{...} sequence.
9414 (validate_switches): Likewise.
9415 (driver::finalize): Call clear_args.
9416
9417 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
9418
9419 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
9420
9421 2018-06-11 Martin Sebor <msebor@redhat.com>
9422
9423 * doc/invoke.texi (-Wall): List -Wc++17-compat.
9424 (Wno-class-memaccess): Add @opindex.
9425 (Wno-templates, Wno-multiple-inheritance): Same.
9426 (Wno-virtual-inheritance, Wno-namespaces): Same.
9427 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
9428 (Wno-format-overflow, Wno-format-truncation): Same.
9429 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
9430 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
9431 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
9432 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
9433 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
9434 (Wno-misspelled-isr): Same.
9435
9436 2018-06-11 Martin Sebor <msebor@redhat.com>
9437
9438 * PR tree-optimization/86083
9439 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
9440
9441 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
9442
9443 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
9444
9445 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
9446
9447 PR target/85755
9448 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
9449 on the correct operand.
9450 (*movdi_internal64): Ditto.
9451
9452 2018-06-11 Martin Liska <mliska@suse.cz>
9453
9454 PR tree-optimization/86089
9455 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
9456
9457 2018-06-11 Julia Koval <julia.koval@intel.com>
9458
9459 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
9460 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
9461 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
9462
9463 2018-06-11 Olivier Hainque <hainque@adacore.com>
9464
9465 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
9466 for Ada with strict dwarf2.
9467
9468 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
9469
9470 PR target/85755
9471 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
9472 addresses.
9473
9474 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
9475
9476 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
9477
9478 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
9479
9480 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
9481 TARGET_ELF.
9482
9483 2018-06-08 Martin Liska <mliska@suse.cz>
9484
9485 * tree-cfg.h (debug_function): Fix argument type to match
9486 implementation.
9487
9488 2018-06-08 Martin Liska <mliska@suse.cz>
9489
9490 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
9491 Remove usage of MPX-related (and removed) fields.
9492 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
9493
9494 2018-06-08 David Malcolm <dmalcolm@redhat.com>
9495
9496 * cfg.c (debug): Use TDF_NONE rather than 0.
9497 * cfghooks.c (debug): Likewise.
9498 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
9499 (struct dump_option_value_info): Convert to...
9500 (struct kv_pair): ...this template type.
9501 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
9502 rather than 0.
9503 (optinfo_verbosity_options): Likewise.
9504 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
9505 OPTGROUP_NONE.
9506 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
9507 than int for "optgroup_flags" param.
9508 (dump_generic_expr_loc): Use dump_flags_t rather than int for
9509 "dump_kind" param.
9510 (dump_dec): Likewise.
9511 (dump_finish): Use TDF_NONE rather than 0.
9512 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
9513 rather than int for "optgroup_flags" param. Use TDF_NONE rather
9514 than 0. Update for change to option_ptr.
9515 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
9516 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
9517 0. Update for changes to optinfo_verbosity_options and
9518 optgroup_options.
9519 (opt_info_switch_p): Convert optgroup_flags from int to
9520 optgroup_flags_t.
9521 (dump_basic_block): Use dump_flags_t rather than int
9522 for "dump_kind" param.
9523 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
9524 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
9525 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
9526 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
9527 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
9528 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
9529 TDF_NONE): Convert from macros to...
9530 (enum dump_flag): ...this new enum.
9531 (dump_flags_t): Update to use enum.
9532 (operator|, operator&, operator~, operator|=, operator&=):
9533 Implement for dump_flags_t.
9534 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
9535 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
9536 Convert from macros to...
9537 (enum optgroup_flag): ...this new enum.
9538 (optgroup_flags_t): New typedef.
9539 (operator|, operator|=): Implement for optgroup_flags_t.
9540 (struct dump_file_info): Convert field "alt_flags" to
9541 dump_flags_t. Convert field "optgroup_flags" to
9542 optgroup_flags_t.
9543 (dump_basic_block): Use dump_flags_t rather than int for param.
9544 (dump_generic_expr_loc): Likewise.
9545 (dump_dec): Likewise.
9546 (dump_register): Convert param "optgroup_flags" to
9547 optgroup_flags_t.
9548 (opt_info_enable_passes): Likewise.
9549 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
9550 than 0.
9551 * gimple-pretty-print.c (debug): Likewise.
9552 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
9553 (merged_store_group::apply_stores): Likewise.
9554 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
9555 * gimple.c (verify_gimple_pp): Likewise.
9556 * graphite-poly.c (print_pbb_body): Likewise.
9557 * passes.c (pass_manager::register_one_dump_file): Convert
9558 local "optgroup_flags" to optgroup_flags_t.
9559 * print-tree.c (print_node): Use TDF_NONE rather than 0.
9560 (debug): Likewise.
9561 (debug_body): Likewise.
9562 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
9563 to optgroup_flags_t.
9564 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
9565 than 0.
9566 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
9567 (convert_mult_to_fma): Likewise.
9568 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
9569 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
9570 * tree-vect-data-refs.c (dump_lower_bound): Convert param
9571 "dump_kind" to dump_flags_t.
9572
9573 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
9574
9575 * config/rs6000/rs6000.c (min, max): Delete.
9576
9577 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
9578
9579 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
9580 -mabi=spe and -mabi=no-spe.
9581
9582 2018-06-08 Martin Liska <mliska@suse.cz>
9583
9584 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
9585 where we expect an existing summary.
9586
9587 2018-06-08 Martin Liska <mliska@suse.cz>
9588
9589 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
9590 * ipa-inline.h (estimate_edge_growth): Likewise.
9591
9592 2018-06-08 Martin Liska <mliska@suse.cz>
9593
9594 * cgraph.c (function_version_hasher::hash): Use
9595 cgraph_node::get_uid ().
9596 (function_version_hasher::equal):
9597 * cgraph.h (cgraph_node::get_uid): New method.
9598 * ipa-inline.c (update_caller_keys): Use
9599 cgraph_node::get_uid ().
9600 (update_callee_keys): Likewise.
9601 * ipa-utils.c (searchc): Likewise.
9602 (ipa_reduced_postorder): Likewise.
9603 * lto-cgraph.c (input_node): Likewise.
9604 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
9605 * symbol-summary.h (symtab_insertion): Likewise.
9606 (symtab_removal): Likewise.
9607 (symtab_duplication): Likewise.
9608 * tree-pretty-print.c (dump_function_header): Likewise.
9609 * tree-sra.c (convert_callers_for_node): Likewise.
9610
9611 2018-06-08 Martin Liska <mliska@suse.cz>
9612
9613 * cgraph.c (symbol_table::create_edge): Always assign a new
9614 unique number.
9615 (symbol_table::free_edge): Do not recycle numbers.
9616 * cgraph.h (cgraph_edge::get): New method.
9617 * symbol-summary.h (symtab_removal): Use it.
9618 (symtab_duplication): Likewise.
9619 (call_summary::hashable_uid): Remove.
9620
9621 2018-06-08 Martin Liska <mliska@suse.cz>
9622
9623 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
9624 (initialize_growth_caches): Remove.
9625 (free_growth_caches): Likewise.
9626 (do_estimate_edge_time): Use edge_growth_cache.
9627 (do_estimate_edge_size): Likewise.
9628 (do_estimate_edge_hints): Likewise.
9629 * ipa-inline.c (reset_edge_caches): Likewise.
9630 (recursive_inlining): Likewise.
9631 (inline_small_functions): Likewise.
9632 * ipa-inline.h (initialize_growth_caches): Remove.
9633 (estimate_edge_size): Likewise.
9634 (estimate_edge_time): Likewise.
9635 (estimate_edge_hints): Likewise.
9636 (reset_edge_growth_cache): Likewise.
9637 * symbol-summary.h (call_summary::remove): New method.
9638
9639 2018-06-08 Martin Liska <mliska@suse.cz>
9640
9641 * ipa-cp.c (class edge_clone_summary): New summary.
9642 (grow_edge_clone_vectors): Remove.
9643 (ipcp_edge_duplication_hook): Remove.
9644 (class edge_clone_summary_t): New call_summary class.
9645 (ipcp_edge_removal_hook): Remove.
9646 (edge_clone_summary_t::duplicate): New function.
9647 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
9648 (create_specialized_node): Likewise.
9649 (ipcp_driver): Initialize edge_clone_summaries and do not
9650 register hooks.
9651
9652 2018-06-08 Martin Liska <mliska@suse.cz>
9653
9654 * symbol-summary.h (get): New function.
9655 (call_summary::m_initialize_when_cloning): New class member.
9656
9657 2018-06-08 Martin Liska <mliska@suse.cz>
9658
9659 * cgraph.c (cgraph_node::remove): Do not recycle uid.
9660 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
9661 (symbol_table::allocate_cgraph_symbol): Do not set uid.
9662 * passes.c (uid_hash_t): Record removed_nodes by their uids.
9663 (remove_cgraph_node_from_order): Use the removed_nodes set.
9664 (do_per_function_toporder): Likwise.
9665 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
9666 instead of summary_uid.
9667 (symtab_removal): Likewise.
9668 (symtab_duplication): Likewise.
9669
9670 2018-06-08 Martin Liska <mliska@suse.cz>
9671
9672 * ipa-cp.c (ipcp_store_bits_results): Use
9673 ipcp_transformation_sum.
9674 (ipcp_store_vr_results): Likewise.
9675 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
9676 to ...
9677 (ipcp_transformation_initialize): ... this.
9678 (ipa_set_node_agg_value_chain):
9679 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
9680 (write_ipcp_transformation_info): Likewise.
9681 (read_ipcp_transformation_info): Likewise.
9682 (ipcp_update_bits): Likewise.
9683 (ipcp_update_vr): Likewise.
9684 (ipcp_transform_function): Likewise.
9685 * ipa-prop.h: Rename ipcp_transformation_summary to
9686 ipcp_transformation.
9687 (class ipcp_transformation_t): New function summary.
9688 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
9689 (ipa_get_agg_replacements_for_node): Likewise.
9690
9691 2018-06-08 Martin Liska <mliska@suse.cz>
9692
9693 * ipa-pure-const.c (struct funct_state_d): Do it class instead
9694 of struct.
9695 (class funct_state_summary_t): New function_summary class.
9696 (has_function_state): Remove.
9697 (get_function_state): Likewise.
9698 (set_function_state): Likewise.
9699 (add_new_function): Likewise.
9700 (funct_state_summary_t::insert): New function.
9701 (duplicate_node_data): Remove.
9702 (remove_node_data): Remove.
9703 (funct_state_summary_t::duplicate): New function.
9704 (register_hooks): Create new funct_state_summaries.
9705 (pure_const_generate_summary): Use it.
9706 (pure_const_write_summary): Likewise.
9707 (pure_const_read_summary): Likewise.
9708 (propagate_pure_const): Likewise.
9709 (propagate_nothrow): Likewise.
9710 (dump_malloc_lattice): Likewise.
9711 (propagate_malloc): Likewise.
9712 (execute): Do not register hooks, just remove summary
9713 instead.
9714 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
9715 constructor.
9716
9717 2018-06-08 Martin Liska <mliska@suse.cz>
9718
9719 * ipa-reference.c (remove_node_data): Remove.
9720 (duplicate_node_data): Likewise.
9721 (class ipa_ref_var_info_summary_t): New class.
9722 (class ipa_ref_opt_summary_t): Likewise.
9723 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
9724 (get_reference_optimization_summary): Use
9725 ipa_ref_opt_sum_summaries.
9726 (set_reference_vars_info): Remove.
9727 (set_reference_optimization_summary): Likewise.
9728 (ipa_init): Create summaries.
9729 (init_function_info): Use function summary.
9730 (ipa_ref_opt_summary_t::duplicate): New function.
9731 (ipa_ref_opt_summary_t::remove): New function.
9732 (get_read_write_all_from_node): Fix GNU coding style.
9733 (propagate): Use function summary.
9734 (write_node_summary_p): Fix GNU coding style.
9735 (stream_out_bitmap): Likewise.
9736 (ipa_reference_read_optimization_summary): Use function summary.
9737 (ipa_reference_c_finalize): Do not release hooks.
9738
9739 2018-06-08 Martin Liska <mliska@suse.cz>
9740
9741 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
9742 (analyze_function_body): Extract multiple calls of get_create.
9743 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
9744 * ipa-inline.c (recursive_inlining): Use ::get method.
9745 * ipa-inline.h (estimate_edge_growth): Likewise.
9746
9747 2018-06-08 Martin Liska <mliska@suse.cz>
9748
9749 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
9750 HSA_INVALID.
9751 (hsa_function_summary::hsa_function_summary): Use the new enum
9752 value.
9753 (hsa_gpu_implementation_p): Use hsa_summaries::get.
9754 * hsa-gen.c (hsa_get_host_function): Likewise.
9755 (get_brig_function_name): Likewise.
9756 * ipa-hsa.c (process_hsa_functions): Likewise.
9757 (ipa_hsa_write_summary): Likewise.
9758 * symbol-summary.h (symtab_duplication): Use ::get function/
9759 (get): New function.
9760
9761 2018-06-08 Martin Liska <mliska@suse.cz>
9762
9763 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
9764 of get.
9765 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
9766 (hsa_register_kernel): Likewise.
9767 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
9768 * hsa-gen.c (hsa_get_host_function): Likewise.
9769 (get_brig_function_name): Likewise.
9770 (generate_hsa): Likewise.
9771 (pass_gen_hsail::execute): Likewise.
9772 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
9773 (devirtualization_time_bonus): Likewise.
9774 (ipcp_propagate_stage): Likewise.
9775 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
9776 (edge_set_predicate): Likewise.
9777 (evaluate_conditions_for_known_args): Likewise.
9778 (evaluate_properties_for_edge): Likewise.
9779 (ipa_fn_summary::reset): Likewise.
9780 (ipa_fn_summary_t::duplicate): Likewise.
9781 (dump_ipa_call_summary): Likewise.
9782 (ipa_dump_fn_summary): Likewise.
9783 (analyze_function_body): Likewise.
9784 (compute_fn_summary): Likewise.
9785 (estimate_edge_devirt_benefit): Likewise.
9786 (estimate_edge_size_and_time): Likewise.
9787 (estimate_calls_size_and_time): Likewise.
9788 (estimate_node_size_and_time): Likewise.
9789 (inline_update_callee_summaries): Likewise.
9790 (remap_edge_change_prob): Likewise.
9791 (remap_edge_summaries): Likewise.
9792 (ipa_merge_fn_summary_after_inlining): Likewise.
9793 (ipa_update_overall_fn_summary): Likewise.
9794 (read_ipa_call_summary): Likewise.
9795 (inline_read_section): Likewise.
9796 (write_ipa_call_summary): Likewise.
9797 (ipa_fn_summary_write): Likewise.
9798 (ipa_free_fn_summary): Likewise.
9799 * ipa-hsa.c (process_hsa_functions): Likewise.
9800 (ipa_hsa_write_summary): Likewise.
9801 (ipa_hsa_read_section): Likewise.
9802 * ipa-icf.c (sem_function::merge): Likewise.
9803 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
9804 (do_estimate_edge_time): Likewise.
9805 (estimate_size_after_inlining): Likewise.
9806 (estimate_growth): Likewise.
9807 (growth_likely_positive): Likewise.
9808 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
9809 (inline_call): Likewise.
9810 * ipa-inline.c (caller_growth_limits): Likewise.
9811 (can_inline_edge_p): Likewise.
9812 (can_inline_edge_by_limits_p): Likewise.
9813 (compute_uninlined_call_time): Likewise.
9814 (compute_inlined_call_time): Likewise.
9815 (want_inline_small_function_p): Likewise.
9816 (edge_badness): Likewise.
9817 (update_caller_keys): Likewise.
9818 (update_callee_keys): Likewise.
9819 (recursive_inlining): Likewise.
9820 (inline_small_functions): Likewise.
9821 (inline_to_all_callers_1): Likewise.
9822 (dump_overall_stats): Likewise.
9823 (early_inline_small_functions): Likewise.
9824 (early_inliner): Likewise.
9825 * ipa-inline.h (estimate_edge_growth): Likewise.
9826 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
9827 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
9828 * ipa-prop.h (IPA_NODE_REF): Likewise.
9829 (IPA_EDGE_REF): Likewise.
9830 * ipa-pure-const.c (malloc_candidate_p): Likewise.
9831 (propagate_malloc): Likewise.
9832 * ipa-split.c (execute_split_functions): Likewise.
9833 * symbol-summary.h: Rename get to get_create.
9834 (get): Likewise.
9835 (get_create): Likewise.
9836 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
9837
9838 2018-06-08 Martin Liska <mliska@suse.cz>
9839
9840 * symbol-summary.h (release): Move definition out of class
9841 declaration.
9842 (symtab_removal): Likewise.
9843 (symtab_duplication): Likewise.
9844
9845 2018-06-08 Martin Liska <mliska@suse.cz>
9846
9847 * symbol-summary.h (function_summary): Move constructor
9848 implementation out of class declaration.
9849 (release): Likewise.
9850 (symtab_insertion): Likewise.
9851 (symtab_removal): Likewise.
9852 (symtab_duplication): Likewise.
9853 (get): Likewise.
9854
9855 2018-06-08 Martin Liska <mliska@suse.cz>
9856
9857 * Makefile.in: Remove support for MPX (macros, related functions,
9858 fields in cgraph_node, ...).
9859 * builtin-types.def (BT_BND): Likewise.
9860 (BT_FN_BND_CONST_PTR): Likewise.
9861 (BT_FN_CONST_PTR_BND): Likewise.
9862 (BT_FN_VOID_PTR_BND): Likewise.
9863 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
9864 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
9865 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
9866 (expand_builtin_mempcpy_with_bounds): Likewise.
9867 (expand_builtin_memset_with_bounds): Likewise.
9868 (expand_builtin_memset_args): Likewise.
9869 (std_expand_builtin_va_start): Likewise.
9870 (expand_builtin): Likewise.
9871 (expand_builtin_with_bounds): Likewise.
9872 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
9873 (DEF_LIB_BUILTIN_CHKP): Likewise.
9874 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
9875 (DEF_CHKP_BUILTIN): Likewise.
9876 (BUILT_IN_MEMCPY): Likewise.
9877 (BUILT_IN_MEMMOVE): Likewise.
9878 (BUILT_IN_MEMPCPY): Likewise.
9879 (BUILT_IN_MEMSET): Likewise.
9880 (BUILT_IN_STPCPY): Likewise.
9881 (BUILT_IN_STRCAT): Likewise.
9882 (BUILT_IN_STRCHR): Likewise.
9883 (BUILT_IN_STRCPY): Likewise.
9884 (BUILT_IN_STRLEN): Likewise.
9885 (BUILT_IN_MEMCPY_CHK): Likewise.
9886 (BUILT_IN_MEMMOVE_CHK): Likewise.
9887 (BUILT_IN_MEMPCPY_CHK): Likewise.
9888 (BUILT_IN_MEMSET_CHK): Likewise.
9889 (BUILT_IN_STPCPY_CHK): Likewise.
9890 (BUILT_IN_STRCAT_CHK): Likewise.
9891 (BUILT_IN_STRCPY_CHK): Likewise.
9892 * calls.c (store_bounds): Likewise.
9893 (emit_call_1): Likewise.
9894 (special_function_p): Likewise.
9895 (maybe_warn_nonstring_arg): Likewise.
9896 (initialize_argument_information): Likewise.
9897 (finalize_must_preallocate): Likewise.
9898 (compute_argument_addresses): Likewise.
9899 (expand_call): Likewise.
9900 * cfgexpand.c (expand_call_stmt): Likewise.
9901 (expand_return): Likewise.
9902 (expand_gimple_stmt_1): Likewise.
9903 (pass_expand::execute): Likewise.
9904 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
9905 (cgraph_node::remove): Likewise.
9906 (cgraph_node::dump): Likewise.
9907 (cgraph_node::verify_node): Likewise.
9908 * cgraph.h (chkp_function_instrumented_p): Likewise.
9909 (symtab_node::get_alias_target): Likewise.
9910 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
9911 (cgraph_local_p): Likewise.
9912 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
9913 (cgraph_edge::rebuild_references): Likewise.
9914 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
9915 (walk_polymorphic_call_targets): Likewise.
9916 (cgraph_node::expand_thunk): Likewise.
9917 (symbol_table::output_weakrefs): Likewise.
9918 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
9919 (ix86_handle_option): Likewise.
9920 * config/i386/constraints.md: Likewise.
9921 * config/i386/i386-builtin-types.def (BND): Likewise.
9922 (VOID): Likewise.
9923 (PVOID): Likewise.
9924 (ULONG): Likewise.
9925 * config/i386/i386-builtin.def (BDESC_END): Likewise.
9926 (BDESC_FIRST): Likewise.
9927 (BDESC): Likewise.
9928 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
9929 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
9930 * config/i386/i386.c (enum reg_class): Likewise.
9931 (ix86_target_string): Likewise.
9932 (ix86_option_override_internal): Likewise.
9933 (ix86_conditional_register_usage): Likewise.
9934 (ix86_valid_target_attribute_inner_p): Likewise.
9935 (ix86_set_indirect_branch_type): Likewise.
9936 (ix86_set_current_function): Likewise.
9937 (ix86_function_arg_regno_p): Likewise.
9938 (init_cumulative_args): Likewise.
9939 (ix86_function_arg_advance): Likewise.
9940 (ix86_function_arg): Likewise.
9941 (ix86_pass_by_reference): Likewise.
9942 (ix86_function_value_regno_p): Likewise.
9943 (ix86_function_value_1): Likewise.
9944 (ix86_function_value_bounds): Likewise.
9945 (ix86_return_in_memory): Likewise.
9946 (ix86_setup_incoming_vararg_bounds): Likewise.
9947 (ix86_va_start): Likewise.
9948 (indirect_thunk_need_prefix): Likewise.
9949 (print_reg): Likewise.
9950 (ix86_print_operand): Likewise.
9951 (ix86_expand_call): Likewise.
9952 (ix86_output_function_return): Likewise.
9953 (reg_encoded_number): Likewise.
9954 (BDESC_VERIFYS): Likewise.
9955 (ix86_init_mpx_builtins): Likewise.
9956 (ix86_init_builtins): Likewise.
9957 (ix86_emit_cmove): Likewise.
9958 (ix86_emit_move_max): Likewise.
9959 (ix86_expand_builtin): Likewise.
9960 (ix86_builtin_mpx_function): Likewise.
9961 (ix86_get_arg_address_for_bt): Likewise.
9962 (ix86_load_bounds): Likewise.
9963 (ix86_store_bounds): Likewise.
9964 (ix86_load_returned_bounds): Likewise.
9965 (ix86_store_returned_bounds): Likewise.
9966 (ix86_class_likely_spilled_p): Likewise.
9967 (ix86_hard_regno_mode_ok): Likewise.
9968 (x86_order_regs_for_local_alloc): Likewise.
9969 (ix86_mitigate_rop): Likewise.
9970 (ix86_bnd_prefixed_insn_p): Likewise.
9971 (ix86_mpx_bound_mode): Likewise.
9972 (ix86_make_bounds_constant): Likewise.
9973 (ix86_initialize_bounds): Likewise.
9974 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
9975 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
9976 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
9977 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
9978 (TARGET_CHKP_BOUND_MODE): Likewise.
9979 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
9980 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
9981 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
9982 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
9983 * config/i386/i386.h (TARGET_MPX): Likewise.
9984 (TARGET_MPX_P): Likewise.
9985 (VALID_BND_REG_MODE): Likewise.
9986 (FIRST_BND_REG): Likewise.
9987 (LAST_BND_REG): Likewise.
9988 (enum reg_class): Likewise.
9989 (BND_REG_P): Likewise.
9990 (BND_REGNO_P): Likewise.
9991 (BNDmode): Likewise.
9992 (ADJUST_INSN_LENGTH): Likewise.
9993 * config/i386/i386.md: Likewise.
9994 * config/i386/i386.opt: Likewise.
9995 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
9996 (defined): Likewise.
9997 (LINK_MPX): Likewise.
9998 (MPX_SPEC): Likewise.
9999 (LIBMPX_SPEC): Likewise.
10000 (LIBMPXWRAPPERS_SPEC): Likewise.
10001 (CHKP_SPEC): Likewise.
10002 * config/i386/predicates.md: Likewise.
10003 * dbxout.c (dbxout_type): Likewise.
10004 * doc/extend.texi: Likewise.
10005 * doc/invoke.texi: Likewise.
10006 * doc/md.texi: Likewise.
10007 * doc/tm.texi: Likewise.
10008 * doc/tm.texi.in: Likewise.
10009 * dwarf2out.c (is_base_type): Likewise.
10010 (gen_formal_types_die): Likewise.
10011 (gen_subprogram_die): Likewise.
10012 (gen_type_die_with_usage): Likewise.
10013 (gen_decl_die): Likewise.
10014 (dwarf2out_late_global_decl): Likewise.
10015 * expr.c (expand_assignment): Likewise.
10016 (emit_storent_insn): Likewise.
10017 (store_expr_with_bounds): Likewise.
10018 (store_expr): Likewise.
10019 (expand_expr_real_1): Likewise.
10020 * expr.h (store_expr_with_bounds): Likewise.
10021 * function.c (use_register_for_decl): Likewise.
10022 (struct bounds_parm_data): Likewise.
10023 (assign_parms_augmented_arg_list): Likewise.
10024 (assign_parm_find_entry_rtl): Likewise.
10025 (assign_parm_is_stack_parm): Likewise.
10026 (assign_parm_load_bounds): Likewise.
10027 (assign_bounds): Likewise.
10028 (assign_parms): Likewise.
10029 (expand_function_start): Likewise.
10030 * gcc.c (CHKP_SPEC): Likewise.
10031 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
10032 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
10033 (wrestrict_dom_walker::check_call): Likewise.
10034 * gimple.c (gimple_build_call_from_tree): Likewise.
10035 * gimple.h (enum gf_mask): Likewise.
10036 (gimple_call_with_bounds_p): Likewise.
10037 (gimple_call_set_with_bounds): Likewise.
10038 * gimplify.c (gimplify_init_constructor): Likewise.
10039 * ipa-cp.c (initialize_node_lattices): Likewise.
10040 (propagate_constants_across_call): Likewise.
10041 (find_more_scalar_values_for_callers_subset): Likewise.
10042 * ipa-hsa.c (process_hsa_functions): Likewise.
10043 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
10044 * ipa-icf.c (sem_function::merge): Likewise.
10045 * ipa-inline.c (early_inliner): Likewise.
10046 * ipa-pure-const.c (warn_function_noreturn): Likewise.
10047 (warn_function_cold): Likewise.
10048 (propagate_pure_const): Likewise.
10049 * ipa-ref.h (enum GTY): Likewise.
10050 * ipa-split.c (find_retbnd): Likewise.
10051 (consider_split): Likewise.
10052 (split_function): Likewise.
10053 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
10054 * ipa.c (walk_polymorphic_call_targets): Likewise.
10055 (symbol_table::remove_unreachable_nodes): Likewise.
10056 (process_references): Likewise.
10057 (cgraph_build_static_cdtor_1): Likewise.
10058 * lto-cgraph.c (lto_output_node): Likewise.
10059 (output_refs): Likewise.
10060 (compute_ltrans_boundary): Likewise.
10061 (input_overwrite_node): Likewise.
10062 (input_node): Likewise.
10063 (input_cgraph_1): Likewise.
10064 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
10065 * passes.c (pass_manager::execute_early_local_passes): Likewise.
10066 (class pass_chkp_instrumentation_passes): Likewise.
10067 (make_pass_chkp_instrumentation_passes): Likewise.
10068 * passes.def: Likewise.
10069 * rtl.h (struct GTY): Likewise.
10070 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
10071 * stor-layout.c (layout_type): Likewise.
10072 * symtab.c: Likewise.
10073 * target.def: Likewise.
10074 * targhooks.c (default_chkp_bound_type): Likewise.
10075 (default_chkp_bound_mode): Likewise.
10076 (default_builtin_chkp_function): Likewise.
10077 (default_chkp_function_value_bounds): Likewise.
10078 (default_chkp_make_bounds_constant): Likewise.
10079 (default_chkp_initialize_bounds): Likewise.
10080 * targhooks.h (default_chkp_bound_type): Likewise.
10081 (default_chkp_bound_mode): Likewise.
10082 (default_builtin_chkp_function): Likewise.
10083 (default_chkp_function_value_bounds): Likewise.
10084 (default_chkp_make_bounds_constant): Likewise.
10085 (default_chkp_initialize_bounds): Likewise.
10086 * toplev.c (compile_file): Likewise.
10087 (process_options): Likewise.
10088 * tree-core.h (DEF_BUILTIN): Likewise.
10089 (DEF_BUILTIN_CHKP): Likewise.
10090 * tree-inline.c (declare_return_variable): Likewise.
10091 (remap_gimple_stmt): Likewise.
10092 (copy_bb): Likewise.
10093 (initialize_inlined_parameters): Likewise.
10094 (expand_call_inline): Likewise.
10095 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
10096 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
10097 (make_pass_ipa_chkp_produce_thunks): Likewise.
10098 (make_pass_chkp): Likewise.
10099 (make_pass_chkp_opt): Likewise.
10100 (make_pass_chkp_instrumentation_passes): Likewise.
10101 * tree-pretty-print.c (dump_generic_node): Likewise.
10102 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
10103 * tree-ssa-dce.c (propagate_necessity): Likewise.
10104 (eliminate_unnecessary_stmts): Likewise.
10105 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
10106 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
10107 * tree-ssa-sccvn.h: Likewise.
10108 * tree-ssa-strlen.c (get_string_length): Likewise.
10109 (valid_builtin_call): Likewise.
10110 (adjust_last_stmt): Likewise.
10111 (handle_builtin_strchr): Likewise.
10112 (handle_builtin_strcpy): Likewise.
10113 (handle_builtin_stxncpy): Likewise.
10114 (handle_builtin_memcpy): Likewise.
10115 (handle_builtin_strcat): Likewise.
10116 (strlen_check_and_optimize_stmt): Likewise.
10117 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
10118 * tree-streamer-in.c: Likewise.
10119 * tree-streamer.c (record_common_node): Likewise.
10120 * tree.c (tree_code_size): Likewise.
10121 (wide_int_to_tree_1): Likewise.
10122 (type_contains_placeholder_1): Likewise.
10123 (build_common_tree_nodes): Likewise.
10124 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
10125 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
10126 (POINTER_BOUNDS_P): Likewise.
10127 (BOUNDED_TYPE_P): Likewise.
10128 (BOUNDED_P): Likewise.
10129 (CALL_WITH_BOUNDS_P): Likewise.
10130 (pointer_bounds_type_node): Likewise.
10131 * value-prof.c (gimple_ic): Likewise.
10132 * var-tracking.c (vt_add_function_parameters): Likewise.
10133 * varasm.c (make_decl_rtl): Likewise.
10134 (assemble_start_function): Likewise.
10135 (output_constant): Likewise.
10136 (maybe_assemble_visibility): Likewise.
10137 * varpool.c (ctor_for_folding): Likewise.
10138 * chkp-builtins.def: Remove.
10139 * ipa-chkp.c: Remove.
10140 * ipa-chkp.h: Remove.
10141 * rtl-chkp.c: Remove.
10142 * rtl-chkp.h: Remove.
10143 * tree-chkp-opt.c: Remove.
10144 * tree-chkp.c: Remove.
10145 * tree-chkp.h: Remove.
10146
10147 2018-06-07 Carl Love <cel@us.ibm.com>
10148
10149 * config/rs6000/vsx.md (vextract_fp_from_shorth,
10150 vextract_fp_from_shortl): Add BE support.
10151
10152 2018-06-07 Paul Koning <ni1d@arrl.net>
10153
10154 * compare-elim.c (try_merge_compare): Don't merge compare if
10155 address contains a side effect.
10156 (try_eliminate_compare): Likewise.
10157
10158 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
10159
10160 * config.gcc: Support "tremont".
10161 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
10162 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
10163 PROCESSOR_TREMONT.
10164 * config/i386/i386.c (m_TREMONT): Define.
10165 (processor_target_table): Add "tremont".
10166 (PTA_TREMONT): Define.
10167 (ix86_lea_outperforms): Add TARGET_TREMONT.
10168 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
10169 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
10170 and M_INTEL_GOLDMONT_PLUS.
10171 (fold_builtin_cpu): Add "tremont".
10172 (ix86_add_stmt_cost): Add TARGET_TREMONT.
10173 (ix86_option_override_internal): Add "tremont".
10174 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
10175 (processor_type): Add PROCESSOR_TREMONT.
10176 * config/i386/x86-tune.def: Add m_TREMONT.
10177 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
10178
10179 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10180
10181 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
10182 symbol defined for msp430i* devices to be lower case.
10183
10184 2018-06-07 Richard Biener <rguenther@suse.de>
10185
10186 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
10187 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
10188 Properly wrap signed arithmetic if overflow wraps.
10189
10190 2018-06-07 Jakub Jelinek <jakub@redhat.com>
10191
10192 PR tree-optimization/69615
10193 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
10194 of a cast from a same precision integral SSA_NAME in a bb dominated
10195 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
10196 cast to utype if rhs2 has already a compatible type.
10197
10198 2018-06-07 Richard Biener <rguenther@suse.de>
10199
10200 PR tree-optimization/85935
10201 * graphite-scop-detection.c (find_params_in_bb): Analyze
10202 condition operands with respect to the correct loop. Assert
10203 the analysis doesn't fail.
10204
10205 2018-06-04 Carl Love <cel@us.ibm.com>
10206
10207 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
10208 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
10209 as it is slightly cheaper.
10210 (first_match_or_eos_index_<mode>):
10211 Calculate index using natural element order.
10212 (first_match_index_<mode>):
10213 Calculate index using natural element order.
10214 (first_match_or_eos_index_<mode>):
10215 Calculate index using natural order.
10216 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
10217 for BE and LE modes.
10218 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
10219 P9V_BUILTIN_VCLZLSBB_V16QI.
10220 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
10221 specific.
10222
10223 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10224
10225 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
10226 indentation and line wrap for many prototypes. Add missing
10227 @smallexample directives around block of prototypes for vec_xl and
10228 vec_xst.
10229
10230 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
10231
10232 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
10233 track if we pass or return IEEE 128-bit floating point.
10234 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
10235 C++ mangling that is compatible with GCC 8.1.
10236 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
10237 (init_cumulative_args): Note if we pass or return IEEE 128-bit
10238 floating point types.
10239 (rs6000_function_arg_advance_1): Likewise.
10240 (rs6000_mangle_type): Optionally generate mangled names that match
10241 what GCC 8.1 generated for IEEE 128-bit floating point types.
10242 (rs6000_globalize_decl_name): If we have an external function that
10243 passes or returns IEEE 128-bit types, generate a weak reference
10244 from the mangled name used in GCC 8.1 to the current mangled
10245 name.
10246 (rs6000_init_builtins): Make __ibm128 use the long double type if
10247 long double is IBM extended double. Make __float128 use the long
10248 double type if long double is IEEE 128-bit.
10249
10250 PR target/85657
10251 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
10252 macro for __ibm128 built-in functions.
10253 (PACK_IF): Add __ibm128 pack/unpack functions.
10254 (UNPACK_IF): Likewise.
10255 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
10256 enable long double built-in functions if long double is IEEE
10257 128-bit floating point.
10258 (rs6000_invalid_builtin): Update long double built-in function
10259 error message.
10260 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
10261 functions, adjust the built-in function to use the long double
10262 built-in function if __ibm128 and long double are the same type.
10263 * doc/extend.texi (PowerPC builtins): Update documention for
10264 __builtin_{,un}pack_longdouble. Add documentation for
10265 __builtin_{,un}pack_ibm128.
10266
10267 2018-06-06 Jim Wilson <jimw@sifive.com>
10268
10269 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
10270 (struct machine_function): New field interrupt_mode.
10271 (riscv_handle_type_attribute): New function. Add forward declaration.
10272 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
10273 (riscv_expand_epilogue): Check interrupt_mode field.
10274 (riscv_set_current_function): Check interrupt attribute args and
10275 set interrupt_mode field.
10276 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
10277 (riscv_sret, riscv_uret): New.
10278 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
10279 new arguments to interrupt attribute.
10280
10281 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
10282
10283 PR target/63177
10284 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
10285 Don't handle -mcpu=power8 if -mpower9-vector is also used.
10286
10287 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
10288
10289 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
10290 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
10291 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
10292 several redundant entries.
10293
10294 2018-06-06 David Malcolm <dmalcolm@redhat.com>
10295
10296 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
10297 type from "rtx" to "rtx_insn *".
10298 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
10299 for local "call_insn", removing cast.
10300 (ix86_expand_call): Likewise, introducing a "call_insn" local.
10301
10302 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
10303
10304 PR tree-optimization/86066
10305 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
10306 for BIT_INSERT_EXPR stores.
10307
10308 2018-06-06 Richard Biener <rguenther@suse.de>
10309
10310 PR tree-optimization/86062
10311 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
10312 component refs ontop
10313 of to be offsetted base.
10314
10315 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10316
10317 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
10318 to be static and remove check on interrupt attribute name.
10319
10320 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
10321
10322 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
10323 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
10324
10325 2018-06-05 Steve Ellcey <sellcey@cavium.com>
10326
10327 PR target/79924
10328 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
10329 second argument.
10330 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
10331 Remove second argument, change how error is called.
10332 (aarch64_layout_arg): Remove second argument from
10333 aarch64_err_no_fpadvsimd call.
10334 (aarch64_init_cumulative_args): Ditto.
10335 (aarch64_gimplify_va_arg_expr): Ditto.
10336 * config/aarch64/aarch64.md (mov<mode>): Ditto.
10337
10338 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
10339
10340 * config/i386/i386.md (simple_return_indirect_internal): New expander.
10341 (*simple_return_indirect_internal<mode>): Rename from
10342 simple_return_indirect_internal. Use W mode iterator.
10343 (rstorssp): New expander.
10344 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
10345 (clrssbsy): New expander.
10346 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
10347
10348 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
10349
10350 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
10351 __typeof__.
10352 (cmse_check_pointed_object): Likewise.
10353
10354 2018-06-05 Martin Liska <mliska@suse.cz>
10355
10356 PR gcov-profile/47618
10357 * doc/invoke.texi: Document how -fprofile-dir format
10358 is extended.
10359
10360 2018-06-05 Richard Biener <rguenther@suse.de>
10361
10362 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
10363 removal pretend DOM info isn't available so we do not update
10364 it and only remove edges, not dominated blocks. Actually free
10365 DOM info in case we removed something. Remove unreachable blocks.
10366 (mfb_keep_latches): Work with either DOM info or marked backedges.
10367 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
10368 first. Mark backedges if DOM info isn't available.
10369 (Re-)compute DOM info after cleanup_control_flow_pre.
10370
10371 2018-06-05 Richard Biener <rguenther@suse.de>
10372
10373 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
10374 (locus_discrim_hasher::hash): Adjust.
10375 (locus_discrim_hasher::equal): Likewise.
10376 (next_discriminator_for_locus): Work on line directly.
10377 (same_line_p): Pass in expanded locus1 as well.
10378 (assign_discriminators): Avoid redundant location expansions.
10379
10380 2018-06-05 Richard Biener <rguenther@suse.de>
10381
10382 PR tree-optimization/86046
10383 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
10384 if required after clearing TREE_ADDRESSABLE.
10385
10386 2018-06-05 Richard Biener <rguenther@suse.de>
10387
10388 PR tree-optimization/86047
10389 * tree-ssa-loop.c (for_each_index): Glob handling of all
10390 decls and constants and really handle all of them.
10391
10392 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10393
10394 PR target/81497
10395 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
10396 qualifier_void_pointer and qualifier_const_void_pointer.
10397 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
10398 (arm_init_builtins): Handle the above.
10399 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
10400 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
10401 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
10402 void intrinsics.
10403
10404 2018-06-05 Martin Liska <mliska@suse.cz>
10405
10406 * auto-profile.c (read_autofdo_file): Do not use
10407 gcov_ctr_summary struct.
10408 (afdo_callsite_hot_enough_for_early_inline): Likewise.
10409 * coverage.c (struct counts_entry): Likewise.
10410 (read_counts_file): Read just single summary entry.
10411 (get_coverage_counts): Use gcov_summary struct.
10412 * coverage.h (get_coverage_counts): Likewise.
10413 * gcov-dump.c (dump_working_sets): Likewise.
10414 (tag_summary): Dump just single summary.
10415 * gcov-io.c (gcov_write_summary): Write just histogram
10416 summary.
10417 (gcov_read_summary): Read just single summary.
10418 (compute_working_sets): Use gcov_summary struct.
10419 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
10420 of GCOV_COUNTERS_SUMMABLE.
10421 (GCOV_COUNTERS_SUMMABLE): Remove.
10422 (GCOV_FIRST_VALUE_COUNTER): Replace with
10423 GCOV_COUNTER_V_INTERVAL.
10424 (struct gcov_ctr_summary): Remove.
10425 (struct gcov_summary): Directly use fields of former
10426 gcov_ctr_summary.
10427 (compute_working_sets): Use gcov_summary struct.
10428 * gcov.c (read_count_file): Do not use ctrs fields.
10429 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
10430 struct.
10431 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
10432 struct.
10433 * profile.c: Likewise.
10434 * profile.h: Likewise.
10435
10436 2018-06-05 Martin Liska <mliska@suse.cz>
10437
10438 PR gcov-profile/84846
10439 * gcov.c (output_lines): Print working directory only
10440 in intermediate format.
10441
10442 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
10443
10444 * config/s390/s390-builtin-types.def: Add void function type.
10445 * config/s390/s390-builtins.def: Use the function type for the
10446 tbeginc builtin.
10447
10448 2018-06-04 Jim Wilson <jimw@sifive.com>
10449
10450 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
10451 to int.
10452 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
10453 and maybe_eh_return. Change regno to unsigned int. Use new args to
10454 handle EH_RETURN_DATA_REGNO registers properly.
10455 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
10456 (riscv_expand_epilogue): Update comment. Change argument name and
10457 type. Update code to use new name and type. Pass new args to
10458 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
10459 EXCEPTION_RETURN.
10460 * config/riscv/riscv.md (NORMAL_RETURN): New.
10461 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
10462 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
10463 (eh_return): Call gen_eh_return_internal and emit barrier.
10464 (eh_return_internal): Call riscv_expand_epilogue.
10465
10466 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
10467
10468 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
10469 bit_insertion field and declare can_be_merged_into method.
10470 (merged_store_group::can_be_merged_into): New method.
10471 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
10472 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
10473 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
10474
10475 2018-06-04 Richard Biener <rguenther@suse.de>
10476
10477 PR tree-optimization/85955
10478 * builtins.c (fold_builtin_sincos): Convert pointers to
10479 destination to appropriate type before dereferencing.
10480
10481 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
10482
10483 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
10484
10485 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
10486
10487 * expr.c (expand_expr_real_1): Force the operand into memory if
10488 its TYPE_MODE is BLKmode and if there is no integer mode for
10489 the number of bits being extracted.
10490
10491 2018-06-04 Jakub Jelinek <jakub@redhat.com>
10492
10493 PR target/85832
10494 PR target/86036
10495 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
10496 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
10497
10498 2018-06-04 Richard Biener <rguenther@suse.de>
10499
10500 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
10501 (cleanup_tree_cfg_noloop): ... single caller. Do
10502 start_recording_case_labels later.
10503
10504 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
10505
10506 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
10507 to _IMMINTRIN_H_INCLUDED.
10508 * config/i386/pconfigintrin.h: Ditto.
10509 * config/i386/waitpkgintrin.h: Ditto.
10510 * config/i386/immintrin.h: Add includes for sgxintrin.h,
10511 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
10512 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
10513 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
10514 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
10515 waitpkgintrin.h and cldemoteintrin.h.
10516
10517 2018-06-04 Richard Biener <rguenther@suse.de>
10518
10519 PR tree-optimization/86038
10520 * tracer.c (find_best_successor): Check probability for
10521 being initialized, bail out if not.
10522
10523 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
10524
10525 PR target/86003
10526 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
10527 of bits to ignore when comparing architectures.
10528
10529 2018-06-04 Jakub Jelinek <jakub@redhat.com>
10530
10531 PR tree-optimization/69615
10532 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
10533 maximum or minimum of the type, try to merge it also as if
10534 range1 is + [-, x - 1] or + [x + 1, -].
10535
10536 PR c++/86025
10537 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
10538
10539 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
10540
10541 PR tree-optimization/86034
10542 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
10543 the unsigned bitfield type in a bit insertion sequence if it does not
10544 have a larger precision than the bitfield size.
10545 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
10546
10547 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
10548
10549 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
10550
10551 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
10552
10553 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
10554 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
10555 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
10556 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
10557
10558 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
10559
10560 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
10561 Disable -fdelete-null-pointer-checks for ELF toolchain.
10562
10563 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
10564 Kito Cheng <kito.cheng@gmail.com>
10565
10566 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
10567 (nds32le-*-*, nds32be-*-*): Integrate checking process.
10568 (nds32*-*-*): Add glibc and uclibc conditions.
10569 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
10570 (TARGET_EXCEPT_UNWIND_INFO): Define.
10571 * config/nds32/elf.h: New file.
10572 * config/nds32/linux.h: New file.
10573 * config/nds32/nds32-elf.opt: New file.
10574 * config/nds32/nds32-linux.opt: New file.
10575 * config/nds32/nds32-fp-as-gp.c
10576 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
10577 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
10578 TARGET_LINUX_ABI.
10579 (nds32_asm_file_end): Ditto.
10580 (nds32_print_operand): Ditto.
10581 (nds32_insert_attributes): Ditto.
10582 (nds32_init_libfuncs): New function.
10583 (TARGET_HAVE_TLS): Define.
10584 (TARGET_INIT_LIBFUNCS): Define.
10585 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
10586 spec content.
10587 (TARGET_ELF): Apply different mcmodel setting.
10588 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
10589 been migrated into elf.h and linux.h files.
10590 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
10591 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
10592 (mcmodel): The content has been migrated into nds32-elf.opt and
10593 nds32-linux.opt files.
10594 * config/nds32/t-elf: New file.
10595 * config/nds32/t-linux: New file.
10596
10597 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
10598 Shiva Chen <shiva0217@gmail.com>
10599
10600 * config/nds32/constants.md (unspec_volatile_element): Add
10601 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
10602 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
10603 optimization.
10604 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
10605 (make_pass_nds32_fp_as_gp): Declare.
10606 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
10607 optmization pass.
10608 (nds32_asm_function_end_prologue): Remove unused asm output.
10609 (nds32_asm_function_begin_epilogue): Remove unused asm output.
10610 (nds32_asm_file_start): Output necessary fp_as_gp information.
10611 (nds32_option_override): Adjust register usage.
10612 (nds32_expand_prologue): Consider fp_as_gp situation.
10613 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
10614 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
10615 (epilogue): Ditto.
10616 (return): Ditto.
10617 (simple_return): Ditto.
10618 (omit_fp_begin): Output special directive for fp_as_gp.
10619 (omit_fp_end): Output special directive for fp_as_gp.
10620 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
10621 mforbid-fp-as-gp): New options.
10622
10623 2018-06-01 Mark Wielaard <mark@klomp.org>
10624
10625 * dwarf2out.c (dwarf2out_finish): Remove generation of
10626 DW_AT_loclists_base.
10627
10628 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
10629
10630 * gimple-ssa-store-merging.c: Include gimple-fold.h.
10631 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
10632 (struct merged_store_group): Add bit_insertion field.
10633 (dump_char_array): Use standard hexadecimal format.
10634 (merged_store_group::merged_store_group): Set bit_insertion to false.
10635 (merged_store_group::apply_stores): Use optimal buffer size. Deal
10636 with BIT_INSERT_EXPR stores. Move up code updating the mask and
10637 also print the mask in the dump file.
10638 (pass_store_merging::gate): Minor tweak.
10639 (imm_store_chain_info::coalesce_immediate): Fix wrong association
10640 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
10641 stores with INTEGER_CST stores.
10642 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
10643 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
10644 and use it throughout. Generate bit insertion sequences if need be.
10645 (pass_store_merging::process_store): Remove redundant condition.
10646 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
10647
10648 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
10649
10650 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
10651 the 128-bit floating point types. Fix function comment.
10652
10653 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10654
10655 * config/aarch64/aarch64-simd.md
10656 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
10657 mnemonics.
10658 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
10659 mnemonics.
10660
10661 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
10662
10663 PR tree-optimization/85989
10664 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
10665 variable.
10666 (backprop::intersect_uses): Check it when deciding whether this
10667 is a backedge reference.
10668 (backprop::process_block): Add each phi to m_visited_phis
10669 after visiting it, then clear it at the end.
10670
10671 2018-06-01 Richard Biener <rguenther@suse.de>
10672
10673 * tree-vectorizer.h (vect_dr_stmt): New function.
10674 (vect_get_load_cost): Adjust.
10675 (vect_get_store_cost): Likewise.
10676 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
10677 Use vect_dr_stmt instead of DR_SMTT.
10678 (vect_record_base_alignments): Likewise.
10679 (vect_calculate_target_alignment): Likewise.
10680 (vect_compute_data_ref_alignment): Likewise and make static.
10681 (vect_update_misalignment_for_peel): Likewise.
10682 (vect_verify_datarefs_alignment): Likewise.
10683 (vector_alignment_reachable_p): Likewise.
10684 (vect_get_data_access_cost): Likewise. Pass down
10685 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
10686 (vect_get_peeling_costs_all_drs): Likewise.
10687 (vect_peeling_hash_get_lowest_cost): Likewise.
10688 (vect_enhance_data_refs_alignment): Likewise.
10689 (vect_find_same_alignment_drs): Likewise.
10690 (vect_analyze_data_refs_alignment): Likewise.
10691 (vect_analyze_group_access_1): Likewise.
10692 (vect_analyze_group_access): Likewise.
10693 (vect_analyze_data_ref_access): Likewise.
10694 (vect_analyze_data_ref_accesses): Likewise.
10695 (vect_vfa_segment_size): Likewise.
10696 (vect_small_gap_p): Likewise.
10697 (vectorizable_with_step_bound_p): Likewise.
10698 (vect_prune_runtime_alias_test_list): Likewise.
10699 (vect_analyze_data_refs): Likewise.
10700 (vect_supportable_dr_alignment): Likewise.
10701 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
10702 (vect_gen_prolog_loop_niters): Likewise.
10703 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
10704 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
10705 modify DR_STMT.
10706 (vect_recog_mask_conversion_pattern): Likewise.
10707 (vect_try_gather_scatter_pattern): Likewise.
10708 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
10709 to vect_get_store_cost.
10710 (vect_get_store_cost): Get stmt_info instead of DR.
10711 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
10712 (vect_get_load_cost): Get stmt_info instead of DR.
10713
10714 2018-06-01 Richard Biener <rguenther@suse.de>
10715
10716 PR middle-end/86017
10717 * gimple-fold.c (var_decl_component_p): Also allow offsetted
10718 vars wrapped in MEM_REFs.
10719
10720 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
10721
10722 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10723 Fix subreg tests so that we only return a choice between
10724 GENERAL_REGS and FP_REGS if the original classes included both.
10725
10726 2018-06-01 Richard Biener <rguenther@suse.de>
10727
10728 PR ipa/85960
10729 * tree-ssa-structalias.c (get_function_part_constraint):
10730 Handle NULL fi->decl.
10731 (find_func_aliases_for_call): Properly handle indirect
10732 fi from direct call.
10733 (find_func_clobbers): Likewise.
10734 (ipa_pta_execute): Likewise.
10735 (create_variable_info_for): For functions that are ifunc_resolver
10736 resolve to a varinfo that contains the result of the resolver call.
10737 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
10738 aliases.
10739
10740 2018-05-31 Michael Collison <michael.collison@arm.com>
10741
10742 * config/aarch64/aarch64.md:
10743 (*fix_to_zero_extenddfdi2): New pattern.
10744 * gcc.target/aarch64/fix_extend1.c: New testcase.
10745
10746 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
10747
10748 PR middle-end/78809
10749 PR middle-end/83026
10750 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
10751 and BUILT_IN_STRNCMP_EQ.
10752 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
10753 BUILT_IN_STRNCMP_EQ.
10754 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
10755 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
10756 (gimple_fold_builtin): Likewise.
10757 * tree-ssa-strlen.c (compute_string_length): New function.
10758 (determine_min_obsize): New function.
10759 (handle_builtin_string_cmp): New function to handle calls to
10760 string compare functions.
10761 (strlen_optimize_stmt): Add handling to builtin string compare
10762 calls.
10763 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
10764 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
10765 * tree.c (build_common_builtin_nodes): Add new defines of
10766 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
10767
10768 2018-05-31 Jakub Jelinek <jakub@redhat.com>
10769
10770 PR target/85984
10771 * bb-reorder.c (pass_partition_blocks::gate): Return false for
10772 functions with naked attribute.
10773
10774 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
10775
10776 * config/i386/sse.md (avx_vec_concat<mode>):
10777 Substitute concat_tg_mode mode attribute with xtg_mode.
10778 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
10779 (concat_tg_mode): Remove mode attribute.
10780
10781 2018-05-31 Martin Sebor <msebor@redhat.com>
10782
10783 PR c/82063
10784 * calls.c (alloc_max_size): Correct a logic error/typo.
10785 Treat excessive arguments as infinite. Warn for invalid arguments.
10786 * doc/invoke.texi (-Walloc-size-larger-than): Update.
10787
10788 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
10789
10790 PR target/85829
10791 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
10792 and movx for Haswell.
10793
10794 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
10795 Cesar Philippidis <cesar@codesourcery.com>
10796
10797 PR middle-end/85879
10798 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
10799 when emitting error on private/firstprivate reductions.
10800 * omp-low.c (lower_omp_target): Avoid reference-type processing
10801 on pointers for firstprivate clause.
10802
10803 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
10804
10805 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
10806 (st1x2): Likewise.
10807 (st1x3): Likewise.
10808 * config/aarch64/aarch64-simd.md
10809 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
10810 (aarch64_ld1_x3_<mode>): Likewise
10811 (aarch64_st1x2<VALLDIF:mode>): Likewise
10812 (aarch64_st1_x2_<mode>): Likewise
10813 (aarch64_st1x3<VALLDIF:mode>): Likewise
10814 (aarch64_st1_x3_<mode>): Likewise
10815 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
10816 (vld1_s8_x3): Likewise.
10817 (vld1_u16_x3): Likewise.
10818 (vld1_s16_x3): Likewise.
10819 (vld1_u32_x3): Likewise.
10820 (vld1_s32_x3): Likewise.
10821 (vld1_u64_x3): Likewise.
10822 (vld1_s64_x3): Likewise.
10823 (vld1_f16_x3): Likewise.
10824 (vld1_f32_x3): Likewise.
10825 (vld1_f64_x3): Likewise.
10826 (vld1_p8_x3): Likewise.
10827 (vld1_p16_x3): Likewise.
10828 (vld1_p64_x3): Likewise.
10829 (vld1q_u8_x3): Likewise.
10830 (vld1q_s8_x3): Likewise.
10831 (vld1q_u16_x3): Likewise.
10832 (vld1q_s16_x3): Likewise.
10833 (vld1q_u32_x3): Likewise.
10834 (vld1q_s32_x3): Likewise.
10835 (vld1q_u64_x3): Likewise.
10836 (vld1q_s64_x3): Likewise.
10837 (vld1q_f16_x3): Likewise.
10838 (vld1q_f32_x3): Likewise.
10839 (vld1q_f64_x3): Likewise.
10840 (vld1q_p8_x3): Likewise.
10841 (vld1q_p16_x3): Likewise.
10842 (vld1q_p64_x3): Likewise.
10843 (vst1_s64_x2): Likewise.
10844 (vst1_u64_x2): Likewise.
10845 (vst1_f64_x2): Likewise.
10846 (vst1_s8_x2): Likewise.
10847 (vst1_p8_x2): Likewise.
10848 (vst1_s16_x2): Likewise.
10849 (vst1_p16_x2): Likewise.
10850 (vst1_s32_x2): Likewise.
10851 (vst1_u8_x2): Likewise.
10852 (vst1_u16_x2): Likewise.
10853 (vst1_u32_x2): Likewise.
10854 (vst1_f16_x2): Likewise.
10855 (vst1_f32_x2): Likewise.
10856 (vst1_p64_x2): Likewise.
10857 (vst1q_s8_x2): Likewise.
10858 (vst1q_p8_x2): Likewise.
10859 (vst1q_s16_x2): Likewise.
10860 (vst1q_p16_x2): Likewise.
10861 (vst1q_s32_x2): Likewise.
10862 (vst1q_s64_x2): Likewise.
10863 (vst1q_u8_x2): Likewise.
10864 (vst1q_u16_x2): Likewise.
10865 (vst1q_u32_x2): Likewise.
10866 (vst1q_u64_x2): Likewise.
10867 (vst1q_f16_x2): Likewise.
10868 (vst1q_f32_x2): Likewise.
10869 (vst1q_f64_x2): Likewise.
10870 (vst1q_p64_x2): Likewise.
10871 (vst1_s64_x3): Likewise.
10872 (vst1_u64_x3): Likewise.
10873 (vst1_f64_x3): Likewise.
10874 (vst1_s8_x3): Likewise.
10875 (vst1_p8_x3): Likewise.
10876 (vst1_s16_x3): Likewise.
10877 (vst1_p16_x3): Likewise.
10878 (vst1_s32_x3): Likewise.
10879 (vst1_u8_x3): Likewise.
10880 (vst1_u16_x3): Likewise.
10881 (vst1_u32_x3): Likewise.
10882 (vst1_f16_x3): Likewise.
10883 (vst1_f32_x3): Likewise.
10884 (vst1_p64_x3): Likewise.
10885 (vst1q_s8_x3): Likewise.
10886 (vst1q_p8_x3): Likewise.
10887 (vst1q_s16_x3): Likewise.
10888 (vst1q_p16_x3): Likewise.
10889 (vst1q_s32_x3): Likewise.
10890 (vst1q_s64_x3): Likewise.
10891 (vst1q_u8_x3): Likewise.
10892 (vst1q_u16_x3): Likewise.
10893 (vst1q_u32_x3): Likewise.
10894 (vst1q_u64_x3): Likewise.
10895 (vst1q_f16_x3): Likewise.
10896 (vst1q_f32_x3): Likewise.
10897 (vst1q_f64_x3): Likewise.
10898 (vst1q_p64_x3): Likewise.
10899
10900 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
10901
10902 * config/msp430/msp430.c (msp430_output_labelref): Prepend
10903 user_label_prefix to name.
10904
10905 * tree-core.h: Update comment about the format of NAME string
10906 passed to handler in attribute_spec.
10907
10908 * config/msp430/msp430.md: Remove erroneous subreg expression from
10909 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
10910 zero_extend{q,h}isi2.
10911
10912 2018-05-30 Borislav Petkov <bp@suse.de>
10913
10914 * doc/extend.texi: Document some architecture specific
10915 constraints and sort entries.
10916
10917 2018-05-30 Martin Sebor <msebor@redhat.com>
10918
10919 PR middle-end/85369
10920 * builtins.c (expand_builtin_stpcpy_1): New function.
10921 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
10922 only if the former succeeds.
10923
10924 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
10925
10926 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
10927 in saphira.
10928
10929 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10930
10931 * doc/invoke.texi (-flinker-output): Document
10932
10933 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10934
10935 * passes.c (ipa_write_summaries): Only modify statements if body
10936 is in memory.
10937 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
10938 incrementally linking.
10939 (ipa_passes): Likewise.
10940 * lto-cgraph.c (lto_output_node): When incrementally linking do not
10941 pass down resolution info.
10942 * common.opt (flag_incremental_link): Update info.
10943 * gcc.c (plugin specs): Turn flinker-output=* to
10944 -plugin-opt=-linker-output-known
10945 * toplev.c (compile_file): Also cut compilation when doing incremental
10946 link.
10947 * flag-types. (enum lto_partition_model): Add
10948 LTO_LINKER_OUTPUT_NOLTOREL.
10949 (invoke.texi): Add -flinker-output docs.
10950 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
10951 link same way as WPA; do not stream in dead initializers.
10952
10953 * dwarf2out.c (dwarf2out_die_ref_for_decl,
10954 darf2out_register_external_decl): Support incremental link.
10955
10956 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10957
10958 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
10959
10960 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
10961
10962 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
10963 it down to simple_object_copy_lto_debug_sections.
10964 (run_gcc): Determine incremental LTO link time and configure
10965 lto1 into non-wpa mode, disable renaming of debug sections.
10966
10967 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
10968
10969 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
10970 descriptions of various incorrectly documented functions.
10971
10972 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
10973
10974 Revert:
10975 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
10976 address check not strict.
10977
10978 2018-05-30 Richard Biener <rguenther@suse.de>
10979
10980 PR tree-optimization/85964
10981 * tracer.c (better_p): Drop initialized count check, we only
10982 call the function with initialized counts now.
10983 (find_best_successor): Do find a best edge if one
10984 has uninitialized count.
10985 (find_best_predecessor): Likewise. Do BB frequency check only
10986 if count is initialized.
10987
10988 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
10989
10990 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
10991 (aarch64_ldrstr_offset_compare): New.
10992 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
10993 load/store orderings.
10994 (aarch64_gen_adjusted_ldpstp): Likewise.
10995
10996 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
10997
10998 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
10999 Check for subset of GENERAL_REGS and FP_REGS.
11000 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
11001 r=w alternative.
11002
11003 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
11004
11005 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
11006 and wi::to_poly_offset. Add the current offset and then check
11007 whether the sum fits, rather than using an unchecked addition of
11008 a checked term. Check for a shwi rather than a uhwi.
11009 * expr.c (get_bit_range): Use tree_to_poly_uint64.
11010 (store_constructor): Use poly_int_tree_p.
11011 (expand_expr_real_1): Likewise.
11012 * function.c (assign_temp): Likewise.
11013 * fold-const.c (const_binop): Use poly_int_tree_p and
11014 wi::to_poly_offset.
11015 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
11016 division.
11017 * ipa-icf-gimple.c (func_checker::compare_operand): Use
11018 to_poly_offset for MEM offsets.
11019 * ipa-icf.c (sem_variable::equals): Likewise.
11020 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
11021 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
11022 wi::to_poly_offset for BIT_FIELD_REF offsets.
11023 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
11024 wi::to_poly_offset.
11025 * var-tracking.c (emit_note_insn_var_location): Use
11026 tree_to_poly_uint64.
11027
11028 2018-05-29 Jim Wilson <jimw@sifive.com>
11029
11030 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
11031
11032 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
11033
11034 PR target/85950
11035 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
11036 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
11037 sequence.
11038 (sse4_1_round<mode>2): Use nonimmediate_operand
11039 for operand 1 predicate.
11040
11041 2018-05-29 Martin Sebor <msebor@redhat.com>
11042 Richard Biener <rguenther@suse.de>
11043
11044 PR testsuite/85888
11045 * calls.c (get_size_range): Call determine_value_range instead
11046 of get_value_range..
11047 * tree-vrp.h (determine_value_range): Declared new function.
11048 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
11049
11050 2018-05-29 Richard Biener <rguenther@suse.de>
11051
11052 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
11053 sure to use non-pattern stmts for get_earlier_stmt arguments.
11054 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
11055 called on pattern stmts.
11056 (get_later_stmt): Likewise.
11057
11058 2018-05-29 Martin Liska <mliska@suse.cz>
11059
11060 PR gcov-profile/85759
11061 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
11062 env variables.
11063
11064 2018-05-29 Jakub Jelinek <jakub@redhat.com>
11065
11066 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
11067 VEC_UNPACK_*_EXPR.
11068 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
11069 VEC_PACK_*_EXPR.
11070
11071 PR target/85918
11072 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
11073 VEC_PACK_FLOAT_EXPR): New tree codes.
11074 * tree-pretty-print.c (op_code_prio): Handle
11075 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
11076 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
11077 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
11078 * tree-inline.c (estimate_operator_cost): Likewise.
11079 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
11080 * fold-const.c (const_binop): Likewise.
11081 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
11082 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
11083 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
11084 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
11085 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
11086 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
11087 * expr.c (expand_expr_real_2): Likewise.
11088 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
11089 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
11090 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
11091 optabs.
11092 * optabs.c (expand_widen_pattern_expr): For
11093 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
11094 sign from result type rather than operand's type.
11095 (expand_binop_directly): For vec_packu_float_optab and
11096 vec_packs_float_optab allow result type to be different from operand's
11097 type.
11098 * optabs-tree.c (optab_for_tree_code): Handle
11099 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
11100 VEC_PACK_FLOAT_EXPR. Formatting fixes.
11101 * tree-vect-generic.c (expand_vector_operations_1): Handle
11102 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
11103 VEC_PACK_FLOAT_EXPR.
11104 * tree-vect-stmts.c (supportable_widening_operation): Handle
11105 FIX_TRUNC_EXPR.
11106 (supportable_narrowing_operation): Handle FLOAT_EXPR.
11107 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
11108 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
11109 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
11110 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
11111 mode attributes.
11112 (vec_pack<floatprefix>_float_<mode>): New expander.
11113 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
11114 attributes.
11115 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
11116 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
11117 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
11118 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
11119 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
11120 Document.
11121 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
11122 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
11123 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
11124 VEC_PACK_FLOAT_EXPR): Document.
11125
11126 2018-05-29 Richard Biener <rguenther@suse.de>
11127
11128 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
11129 member.
11130 (stmt_vec_info_vec): Make pointer.
11131 (init_stmt_vec_info_vec): Remove.
11132 (free_stmt_vec_info_vec): Likewise.
11133 (set_stmt_vec_info_vec): New function.
11134 (free_stmt_vec_infos): Likewise.
11135 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
11136 (set_vinfo_for_stmt): Likewise.
11137 (get_earlier_stmt): Likewise.
11138 (get_later_stmt): Likewise.
11139 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
11140 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
11141 (vec_info::~vec_info): Free stmt_vec_infos.
11142 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
11143 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
11144 (pass_slp_vectorize::execute): Likewise.
11145 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
11146 (free_stmt_vec_info_vec): Likewise.
11147 (set_stmt_vec_info_vec): New function.
11148 (free_stmt_vec_infos): Likewise.
11149 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
11150 the global stmt_vec_info_vec.
11151 * tree-parloops.c (gather_scalar_reductions): Use
11152 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
11153 vector.
11154
11155 2018-05-29 Richard Biener <rguenther@suse.de>
11156
11157 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
11158
11159 2018-05-29 Martin Liska <mliska@suse.cz>
11160 David Malcolm <dmalcolm@redhat.com>
11161
11162 * vec.c (test_reverse): New.
11163 (vec_c_tests): Add new test.
11164 * vec.h (vl_ptr>::reverse): New function.
11165
11166 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
11167
11168 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
11169
11170 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
11171 and later.
11172
11173 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
11174
11175 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
11176
11177 2018-05-28 Richard Biener <rguenther@suse.de>
11178
11179 PR tree-optimization/85933
11180 * tree-vect-data-refs.c (vect_record_base_alignments): Only
11181 look at stmts marked as vectorizable.
11182
11183 2018-05-28 Richard Biener <rguenther@suse.de>
11184
11185 PR tree-optimization/85934
11186 * tree-vect-generic.c (expand_vector_operations_1): Hoist
11187 vector boolean check before scalar optimization.
11188
11189 2018-05-28 Jakub Jelinek <jakub@redhat.com>
11190
11191 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
11192 for armv5te.
11193
11194 2018-05-28 Mark Wielaard <mark@klomp.org>
11195
11196 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
11197 if it is an expression containing a minus sign.
11198
11199 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
11200
11201 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
11202
11203 2018-05-27 Paul Koning <ni1d@arrl.net>
11204
11205 * config/pdp11/pdp11.md (truncsihi2): Remove.
11206
11207 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
11208 Chung-Ju Wu <jasonwucj@gmail.com>
11209
11210 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
11211 implementation.
11212 (unaligned_store_dw): Ditto.
11213 * config/nds32/nds32-memory-manipulation.c
11214 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
11215 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
11216 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
11217 (emit_setmem_word_loop): Rename to ...
11218 (emit_setmem_doubleword_loop): ... this.
11219 (nds32_gen_dup_4_byte_to_word_value): New function.
11220 (nds32_gen_dup_8_byte_to_double_word_value): New function.
11221 (nds32_expand_setmem_loop): Refine implementation.
11222 (nds32_expand_setmem_loop_v3m): Ditto.
11223 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
11224 pattern.
11225
11226 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11227
11228 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
11229
11230 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
11231
11232 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
11233 (nds32_init_machine_status): Initialize machine->attr_naked_p and
11234 machine->attr_no_prologue_p.
11235 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
11236 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
11237 (nds32_expand_epilogue): Consider attr_naked_p.
11238 (nds32_expand_epilogue_v3pop): Likewise.
11239 (nds32_can_use_return_insn): Likewise.
11240 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
11241 attr_no_prologue_p fields.
11242 * config/nds32/nds32.opt (mret-in-naked-func): New option.
11243
11244 2018-05-27 Jakub Jelinek <jakub@redhat.com>
11245
11246 PR target/85918
11247 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
11248 attributes.
11249 * config/i386/sse.md
11250 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
11251 Rename to ...
11252 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
11253 ... this.
11254 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
11255 Rename to ...
11256 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
11257 ... this.
11258 (*<floatsuffix>floatv2div2sf2): Rename to ...
11259 (*float<floatunssuffix>v2div2sf2): ... this.
11260 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
11261 (float<floatunssuffix>v2div2sf2_mask): ... this.
11262 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
11263 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
11264 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
11265 to ...
11266 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
11267 ... this.
11268 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11269 Rename to ...
11270 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
11271 ... this.
11272 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11273 Rename to ...
11274 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
11275 ... this.
11276 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
11277 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
11278 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
11279 gen_ufix_truncv8dfv8si2.
11280 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
11281 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
11282 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
11283 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
11284 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
11285 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
11286 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
11287 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
11288
11289 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
11290
11291 PR target/85900
11292 PR target/85345
11293 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
11294
11295 2018-05-25 Jim Wilson <jimw@sifive.com>
11296
11297 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
11298 * config/riscv/riscv.c (struct machine_function): Add
11299 interrupt_handler_p and attribute_checked_p fields.
11300 (riscv_attribute_table): Add interrupt.
11301 (riscv_interrupt_type_p): New.
11302 (riscv_save_reg_p): Save extra regs for interrupt handler.
11303 (riscv_use_save_libcall): Return false for interrupt handler.
11304 (riscv_first_stack_step): Add forward declaration.
11305 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
11306 for interrupt handler with large frame. Use it for saved reg list.
11307 (riscv_expand_prologue): Move flag_stack_usage_info support to
11308 eliminate duplication.
11309 (riscv_expand_epilogue): Generate mret for interrupt handler.
11310 (riscv_epilogue_uses): New.
11311 (riscv_can_use_return_insn): Return false for interrupt handler.
11312 (riscv_function_ok_for_sibcall): Likewise.
11313 (riscv_set_current_function): Add interrupt handler support.
11314 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
11315 * config/riscv/riscv.md (UNSPECV_MRET): New.
11316 (GP_REGNUM): New.
11317 (riscv_frflags, riscv_fsflags): Use tab after opcode.
11318 (riscv_mret): New.
11319 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
11320
11321 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
11322
11323 PR tree-optimization/85712
11324 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
11325 this candidate has already been replaced in-situ by a copy.
11326
11327 2018-05-25 Jason Merrill <jason@redhat.com>
11328
11329 PR c++/80485 - inline function non-zero address.
11330 * symtab.c (nonzero_address): Check DECL_COMDAT.
11331
11332 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
11333
11334 PR target/83628
11335 * config/alpha/alpha.md (ashlsi3): New insn pattern.
11336 (*ashlsi_se): Rename from *ashldi_se. Define as sign
11337 extension of SImode operation. Use const123_operand predicate.
11338 (*saddsi_1): Remove.
11339 (*saddl_se_1): Ditto.
11340 (*ssubsi_1): Ditto.
11341 (*ssubl_se_1): Ditto.
11342 * config/alpha/predicates.md (const123_operand): New predicate.
11343 * config/alpha/constraints.md (P): Use IN_RANGE.
11344
11345 2018-05-25 Richard Biener <rguenther@suse.de>
11346
11347 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
11348 defaulted to true.
11349 (ref_maybe_used_by_stmt_p): Likewise.
11350 (stmt_may_clobber_ref_p): Likewise.
11351 (stmt_may_clobber_ref_p_1): Likewise.
11352 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
11353 and pass it along.
11354 (ref_maybe_used_by_stmt_p): Likewise.
11355 (stmt_may_clobber_ref_p): Likewise.
11356 (stmt_may_clobber_ref_p_1): Likewise.
11357 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
11358 the alias oracle to disambiguate DRs with stmts DR analysis
11359 couldn't handle.
11360 (vect_analyze_data_refs): Do not give up on not analyzable
11361 DRs for BB vectorization. Remove code truncating the dataref
11362 vector.
11363
11364 2018-05-25 Jakub Jelinek <jakub@redhat.com>
11365
11366 PR target/85832
11367 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
11368 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
11369 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
11370
11371 2018-05-25 Richard Biener <rguenther@suse.de>
11372
11373 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
11374 function, combining stmt data ref gathering and fatal analysis
11375 parts.
11376 (vect_analyze_data_refs): Remove now redudnant code and simplify.
11377 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
11378 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
11379 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
11380 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
11381
11382 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11383
11384 PR tree-optimization/85720
11385 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
11386 SCC if all partitions are builtins.
11387 (version_loop_by_alias_check): New parameter. Generate cancelable
11388 runtime alias check if all partitions are builtins.
11389 (distribute_loop): Update call to above function.
11390
11391 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11392
11393 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
11394 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
11395 (parm_default_def_partition_arg): Ditto.
11396 (set_parm_default_def_partition): Ditto.
11397 (get_parm_default_def_partitions): Ditto and make it static.
11398 (get_undefined_value_partitions): Ditto and make it static.
11399 (remove_ssa_form): Refactor call to init_var_map here.
11400 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
11401 computation for loop region.
11402 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
11403 (register_default_def): Delete.
11404 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
11405 (parm_default_def_partition_arg): Ditto.
11406 (set_parm_default_def_partition): Ditto.
11407 (get_parm_default_def_partitions): Ditto and make it static.
11408 (get_undefined_value_partitions): Ditto and make it static.
11409 (coalesce_with_default, coalesce_with_default): Update comment.
11410 (create_coalesce_list_for_region): New func factored out from
11411 create_outofssa_var_map.
11412 (populate_coalesce_list_for_outofssa): New func factored out from
11413 create_outofssa_var_map and coalesce_ssa_name.
11414 (create_outofssa_var_map): Delete.
11415 (coalesce_ssa_name): Refactor to support live range computation.
11416 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
11417 (get_parm_default_def_partitions): Delete.
11418 (get_undefined_value_partitions): Ditto.
11419 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
11420 computation for loop region.
11421 (new_tree_live_info, loe_visit_block): Ditto.
11422 (live_worklist, set_var_live_on_entry): Ditto.
11423 (calculate_live_on_exit, verify_live_on_entry): Ditto.
11424 * tree-ssa-live.h (struct _var_map): New fields.
11425 (init_var_map): Change decl.
11426 (region_contains_p): New.
11427
11428 2018-05-25 Bin Cheng <bin.cheng@arm.com>
11429
11430 * tree-ssa-live.h (live_merge_and_clear): Delete.
11431
11432 2018-05-25 Richard Biener <rguenther@suse.de>
11433
11434 PR c++/85912
11435 * tree-dump.c (dequeue_and_dump): Remove access to removed
11436 operand 2 of a SWITCH_EXPR.
11437
11438 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11439
11440 * doc/sourcebuild.texi (vect_double_cond_arith): Include
11441 multiplication and division.
11442 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
11443 (cond_udiv@var{m}, cond_umod@var{m}): Document.
11444 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
11445 (cond_udiv_optab, cond_umod_optab): New optabs.
11446 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
11447 (IFN_COND_RDIV): New internal functions.
11448 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
11449 TRUNC_MOD_EXPR and RDIV_EXPR.
11450 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
11451 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
11452 New unspecs.
11453 (SVE_INT_BINARY): Include mult.
11454 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
11455 (optab, sve_int_op): Handle mult.
11456 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
11457 UNSPEC_COND_DIV.
11458 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
11459 for SVE_INT_BINARY_SD.
11460
11461 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11462
11463 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
11464 (optab, sve_int_op): Handle div and udiv.
11465 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
11466 for SVE_INT_BINARY_SD.
11467 (*<optab><mode>3): New insn for the same.
11468
11469 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11470
11471 * tree-vect-patterns.c: Include predict.h.
11472 (vect_recog_divmod_pattern): Restrict check for division support
11473 to when optimizing for size.
11474
11475 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11476
11477 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
11478 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
11479 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
11480 (gimple_match_op::set_op): Likewise.
11481 (gimple_resimplify4): Declare.
11482 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
11483 (expr::gen_transform): Likewise.
11484 (decision_tree::gen): Generate a simplification routine for 4 operands.
11485 * gimple-match-head.c (gimple_simplify): Add an overload for
11486 4 operands. In the top-level function, handle up to 4 call
11487 arguments and call gimple_resimplify4.
11488 (gimple_resimplify4): New function.
11489 (build_call_internal): Pass a fourth operand.
11490 (maybe_push_to_seq): Likewise.
11491 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
11492 Fold VEC_COND_EXPRs of an operation and a default value into
11493 an IFN_COND_* function if possible.
11494 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
11495 New unspecs.
11496 (SVE_COND_FP_BINARY): Include them.
11497 (optab, sve_fp_op): Handle them.
11498 (SVE_INT_BINARY_REV): New code iterator.
11499 (SVE_COND_FP_BINARY_REV): New int iterator.
11500 (commutative): New int attribute.
11501 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
11502 Declare.
11503 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
11504 function.
11505 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
11506 (*cond_<optab><mode>): New patterns for reversed operands.
11507
11508 2018-05-25 Richard Biener <rguenther@suse.de>
11509
11510 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
11511 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
11512 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
11513 (STMT_VINFO_GROUPED_ACCESS): Adjust.
11514 * tree-vect-data-refs.c (everywhere): Adjust users.
11515 * tree-vect-loop.c (everywhere): Likewise.
11516 * tree-vect-slp.c (everywhere): Likewise.
11517 * tree-vect-stmts.c (everywhere): Likewise.
11518 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
11519
11520 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
11521
11522 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
11523 Rename to...
11524 (gcc_cv_as_section_exclude): ... this.
11525 Try Solaris as #exclude syntax.
11526 * configure: Regenerate.
11527 * config.in: Regenerate.
11528 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
11529 SECTION_EXCLUDE.
11530 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
11531 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
11532
11533 * varasm.c (default_elf_asm_named_section): Don't check if
11534 HAVE_GAS_SECTION_EXCLUDE is defined.
11535
11536 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11537
11538 * doc/md.texi: Update the documentation of the cond_* optabs
11539 to mention the new final operand. Fix GET_MODE_NUNITS call.
11540 Describe the scalar case too.
11541 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
11542 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
11543 instead of 2.
11544 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
11545 (get_conditional_internal_fn): Update comment.
11546 * tree-vect-loop.c (vectorizable_reduction): Pass the original
11547 accumulator value as a final argument to conditional functions.
11548 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
11549 a define_expand and add an "else" operand. Assert for now that
11550 the else operand is equal to operand 2. Use SVE_INT_BINARY and
11551 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
11552 (*cond_<optab><mode>): New patterns.
11553 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
11554 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
11555 (UNSPEC_COND_EOR): Delete.
11556 (optab): Remove associated mappings.
11557 (SVE_INT_BINARY): New code iterator.
11558 (sve_int_op): Remove int attribute and add "minus" to the code
11559 attribute.
11560 (SVE_COND_INT_OP): Delete.
11561 (SVE_COND_FP_OP): Rename to...
11562 (SVE_COND_FP_BINARY): ...this.
11563
11564 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
11565
11566 * optabs.c (can_reuse_operands_p): New function.
11567 (maybe_legitimize_operands): Try to reuse the results for
11568 earlier operands.
11569
11570 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
11571
11572 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
11573 Add {q} suffix to insn mnemonic.
11574
11575 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11576
11577 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
11578 (msp430_warn_func_return): New.
11579
11580 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
11581
11582 * fold-const.c (tree_nonzero_bits): New function.
11583 * fold-const.h (tree_nonzero_bits): Likewise.
11584 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
11585 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
11586
11587 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
11588
11589 PR target/85900
11590 PR target/85345
11591 * varasm.c (assemble_alias): Check ifunc_resolver only on
11592 FUNCTION_DECL.
11593
11594 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
11595
11596 PR target/85903
11597 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
11598 when memory input operand is handled.
11599
11600 2018-05-24 Luis Machado <luis.machado@linaro.org>
11601
11602 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
11603 global.
11604 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
11605
11606 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
11607
11608 * match.pd: Delay FMA folds until after vectorization.
11609
11610 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
11611
11612 PR target/83009
11613 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
11614 address check not strict.
11615
11616 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
11617
11618 * gimple-match.h (gimple_match_op): New class.
11619 (mprts_hook): Replace parameters with a gimple_match_op *.
11620 (maybe_build_generic_op): Likewise.
11621 (gimple_simplified_result_is_gimple_val): Replace parameters with
11622 a const gimple_match_op *.
11623 (gimple_simplify): Replace code_helper * and tree * parameters with
11624 a gimple_match_op * parameter.
11625 (gimple_resimplify1): Replace code_helper *, tree and tree *
11626 parameters with a gimple_match_op * parameter.
11627 (gimple_resimplify2): Likewise.
11628 (gimple_resimplify3): Likewise.
11629 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
11630 parameters with a gimple_match_op * parameter.
11631 * gimple-match-head.c (gimple_simplify): Change prototypes of
11632 auto-generated functions to take a gimple_match_op * instead of
11633 separate code_helper * and tree * parameters. Make the same
11634 change in the top-level overload and update calls to the
11635 gimple_resimplify routines. Update calls to the auto-generated
11636 functions and to maybe_push_res_to_seq in the publicly-facing
11637 operation-specific gimple_simplify overloads.
11638 (gimple_match_op::MAX_NUM_OPS): Define.
11639 (gimple_resimplify1): Replace rcode and ops with a single res_op
11640 parameter. Update call to gimple_simplify.
11641 (gimple_resimplify2): Likewise.
11642 (gimple_resimplify3): Likewise.
11643 (mprts_hook): Replace parameters with a gimple_match_op *.
11644 (maybe_build_generic_op): Likewise.
11645 (build_call_internal): Replace type, nargs and ops with
11646 a gimple_match_op *.
11647 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
11648 with a single gimple_match_op *. Update calls to mprts_hook,
11649 build_call_internal and gimple_simplified_result_is_gimple_val.
11650 Factor out code that is common to the tree_code and combined_fn cases.
11651 * genmatch.c (expr::gen_transform): Replace tem_code and
11652 tem_ops with a gimple_match_op called tem_op. Update calls
11653 to the gimple_resimplify functions and maybe_push_res_to_seq.
11654 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
11655 res_ops. Update call to the gimple_resimplify functions.
11656 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
11657 (decision_tree::gen): Make the functions take a gimple_match_op *
11658 called res_op instead of separate res_code and res_ops parameters.
11659 Update call accordingly.
11660 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
11661 and ops with a single res_op parameter. Update calls to
11662 maybe_build_generic_op and maybe_push_res_to_seq.
11663 (fold_stmt_1): Update calls to gimple_simplify and
11664 replace_stmt_with_simplification.
11665 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
11666 and gimple_simplified_result_is_gimple_val.
11667 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
11668 gimple_simplify.
11669 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
11670 with a gimple_match_op *.
11671 (vn_nary_build_or_lookup): Likewise. Update call to
11672 vn_nary_build_or_lookup_1.
11673 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
11674 gimple_match_op *. Update calls to the gimple_resimplify routines
11675 and to gimple_simplified_result_is_gimple_val.
11676 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
11677 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
11678 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
11679 (visit_nary_op): Likewise.
11680 (visit_reference_op_load): Likewise.
11681
11682 2018-05-23 Luis Machado <luis.machado@linaro.org>
11683
11684 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
11685 modifier for printing the step amount.
11686
11687 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
11688
11689 PR target/78849
11690 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
11691 types.
11692
11693 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
11694
11695 * doc/sourcebuild.texi (Endianness): New subsubsection.
11696
11697 2018-05-23 Luis Machado <luis.machado@linaro.org>
11698
11699 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11700 <prefetch_dynamic_strides>: New const bool field.
11701 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11702 prefetch_dynamic_strides.
11703 (exynosm1_prefetch_tune): Likewise.
11704 (thunderxt88_prefetch_tune): Likewise.
11705 (thunderx_prefetch_tune): Likewise.
11706 (thunderx2t99_prefetch_tune): Likewise.
11707 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
11708 false.
11709 (aarch64_override_options_internal): Update to set
11710 PARAM_PREFETCH_DYNAMIC_STRIDES.
11711 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
11712 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
11713 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
11714 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
11715 prefetch-dynamic-strides setting.
11716
11717 2018-05-23 Luis Machado <luis.machado@linaro.org>
11718
11719 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
11720 <minimum_stride>: New const int field.
11721 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
11722 minimum_stride field defaulting to -1.
11723 (exynosm1_prefetch_tune): Likewise.
11724 (thunderxt88_prefetch_tune): Likewise.
11725 (thunderx_prefetch_tune): Likewise.
11726 (thunderx2t99_prefetch_tune): Likewise.
11727 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
11728 <default_opt_level>: Set to 3.
11729 (aarch64_override_options_internal): Update to set
11730 PARAM_PREFETCH_MINIMUM_STRIDE.
11731 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
11732 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
11733 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
11734 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
11735 stride is constant and is below the minimum stride threshold.
11736
11737 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11738
11739 * config/arm/arm-cpus.in (mode26): Delete.
11740 (armv4): Delete mode26 reference.
11741 * config/arm/arm.c (arm_configure_build_target): Delete use of
11742 isa_bit_mode26.
11743
11744 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
11745
11746 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
11747 New insn pattern.
11748 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
11749 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
11750 for non-SSE modes.
11751 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
11752 (floatunsdidf2): Ditto.
11753
11754 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
11755
11756 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
11757 (fixuns_trunc<mode>si2_avx512f): Ditto.
11758 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
11759 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
11760 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
11761
11762 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
11763
11764 PR rtl-optimization/79985
11765 * df-scan.c (df_insn_refs_collect): Remove special case for
11766 global registers and asm statements.
11767
11768 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
11769
11770 * extend.texi (Global Register Variables): Rewrite the bullet list.
11771 Note that the register is available for allocation. Note that access
11772 via inline asm must use constraints. Add note about async-signal
11773 handlers. Remove paragraph about automagic register selection.
11774
11775 2018-05-23 Richard Biener <rguenther@suse.de>
11776
11777 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
11778 of fixed offset from memset VN.
11779
11780 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
11781
11782 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
11783 first_interp field.
11784 (alloc_cand_and_find_basis): Initialize first_interp field.
11785 (slsr_process_mul): Modify first_interp field.
11786 (slsr_process_add): Likewise.
11787 (slsr_process_cast): Modify first_interp field for each new
11788 interpretation.
11789 (slsr_process_copy): Likewise.
11790 (dump_candidate): Dump first_interp field.
11791 (replace_mult_candidate): Process all interpretations, not just
11792 subsequent ones.
11793 (replace_rhs_if_not_dup): Likewise.
11794 (replace_one_candidate): Likewise.
11795
11796 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
11797
11798 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
11799 Add new boolean.
11800 (aarch64_needs_frame_chain): New function.
11801 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
11802
11803 2018-05-23 Sudakshina Das <sudi.das@arm.com>
11804
11805 PR target/84882
11806 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
11807 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
11808 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
11809 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
11810 as true for strict-align.
11811 (aarch64_can_inline_p): Perform checks even when callee has no
11812 attributes to check for strict alignment.
11813 * doc/extend.texi (AArch64 Function Attributes): Document
11814 no-strict-align.
11815 * doc/invoke.texi: (AArch64 Options): Likewise.
11816
11817 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
11818
11819 PR tree-optimization/85853
11820 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
11821 the handling of the root of the node to...
11822 (vect_slp_analyze_node_operations_1): ...this new function,
11823 and run the whole thing with the child nodes' def types
11824 set according to their SLP node's def type.
11825
11826 2018-05-23 Richard Biener <rguenther@suse.de>
11827
11828 PR middle-end/85874
11829 * tree-data-ref.c (create_runtime_alias_checks): Defer
11830 and ignore overflow warnings.
11831
11832 2018-05-23 Yury Gribov <tetra2005@gmail.com>
11833
11834 PR tree-optimization/85822
11835 * tree-vrp.c (is_masked_range_test): Fix handling of negative
11836 constants.
11837
11838 2018-05-23 Richard Biener <rguenther@suse.de>
11839
11840 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
11841 memset constants via native_interpret_expr.
11842
11843 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
11844
11845 PR target/85345
11846 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
11847 attribute.
11848 (cgraph_node::create_alias): Likewise.
11849 (cgraph_node::get_availability): Check ifunc_resolver instead
11850 of looking up ifunc attribute.
11851 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
11852 * varasm.c (do_assemble_alias): Likewise.
11853 (assemble_alias): Likewise.
11854 (default_binds_local_p_3): Likewise.
11855 * cgraph.h (cgraph_node): Add ifunc_resolver.
11856 (cgraph_node::only_called_directly_or_aliased_p): Return false
11857 for IFUNC resolver.
11858 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
11859 attribute.
11860 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
11861 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
11862 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
11863 instead of looking up ifunc attribute.
11864
11865 2018-05-22 Luis Machado <luis.machado@linaro.org>
11866
11867 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
11868
11869 2018-05-22 Martin Sebor <msebor@redhat.com>
11870
11871 PR middle-end/85359
11872 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
11873 only when expasion succeeds.
11874 (expand_builtin_strcmp): Same.
11875 (expand_builtin_strncmp): Same.
11876
11877 2018-05-22 Martin Sebor <msebor@redhat.com>
11878
11879 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
11880
11881 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
11882 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11883
11884 * config/aarch64/aarch64-ldpstp.md: Replace uses of
11885 aarch64_mem_pair_operand with memory_operand and delete operand swapping
11886 code.
11887 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
11888 Add check for legitimate_address.
11889 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
11890 (aarch64_swap_ldrstr_operands): New.
11891 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
11892 Define prototype.
11893
11894 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
11895 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11896
11897 * config/aarch64/aarch64.md: New patterns to generate stp
11898 and ldp.
11899 (store_pair_sw, store_pair_dw): New patterns to generate stp for
11900 single words and double words.
11901 (load_pair_sw, load_pair_dw): Likewise.
11902 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
11903 Delete.
11904 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
11905 Delete.
11906 * config/aarch64/aarch64-ldpstp.md: Modify peephole
11907 for different mode ldpstp and add peephole for merged zero stores.
11908 Likewise for loads.
11909 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
11910 Add size check.
11911 (aarch64_gen_store_pair): Rename calls to match new patterns.
11912 (aarch64_gen_load_pair): Rename calls to match new patterns.
11913 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
11914 (load_pair<DREG:mode><DREG2:mode>): ... This.
11915 (store_pair<mode>): Rename to...
11916 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
11917 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
11918 New mode iterators.
11919 (V_INT_EQUIV): Handle SImode.
11920 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
11921 New predicate.
11922
11923 2018-05-22 Martin Sebor <msebor@redhat.com>
11924
11925 PR c/85623
11926 * calls.c (maybe_warn_nonstring_arg): Use string length to set
11927 or ajust the presumed bound on an operation to avoid unnecessary
11928 warnings.
11929
11930 2018-05-22 Martin Sebor <msebor@redhat.com>
11931
11932 PR tree-optimization/85826
11933 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
11934 assuming that a DECL necesarily has a constant size.
11935
11936 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
11937
11938 PR middle-end/85862
11939 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
11940
11941 2018-05-22 Richard Biener <rguenther@suse.de>
11942
11943 PR tree-optimization/85834
11944 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
11945 non-constant and non-zero memset arguments.
11946
11947 2018-05-22 Martin Liska <mliska@suse.cz>
11948
11949 PR ipa/85607
11950 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
11951
11952 2018-05-22 Richard Biener <rguenther@suse.de>
11953
11954 PR tree-optimization/85863
11955 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
11956 comparisons when vectype is specified.
11957 (vectorizable_condition): Do not specify vectype for
11958 vect_is_simple_cond when SLP vectorizing.
11959
11960 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
11961
11962 PR target/85657
11963 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
11964 define __ibm128 as long double.
11965 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
11966 as a distinct type when IEEE 128-bit support is enabled.
11967 (init_float128_ieee): Fix up conversions between IFmode and IEEE
11968 128-bit types to use the correct functions.
11969 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
11970 convert between 128-bit floating point types that have different
11971 modes but the same representation, instead of using gen_lowpart to
11972 makean alias.
11973 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
11974 KFmode.
11975 (IFKF_reg): New attributes to give the register constraints for
11976 IFmode and KFmode.
11977 (extend<mode>tf2_internal): New insns to mark an explicit
11978 conversion between 128-bit floating point types that have a
11979 different mode but share the same representation.
11980
11981 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
11982
11983 PR tree-optimization/85814
11984 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
11985 a null return from get_strinfo when unsharing the next
11986 strinfo in the chain.
11987
11988 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
11989
11990 PR gcc/84923
11991 * varasm.c (weak_finish): Clean up weak_decls.
11992
11993 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11994
11995 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
11996 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
11997 UNSPEC_UADALP values.
11998 * config/aarch64/iterators.md (ABAL): New int iterator.
11999 (ABDL2): Likewise.
12000 (ADALP): Likewise.
12001 (sur): Add mappings for the above.
12002 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
12003 New define_insn.
12004 (aarch64_<sur>abal<mode>_4): Likewise.
12005 (aarch64_<sur>adalp<mode>_3): Likewise.
12006 (<sur>sadv16qi): New define_expand.
12007
12008 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
12009
12010 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
12011 (*movdf_internal): Ditto.
12012 (*rcpsf2_sse): Ditto.
12013 (*rsqrtsf2_sse): Ditto.
12014 (*sqrt<mode>2_sse): Ditto.
12015
12016 2018-05-21 Tamar Christina <tamar.christina@arm.com>
12017
12018 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
12019 eor3q<mode>4.
12020 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
12021 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
12022 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
12023 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
12024 vbcaxq_s64): New.
12025 * config/aarch64/arm_neon.h: Likewise.
12026 * config/aarch64/iterators.md (VQ_I): New.
12027
12028 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
12029
12030 * config.gcc: Add arc/t-multilib-linux to tmake_file for
12031 arc*-*-linux*.
12032 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
12033 MULTILIB_DIRNAMES
12034
12035 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
12036
12037 * config/nds32/constraints.md (S): New constraint.
12038 * config/nds32/nds32.md (call_internal): Use constraint S.
12039 (call_value_internal): Likewise.
12040 (sibcall_internal): Likewise.
12041 (sibcall_value_internal): Likewise.
12042
12043 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
12044 Chung-Ju Wu <jasonwucj@gmail.com>
12045
12046 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
12047 into consideration.
12048
12049 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
12050 Chung-Ju Wu <jasonwucj@gmail.com>
12051
12052 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
12053 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
12054 (nds32_rtx_costs_impl): Simplify.
12055 (nds32_address_cost_impl): Simplify.
12056 (nds32_init_rtx_costs): New function.
12057 (nds32_rtx_costs_speed_prefer): Likewise.
12058 (nds32_rtx_costs_size_prefer): Likewise.
12059 (nds32_address_cost_speed_prefer): Likewise.
12060 (nds32_address_cost_speed_fwprop): Likewise.
12061 (nds32_address_cost_size_prefer): Likewise.
12062 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
12063 * config/nds32/nds32.c (nds32_option_override): Use
12064 nds32_init_rtx_costs function.
12065
12066 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
12067
12068 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
12069 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
12070 (TARGET_PIPELINE_N8): Likewise.
12071 (TARGET_PIPELINE_N10): Likewise.
12072 (TARGET_PIPELINE_N13): Likewise.
12073 (TARGET_PIPELINE_GRAYWOLF): Likewise.
12074
12075 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
12076
12077 * config/nds32/nds32-fpu.md: Update copyright year.
12078
12079 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12080
12081 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
12082
12083 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12084
12085 * config/nds32/nds32.c
12086 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
12087 * config/nds32/nds32.opt (minline-asm-r15): New option.
12088
12089 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12090
12091 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
12092 MASK_HW_ABS.
12093 * config/nds32/nds32.md (abssi2): New pattern.
12094
12095 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
12096
12097 * config/i386/i386.md (rex64namesuffix): New mode attribute.
12098 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
12099 Merge insn pattern from sse_cvtsi2ss<round_name> and
12100 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
12101 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
12102 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
12103 using SWI48 mode iterator.
12104 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
12105 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
12106 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
12107 pattern from sse_cvttss2si<round_saeonly_name>
12108 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
12109 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
12110 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
12111 using SWI48 mode iterator.
12112 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
12113 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
12114 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
12115 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
12116 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
12117 using SWI48 mode iterator.
12118 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
12119 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
12120 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
12121 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
12122 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
12123 SWI48 mode iterator.
12124 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
12125 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
12126 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
12127 pattern from sse_cvttsd2si<round_saeonly_name>
12128 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
12129
12130 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12131
12132 * config/nds32/nds32-md-auxiliary.c
12133 (nds32_valid_smw_lwm_base_p): Refine.
12134 (nds32_output_smw_single_word): Refine.
12135 (nds32_output_smw_double_word): New.
12136 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
12137
12138 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12139
12140 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
12141 (nds32_output_stack_pop): Refine.
12142 (nds32_expand_unaligned_load): Refine.
12143 (nds32_expand_unaligned_store): Refine.
12144
12145 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
12146 Chung-Ju Wu <jasonwucj@gmail.com>
12147
12148 * config/nds32/constants.md: Add TP_REGNUM constant.
12149 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
12150 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
12151 UNSPEC_ADD32.
12152 * config/nds32/nds32-doubleword.md: Consider flag_pic.
12153 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
12154 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
12155 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
12156 and PIC code generation.
12157 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
12158 code generation.
12159 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
12160 optimization.
12161 * config/nds32/nds32.md: Support TLS and PIC.
12162 * config/nds32/nds32.c: Support TLS and PIC.
12163 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
12164 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
12165 predicate.
12166
12167 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
12168
12169 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
12170 mode with E_ prefix.
12171
12172 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
12173 Chung-Ju Wu <jasonwucj@gmail.com>
12174
12175 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
12176 * config/nds32/nds32-md-auxiliary.c
12177 (symbolic_reference_mentioned_p): New.
12178 (nds32_legitimize_ict_address): New.
12179 (nds32_expand_ict_move): New.
12180 (nds32_indirect_call_referenced_p): New.
12181 (nds32_symbol_binds_local_p): Delete.
12182 (nds32_long_call_p): Modify.
12183 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
12184 * config/nds32/nds32-protos.h
12185 (symbolic_reference_mentioned_p): Declare.
12186 (nds32_legitimize_ict_address): Declare.
12187 (nds32_expand_ict_move): Declare.
12188 (nds32_indirect_call_referenced_p): Declare.
12189 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
12190 (nds32_relax_group): Use nds32_ict_const_p as condition.
12191 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
12192 (nds32_asm_file_start): Output ict_model directive in asm code.
12193 (nds32_legitimate_address_p): Consider indirect call.
12194 (nds32_print_operand): Consider indirect call.
12195 (nds32_print_operand_address): Consider indirect call.
12196 (nds32_insert_attributes): Handle "indirect_call" attribute.
12197 (TARGET_LEGITIMATE_ADDRESS_P): Define.
12198 (TARGET_LEGITIMATE_CONSTANT_P): Define.
12199 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
12200 (TARGET_DELEGITIMIZE_ADDRESS): Define.
12201 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
12202 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
12203 (TARGET_ICT_MODEL_SMALL): Define.
12204 (TARGET_ICT_MODEL_LARGE): Define.
12205 * config/nds32/nds32.md (movsi): Consider ict model.
12206 (call, call_value): Consider ict model.
12207 (sibcall, sibcall_value): Consider ict model.
12208 * config/nds32/nds32.opt (mict-model): New option.
12209 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
12210 model.
12211
12212 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
12213 Monk Chiang <sh.chiang04@gmail.com>
12214 Jim Wilson <jimw@sifive.com>
12215
12216 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
12217 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
12218 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
12219 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
12220 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
12221 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
12222 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
12223 compute save_libcall_adjustment properly.
12224 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
12225 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
12226 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
12227 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
12228 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
12229 (ABI_SPEC): Handle mabi=ilp32e.
12230 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
12231 (RVE): Add RVE mask.
12232 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
12233 <-march>: Add rv32e as an example.
12234
12235 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
12236
12237 PR c++/82899
12238 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
12239 (intra_create_variable_infos): Handle C++ constructors.
12240
12241 2018-05-18 Martin Liska <mliska@suse.cz>
12242
12243 * passes.def: Remove a redundant pass.
12244
12245 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
12246
12247 PR bootstrap/85838
12248 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
12249
12250 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12251
12252 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
12253 (ARMv4): Update.
12254 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
12255 (ARMv6m): Update.
12256 (armv2, armv2a, armv3, armv3m): Delete architectures.
12257 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
12258 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
12259 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
12260 Delete cpus.
12261 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
12262 (*mulsidi3adddi): Likewise.
12263 (mulsidi3): Likewise.
12264 (*mulsidi3_nov6): Likewise.
12265 (umulsidi3): Likewise.
12266 (umulsidi3_nov6): Likewise.
12267 (umaddsidi4): Likewise.
12268 (*umulsidi3adddi): Likewise.
12269 (smulsi3_highpart): Likewise.
12270 (*smulsi3_highpart_nov6): Likewise.
12271 (umulsi3_highpart): Likewise.
12272 (*umulsi3_highpart_nov6): Likewise.
12273 * config/arm/arm.h (arm_arch3m): Delete.
12274 * config/arm/arm.c (arm_arch3m): Delete.
12275 (arm_option_override_internal): Update armv3-related comment.
12276 (arm_configure_build_target): Delete use of isa_bit_mode32.
12277 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
12278 (arm_rtx_costs_internal): Delete check of arm_arch3m.
12279 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
12280 (mulsa3): Likewise.
12281 (mulusa3): Likewise.
12282 * config/arm/arm-protos.h (arm_arch3m): Delete.
12283 * config/arm/arm-tables.opt: Regenerate.
12284 * config/arm/arm-tune.md: Likewise.
12285 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
12286 deleted architectures.
12287
12288 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12289
12290 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
12291 (armv5t, armv5te): New features.
12292 (ARMv5, ARMv5e): Delete fgroups.
12293 (ARMv5t, ARMv5te): Adjust for above changes.
12294 (ARMv6m): Likewise.
12295 (armv5, armv5e): Delete arches.
12296 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
12297 arm_arch5.
12298 (*call_reg_arm): Likewise.
12299 (*call_value_reg_armv5): Likewise.
12300 (*call_value_reg_arm): Likewise.
12301 (*call_symbol): Likewise.
12302 (*call_value_symbol): Likewise.
12303 (*sibcall_insn): Likewise.
12304 (*sibcall_value_insn): Likewise.
12305 (clzsi2): Likewise.
12306 (prefetch): Likewise.
12307 (define_split and define_peephole2 dependent on arm_arch5):
12308 Likewise.
12309 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
12310 arm_arch5e.
12311 (TARGET_ARM_QBIT): Likewise.
12312 (TARGET_DSP_MULTIPLY): Likewise.
12313 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
12314 (arm_arch5, arm_arch5e): Delete.
12315 (arm_arch5t, arm_arch5te): Declare.
12316 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
12317 (arm_arch5t): Declare.
12318 (arm_option_reconfigure_globals): Update for the above.
12319 (arm_options_perform_arch_sanity_checks): Update comment, replace
12320 use of arm_arch5 with arm_arch5t.
12321 (use_return_insn): Likewise.
12322 (arm_emit_call_insn): Likewise.
12323 (output_return_instruction): Likewise.
12324 (arm_final_prescan_insn): Likewise.
12325 (arm_coproc_builtin_available): Likewise.
12326 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
12327 arm_arch5e with arm_arch5t and arm_arch5te.
12328 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
12329 (arm_arch5t, arm_arch5te): Declare.
12330 * config/arm/arm-tables.opt: Regenerate.
12331 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
12332 * config/arm/t-multilib: Likewise.
12333 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
12334 instead of arm_arch5.
12335 (*call_reg_thumb1): Likewise.
12336 (*call_value_reg_thumb1_v5): Likewise.
12337 (*call_value_reg_thumb1): Likewise.
12338 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
12339 unreachable path.
12340 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
12341
12342 2018-05-18 Martin Liska <mliska@suse.cz>
12343
12344 PR gcov-profile/84846
12345 * doc/gcov.texi: Document -t option of gcov tool.
12346
12347 2018-05-18 Martin Liska <mliska@suse.cz>
12348
12349 PR gcov-profile/84846
12350 * gcov.c (print_usage): Add new -t option.
12351 (process_args): Handle the option.
12352 (generate_results): Use stdout as output when requested by
12353 the option.
12354
12355 2018-05-18 Martin Liska <mliska@suse.cz>
12356
12357 PR gcov-profile/84846
12358 * coverage.c (coverage_init): Write PWD to .gcno file.
12359 * doc/gcov.texi: Document how working directory is printed.
12360 * gcov-dump.c (dump_gcov_file): Print PWD.
12361 * gcov.c (output_intermediate_file): Likewise.
12362 (read_graph_file): Read PWD string.
12363 (output_lines): Print PWD.
12364
12365 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12366
12367 PR middle-end/85817
12368 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
12369 for retval and return false if all args to phi are zero.
12370
12371 2018-05-18 Richard Biener <rguenther@suse.de>
12372
12373 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
12374 method.
12375 (evrp_dom_walker::before_dom_children): Call it.
12376
12377 2018-05-18 Richard Biener <rguenther@suse.de>
12378
12379 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
12380 results when processing array refs with variable index.
12381
12382 2018-05-18 Toon Moene <toon@moene.org>
12383
12384 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
12385 directly after that of -floop-interchange. Indicate that both
12386 options are enabled by default when specifying -O3.
12387
12388 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12389
12390 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
12391 iterator. Delete separate integer-mode vec_set<mode> expander.
12392 (aarch64_simd_vec_setv2di): Delete.
12393 (vec_setv2di): Delete.
12394 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
12395 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
12396 the "w, r" alternative.
12397
12398 2018-05-18 Martin Liska <mliska@suse.cz>
12399
12400 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
12401 * tree-pass.h (make_pass_lower_switch_O0): New function.
12402 * tree-switch-conversion.c (node_has_low_bound): Remove.
12403 (node_has_high_bound): Likewise.
12404 (node_is_bounded): Likewise.
12405 (class pass_lower_switch): Make it a template type and create
12406 two instances.
12407 (pass_lower_switch::execute): Add template argument.
12408 (make_pass_lower_switch): New function.
12409 (make_pass_lower_switch_O0): New function.
12410 (do_jump_if_equal): Remove.
12411 (emit_case_nodes): Simplify to just handle all 3 cases and leave
12412 all the hard work to tree optimization passes.
12413
12414 2018-05-18 Martin Liska <mliska@suse.cz>
12415
12416 * dbgcnt.c (limit_low): Renamed from limit.
12417 (limit_high): New variable.
12418 (dbg_cnt_is_enabled): Check for upper limit.
12419 (dbg_cnt): Adjust dumping.
12420 (dbg_cnt_set_limit_by_index): Add new argument for high
12421 value.
12422 (dbg_cnt_set_limit_by_name): Likewise.
12423 (dbg_cnt_process_single_pair): Parse new format.
12424 (dbg_cnt_process_opt): Use strtok.
12425 (dbg_cnt_list_all_counters): Remove 'value' and add
12426 'limit_high'.
12427 * doc/invoke.texi: Document changes.
12428
12429 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
12430
12431 * doc/sourcebuild.texi (scalar_all_fma): Document.
12432 * tree.def (FMA_EXPR): Delete.
12433 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
12434 * internal-fn.c (ternary_direct): New macro.
12435 (expand_ternary_optab_fn): Likewise.
12436 (direct_ternary_optab_supported_p): Likewise.
12437 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
12438 * builtins.c (fold_builtin_fma): Delete.
12439 (fold_builtin_3): Don't call it.
12440 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
12441 * expr.c (expand_expr_real_2): Likewise.
12442 * fold-const.c (operand_equal_p): Likewise.
12443 (fold_ternary_loc): Likewise.
12444 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
12445 * gimple.c (DEFTREECODE): Likewise.
12446 * gimplify.c (gimplify_expr): Likewise.
12447 * optabs-tree.c (optab_for_tree_code): Likewise.
12448 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
12449 * tree-eh.c (operation_could_trap_p): Likewise.
12450 (stmt_could_throw_1_p): Likewise.
12451 * tree-inline.c (estimate_operator_cost): Likewise.
12452 * tree-pretty-print.c (dump_generic_node): Likewise.
12453 (op_code_prio): Likewise.
12454 * tree-ssa-loop-im.c (stmt_cost): Likewise.
12455 * tree-ssa-operands.c (get_expr_operands): Likewise.
12456 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
12457 * fold-const-call.h (fold_fma): Delete.
12458 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
12459 CFN_FNMA and CFN_FNMS.
12460 (fold_fma): Delete.
12461 * genmatch.c (combined_fn): New enum.
12462 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
12463 (commutative_op): New function.
12464 (commutate): Use it. Handle more than 2 operands.
12465 (dt_operand::gen_gimple_expr): Use commutative_op.
12466 (parser::parse_expr): Allow :c to be used with non-binary
12467 operators if the commutative operand is known.
12468 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
12469 CFN_FMS, CFN_FNMA and CFN_FNMS.
12470 (backprop::process_assign_use): Remove FMA_EXPR handling.
12471 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
12472 (gen_hsa_fma): New function.
12473 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
12474 IFN_FNMA and IFN_FNMS.
12475 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
12476 * gimple-fold.h (follow_all_ssa_edges): Declare.
12477 * gimple-fold.c (follow_all_ssa_edges): New function.
12478 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
12479 gimple_build interface and use follow_all_ssa_edges to fold the result.
12480 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
12481 instead of checking for optabs directly.
12482 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
12483 rather than FMA_EXPRs.
12484 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
12485 call to IFN_FMA instead of an FMA_EXPR.
12486
12487 2018-05-17 Jim Wilson <jimw@sifive.com>
12488
12489 * expr.c (do_tablejump): When converting index to Pmode, if we have a
12490 sign extended promoted subreg, and the range does not have the sign bit
12491 set, then do a sign extend.
12492
12493 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
12494 test, check for sign extended subreg and/or constant operands, and
12495 do a sign extend in that case.
12496
12497 2018-05-17 Steve Ellcey <sellcey@cavium.com>
12498
12499 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
12500 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
12501 Add untyped.
12502 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
12503 Change logics_shift_reg to logics_shift_imm.
12504 (thunderx2t99_fp_loadpair_basic): Delete.
12505 (thunderx2t99_fp_storepair_basic): Delete.
12506 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
12507 (thunderx2t99_asimd_polynomial): Delete.
12508 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
12509 and neon_fp_mul_d_scalar_q.
12510 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
12511 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
12512 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
12513 (thunderx2t99_asimd_lut): Add missing tbl types.
12514 (thunderx2t99_asimd_ext): Delete.
12515 (thunderx2t99_asimd_load1_1_mult): Delete.
12516 (thunderx2t99_asimd_load1_2_mult): Delete.
12517 (thunderx2t99_asimd_load1_ldp): New.
12518 (thunderx2t99_asimd_load1): New.
12519 (thunderx2t99_asimd_load2): Add missing *load2* types.
12520 (thunderx2t99_asimd_load3): New.
12521 (thunderx2t99_asimd_load4): New.
12522 (thunderx2t99_asimd_store1_1_mult): Delete.
12523 (thunderx2t99_asimd_store1_2_mult): Delete.
12524 (thunderx2t99_asimd_store2_mult): Delete.
12525 (thunderx2t99_asimd_store2_onelane): Delete.
12526 (thunderx2t99_asimd_store_stp): New.
12527 (thunderx2t99_asimd_store1): New.
12528 (thunderx2t99_asimd_store2): New.
12529 (thunderx2t99_asimd_store3): New.
12530 (thunderx2t99_asimd_store4): New.
12531
12532 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
12533
12534 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
12535 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
12536
12537 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
12538 Segher Boessenkool <segher@kernel.crashing.org>
12539
12540 PR target/85698
12541 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
12542 operand.
12543
12544 2018-05-17 Richard Biener <rguenther@suse.de>
12545
12546 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
12547 for pruning loop and prune defs feeding only already visited PHIs.
12548
12549 2018-05-17 Richard Biener <rguenther@suse.de>
12550
12551 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
12552
12553 2018-05-17 Bin Cheng <bin.cheng@arm.com>
12554 Richard Biener <rguenther@suse.de>
12555
12556 PR tree-optimization/85793
12557 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
12558 for VMAT_ELEMENTWISE.
12559
12560 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
12561
12562 * internal-fn.h (lookup_internal_fn): Declare
12563 * internal-fn.c (lookup_internal_fn): New function.
12564 * gimple.c (gimple_build_call_from_tree): Handle calls to
12565 internal functions.
12566 * gimple-pretty-print.c (dump_gimple_call): Print "." before
12567 internal function names.
12568 * tree-pretty-print.c (dump_generic_node): Likewise.
12569 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
12570
12571 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
12572
12573 * gimple-fold.h (gimple_build): Make the function forms take
12574 combined_fn rather than built_in_function.
12575 (gimple_simplify): Likewise.
12576 * gimple-match-head.c (gimple_simplify): Likewise.
12577 * gimple-fold.c (gimple_build): Likewise.
12578 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
12579 rather than gimple_build_call_internal.
12580 (get_initial_defs_for_reduction): Likewise.
12581 (vect_create_epilog_for_reduction): Likewise.
12582 (vectorizable_live_operation): Likewise.
12583
12584 2018-05-17 Martin Liska <mliska@suse.cz>
12585
12586 * gimple-ssa-sprintf.c (format_directive): Do not use
12587 space in between 'G_' and '('.
12588
12589 2018-05-17 Jakub Jelinek <jakub@redhat.com>
12590
12591 PR target/85323
12592 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
12593 even if the mask is not all ones.
12594
12595 PR target/85323
12596 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
12597 vector.
12598 (ix86_gimple_fold_builtin): Likewise.
12599
12600 PR target/85323
12601 * config/i386/i386.c: Include tree-vector-builder.h.
12602 (ix86_vector_shift_count): New function.
12603 (ix86_fold_builtin): Fold shift builtins by scalar count.
12604 (ix86_gimple_fold_builtin): Likewise.
12605
12606 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
12607 _mm512_setzero): New intrinsics.
12608
12609 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
12610 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12611
12612 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
12613 code generation for cases where splatting a value is not useful.
12614 * simplify-rtx.c (simplify_ternary_operation): Simplify
12615 vec_merge across a vec_duplicate and a paradoxical subreg forming
12616 a vector mode to a vec_concat.
12617
12618 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
12619
12620 * config.gcc: Support "goldmont-plus".
12621 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12622 "goldmont-plus".
12623 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12624 PROCESSOR_GOLDMONT_PLUS.
12625 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
12626 (processor_target_table): Add "goldmont-plus".
12627 (PTA_GOLDMONT_PLUS): Define.
12628 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
12629 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
12630 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
12631 (fold_builtin_cpu): Add "goldmont-plus".
12632 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
12633 (ix86_option_override_internal): Add "goldmont-plus".
12634 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
12635 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
12636 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
12637 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
12638
12639 2018-05-17 Richard Biener <rguenther@suse.de>
12640
12641 PR tree-optimization/85757
12642 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
12643 remove defs that only feed that PHI from further processing.
12644
12645 2018-05-16 Jim Wilson <jimw@sifive.com>
12646
12647 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
12648 asterisk to name.
12649 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
12650
12651 2018-05-16 Mark Wielaard <mark@klomp.org>
12652
12653 * dwarf2out.c (count_index_strings): New function.
12654 (output_indirect_strings): Call count_index_strings and generate
12655 header for dwarf_version >= 5.
12656
12657 2018-05-16 Mark Wielaard <mark@klomp.org>
12658
12659 * dwarf2out.c (dwarf_FORM): New function.
12660 (set_indirect_string): Use dwarf_FORM.
12661 (reset_indirect_string): Likewise.
12662 (size_of_die): Likewise.
12663 (value_format): Likewise.
12664 (output_die): Likewise.
12665 (add_skeleton_AT_string): Likewise.
12666 (output_macinfo_op): Likewise.
12667 (index_string): Likewise.
12668 (output_index_string_offset): Likewise.
12669 (output_index_string): Likewise.
12670 (count_index_strings): Likewise.
12671
12672 2018-05-16 Carl Love <cel@us.ibm.com>
12673
12674 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
12675 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
12676
12677 2018-05-16 Martin Jambor <mjambor@suse.cz>
12678
12679 * ipa-prop.c (ipa_free_all_edge_args): Remove.
12680 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
12681
12682 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
12683
12684 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
12685 (fnma<mode>4): Likewise.
12686 (fms<mode>4): Likewise.
12687 (fnms<mode>4): Likewise.
12688 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
12689 (aarch64_fnma<mode>4): Likewise.
12690 (aarch64_fms<mode>4): Likewise.
12691 (aarch64_fnms<mode>4): Likewise.
12692 (aarch64_fnmadd<mode>4): Likewise.
12693
12694 2018-05-16 Jason Merrill <jason@redhat.com>
12695
12696 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
12697
12698 2018-05-16 Richard Biener <rguenther@suse.de>
12699
12700 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
12701 (dump_stmt_cost): Declare.
12702 (add_stmt_cost): Dump cost we add.
12703 (add_stmt_costs): New function.
12704 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
12705 No longer exported.
12706 (vect_analyze_stmt): Adjust prototype.
12707 (vectorizable_condition): Likewise.
12708 (vectorizable_live_operation): Likewise.
12709 (vectorizable_reduction): Likewise.
12710 (vectorizable_induction): Likewise.
12711 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
12712 cost vector to pass to vectorizable_ and record afterwards.
12713 (vect_model_reduction_cost): Take cost vector argument and adjust.
12714 (vect_model_induction_cost): Likewise.
12715 (vectorizable_reduction): Likewise.
12716 (vectorizable_induction): Likewise.
12717 (vectorizable_live_operation): Likewise.
12718 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
12719 SLP_TREE_NUMBER_OF_VEC_STMTS.
12720 (vect_analyze_slp_cost_1): Remove.
12721 (vect_analyze_slp_cost): Likewise.
12722 (vect_slp_analyze_node_operations): Take visited args and
12723 a target cost vector. Avoid processing already visited stmt sets.
12724 (vect_slp_analyze_operations): Use a local cost vector to gather
12725 costs and register those of non-discarded instances.
12726 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
12727 (vect_schedule_slp_instance): Remove copying of
12728 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
12729 zero.
12730 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
12731 adding cost. Record cost entry location.
12732 (vect_prologue_cost_for_slp_op): Function to compute cost of
12733 a constant or invariant generated for SLP vect in the prologue,
12734 split out from vect_analyze_slp_cost_1.
12735 (vect_model_simple_cost): Make static. Adjust for SLP costing.
12736 (vect_model_promotion_demotion_cost): Likewise.
12737 (vect_model_store_cost): Likewise, make static.
12738 (vect_model_load_cost): Likewise.
12739 (vectorizable_bswap): Add cost vector arg and adjust.
12740 (vectorizable_call): Likewise.
12741 (vectorizable_simd_clone_call): Likewise.
12742 (vectorizable_conversion): Likewise.
12743 (vectorizable_assignment): Likewise.
12744 (vectorizable_shift): Likewise.
12745 (vectorizable_operation): Likewise.
12746 (vectorizable_store): Likewise.
12747 (vectorizable_load): Likewise.
12748 (vectorizable_condition): Likewise.
12749 (vectorizable_comparison): Likewise.
12750 (can_vectorize_live_stmts): Likewise.
12751 (vect_analyze_stmt): Likewise.
12752 (vect_transform_stmt): Adjust calls to vectorizable_*.
12753 * tree-vectorizer.c: Include gimple-pretty-print.h.
12754 (dump_stmt_cost): New function.
12755
12756 2018-05-16 Richard Biener <rguenther@suse.de>
12757
12758 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
12759 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
12760 * tree-ssa-dse.c: Include tree-ssa-loop.h.
12761 (check_name): New callback.
12762 (dse_classify_store): Track cycles via a visited bitmap of PHI
12763 defs and simplify handling of in-loop and across loop dead stores
12764 and properly fail for loop-variant refs. Handle byte-tracking with
12765 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
12766 limiting the walk.
12767
12768 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
12769
12770 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
12771 (vect_get_mask_type_for_stmt): Likewise.
12772 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
12773 split out from...
12774 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
12775 to determine the statement's vector type and the vector type that
12776 should be used for calculating nunits. Deal with cases in which
12777 the type has to be deferred.
12778 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
12779 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
12780 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
12781 (vect_determine_vf_for_stmt): New functions, split out from...
12782 (vect_determine_vectorization_factor): ...here.
12783 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
12784 (vect_get_mask_type_for_stmt): New functions, split out from
12785 vect_determine_vectorization_factor.
12786
12787 2018-05-16 Richard Biener <rguenther@suse.de>
12788
12789 * tree-cfg.c (verify_gimple_assign_ternary): Properly
12790 verify the [VEC_]COND_EXPR embedded comparison.
12791
12792 2018-05-15 Martin Sebor <msebor@redhat.com>
12793
12794 PR tree-optimization/85753
12795 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
12796 RECORD_TYPE in addition to ARRAY_TYPE.
12797
12798 2018-05-15 Martin Sebor <msebor@redhat.com>
12799
12800 PR middle-end/85643
12801 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
12802
12803 2018-05-15 Richard Biener <rguenther@suse.de>
12804
12805 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
12806 add by_clobber_p one. Change algorithm to collect all defs
12807 representing uses we need to walk and try reducing them to
12808 a single one before failing.
12809 (dse_dom_walker::dse_optimize_stmt): Adjust.
12810
12811 2018-05-13 Mark Wielaard <mark@klomp.org>
12812
12813 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
12814 (size_of_loc_descr): Likewise.
12815 (output_loc_operands): Likewise.
12816 (output_loc_operands_raw): Likewise.
12817 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
12818 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
12819 (hash_loc_operands): Likewise.
12820 (compare_loc_operands): Likewise.
12821
12822 2018-05-14 Mark Wielaard <mark@klomp.org>
12823
12824 * dwarf2out.c (count_index_addrs): New function.
12825 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
12826
12827 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12828
12829 PR tree-optimization/83648
12830 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
12831 return value as malloc candidate.
12832
12833 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
12834
12835 PR ipa/85734
12836 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
12837 param as true in call to suggest_attribute.
12838
12839 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
12840
12841 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
12842 -mreadonly-in-sdata.
12843
12844 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12845
12846 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
12847 New pattern.
12848 (aarch64_crypto_aesd_fused): Likewise.
12849
12850 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
12851
12852 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
12853 (movsi_aarch64): Likewise.
12854 (load_pairsi): Likewise.
12855 (load_pairdi): Likewise.
12856 (store_pairsi): Likewise.
12857 (store_pairdi): Likewise.
12858 (load_pairsf): Likewise.
12859 (load_pairdf): Likewise.
12860 (store_pairsf): Likewise.
12861 (store_pairdf): Likewise.
12862 (zero_extend): Likewise.
12863 (trunc): Swap alternatives.
12864 (fcvt_target): Add '?' to prefer w over r.
12865
12866 2018-05-14 Jakub Jelinek <jakub@redhat.com>
12867
12868 PR target/85756
12869 * config/i386/i386.md: Disallow non-commutative arithmetics in
12870 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
12871 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
12872 in the peephole2 before it.
12873
12874 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
12875
12876 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
12877 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
12878 (ix86_handle_option): Handle -mcldemote.
12879 * config.gcc: New header.
12880 * config/i386/cldemoteintrin.h: New file.
12881 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
12882 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
12883 -mcldemote.
12884 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
12885 OPTION_MASK_ISA_CLDEMOTE.
12886 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
12887 (ix86_valid_target_attribute_inner_p): Ditto.
12888 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
12889 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
12890 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
12891 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
12892 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
12893 (cldemote): New.
12894 * config/i386/i386.opt: Add -mcldemote.
12895 * config/i386/x86intrin.h: New header.
12896 * doc/invoke.texi: Add -mcldemote.
12897
12898 2018-05-14 Richard Biener <rguenther@suse.de>
12899
12900 * doc/match-and-simplify.texi: Adjust :s documentation.
12901
12902 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
12903
12904 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
12905 intended memcpy size.
12906 (REORDER_45): Likewise.
12907
12908 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
12909
12910 * sort.cc: New file.
12911 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
12912 * vec.c (qsort_chk): Use gcc_qsort.
12913 * Makefile.in (OBJS-libcommon): Add sort.o.
12914 (build/sort.o): New target. Use it...
12915 (BUILD_RTL): ... here, and...
12916 (build/gencfn-macros): ... here, and...
12917 (build/genmatch): ... here.
12918
12919 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12920 Chung-Ju Wu <jasonwucj@gmail.com>
12921
12922 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
12923 * config/nds32/nds32-graywolf.md: New file.
12924 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
12925 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
12926 pipeline.
12927 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
12928 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
12929 * config/nds32/nds32.md (pipeline_model): Add graywolf.
12930 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
12931 * config/nds32/pipelines.md: Include n15 settings.
12932
12933 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12934 Chung-Ju Wu <jasonwucj@gmail.com>
12935
12936 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
12937 * config/nds32/nds32-n13.md: New file.
12938 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
12939 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
12940 pipeline.
12941 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
12942 * config/nds32/nds32.md (pipeline_model): Add n13.
12943 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
12944 * config/nds32/pipelines.md: Include n13 settings.
12945
12946 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
12947 Chung-Ju Wu <jasonwucj@gmail.com>
12948
12949 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
12950 * config/nds32/nds32-n10.md: New file.
12951 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
12952 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
12953 pipeline.
12954 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
12955 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
12956 * config/nds32/nds32.md (pipeline_model): Add n10.
12957 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
12958 * config/nds32/pipelines.md: Include n10 settings.
12959
12960 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
12961 Kito Cheng <kito.cheng@gmail.com>
12962 Chung-Ju Wu <jasonwucj@gmail.com>
12963
12964 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
12965 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
12966 Add enum values for DSP extension instructions.
12967 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
12968 New constraints.
12969 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
12970 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
12971 New code iterators.
12972 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
12973 * config/nds32/nds32-dspext.md: New file for DSP implementation.
12974 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
12975 * config/nds32/nds32-intrinsic.md: Likewise.
12976 * config/nds32/nds32_intrinsic.h: Likewise.
12977 * config/nds32/nds32-md-auxiliary.c: Likewise.
12978 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
12979 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
12980 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
12981 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
12982 * config/nds32/nds32-protos.h: New declarations for DSP extension.
12983 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
12984 TYPE_DMAC in switch statement.
12985 * config/nds32/nds32.c: New checking and implementation for DSP
12986 extension instructions.
12987 * config/nds32/nds32.h: Likewise.
12988 * config/nds32/nds32.md: Likewise.
12989 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
12990 * config/nds32/predicates.md: Implement new predicates for DSP
12991 extension.
12992
12993 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
12994
12995 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
12996 Reformat alternatives and attributes so it is easier to identify
12997 which constraints/attributes go with which instruction.
12998 (mov<mode>_hardfloat32, FMOVE64): Likewise.
12999 (mov<mode>_softfloat32, FMOVE64): Likewise.
13000 (mov<mode>_hardfloat64, FMOVE64): Likewise.
13001 (mov<mode>_softfloat64, FMOVE64): Likewise.
13002
13003 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
13004
13005 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13006 subsection.
13007 (Basic PowerPC Built-in Functions): The new name of the
13008 subsection previously known as "PowerPC Built-in Functions".
13009 (Basic PowerPC Built-in Functions Available on all Configurations):
13010 New subsubsection.
13011 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
13012 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
13013 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
13014 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
13015
13016 2018-05-11 Martin Jambor <mjambor@suse.cz>
13017
13018 PR ipa/85655
13019 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
13020 single const.
13021
13022 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
13023
13024 PR target/85733
13025 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
13026
13027 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
13028
13029 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
13030 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
13031 (ix86_handle_option): Handle -mwaitpkg.
13032 * config.gcc: New header.
13033 * config/i386/cpuid.h (bit_WAITPKG): New bit.
13034 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
13035 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
13036 function type.
13037 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13038 OPTION_MASK_ISA_WAITPKG.
13039 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
13040 (ix86_option_override_internal): Add PTA_WAITPKG.
13041 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
13042 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
13043 IX86_BUILTIN_TPAUSE.
13044 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
13045 __builtin_ia32_umwait and __builtin_ia32_tpause.
13046 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
13047 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
13048 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
13049 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
13050 UNSPECV_TPAUSE): New.
13051 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
13052 * config/i386/i386.opt: Add -mwaitpkg.
13053 * config/i386/waitpkgintrin.h: New file.
13054 * config/i386/x86intrin.h: New header.
13055 * doc/invoke.texi: Add -mwaitpkg.
13056
13057 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
13058
13059 PR target/85606
13060 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
13061 equivalent.
13062 (cortex-m0): Use armv6s-m isa.
13063 (cortex-m0plus): Likewise.
13064 (cortex-m1): Likewise.
13065 (cortex-m0.small-multiply): Likewise.
13066 (cortex-m0plus.small-multiply): Likewise.
13067 (cortex-m1.small-multiply): Likewise.
13068
13069 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
13070 Jakub Jelinek <jakub@redhat.com>
13071
13072 PR tree-optimization/85692
13073 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
13074 source permute as well.
13075
13076 2018-05-11 Martin Liska <mliska@suse.cz>
13077
13078 PR sanitizer/85556
13079 * doc/extend.texi: Document LLVM style format for no_sanitize
13080 attribute.
13081
13082 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
13083
13084 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
13085 mode_supports_vsx_dform_quad to mode_supports_dq_form.
13086 (mode_supports_vsx_dform_quad): Likewise.
13087 (mode_supports_vmx_dform): Move these functions to be next to the
13088 other mode_supports functions.
13089 (mode_supports_dq_form): Likewise.
13090 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
13091 mode_supports_dq_form.
13092 (reg_offset_addressing_ok_p): Likewise.
13093 (offsettable_ok_by_alignment): Likewise.
13094 (rs6000_legitimate_offset_address_p): Likewise.
13095 (legitimate_lo_sum_address_p): Likewise.
13096 (rs6000_legitimize_address): Likewise.
13097 (rs6000_legitimize_reload_address): Likewise.
13098 (rs6000_secondary_reload_inner): Likewise.
13099 (rs6000_preferred_reload_class): Likewise.
13100 (rs6000_output_move_128bit): Likewise.
13101
13102 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
13103
13104 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
13105 Generate SImode target register for null target.
13106 <case IX86_BUILTIN_XGETBV>: Ditto.
13107 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
13108 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
13109
13110 2018-05-10 Carl Love <cel@us.ibm.com>
13111
13112 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
13113 dcbtt and dcbtstt if operands[2] is 0.
13114
13115 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
13116
13117 PR target/85693
13118 * config/i386/sse.md (usadv64qi): New expander.
13119
13120 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
13121
13122 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
13123 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
13124 -maltivec=be support.
13125 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
13126 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
13127 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
13128 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
13129 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
13130 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
13131 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
13132 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
13133 altivec_vsumsws): Adjust.
13134 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
13135 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
13136 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
13137 support.
13138 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
13139 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
13140 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
13141 (altivec_lve<VI_char>x): Delete expand.
13142 (*altivec_lve<VI_char>x_internal): Rename to...
13143 (altivec_lve<VI_char>x): ... this.
13144 (altivec_lvxl_<mode>): Delete expand.
13145 (*altivec_lvxl_<mode>_internal): Rename to ...
13146 (altivec_lvxl_<mode>): ... this.
13147 (altivec_stvxl_<mode>): Delete expand.
13148 (*altivec_stvxl_<mode>_internal): Rename to ...
13149 (altivec_stvxl_<mode>): ... this.
13150 (altivec_stve<VI_char>x): Delete expand.
13151 (*altivec_stve<VI_char>x_internal): Rename to ...
13152 (altivec_stve<VI_char>x): ... this.
13153 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
13154 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
13155 reduc_plus_scal_<mode>): Adjust.
13156 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
13157 comment.
13158 (rs6000_cpu_cpp_builtins): Adjust.
13159 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
13160 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
13161 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
13162 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
13163 -maltivec=be support.
13164 (rs6000_split_vec_extract_var): Adjust.
13165 (rs6000_split_v4si_init): Adjust.
13166 (swap_selector_for_mode): Delete.
13167 (altivec_expand_lvx_be, altivec_expand_stvx_be,
13168 altivec_expand_stvex_be): Delete.
13169 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
13170 -maltivec=be support.
13171 (rs6000_gimple_fold_builtin): Ditto.
13172 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
13173 Adjust.
13174 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
13175 (TARGET_DIRECT_MOVE_64BIT): Adjust.
13176 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
13177 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
13178 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
13179 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
13180 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
13181 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
13182 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
13183 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
13184 anonymous split): Adjust.
13185 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
13186 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
13187
13188 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
13189
13190 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
13191 when --with-gxx-include-dir is also specified.
13192 * configure: Regenerate.
13193
13194 2018-05-09 Jim Wilson <jimw@sifive.com>
13195
13196 PR target/84797
13197 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
13198 * config/riscv/t-withmultilib: New.
13199 * config/riscv/withmultilib.h: New.
13200 * doc/install.texi: Document RISC-V --with-multilib-list support.
13201
13202 2018-05-09 Richard Biener <rguenther@suse.de>
13203
13204 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
13205 vector.
13206 (vect_bb_vectorization_profitable_p): Adjust. Compute
13207 actual scalar cost using the cost vector and the add_stmt_cost
13208 machinery.
13209
13210 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13211
13212 PR rtl-optimization/85645
13213 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
13214 in the REG_CFA_REGISTER note for LR, don't leave it empty.
13215
13216 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13217
13218 PR rtl-optimization/85645
13219 * shrink-wrap.c (spread_components): Return a boolean saying if
13220 anything was changed.
13221 (try_shrink_wrapping_separate): Iterate spread_components until
13222 nothing changes anymore.
13223
13224 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13225
13226 PR rtl-optimization/85645
13227 * regrename.c (build_def_use): Also kill the chains that include the
13228 destination of a REG_CFA_REGISTER note.
13229
13230 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
13231
13232 PR rtl-optimization/85645
13233 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
13234 insn that has a REG_CFA_REGISTER note.
13235
13236 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
13237
13238 * cfgexpand.c (expand_clobber): New function.
13239 (expand_gimple_stmt_1): Use it.
13240 * tree-vect-stmts.c (vect_clobber_variable): New function,
13241 split out from...
13242 (vectorizable_simd_clone_call): ...here.
13243 (vectorizable_store): Emit a clobber either side of an
13244 IFN_STORE_LANES sequence.
13245 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
13246
13247 2018-05-09 Tom de Vries <tom@codesourcery.com>
13248
13249 PR target/85626
13250 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
13251 (define_insn "trap_if_false"): Add exit after trap.
13252
13253 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
13254
13255 PR rtl-optimization/85638
13256 * bb-reorder.c: Include common/common-target.h.
13257 (create_forwarder_block): New function extracted from...
13258 (fix_up_crossing_landing_pad): ...here. Rename into...
13259 (dw2_fix_up_crossing_landing_pad): ...this.
13260 (sjlj_fix_up_crossing_landing_pad): New function.
13261 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
13262 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
13263 from both partitions and exit the loop after one iteration.
13264
13265 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13266
13267 Revert:
13268 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13269 subsection.
13270 (Basic PowerPC Built-in Functions): The new name of the
13271 subsection previously known as "PowerPC Built-in Functions".
13272 (Basic PowerPC Built-in Functions Available on all Configurations):
13273 New subsubsection.
13274 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13275 subsubsection.
13276 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13277 subsubsection.
13278 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13279 subsubsection.
13280 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13281 subsubsection.
13282
13283 2018-05-08 Jim Wilson <jimw@sifive.com>
13284
13285 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
13286 (LD_EMUL_SUFFIX): New.
13287 (LINK_SPEC): Use it.
13288
13289 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
13290
13291 * doc/extend.texi (PowerPC Built-in Functions): Rename this
13292 subsection.
13293 (Basic PowerPC Built-in Functions): The new name of the
13294 subsection previously known as "PowerPC Built-in Functions".
13295 (Basic PowerPC Built-in Functions Available on all Configurations):
13296 New subsubsection.
13297 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
13298 subsubsection.
13299 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
13300 subsubsection.
13301 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
13302 subsubsection.
13303 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
13304 subsubsection.
13305
13306 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13307
13308 PR target/85683
13309 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
13310 after cmpelim optimization.
13311
13312 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
13313
13314 * config.gcc: Support "goldmont".
13315 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
13316 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
13317 PROCESSOR_GOLDMONT.
13318 * config/i386/i386.c (m_GOLDMONT): Define.
13319 (processor_target_table): Add "goldmont".
13320 (PTA_GOLDMONT): Define.
13321 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
13322 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
13323 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
13324 (fold_builtin_cpu): Add "goldmont".
13325 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
13326 (ix86_option_override_internal): Add "goldmont".
13327 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
13328 (processor_type): Add PROCESSOR_GOLDMONT.
13329 * config/i386/i386.md: Add CPU "glm".
13330 * config/i386/glm.md: New file.
13331 * config/i386/x86-tune.def: Add m_GOLDMONT.
13332 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
13333
13334 2018-05-08 Jakub Jelinek <jakub@redhat.com>
13335
13336 PR target/85572
13337 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
13338 E_V4DImode.
13339 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
13340 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
13341 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
13342
13343 PR target/85317
13344 * config/i386/i386.c (ix86_fold_builtin): Handle
13345 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
13346
13347 PR target/85480
13348 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
13349 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
13350
13351 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
13352
13353 PR target/85658
13354 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
13355 (check_arch): Likewise.
13356 (check_fpu): Return the result rather than printing it.
13357 (end arch): Fix operator precedence.
13358 (end cpu): Likewise.
13359 (END): Print the result from check_fpu.
13360
13361 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13362 Alan Hayward <alan.hayward@arm.com>
13363 David Sherwood <david.sherwood@arm.com>
13364
13365 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
13366 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
13367 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
13368 (*fcmuo<mode>_and): New patterns.
13369
13370 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
13371
13372 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
13373 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
13374 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
13375 (cmp_op, sve_imm_con): New code attributes.
13376 (SVE_COND_INT_CMP, imm_con): Delete.
13377 (cmp_op): Remove above unspecs from int attribute.
13378 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
13379 to...
13380 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
13381 comparison-specific unspecs.
13382 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
13383 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
13384 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
13385 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
13386 (*vec_fcm<cmp_op><mode>): Rename to...
13387 (*fcm<cmp_op><mode>): ...this and adjust likewise.
13388 (*vec_fcmuo<mode>): Rename to...
13389 (*fcmuo<mode>): ...this and adjust likewise.
13390 (*pred_fcm<cmp_op><mode>): New pattern.
13391 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
13392 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
13393 functions.
13394 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
13395 and UNORDERED.
13396 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
13397 (aarch64_emit_sve_predicated_cond): New function.
13398 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
13399 (aarch64_emit_unspec_cond_or): Replace with...
13400 (aarch64_emit_sve_or_conds): ...this new function. Use
13401 aarch64_emit_sve_ptrue_op for the individual comparisons and
13402 aarch64_emit_binop to OR them together.
13403 (aarch64_emit_inverted_unspec_cond): Replace with...
13404 (aarch64_emit_sve_inverted_cond): ...this new function. Use
13405 aarch64_emit_sve_ptrue_op for the comparison and
13406 aarch64_emit_unop to invert the result.
13407 (aarch64_expand_sve_vec_cmp_float): Update after the above
13408 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
13409
13410 2018-05-07 Nathan Sidwell <nathan@acm.org>
13411
13412 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
13413 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
13414 (Backwards Compatibility): Likewise.
13415
13416 2018-05-07 Luis Machado <luis.machado@linaro.org>
13417
13418 PR bootstrap/85681
13419 Revert:
13420 2018-05-07 Luis Machado <luis.machado@linaro.org>
13421
13422 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13423 <prefetch_dynamic_strides>: New const bool field.
13424 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13425 prefetch_dynamic_strides.
13426 (exynosm1_prefetch_tune): Likewise.
13427 (thunderxt88_prefetch_tune): Likewise.
13428 (thunderx_prefetch_tune): Likewise.
13429 (thunderx2t99_prefetch_tune): Likewise.
13430 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13431 to false.
13432 (aarch64_override_options_internal): Update to set
13433 PARAM_PREFETCH_DYNAMIC_STRIDES.
13434 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13435 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13436 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13437 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13438 prefetch-dynamic-strides setting.
13439
13440 2018-05-07 Luis Machado <luis.machado@linaro.org>
13441
13442 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13443 <minimum_stride>: New const int field.
13444 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13445 minimum_stride field.
13446 (exynosm1_prefetch_tune): Likewise.
13447 (thunderxt88_prefetch_tune): Likewise.
13448 (thunderx_prefetch_tune): Likewise.
13449 (thunderx2t99_prefetch_tune): Likewise.
13450 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
13451 (aarch64_override_options_internal): Update to set
13452 PARAM_PREFETCH_MINIMUM_STRIDE.
13453 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
13454 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
13455 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
13456 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
13457 stride is constant and is below the minimum stride threshold.
13458
13459 2018-05-07 Luis Machado <luis.machado@linaro.org>
13460
13461 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
13462 to 512.
13463
13464 2018-05-07 Luis Machado <luis.machado@linaro.org>
13465
13466 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13467 <prefetch_dynamic_strides>: New const bool field.
13468 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13469 prefetch_dynamic_strides.
13470 (exynosm1_prefetch_tune): Likewise.
13471 (thunderxt88_prefetch_tune): Likewise.
13472 (thunderx_prefetch_tune): Likewise.
13473 (thunderx2t99_prefetch_tune): Likewise.
13474 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
13475 to false.
13476 (aarch64_override_options_internal): Update to set
13477 PARAM_PREFETCH_DYNAMIC_STRIDES.
13478 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
13479 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
13480 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
13481 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
13482 prefetch-dynamic-strides setting.
13483
13484 2018-05-07 Luis Machado <luis.machado@linaro.org>
13485
13486 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
13487 <minimum_stride>: New const int field.
13488 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
13489 minimum_stride field.
13490 (exynosm1_prefetch_tune): Likewise.
13491 (thunderxt88_prefetch_tune): Likewise.
13492 (thunderx_prefetch_tune): Likewise.
13493 (thunderx2t99_prefetch_tune): Likewise.
13494 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
13495 (aarch64_override_options_internal): Update to set
13496 PARAM_PREFETCH_MINIMUM_STRIDE.
13497 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
13498 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
13499 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
13500 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
13501 stride is constant and is below the minimum stride threshold.
13502
13503 2018-05-06 Jakub Jelinek <jakub@redhat.com>
13504
13505 PR c++/85659
13506 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
13507 the type is addressable. Don't force op into register if it has
13508 BLKmode.
13509
13510 2018-05-05 Roland McGrath <mcgrathr@google.com>
13511
13512 PR other/77609
13513 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
13514 any section for which we don't know a specific type it should have,
13515 regardless of name. Previously this was done only for the exact
13516 names ".init_array", ".fini_array", and ".preinit_array".
13517 (default_elf_asm_named_section): Add comment about
13518 relationship with default_section_type_flags and SECTION_NOTYPE.
13519 (get_section): Don't consider it a type conflict if one side has
13520 SECTION_NOTYPE and the other doesn't, as long as neither has the
13521 SECTION_BSS et al used in the default_section_type_flags logic.
13522
13523 2018-05-05 Tom de Vries <tom@codesourcery.com>
13524
13525 PR target/85653
13526 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
13527 (workaround_barsyncs): New function.
13528 (nvptx_reorg): Use workaround_barsyncs.
13529 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
13530 (define_expand "nvptx_membar_cta"): New define_expand.
13531 (define_insn "*nvptx_membar_cta"): New insn.
13532
13533 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
13534
13535 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
13536 To improve optimization opportunities.
13537 * builtin-types.def: The new needed builtin types for the above.
13538
13539 2018-05-04 Richard Biener <rguenther@suse.de>
13540
13541 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
13542 * gimple-ssa-store-merging.c
13543 (imm_store_chain_info::output_merged_store): Remove redundant create,
13544 release split_store vector contents on failure.
13545 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
13546 scalar stmt vector on cache hit.
13547
13548 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
13549
13550 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
13551 Xilinx FP support.
13552 * config.gcc (powerpc-xilinx-eabi*): Remove.
13553 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
13554 support.
13555 (fusion_addis_mem_combo_load): Ditto.
13556 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
13557 FP support.
13558 (rs6000_cpu_cpp_builtins): Ditto.
13559 * config/rs6000/rs6000-linux.c
13560 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
13561 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
13562 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
13563 support.
13564 (rs6000_setup_reg_addr_masks): Ditto.
13565 (rs6000_init_hard_regno_mode_ok): Ditto.
13566 (rs6000_option_override_internal): Ditto.
13567 (legitimate_lo_sum_address_p): Ditto.
13568 (rs6000_legitimize_address): Ditto.
13569 (rs6000_legitimize_reload_address): Ditto.
13570 (rs6000_legitimate_address_p): Ditto.
13571 (abi_v4_pass_in_fpr): Ditto.
13572 (setup_incoming_varargs): Ditto.
13573 (rs6000_gimplify_va_arg): Ditto.
13574 (rs6000_split_multireg_move): Ditto.
13575 (rs6000_savres_strategy): Ditto.
13576 (rs6000_emit_prologue_components): Ditto.
13577 (rs6000_emit_epilogue_components): Ditto.
13578 (rs6000_emit_prologue): Ditto.
13579 (rs6000_emit_epilogue): Ditto.
13580 (rs6000_elf_file_end): Ditto.
13581 (rs6000_function_value): Ditto.
13582 (rs6000_libcall_value): Ditto.
13583 * config/rs6000/rs6000.h: Ditto.
13584 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
13585 (TARGET_MINMAX): ... this. New.
13586 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
13587 * config/rs6000/rs6000.md: Remove Xilinx FP support.
13588 (*movsi_internal1_single): Delete.
13589 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
13590 mfpu=, mxilinx-fpu): Delete.
13591 * config/rs6000/singlefp.h: Delete.
13592 * config/rs6000/sysv4.h: Remove Xilinx FP support.
13593 * config/rs6000/t-rs6000: Ditto.
13594 * config/rs6000/t-xilinx: Delete.
13595 * config/rs6000/titan.md: Adjust for fp_type removal.
13596 * config/rs6000/vsx.md: Remove Xilinx FP support.
13597 (VStype_simple): Delete.
13598 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
13599 * config/rs6000/xfpu.h: Delete.
13600 * config/rs6000/xfpu.md: Delete.
13601 * config/rs6000/xilinx.h: Delete.
13602 * config/rs6000/xilinx.opt: Delete.
13603 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
13604 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
13605
13606 2018-05-04 Tom de Vries <tom@codesourcery.com>
13607
13608 PR libgomp/85639
13609 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
13610 if ignore == 0.
13611
13612 2018-05-04 Richard Biener <rguenther@suse.de>
13613
13614 PR middle-end/85627
13615 * tree-complex.c (update_complex_assignment): We are always in SSA form.
13616 (expand_complex_div_wide): Likewise.
13617 (expand_complex_operations_1): Likewise.
13618 (expand_complex_libcall): Preserve EH info of the original stmt.
13619 (tree_lower_complex): Handle removed blocks.
13620 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
13621 on complex multiplication and division libcall builtins.
13622
13623 2018-05-04 Richard Biener <rguenther@suse.de>
13624
13625 PR middle-end/85574
13626 * fold-const.c (negate_expr_p): Restrict negation of operand
13627 zero of a division to when we know that can happen without
13628 overflow.
13629 (fold_negate_expr_1): Likewise.
13630
13631 2018-05-04 Jakub Jelinek <jakub@redhat.com>
13632
13633 PR libstdc++/85466
13634 * real.h (real_nextafter): Declare.
13635 * real.c (real_nextafter): New function.
13636 * fold-const-call.c (fold_const_nextafter): New function.
13637 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
13638 CASE_CFN_NEXTTOWARD.
13639 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
13640 even when arg1_mode is different from arg0_mode.
13641
13642 2018-05-03 Nathan Sidwell <nathan@acm.org>
13643
13644 * doc/extend.texi (Deprecated Features): Remove
13645 -ffriend-injection.
13646 (Backwards Compatibility): Likewise.
13647 * doc/invoke.texi (C++ Language Options): Likewise.
13648 (C++ Dialect Options): Likewise.
13649
13650 2018-05-03 Jakub Jelinek <jakub@redhat.com>
13651
13652 PR target/85530
13653 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
13654 _mm512_mask_mullox_epi64): New intrinsics.
13655
13656 2018-05-03 Tom de Vries <tom@codesourcery.com>
13657
13658 PR testsuite/85106
13659 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13660 dump files): Add offload-tree.
13661
13662 2018-05-03 Richard Biener <rguenther@suse.de>
13663
13664 PR tree-optimization/85615
13665 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
13666 to loops not nested in BBs loop father to avoid creating multi-entry
13667 loops.
13668
13669 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13670
13671 PR tree-optimization/70291
13672 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
13673 arguments. Change return type to tree. Emit libcall as a new
13674 statement rather than replacing existing one when inplace_p is true.
13675 (expand_complex_multiplication_components): New function.
13676 (expand_complex_multiplication): Expand floating-point complex
13677 multiplication using the above.
13678 (expand_complex_division): Rename inner_type parameter to type.
13679 Update expand_complex_libcall call-site.
13680 (expand_complex_operations_1): Update expand_complex_multiplication
13681 and expand_complex_division call-sites.
13682
13683 2018-05-02 Jakub Jelinek <jakub@redhat.com>
13684
13685 PR target/85582
13686 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
13687 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
13688 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
13689 the highest significant bit of the shift count mask is clear. In
13690 check whether and[sq]i3 is needed verify that all significant bits
13691 of the shift count other than the highest are set.
13692
13693 2018-05-02 Tom de Vries <tom@codesourcery.com>
13694
13695 PR libgomp/82428
13696 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
13697 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
13698 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
13699 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
13700 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
13701 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
13702 __builtin_goacc_parlevel_size.
13703
13704 2018-05-02 Richard Biener <rguenther@suse.de>
13705
13706 PR tree-optimization/85597
13707 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
13708 do not use split vect_get_vec_defs call but call vect_get_slp_defs
13709 directly.
13710
13711 2018-05-02 Tom de Vries <tom@codesourcery.com>
13712
13713 PR testsuite/85106
13714 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13715 dump files): Add ltrans-tree.
13716
13717 2018-05-02 Tom de Vries <tom@codesourcery.com>
13718
13719 PR testsuite/85106
13720 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
13721 dump files): Add wpa-ipa.
13722
13723 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
13724
13725 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
13726 powerpc*-*-linux*paired* target.
13727 * config/rs6000/750cl.h: Delete.
13728 * config/rs6000/paired.h: Delete.
13729 * config/rs6000/paired.md: Delete.
13730 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
13731 float support.
13732 * config/rs6000/rs6000-builtin.def: Remove paired float support.
13733 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
13734 comment. Remove paired float support.
13735 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
13736 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
13737 VECTOR_PAIRED.
13738 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
13739 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
13740 declarations.
13741 * config/rs6000/rs6000.c: Remove paired float support.
13742 (paired_expand_vector_init, paired_expand_vector_move,
13743 paired_emit_vector_compare, paired_emit_vector_cond_expr,
13744 (paired_expand_lv_builtin, paired_expand_stv_builtin,
13745 paired_expand_builtin, paired_expand_predicate_builtin,
13746 paired_init_builtins): Delete.
13747 * config/rs6000/rs6000.h: Remove paired float support.
13748 * config/rs6000/rs6000.md: Remove paired float support.
13749 (move_from_CR_ov_bit): Delete.
13750 * config/rs6000/rs6000.opt (mpaired): Delete.
13751 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
13752 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
13753
13754 2018-05-02 Richard Biener <rguenther@suse.de>
13755
13756 PR middle-end/85567
13757 * gimplify.c (gimplify_save_expr): When in SSA form allow
13758 SAVE_EXPRs to compute to SSA vars.
13759
13760 2018-05-02 Jakub Jelinek <jakub@redhat.com>
13761
13762 PR target/85582
13763 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
13764 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
13765 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
13766 clobber operands[2], instead use a new pseudo. Formatting fixes.
13767
13768 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
13769
13770 PR tree-optimization/85586
13771 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
13772 exit early for statements in the same group if the accesses are
13773 not strided.
13774
13775 2018-05-02 Tom de Vries <tom@codesourcery.com>
13776
13777 PR lto/85451
13778 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
13779 error message.
13780
13781 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
13782
13783 PR tree-optimization/85143
13784 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
13785
13786 2018-05-01 Tom de Vries <tom@codesourcery.com>
13787
13788 PR lto/85451
13789 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
13790 not found" error message.
13791
13792 2018-05-01 Tom de Vries <tom@codesourcery.com>
13793
13794 PR other/83786
13795 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
13796 * vec.c (test_ordered_remove_if): New function.
13797 (vec_c_tests): Call test_ordered_remove_if.
13798 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
13799 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
13800 * tree-vect-patterns.c (vect_pattern_recog_1): Use
13801 VEC_ORDERED_REMOVE_IF.
13802
13803 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13804
13805 PR tree-optimization/82665
13806 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
13807 pointer subtraction where arguments come from a memchr call.
13808
13809 2018-05-01 Jakub Jelinek <jakub@redhat.com>
13810
13811 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
13812 --push-state --as-needed and --pop-state instead of --as-needed and
13813 --no-as-needed if ld supports it.
13814 * configure: Regenerated.
13815
13816 PR web/85578
13817 * doc/install.texi2html: Replace _002d with - and _002a with * in
13818 generated html files using sed.
13819
13820 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13821
13822 PR c++/85523
13823 * gcc-rich-location.c (blank_line_before_p): New function.
13824 (use_new_line): New function.
13825 (gcc_rich_location::add_fixit_insert_formatted): New function.
13826 * gcc-rich-location.h
13827 (gcc_rich_location::add_fixit_insert_formatted): New function.
13828
13829 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13830
13831 * selftest.c (assert_streq): Rename "expected" and "actual" to
13832 "val1" and "val2". Extend NULL-handling to cover both inputs
13833 symmetrically, while still requiring both to be non-NULL for a pass.
13834 * selftest.h (assert_streq): Rename "expected" and "actual" to
13835 "val1" and "val2".
13836 (ASSERT_EQ): Likewise.
13837 (ASSERT_EQ_AT): Likewise.
13838 (ASSERT_KNOWN_EQ): Likewise.
13839 (ASSERT_KNOWN_EQ_AT): Likewise.
13840 (ASSERT_NE): Likewise.
13841 (ASSERT_MAYBE_NE): Likewise.
13842 (ASSERT_MAYBE_NE_AT): Likewise.
13843 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
13844 the assertion to pass.
13845 (ASSERT_STREQ_AT): Likewise.
13846
13847 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
13848
13849 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
13850 interaction with -pie.
13851
13852 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13853
13854 * selftest.h: Fix alphabetization of per-source-file selftest
13855 declarations.
13856
13857 2018-04-30 Jason Merrill <jason@redhat.com>
13858
13859 PR c++/61982 - dead stores to destroyed objects.
13860 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
13861 of clobber.
13862
13863 2018-04-30 Jason Merrill <jason@redhat.com>
13864
13865 * tree.c (build_clobber): New.
13866 * tree.h: Declare it.
13867 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
13868
13869 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13870
13871 * diagnostic-show-locus.c (layout::layout): Update for
13872 location_get_source_line returning a char_span.
13873 (struct char_span): Move to input.h.
13874 (struct correction): Update for fields in char_span becoming
13875 private.
13876 (struct source_line): Update for location_get_source_line
13877 returning a char_span.
13878 (layout::print_line): Likewise.
13879 * edit-context.c (edited_file::print_content): Likewise.
13880 (edited_file::print_diff_hunk): Likewise.
13881 (edited_file::print_run_of_changed_lines): Likewise.
13882 (edited_file::get_num_lines): Likewise.
13883 (edited_line::edited_line): Likewise.
13884 * final.c (asm_show_source): Likewise.
13885 * input.c (location_get_source_line): Convert return type
13886 from const char * to char_span, losing the final "line_len"
13887 param.
13888 (dump_location_info): Update for the above.
13889 (get_substring_ranges_for_loc): Likewise. Use a char_span
13890 when handling the literal within the line.
13891 (test_reading_source_line): Update for location_get_source_line
13892 returning a char_span.
13893 * input.h (class char_span): Move here from
13894 diagnostic-show-locus.c, converting from a struct to a class.
13895 Make data members private.
13896 (char_span::operator bool): New.
13897 (char_span::length): New.
13898 (char_span::get_buffer): New.
13899 (char_span::operator[]): New.
13900 (char_span::subspan): Make const.
13901 (char_span::xstrdup): New.
13902 (location_get_source_line): Convert return type from const char *
13903 to char_span, losing the final "line_size" param.
13904
13905 2018-04-30 Jan Hubicka <jh@suse.cz>
13906
13907 * lto-wrapper.c (ltrans_priorities): New static var.
13908 (cmp_priority): New.
13909 (run_gcc): Read priorities and if doing parallel build order
13910 the Makefile by them.
13911
13912 2018-04-30 David Malcolm <dmalcolm@redhat.com>
13913
13914 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
13915
13916 2018-04-30 Richard Biener <rguenther@suse.de>
13917
13918 * tree-cfg.c (verify_address): Remove base argument, add
13919 flag whether to check TREE_ADDRESSABLE and do that.
13920 (verify_expr): Remove.
13921 (verify_types_in_gimple_reference): Add pieces from verify_expr.
13922 (verify_gimple_assign_single): Likewise.
13923 (verify_gimple_switch): Likewise.
13924 (verify_expr_location_1): Dereference tp once. Add (disabled)
13925 piece from verify_expr.
13926 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
13927
13928 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13929
13930 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
13931
13932 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13933
13934 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
13935 (small_data_pattern): Likewise.
13936 (arc_rewrite_small_data): Likewise.
13937 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
13938 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
13939 (get_symbol_alignment): New function.
13940 (legitimate_small_data_address_p): Likewise.
13941 (legitimate_scaled_address): Update, call
13942 legitimate_small_data_address_p.
13943 (output_sdata): New static variable.
13944 (arc_print_operand): Update how we handle small data operands.
13945 (arc_print_operand_address): Likewise.
13946 (arc_legitimate_address_p): Update, use
13947 legitimate_small_data_address_p.
13948 (arc_rewrite_small_data_p): Remove.
13949 (arc_rewrite_small_data_1): Likewise.
13950 (arc_rewrite_small_data): Likewise.
13951 (small_data_pattern): Likewise.
13952 (compact_sda_memory_operand): Update to use
13953 legitimate_small_data_address_p and get_symbol_alignment.
13954 (prepare_move_operands): Don't rewite sdata pattern.
13955 (prepare_extend_operands): Remove.
13956 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
13957 pattern.
13958 (zero_extendqisi2): Likewise.
13959 (zero_extendhisi2): Likewise.
13960 (extendqihi2): Likewise.
13961 (extendqisi2): Likewise.
13962 (extendhisi2): Likewise.
13963 (addsi3): Likewise.
13964 (subsi3): Likewise.
13965 (andsi3): Likewise.
13966 * config/arc/constraints.md (Usd): Change it to memory constraint.
13967
13968 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
13969
13970 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
13971 as source of std instructions.
13972 * config/arc/arc.md (movsi_insn): Update pattern predicate to
13973 allow 6-bit constants as source for store instructions.
13974 (movdi_insn): Update instruction pattern to allow 6-bit constants
13975 as source for store instructions.
13976
13977 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
13978
13979 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
13980
13981 2018-04-30 Nathan Sidwell <nathan@acm.org>
13982 Sandra Loosemore <sandra@codesourcery.com>
13983
13984 * dumpfile.c (dump_open): Allow '-' for stdout.
13985 * doc/invoke.texi (Developer Options): Document dump filename
13986 determination early. Document stdin/stdout selection.
13987
13988 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
13989
13990 Microblaze Target: PIC data text relative
13991
13992 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
13993 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
13994 Add declaration.
13995 * config/microblaze/microblaze.h (microblaze_constant_address_p):
13996 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
13997 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
13998 New addressing mode for data-text relative position indepenedent code.
13999 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
14000 'ADDRESS_SYMBOLIC_TXT_REL'.
14001 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
14002 (microblaze_legitimate_pic_operand): Exclude function calls from
14003 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
14004 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
14005 addresses cases.
14006 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
14007 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
14008 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
14009 for 'address + offset'.
14010 (microblaze_expand_prologue): Add new function prologue call for
14011 'r20' assignation.
14012 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
14013 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
14014 table in case of TARGET_PIC_DATA_TEXT_REL.
14015 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
14016 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
14017 Add new macros 'UNSPEC_TEXT',
14018 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
14019 + exclude function calls from 'UNSPEC_PLT' in case of data text
14020 relative mode.
14021 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
14022 new target hook for generating address diff vector tables in case of
14023 flag_pic.
14024 * doc/tm.texi : Regenerate.
14025 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
14026 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
14027 of addr diff vector generation.
14028 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
14029 target hook definition.
14030 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
14031 Add default function for generate_pic_addr_diff_vec -> flag_pic.
14032 * doc/invoke.texi (Add new pic option): Add new microblaze pic
14033 option for data text relative.
14034
14035 2018-04-30 Richard Biener <rguenther@suse.de>
14036
14037 * tree-chrec.h (evolution_function_is_constant_p): Remove
14038 redundant check.
14039 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
14040
14041 2018-04-30 Richard Biener <rguenther@suse.de>
14042
14043 PR bootstrap/85571
14044 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
14045
14046 2018-04-30 Richard Biener <rguenther@suse.de>
14047
14048 PR tree-optimization/28364
14049 PR tree-optimization/85275
14050 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
14051 copying first exit test.
14052
14053 2018-04-28 Mark Wielaard <mark@klomp.org>
14054
14055 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
14056 dwarf_version >= 5.
14057 (dwarf_AT): Handle DW_AT_addr_base.
14058 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
14059
14060 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
14061
14062 PR target/84431
14063 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
14064 (*ashl<dwi>3_doubleword_mask_1): Ditto.
14065 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
14066 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
14067
14068 2018-04-28 Richard Biener <rguenther@suse.de>
14069
14070 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
14071 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
14072 to reflect use. Only add interesting stmts.
14073
14074 2018-04-27 Martin Jambor <mjambor@suse.cz>
14075
14076 PR ipa/85549
14077 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
14078 the jump function allows for passing through aggregate values.
14079
14080 2018-04-27 David Malcolm <dmalcolm@redhat.com>
14081
14082 * input.h (in_system_header_at): Convert from macro to inline
14083 function.
14084 (from_macro_expansion_at): Likewise.
14085 (from_macro_definition_at): Likewise.
14086
14087 2018-04-27 Jeff Law <law@redhat.com>
14088
14089 * config.gcc: Mark tile* targets as deprecated/obsolete.
14090
14091 2018-04-27 Richard Biener <rguenther@suse.de>
14092
14093 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
14094 fix for ILP32.
14095
14096 2018-04-27 Richard Biener <rguenther@suse.de>
14097
14098 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
14099
14100 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
14101
14102 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
14103 with Yd constraint. Set "preferred_for_speed" attribute from
14104 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
14105 with Yd constraint.
14106 (*movdi_internal): Ditto.
14107 (movti_interunit splitters): Remove
14108 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
14109 (movdi_interunit splitters): Ditto.
14110 * config/i386/constraints.md (Ye): Remove.
14111 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
14112
14113 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14114
14115 PR target/85512
14116 * config/aarch64/constraints.md (Usg): Limit to 31.
14117 (Usj): Limit to 63.
14118
14119 2018-04-27 Jakub Jelinek <jakub@redhat.com>
14120
14121 PR tree-optimization/85529
14122 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
14123 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
14124 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
14125 zero extension or masking of the MSB bit.
14126 (optimize_range_tests): Add FIRST_BB argument, pass it through
14127 to optimize_range_tests_var_bound.
14128 (maybe_optimize_range_tests, reassociate_bb): Adjust
14129 optimize_range_tests callers.
14130
14131 2018-04-26 Richard Biener <rguenther@suse.de>
14132 Jakub Jelinek <jakub@redhat.com>
14133
14134 * cgraph.h (symbol_table): Just declare debug method here.
14135 * symtab.c (symbol_table::debug): Define.
14136
14137 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
14138
14139 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
14140
14141 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
14142
14143 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
14144 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
14145 (*movdi_internal): Substitute Yi and Yj constraint with x
14146 and Ym and Yn constraint with y constraint. Update "isa"
14147 attribute and set "preferred_for_speed" attribute from
14148 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
14149 (*movsi_internal): Ditto.
14150 (*movdf_internal): Ditto.
14151 (*movsf_internal): Ditto.
14152 (*zero_extendsidi2): Ditto.
14153 * config/i386/sse.md (vec_set<mode>_0): Ditto.
14154 (sse2_loadld): Ditto.
14155 (*vec_extract<ssevecmodelower>_0): Ditto.
14156 (*vec_extractv4si_0_zext_sse4): Ditto.
14157 (vec_concatv2di): Ditto.
14158 (*vec_dup<mode>): Ditto.
14159 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
14160 * config/i386/constraints.md (Yi): Remove.
14161 (Yj): Remove.
14162 (Ym): Remove.
14163 (Yn): Remove.
14164
14165 2018-04-26 Nathan Sidwell <nathan@acm.org>
14166
14167 * dumpfile.c (dump_open): New.
14168 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
14169 (dump_finish): Detect stdio/stderr by value not name.
14170
14171 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
14172
14173 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
14174
14175 2018-04-26 Tom de Vries <tom@codesourcery.com>
14176
14177 PR target/84952
14178 * config/nvptx/nvptx.c (verify_neutering_jumps)
14179 (verify_neutering_labels): New function
14180 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
14181
14182 2018-04-26 Tom de Vries <tom@codesourcery.com>
14183
14184 PR target/84025
14185 * config/nvptx/nvptx.c (needs_neutering_p): New function.
14186 (nvptx_single): Use needs_neutering_p to skip over insns that do not
14187 need neutering.
14188
14189 2018-04-26 Richard Biener <rguenther@suse.de>
14190 Tom de Vries <tom@codesourcery.com>
14191
14192 PR lto/85422
14193 * lto-streamer-out.c (output_function): Fixup loops if required to match
14194 discovery done in the reader.
14195
14196 2018-04-26 Richard Biener <rguenther@suse.de>
14197
14198 PR tree-optimization/85116
14199 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
14200 have a loop exit from the single latch predecessor. Remove
14201 case of header with just condition.
14202 (ch_base::copy_headers): Exclude infinite loops from any
14203 processing.
14204 (pass_ch::execute): Record exits.
14205
14206 2018-04-26 Richard Biener <rguenther@suse.de>
14207
14208 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
14209 prologue cost vector and pass it to vect_get_load_cost.
14210 (vect_get_peeling_costs_all_drs): Likewise.
14211 (vect_peeling_hash_get_lowest_cost): Likewise.
14212 (vect_enhance_data_refs_alignment): Likewise.
14213
14214 2018-04-26 Richard Biener <rguenther@suse.de>
14215
14216 PR middle-end/85450
14217 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
14218 checking of integer<->pointer conversions.
14219 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
14220 sign-/zero-extending pointer types.
14221 (expand_omp_for_static_chunk): Likewise.
14222
14223 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
14224 Jean Lee <xiaoyur347@gmail.com>
14225
14226 * config/mips/mips.c (mips_asan_shadow_offset): New function.
14227 (TARGET_ASAN_SHADOW_OFFSET): Define.
14228 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
14229 true for -fsanitize=address.
14230
14231 2018-04-25 Mark Wielaard <mark@klomp.org>
14232
14233 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
14234 shorter ones.
14235
14236 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14237
14238 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
14239 than "alu", remove explicit "memory" and "imm_disp" attributes.
14240 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
14241
14242 PR middle-end/85414
14243 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
14244 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
14245 gen_lowpart_no_emit.
14246
14247 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
14248
14249 PR target/85473
14250 * config/i386/i386.c (ix86_expand_builtin): Change memory
14251 operand to XI, extend p0 to Pmode.
14252 * config/i386/i386.md: Change unspec volatile and operand
14253 1 mode to XI, change operand 0 mode to P.
14254
14255 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14256
14257 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
14258 GET_MODE_MASK before any checking.
14259 (nds32_can_use_bset_p): Likewise.
14260 (nds32_can_use_btgl_p): Likewise.
14261
14262 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14263
14264 * config/nds32/nds32-doubleword.md: New define_split pattern for
14265 illegal register number.
14266
14267 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14268
14269 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
14270
14271 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
14272
14273 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
14274
14275 2018-04-25 Richard Biener <rguenther@suse.de>
14276
14277 * lto-streamer.h (LTO_major_version): Bump to 8.
14278
14279 2018-04-25 Jakub Jelinek <jakub@redhat.com>
14280
14281 * BASE-VER: Set to 9.0.0.
14282
14283 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
14284
14285 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
14286 in __abskf2 and __powikf2.
14287
14288 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14289
14290 PR target/85512
14291 * config/aarch64/constraints.md (Usg, Usj): New constraints.
14292 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
14293 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
14294 Use the above on operand 2. Reindent.
14295 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
14296
14297 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14298
14299 PR target/85485
14300 * common/config/i386/i386-common.c (ix86_handle_option): Don't
14301 handle OPT_mcet.
14302 * config/i386/i386.opt (mcet): Removed.
14303 * doc/install.texi: Remove -mcet documentation.
14304 * doc/invoke.texi: Likewise.
14305
14306 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
14307
14308 PR target/85485
14309 * doc/install.texi: Remove -mcet from bootstrap-cet.
14310
14311 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14312
14313 PR target/85511
14314 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
14315 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
14316 if TARGET_64BIT.
14317
14318 PR target/85503
14319 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
14320 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
14321 containing a CONST_VECTOR.
14322
14323 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
14324
14325 * doc/install.texi: Update newlib dependency for nvptx.
14326
14327 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14328
14329 PR target/85508
14330 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
14331 instead of INTVAL when shifting x left.
14332
14333 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
14334
14335 PR tree-optimization/85478
14336 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
14337 vect_grouped_store_supported for single element vectors.
14338
14339 2018-04-24 Richard Biener <rguenther@suse.de>
14340
14341 PR target/85491
14342 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
14343 load cost increase to the case of non-constant step.
14344
14345 2018-04-24 Jakub Jelinek <jakub@redhat.com>
14346
14347 PR target/84828
14348 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
14349 destination if any_malformed_asm.
14350
14351 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
14352
14353 PR middle-end/85496
14354 * expr.c (store_field): In the bitfield case, if the value comes from
14355 a function call and is returned in registers by means of a PARALLEL,
14356 do not change the mode of the temporary unless BLKmode and VOIDmode.
14357
14358 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
14359
14360 PR rtl-optimization/85423
14361 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
14362 dependencies to debug insns when the previous insn is non-debug.
14363
14364 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14365
14366 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
14367 enums into a single definition.
14368 (fls): Fix predicates and printing.
14369 (seti): Likewise.
14370
14371 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
14372
14373 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
14374 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
14375 and short u6 immediate.
14376 (check_if_valid_sleep_operand): Remove.
14377 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
14378
14379 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14380
14381 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
14382 flag_always_save_lp condition.
14383 * config/nds32/nds32.opt (malways-save-lp): New option.
14384
14385 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14386
14387 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
14388 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
14389 * config/nds32/nds32.h
14390 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
14391 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
14392
14393 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14394
14395 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
14396 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
14397
14398 2018-04-22 Shiva Chen <shiva0217@gmail.com>
14399 Chung-Ju Wu <jasonwucj@gmail.com>
14400
14401 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
14402 Declare.
14403 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
14404 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
14405
14406 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14407
14408 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
14409
14410 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14411
14412 * config/nds32/nds32-protos.h (nds32_data_alignment,
14413 nds32_local_alignment): Declare.
14414 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
14415 nds32_local_alignment): New functions.
14416 (TARGET_CONSTANT_ALIGNMENT): Define.
14417 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
14418
14419 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14420
14421 * config/nds32/nds32.c
14422 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
14423 (TARGET_MODES_TIEABLE_P): Likewise.
14424
14425 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
14426
14427 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
14428 level Ofast and Og.
14429
14430 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
14431 Chung-Ju Wu <jasonwucj@gmail.com>
14432
14433 * config/nds32/constants.md (unspec_volatile_element): Add enum values
14434 for unaligned access.
14435 * config/nds32/nds32-intrinsic.c: Implementation of expanding
14436 unaligned access.
14437 * config/nds32/nds32-intrinsic.md: Likewise.
14438 * config/nds32/nds32_intrinsic.h: Likewise.
14439 * config/nds32/nds32.h (nds32_builtins): Likewise.
14440 * config/nds32/nds32.opt (munaligned-access): New option.
14441 * config/nds32/nds32.c (nds32_asm_file_start): Display
14442 flag_unaligned_access status.
14443
14444 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14445
14446 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
14447 -mno-relax is present.
14448 * config/riscv/linux.h (LINK_SPEC): Ditto.
14449
14450 2018-04-20 Martin Sebor <msebor@redhat.com>
14451
14452 PR c/85365
14453 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
14454 for null pointers.
14455 (gimple_fold_builtin_stxcpy_chk): Same.
14456 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
14457
14458 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
14459
14460 PR target/85456
14461 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
14462 __powikf2 when long double is IEEE 128-bit.
14463
14464 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
14465
14466 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
14467 step to make sure stack always aligned.
14468
14469 2018-04-20 Carl Love <cel@us.ibm.com>
14470
14471 PR target/83402
14472 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
14473 size check for arg0.
14474
14475 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
14476 Tom de Vries <tom@codesourcery.com>
14477
14478 PR target/85445
14479 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
14480 Emit insns for calls too.
14481 (nvptx_find_par): Always look for worker-level predecessor insn.
14482 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
14483 calls.
14484 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
14485 (nvptx_process_pars): Propagate frames for calls.
14486
14487 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
14488
14489 PR target/85469
14490 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
14491 Removed.
14492 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
14493 (ix86_handle_option): Don't handle OPT_mibt.
14494 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
14495 __SHSTK__.
14496 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
14497 has_ibt and ibt.
14498 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
14499 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
14500 (ix86_target_macros): Define __CET__ with flag_cf_protection
14501 for -fcf-protection.
14502 * config/i386/i386.c (isa2_opts): Remove -mibt.
14503 * config/i386/i386.h (TARGET_IBT): Removed.
14504 (TARGET_IBT_P): Likewise.
14505 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
14506 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
14507 * config/i386/i386.opt (mcet): Update help message.
14508 (mshstk): Likewise.
14509 (mibt): Removed.
14510 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
14511 -mcet as an alias for -mshstk.
14512
14513 2018-04-20 Richard Biener <rguenther@suse.de>
14514
14515 PR middle-end/85475
14516 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
14517 complexity by forcing a single use of the multiply operand.
14518
14519 2018-04-20 Martin Jambor <mjambor@suse.cz>
14520
14521 ipa/85449
14522 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
14523 recursion dependency to only apply to non-clones.
14524
14525 2018-04-20 Martin Jambor <mjambor@suse.cz>
14526
14527 ipa/85447
14528 * ipa-cp.c (create_specialized_node): Check that clones of
14529 self-recursive edges exist during IPA-CP.
14530
14531 2018-04-19 Toon Moene <toon@moene.org>
14532
14533 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
14534 by -O3.
14535
14536 2018-04-19 Jakub Jelinek <jakub@redhat.com>
14537
14538 PR tree-optimization/85467
14539 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
14540 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
14541 VECTOR_CST element to type.
14542
14543 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14544
14545 PR target/85397
14546 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
14547 * config/i386/i386.md (builtin_setjmp_setup): Removed.
14548 (builtin_longjmp): Likewise.
14549 (save_stack_nonlocal): New pattern.
14550 (restore_stack_nonlocal): Likewise.
14551
14552 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14553
14554 PR target/85404
14555 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
14556 Replace ASM_OUTPUT_LABEL with fprintf.
14557
14558 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
14559
14560 PR target/85417
14561 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
14562 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
14563 * config/i386/i386-c.c (ix86_target_macros_internal): Also
14564 define __IBT__ and __SHSTK__ for -fcf-protection.
14565 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
14566 TARGET_IBT.
14567 (ix86_trampoline_init): Likewise.
14568 (x86_output_mi_thunk): Likewise.
14569 (ix86_notrack_prefixed_insn_p): Likewise.
14570 (ix86_option_override_internal): Don't disallow -fcf-protection.
14571 * config/i386/i386.md (rdssp<mode>): Also enable for
14572 -fcf-protection.
14573 (incssp<mode>): Likewise.
14574 (nop_endbr): Likewise.
14575 * config/i386/i386.opt (mcet): Change help message to built-in
14576 functions only.
14577 (mibt): Likewise.
14578 (mshstk): Likewise.
14579 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
14580 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
14581 enable CET built-in functions.
14582
14583 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
14584
14585 * common/config/i386/i386-common.c
14586 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
14587 OPTION_MASK_ISA_MOVDIRI_UNSET,
14588 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
14589 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
14590 * config.gcc (movdirintrin.h): New header.
14591 * config/i386/cpuid.h (bit_MOVDIRI,
14592 bit_MOVDIR64B): New bits.
14593 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
14594 and -mmvodir64b.
14595 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
14596 (VOID, PVOID, PCVOID)): New function types.
14597 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
14598 __builtin_ia32_directstoreu_u64,
14599 __builtin_ia32_movdir64b): New builtins.
14600 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
14601 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
14602 and -mmovdiri.
14603 (ix86_valid_target_attribute_inner_p): Ditto.
14604 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
14605 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
14606 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
14607 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
14608 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
14609 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
14610 (movdiri<mode>, movdir64b_<mode>): New.
14611 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
14612 * config/i386/immintrin.h: Include movdirintrin.h.
14613 * config/i386/movdirintrin.h: New file.
14614 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
14615
14616 2018-04-19 Richard Biener <rguenther@suse.de>
14617
14618 PR middle-end/85455
14619 * cfg.c (clear_bb_flags): When loop state says we have
14620 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
14621
14622 2018-04-19 Richard Biener <rguenther@suse.de>
14623
14624 PR tree-optimization/84737
14625 * tree-vect-data-refs.c (vect_copy_ref_info): New function
14626 copying restrict info.
14627 (vect_setup_realignment): Use it.
14628 * tree-vectorizer.h (vect_copy_ref_info): Declare.
14629 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
14630 the first DR to all generated stores.
14631 (vectorizable_load): Likewise for loads.
14632
14633 2018-04-19 Jakub Jelinek <jakub@redhat.com>
14634
14635 PR tree-optimization/85446
14636 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
14637 the integral and pointer types to have the same precision.
14638
14639 * doc/install.texi: Document --disable-cet being the default and
14640 --enable-cet=auto.
14641
14642 2018-04-18 Martin Liska <mliska@suse.cz>
14643
14644 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
14645 style.
14646
14647 2018-04-18 Martin Liska <mliska@suse.cz>
14648
14649 Revert
14650 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
14651
14652 PR ipa/83983
14653 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
14654 arguments if they are comparable.
14655
14656 2018-04-18 Martin Liska <mliska@suse.cz>
14657
14658 Revert
14659 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
14660
14661 PR lto/84805
14662 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
14663 incomplete types.
14664
14665 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
14666
14667 PR target/85388
14668 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
14669 ENDBR after calling __morestack.
14670
14671 2018-04-18 David Malcolm <dmalcolm@redhat.com>
14672
14673 PR jit/85384
14674 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
14675 by using gcc_base_ver to generate a gcc_driver_version, and use
14676 it when generating GCC_DRIVER_NAME.
14677 * configure: Regenerate.
14678
14679 2018-04-18 Jakub Jelinek <jakub@redhat.com>
14680
14681 PR target/81084
14682 * config.gcc: Obsolete powerpc*-*-*spe*.
14683
14684 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14685
14686 PR debug/84637
14687 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
14688 (stabstr_D): Change type of unum from unsigned int to
14689 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
14690 type.
14691
14692 2018-04-17 Jim Wilson <jimw@sifive.com>
14693
14694 PR 84856
14695 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
14696 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
14697 Set arg_pointer_offset after using pretend_args_size.
14698
14699 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14700
14701 PR rtl-optimization/85431
14702 * dse.c (record_store): Ignore zero width stores.
14703
14704 PR sanitizer/85230
14705 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
14706 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
14707 __builtin_stack_restore rather than after it.
14708 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
14709 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
14710 argument instead of virtual_dynamic_stack_rtx.
14711
14712 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
14713
14714 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
14715 New prototype.
14716 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
14717 Add note to error message to explain internal mapping of overloaded
14718 built-in function name to non-overloaded built-in function name.
14719 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
14720 function.
14721
14722 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
14723
14724 PR target/85424
14725 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
14726 where the inputs overlap with the output.
14727
14728 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14729
14730 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
14731 (=v, v) alternative and explicit "memory" attribute.
14732 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
14733 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14734 attributes.
14735 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
14736 "sselog1" type instead of "sselog".
14737 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
14738 "sselog". Remove explicit "memory" attribute.
14739 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
14740 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14741 attributes.
14742 (vec_extract_hi_v32hi): Merge all alternatives into one, use
14743 "sselog1" type instead of "sselog". Remove explicit "memory"
14744 attribute.
14745 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
14746 use "sselog1" type instead of "sselog". Remove explicit "memory"
14747 attribute.
14748 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
14749 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
14750 attributes.
14751 (vec_extract_hi_v64qi): Merge all alternatives into one, use
14752 "sselog1" type instead of "sselog". Remove explicit "memory"
14753 attribute.
14754 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
14755 use "sselog1" type instead of "sselog". Remove explicit "memory"
14756 attribute.
14757
14758 PR target/85430
14759 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
14760
14761 PR middle-end/85414
14762 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
14763 on a SUBREG.
14764
14765 2018-04-17 Martin Jambor <mjambor@suse.cz>
14766
14767 PR ipa/85421
14768 * ipa-cp.c (create_specialized_node): Call
14769 expand_all_artificial_thunks if necessary.
14770
14771 2018-04-17 Martin Liska <mliska@suse.cz>
14772
14773 PR lto/85405
14774 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
14775 in message, remote space in between '_G' and '('.
14776
14777 2018-04-17 Jakub Jelinek <jakub@redhat.com>
14778
14779 PR target/85281
14780 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
14781 avx512f_vmcmp<mode>3<round_saeonly_name>,
14782 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
14783 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
14784 avx512f_rndscale<mode><round_saeonly_name>,
14785 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
14786 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
14787 Use %<iptr>2 instead of %2 for -masm=intel.
14788 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
14789 avx512f_vcvttss2usi<round_saeonly_name>,
14790 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
14791 -masm=intel.
14792 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
14793 avx512f_vcvttsd2usi<round_saeonly_name>,
14794 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
14795 Use %q1 instead of %1 for -masm=intel.
14796 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
14797 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
14798 of %3 for -masm=intel.
14799 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
14800 -masm=intel.
14801 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
14802 -masm=intel.
14803 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
14804 -masm=intel.
14805 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
14806 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
14807 %g1.
14808 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
14809 -masm=intel.
14810 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
14811 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
14812 %g1 and one with %0 and %1.
14813 (avx512er_vmrcp28<mode><round_saeonly_name>,
14814 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
14815 %1 for -masm=intel.
14816 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
14817 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
14818 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
14819 of %0 and %{%4%} for -masm=intel.
14820 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
14821 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
14822 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
14823 order of %0 and %{%5%}%{z%} for -masm=intel.
14824
14825 2018-04-17 Jan Hubicka <jh@suse.cz>
14826
14827 PR lto/85405
14828 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
14829
14830 2018-04-17 Martin Liska <mliska@suse.cz>
14831
14832 PR ipa/85329
14833 * multiple_target.c (create_dispatcher_calls): Set apostrophes
14834 for target_clone error message. Make default implementation
14835 clone to be a local declaration.
14836 (separate_attrs): Add new argument and check for an empty
14837 string.
14838 (expand_target_clones): Handle it.
14839 (ipa_target_clone): Make redirection just for target_clones
14840 functions.
14841
14842 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
14843 Tom de Vries <tom@codesourcery.com>
14844
14845 PR middle-end/84955
14846 * omp-expand.c (expand_oacc_for): Add dummy false branch for
14847 tiled basic blocks without omp continue statements.
14848
14849 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
14850
14851 PR target/83660
14852 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
14853 vec_extract expression as having side effects to make sure it gets
14854 a cleanup point.
14855
14856 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
14857
14858 PR target/85403
14859 * config/i386/i386.c (get_builtin_code_for_version): Check
14860 error_mark_node.
14861
14862 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
14863
14864 PR target/84331
14865 * config.gcc: Support "skylake".
14866 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
14867 PROCESSOR_SKYLAKE.
14868 * config/i386/i386.c (m_SKYLAKE): Define.
14869 (processor_target_table): Add "skylake".
14870 (ix86_option_override_internal): Add "skylake".
14871 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
14872 PROCESSOR_CANNONLAKE.
14873 (get_builtin_code_for_version): Fix priority for
14874 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
14875 PROCESSOR_SKYLAKE-AVX512.
14876 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
14877 (processor_type): Add PROCESSOR_SKYLAKE.
14878
14879 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
14880 Jason Merrill <jason@redhat.com>
14881
14882 PR c++/85112
14883 * convert.c (convert_to_integer_1): Use direct recursion for
14884 enumeral types and types with a precision less than the number
14885 of bits in their mode.
14886
14887 2018-04-16 Julia Koval <julia.koval@intel.com>
14888
14889 PR target/84413
14890 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
14891 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
14892
14893 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
14894
14895 PR target/85293
14896 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
14897 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
14898 and -mno-direct-move.
14899
14900 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
14901
14902 PR target/83402
14903 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
14904 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
14905 Ensure negative shifts result in {0}.
14906
14907 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
14908
14909 PR rtl-optimization/79916
14910 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
14911 regs (if any) to define how to gnerate SD moves when LRA is in
14912 progress.
14913
14914 2018-04-13 Jakub Jelinek <jakub@redhat.com>
14915
14916 PR rtl-optimization/85393
14917 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
14918 * except.c (expand_dw2_landing_pad_for_region): Make static.
14919 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
14920 a label and unconditional jump to old_bb, rather than
14921 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
14922 basic block.
14923
14924 PR rtl-optimization/85376
14925 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
14926 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
14927 instead of a specific value.
14928
14929 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
14930 Bin Cheng <bin.cheng@arm.com>
14931
14932 PR tree-optimization/82965
14933 PR tree-optimization/83991
14934 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
14935 by_profile_only parameter.
14936 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
14937 information if the loop was predicted to iterate too many times.
14938 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
14939
14940 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
14941
14942 PR lto/71991
14943 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
14944 always inline.
14945
14946 2018-04-13 Martin Liska <mliska@suse.cz>
14947 Jakub Jelinek <jakub@redhat.com>
14948
14949 PR middle-end/81657
14950 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
14951 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
14952 * builtins.c (expand_builtin_memory_copy_args): Use
14953 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
14954 handle dest_addr == pc_rtx.
14955
14956 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
14957
14958 PR target/85291
14959 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
14960 asked to not generate direct moves.
14961 (fix_trunc<mode>si2_stfiwx): Similar.
14962 (fix_trunc<mode>si2_internal): Similar.
14963
14964 2018-04-12 Jakub Jelinek <jakub@redhat.com>
14965
14966 PR debug/83157
14967 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
14968 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
14969 lookup if dest in some wider mode is known to be const0_rtx and
14970 if so, record permanent equivalence for it to be ZERO_EXTEND of
14971 the narrower mode destination.
14972
14973 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
14974
14975 * lto-streamer-out.c (output_function): Revert 259346.
14976 * omp-expand.c (expand_oacc_for): Likewise.
14977
14978 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
14979
14980 PR rtl-optimization/85354
14981 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
14982 * sel-sched.c (sel_global_init): ... here.
14983
14984 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
14985
14986 PR target/85238
14987 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
14988 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
14989 mode for PE-COFF targets.
14990 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
14991 (i386_pe_asm_lto_end): Likewise.
14992 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
14993 (TARGET_ASM_LTO_END): Likewise.
14994 * config/i386/winnt.c (saved_debug_info_level): New static variable.
14995 (i386_pe_asm_lto_start): New function.
14996 (i386_pe_asm_lto_end): Likewise.
14997
14998 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
14999 Richard Biener <rguenther@suse.de>
15000
15001 PR middle-end/84955
15002 * lto-streamer-out.c (output_function): Fix CFG loop state before
15003 streaming out.
15004 * omp-expand.c (expand_oacc_for): Handle calls to internal
15005 functions like regular functions.
15006
15007 2018-04-12 Richard Biener <rguenther@suse.de>
15008
15009 PR lto/85371
15010 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
15011 for the early LTO debug to properly generate references to it
15012 during DIE emission. Do not re-use that for the skeleton for
15013 split-dwarf.
15014 (dwarf2out_early_finish): Likewise.
15015
15016 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15017
15018 PR target/85328
15019 * config/i386/sse.md
15020 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
15021 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
15022 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
15023 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
15024 and output is a reg, avoid creating invalid lowpart subreg, but
15025 instead split into a 512-bit move. Don't split if not AVX512VL,
15026 input is xmm16+ reg and output is a mem.
15027 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
15028 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
15029 xmm16+ reg and output is a mem.
15030
15031 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15032
15033 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
15034 also for flag_dwarf2_cfi_asm.
15035
15036 2018-04-12 Jakub Jelinek <jakub@redhat.com>
15037
15038 PR rtl-optimization/85342
15039 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
15040 a bool scalar var inside of the loop instead. Don't try to update
15041 recog_data.operand after failed apply_change_group.
15042
15043 2018-04-12 Tom de Vries <tom@codesourcery.com>
15044
15045 PR target/85296
15046 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
15047 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
15048 array with flexible array member as array without given dimension.
15049 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
15050 argument for undefined param to true.
15051
15052 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
15053
15054 PR target/85321
15055 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
15056 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
15057 from PowerPC section.
15058 * config/rs6000/sysv4.opt (mcall-): Improve help text.
15059 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
15060 help text that is too long.
15061 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
15062 help text that is too long.
15063 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
15064 help text that is too long.
15065
15066 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
15067
15068 * config/alpha/alpha.md (stack_probe_internal): Rename
15069 from "probe_stack". Update all callers.
15070
15071 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
15072
15073 PR rtl-optimization/84566
15074 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
15075 sched_macro_fuse_insns.
15076
15077 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
15078
15079 PR target/84301
15080 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
15081 (compute_block_dependences): ... from here.
15082
15083 2018-04-11 Jakub Jelinek <jakub@redhat.com>
15084
15085 PR tree-optimization/85331
15086 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
15087 from int to HOST_WIDE_INT.
15088
15089 2018-04-11 Martin Jambor <mjambor@suse.cz>
15090
15091 PR ipa/84149
15092 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
15093 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
15094 not the same as the source val.
15095 (cgraph_edge_brings_value_p): New parameter.
15096 (gather_edges_for_value): Pass destination value to
15097 cgraph_edge_brings_value_p.
15098 (perhaps_add_new_callers): Likewise.
15099 (get_info_about_necessary_edges): Likewise and exclude values brought
15100 only by self-recursive edges.
15101 (create_specialized_node): Redirect only clones of self-calling edges.
15102 (+self_recursive_pass_through_p): New function.
15103 (find_more_scalar_values_for_callers_subset): Use it.
15104 (find_aggregate_values_for_callers_subset): Likewise.
15105 (known_aggs_to_agg_replacement_list): Removed.
15106 (decide_whether_version_node): Re-calculate known constants for all
15107 remaining context clones.
15108
15109 2018-04-11 Richard Biener <rguenther@suse.de>
15110
15111 PR lto/85339
15112 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
15113 from early DWARF output.
15114 (dwarf2out_early_finish): Output line info unconditionally into
15115 early DWARF and add reference to it.
15116
15117 2018-04-11 Jakub Jelinek <jakub@redhat.com>
15118
15119 PR target/85281
15120 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
15121 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
15122 other than V2DFmode using iptr mode attribute.
15123 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
15124
15125 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
15126
15127 PR rtl-optimization/84659
15128 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
15129
15130 2018-04-11 Jakub Jelinek <jakub@redhat.com>
15131
15132 PR debug/85302
15133 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
15134 SIZEP is NULL.
15135 (output_loc_list): Pass address of a dummy size variable even in the
15136 locview handling loop.
15137 (index_location_lists): Add comment on why skip_loc_list_entry can't
15138 call size_of_locs.
15139
15140 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
15141
15142 PR target/85261
15143 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
15144 into register.
15145
15146 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
15147
15148 PR target/85321
15149 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
15150 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
15151 and -mstring-compare-inline-limit.
15152
15153 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15154
15155 PR target/85287
15156 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
15157 for stack clash protection in a register whenever we need it to be in
15158 a register.
15159
15160 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15161
15162 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
15163 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
15164
15165 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
15166
15167 PR target/85321
15168 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
15169 the help text.
15170 (mlong-double-): Ditto.
15171 * config/rs6000/sysv4.opt (msdata=): Ditto.
15172 (mtls-size=): Ditto.
15173
15174 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
15175
15176 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
15177 erroneous entries for
15178 "vector int vec_ldl (int, long int *)", and
15179 "vector unsigned int vec_ldl (int, unsigned long int *)".
15180 Add comments and entries for
15181 "vector bool char vec_ldl (int, bool char *)",
15182 "vector bool short vec_ldl (int, bool short *)",
15183 "vector bool int vec_ldl (int, bool int *)",
15184 "vector bool long long vec_ldl (int, bool long long *)",
15185 "vector pixel vec_ldl (int, pixel *)",
15186 "vector long long vec_ldl (int, long long *)",
15187 "vector unsigned long long vec_ldl (int, unsigned long long *)".
15188 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
15189 type tree bool_long_long_type_node and correct definition of
15190 bool_V2DI_type_node to make reference to this new type tree.
15191 (rs6000_mangle_type): Replace erroneous reference to
15192 bool_long_type_node with bool_long_long_type_node.
15193 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
15194 comments to emphasize sign distinctions for char and int types and
15195 replace RS6000_BTI_bool_long constant with
15196 RS6000_BTI_bool_long_long constant. Also add comment to restrict
15197 use of RS6000_BTI_pixel.
15198 (bool_long_type_node): Remove this macro definition.
15199 (bool_long_long_type_node): New macro definition
15200
15201 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15202
15203 PR rtl-optimization/85300
15204 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
15205 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
15206 simplify_unary_operation fails.
15207
15208 2018-04-10 Martin Liska <mliska@suse.cz>
15209
15210 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
15211 cgraph_edge and ipa_ref.
15212
15213 2018-04-10 Jakub Jelinek <jakub@redhat.com>
15214
15215 PR target/85177
15216 PR target/85255
15217 * config/i386/sse.md
15218 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
15219 computation of the VEC_MERGE selector from mask.
15220 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
15221 Fix decoding of the VEC_MERGE selector into mask.
15222
15223 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15224
15225 PR tree-optimization/85286
15226 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
15227
15228 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
15229
15230 * final.c (final_1): Set insn_last_address as well as
15231 insn_current_address.
15232
15233 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15234
15235 PR target/85173
15236 * explow.c (emit_stack_probe): Call validize_mem on memory location
15237 before passing it to gen_probe_stack. Create address operand and
15238 legitimize it for the probe_stack_address case.
15239
15240 2018-04-09 Jan Hubicka <jh@suse.cz>
15241
15242 PR lto/85078
15243 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
15244 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
15245 * tree.c (free_lang_data_in_type): Fix handling of binfos;
15246 walk basetypes.
15247 (free_lang_data): Rebuild type inheritance graph.
15248
15249 2018-04-09 Martin Sebor <msebor@redhat.com>
15250
15251 * invoke.texi (-finline-small-functions): Mention other optimization
15252 options.
15253 (-findirect-inlining, -fpartial-inlining): Same.
15254 (-finline-functions-called-once): Same.
15255 (-freorder-blocks-and-partition): Same.
15256
15257 2018-04-09 Jan Hubicka <jh@suse.cz>
15258
15259 PR rtl/84058
15260 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
15261 jumps; choose last target that matches the criteria (i.e.
15262 no partition changes for non-crossing jumps).
15263 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
15264 support for redirecting crossing jumps to non-crossing.
15265
15266 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
15267
15268 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
15269 also for naked functions.
15270
15271 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
15272
15273 * config/arc/arc.md (add_shift): New pattern.
15274 (add_shift2): Likewise.
15275 (sub_shift): Likewise.
15276 (sub_shift_cmp0_noout): Likewise.
15277 (compare_si_ashiftsi): Likewise.
15278 (xbfu_cmp0_noout): New combine pattern.
15279 (xbfu_cmp0"): Likewise.
15280 (movsi_set_cc_insn): Place the predicable variant first.
15281 (commutative_binary_cmp0_noout): Remove clobber.
15282 (commutative_binary_cmp0): New pattern.
15283 (noncommutative_binary_cmp0): Likewise.
15284 (noncommutative_binary_cmp0_noout): Likewise.
15285 (noncommutative_binary_comparison_result_used): Removed.
15286 (rsub_cmp0): New pattern.
15287 (rsub_cmp0_noout): Likewise.
15288 (extzvsi): Changed, keep only meaningful variants.
15289 (SQH, SEZ): New iterators.
15290 (SQH_postfix): New mode attribute.
15291 (SEZ_prefix): New code attribute.
15292 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
15293 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
15294 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
15295 of numerical value.
15296 (noncommutative_operator): Check the availability of barrel
15297 shifter option.
15298
15299 2018-04-09 Richard Biener <rguenther@suse.de>
15300
15301 PR tree-optimization/85284
15302 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
15303 Only use the niter constraining form of simple_iv when the exit
15304 is always executed.
15305
15306 2018-04-09 Tom de Vries <tom@codesourcery.com>
15307
15308 PR target/84041
15309 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
15310 (define_expand "*memory_barrier"): New define_expand.
15311 (define_insn "memory_barrier"): New insn.
15312
15313 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15314
15315 PR rtl-optimization/80463
15316 PR rtl-optimization/83972
15317 PR rtl-optimization/83480
15318
15319 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
15320 correct producer for the insn.
15321 (tidy_control_flow): Fixup seqnos in case of debug insns.
15322
15323 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15324
15325 PR rtl-optimization/83913
15326
15327 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
15328 different sched-times when merging exprs.
15329
15330 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15331
15332 PR rtl-optimization/83962
15333
15334 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
15335 tidy_fallthru_edge and tidy_control_flow.
15336
15337 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
15338
15339 PR rtl-optimization/83530
15340
15341 * sel-sched.c (force_next_insn): New global variable.
15342 (remove_insn_for_debug): When force_next_insn is true, also leave only
15343 next insn in the ready list.
15344 (sel_sched_region): When the region wasn't scheduled, make another pass
15345 over it with force_next_insn set to 1.
15346
15347 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15348
15349 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
15350 into tm_file.
15351 * config/nds32/constants.md (unspec_volatile_element): Add enum values
15352 for interrupt control.
15353 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15354 functions for interrupt control.
15355 * config/nds32/nds32-intrinsic.md: Likewise.
15356 * config/nds32/nds32_intrinsic.h: Likewise.
15357 * config/nds32/nds32.h (nds32_builtins): Likewise.
15358
15359 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15360
15361 * config/nds32/nds32.c (nds32_init_machine_status,
15362 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
15363 strict_aligned_p field.
15364 (nds32_expand_to_rtl_hook): New function.
15365 (TARGET_EXPAND_TO_RTL_HOOK): Define.
15366 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
15367
15368 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15369 Chung-Ju Wu <jasonwucj@gmail.com>
15370
15371 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
15372 * config/nds32/nds32-n7.md: New file.
15373 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
15374 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
15375 pipeline.
15376 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
15377 * config/nds32/nds32.md (pipeline_model): Add n7.
15378 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
15379 * config/nds32/pipelines.md: Include n7 settings.
15380
15381 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15382 Chung-Ju Wu <jasonwucj@gmail.com>
15383
15384 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
15385 * config/nds32/nds32-e8.md: New file.
15386 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
15387 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
15388 pipeline.
15389 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
15390 * config/nds32/nds32.md (pipeline_model): Add e8.
15391 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
15392 * config/nds32/pipelines.md: Include e8 settings.
15393
15394 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15395 Chung-Ju Wu <jasonwucj@gmail.com>
15396
15397 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
15398 * config/nds32/nds32-n8.md: New file.
15399 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
15400 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
15401 pipeline.
15402 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
15403 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
15404 * config/nds32/nds32.md (pipeline_model): Add n8.
15405 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
15406 * config/nds32/pipelines.md: Include n8 settings.
15407
15408 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
15409 Chung-Ju Wu <jasonwucj@gmail.com>
15410
15411 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
15412 * config/nds32/nds32-n9-2r1w.md: New file.
15413 * config/nds32/nds32-n9-3r2w.md: New file.
15414 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
15415 nds32_register_ports): New or modify for cpu n9.
15416 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
15417 pipeline.
15418 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
15419 * config/nds32/nds32-utils.c: New file.
15420 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
15421 TARGET_MUL_SLOW): Define.
15422 * config/nds32/nds32.md (pipeline_model): New attribute.
15423 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
15424 New options that support cpu n9.
15425 * config/nds32/pipelines.md: Include n9 settings.
15426 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
15427
15428 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
15429
15430 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
15431 information if necessary.
15432 (output_cond_branch_compare_zero): Likewise.
15433 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
15434 (nds32_target_alignment): Refine for alignment.
15435 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
15436 (FUNCTION_BOUNDARY): Modify.
15437 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
15438 align case.
15439 * config/nds32/nds32.opt (malways-align, malign-functions): New.
15440
15441 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
15442
15443 * config/nds32/constants.md (unspec_volatile_element): Add values for
15444 TLB operation and data prefetch.
15445 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
15446 functions for TLB operation and data prefetch.
15447 * config/nds32/nds32-intrinsic.md: Likewise.
15448 * config/nds32/nds32_intrinsic.h: Likewise.
15449 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
15450 (nds32_print_operand): Likewise.
15451 * config/nds32/nds32.h (nds32_builtins): Likewise.
15452
15453 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
15454 Andrew Pinski <pinsika@gcc.gnu.org>
15455
15456 PR middle-end/82976
15457 * match.pd: Use constant_boolean_node of correct type instead of
15458 boolean_true_node or boolean_false_node for simplifying
15459 pointer comparisons to zero.
15460
15461 2018-04-07 Jakub Jelinek <jakub@redhat.com>
15462
15463 PR tree-optimization/80021
15464 * tree.c (verify_type_variant): Make error call in verify_variant_match
15465 translatable and remove final full stop.
15466
15467 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15468
15469 * config/nds32/constants.md (unspec_volatile_element): Add
15470 UNSPEC_VOLATILE_EH_RETURN.
15471 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
15472 nds32_output_stack_pop): Support dwarf exception handling process.
15473 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
15474 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
15475 exception handling process.
15476 (nds32_compute_stack_frame): Likewise.
15477 (nds32_return_addr_rtx): Likewise.
15478 (nds32_initial_elimination_offset): Likewise.
15479 (nds32_expand_prologue): Likewise.
15480 (nds32_expand_epilogue): Likewise.
15481 (nds32_dynamic_chain_address): New function.
15482 * config/nds32/nds32.h (machine_function): Add fields for dwarf
15483 exception handling.
15484 (DYNAMIC_CHAIN_ADDRESS): Define.
15485 (EH_RETURN_DATA_REGNO): Define.
15486 (EH_RETURN_STACKADJ_RTX): Define.
15487 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
15488 patterns for dwarf exception handling.
15489
15490 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15491
15492 * config/nds32/nds32.h: Clean up obsolete macros.
15493
15494 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15495
15496 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15497 Add enum values for particular instructions.
15498 * config/nds32/nds32-intrinsic.c: Implementation of expanding
15499 particular intrinsic functions.
15500 * config/nds32/nds32-intrinsic.md: Likewise.
15501 * config/nds32/nds32_intrinsic.h: Likewise.
15502 * config/nds32/nds32.h (nds32_builtins): Likewise.
15503 * config/nds32/nds32.md (type): Add pbsad and pbsada.
15504 (btst, ave): New patterns for particular instructions.
15505
15506 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15507
15508 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
15509 Add enum values for atomic load/store and memory sync.
15510 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
15511 and memory sync.
15512 * config/nds32/nds32-intrinsic.md: Likewise.
15513 * config/nds32/nds32_intrinsic.h: Likewise.
15514 * config/nds32/nds32.h (nds32_builtins): Likewise.
15515
15516 2018-04-07 Jakub Jelinek <jakub@redhat.com>
15517
15518 PR tree-optimization/85257
15519 * fold-const.c (native_encode_vector): If not all elts could fit
15520 and off is -1, return 0 rather than offset.
15521 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
15522 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
15523 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
15524 adjust buffer in native_interpret_expr call.
15525
15526 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15527
15528 * config/nds32/constants.md (unspec_volatile_element): Add cache
15529 control enum values.
15530 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
15531 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
15532 * config/nds32/nds32.c (nds32_cctl_names): New.
15533 (nds32_print_operand): Handle cache control register names.
15534 * config/nds32/nds32.h (nds32_builtins): New enum values.
15535 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
15536 macros.
15537 * config/nds32/nds32.md (type): Add mmu.
15538 * config/nds32/pipelines.md (simple_insn): Add mmu.
15539
15540 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15541
15542 * config/nds32/nds32.md (type): Remove call.
15543 * config/nds32/pipelines.md (simple_insn): Likewise.
15544
15545 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15546
15547 * config/nds32/constants.md (unspec_volatile_element): Add
15548 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
15549 UNSPEC_VOLATILE_FMFCFG.
15550 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
15551 description for fmfcfg and fmfcsr.
15552 (bdesc_1arg): Add fmtcsr.
15553 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
15554 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
15555 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
15556 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
15557 unspec_fmfcfg): New patterns.
15558 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
15559 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
15560 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
15561 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
15562 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
15563 __nds32__fmfcfg): Define.
15564
15565 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
15566
15567 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
15568 intrinsic register names.
15569 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
15570 intrinsic register enum values and macros.
15571
15572 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
15573
15574 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
15575 for load/store addressing form.
15576 (nds32_print_operand_address): Likewise.
15577
15578 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
15579
15580 PR target/85196
15581 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
15582 based on LABEL_REF. Remove useless assertion.
15583 (pic_address_needs_scratch): Fix formatting.
15584 (sparc_legitimize_pic_address): Minor tweaks.
15585 (sparc_delegitimize_address): Adjust assertion accordingly.
15586 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
15587 into symbolic_operand.
15588 (movsi_high_pic_label_ref): Likewise.
15589 (movsi_lo_sum_pic_label_ref): Likewise.
15590 (movdi_pic_label_ref): Likewise.
15591 (movdi_high_pic_label_ref): Likewise.
15592 (movdi_lo_sum_pic_label_ref): Likewise.
15593
15594 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
15595
15596 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
15597 custom LIB_SPEC setup.
15598
15599 2018-04-06 Ruslan Bukin <br@bsdpad.com>
15600 Kito Cheng <kito.cheng@gmail.com>
15601
15602 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
15603 * config/riscv/freebsd.h: New.
15604
15605 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15606
15607 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
15608 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
15609 file.
15610
15611 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15612 Kito Cheng <kito.cheng@gmail.com>
15613
15614 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
15615 nds32_output_call, nds32_symbol_binds_local_p): New functions.
15616 * config/nds32/nds32-protos.h (nds32_output_call,
15617 nds32_output_return): Declare.
15618 * config/nds32/nds32.md: Refine all the call and return patterns.
15619
15620 2018-04-06 Jakub Jelinek <jakub@redhat.com>
15621
15622 PR debug/85252
15623 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
15624 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
15625
15626 PR rtl-optimization/84872
15627 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
15628 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
15629 EDGE_CROSSING edge.
15630
15631 2018-04-06 Tamar Christina <tamar.christina@arm.com>
15632
15633 * expr.c (copy_blkmode_to_reg): Revert 254862.
15634 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
15635
15636 2018-04-06 Richard Biener <rguenther@suse.de>
15637
15638 PR middle-end/85244
15639 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
15640 after seeing a component reference with an adjacent field. Treat
15641 refs to arrays at struct end of external decls similar to
15642 refs to unconstrained commons.
15643
15644 2018-04-06 Jakub Jelinek <jakub@redhat.com>
15645
15646 PR sanitizer/85213
15647 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
15648 look through SAVE_EXPRs with non-side-effects argument. Adjust
15649 recursive calls.
15650 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
15651 save_p here.
15652
15653 2018-04-06 Richard Biener <rguenther@suse.de>
15654
15655 PR middle-end/85180
15656 * alias.c (find_base_term): New wrapper around find_base_term
15657 unwinding CSELIB_VAL_PTR changes.
15658 (find_base_term): Do not restore CSELIB_VAL_PTR during the
15659 recursion.
15660
15661 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15662
15663 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
15664 instructions.
15665 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
15666 constant definitions.
15667 ("nop"): lr 0,0 -> nopr r0
15668 ("nop_lr0", "nop_lr1"): New insn definitions.
15669
15670 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
15671
15672 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
15673 NDS32_V3PUSH_AVAILABLE_P macro.
15674
15675 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
15676 Chung-Ju Wu <jasonwucj@gmail.com>
15677
15678 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
15679 (nds32*-*-*): Add float and fpu_config into supported_defaults.
15680 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
15681 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
15682 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
15683 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
15684 * config/nds32/constraints.md: New constraints and checking for hard
15685 float configuration.
15686 * config/nds32/iterators.md: New mode iterator and attribute for hard
15687 float configuration.
15688 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
15689 patterns.
15690 * config/nds32/nds32-fpu.md: New file.
15691 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
15692 deal with hard float code generation.
15693 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
15694 ARCH_V3S.
15695 (abi_type, float_reg_number): New enum type.
15696 * config/nds32/nds32-predicates.c: New predicates for hard float.
15697 * config/nds32/nds32-protos.h: Declare functions for hard float.
15698 * config/nds32/nds32.c: Implementation for hard float configuration.
15699 * config/nds32/nds32.h: Definitions for hard float configuration.
15700 * config/nds32/nds32.md: Include hard float machine description and
15701 modify patterns for hard float configuration.
15702 * config/nds32/nds32.opt: New options for hard float configuration.
15703 * config/nds32/predicates.md: New predicates for hard float
15704 configuration.
15705
15706 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
15707
15708 * common/config/nds32/nds32-common.c
15709 (nds32_option_optimization_table): Enable -mreleax-hint by default.
15710
15711 2018-04-05 Jakub Jelinek <jakub@redhat.com>
15712
15713 PR middle-end/85195
15714 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
15715 CONSTRUCTOR_ELT (ctor, ...)->value.
15716
15717 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
15718
15719 PR target/85193
15720 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
15721
15722 2018-04-05 Tom de Vries <tom@codesourcery.com>
15723
15724 PR target/85204
15725 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
15726 cond jump.
15727
15728 2018-04-05 Shiva Chen <shiva0217@gmail.com>
15729 Kito Cheng <kito.cheng@gmail.com>
15730
15731 * config/nds32/constraints.md (U33): Fine-tune checking condition.
15732 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
15733 * config/nds32/nds32.h (nds32_16bit_address_type): Add
15734 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
15735
15736 2018-04-05 Shiva Chen <shiva0217@gmail.com>
15737 Kito Cheng <kito.cheng@gmail.com>
15738
15739 * config/nds32/constraints.md (Ufe): New memory constraint.
15740 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
15741 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
15742 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
15743 operands.
15744 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
15745 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
15746
15747 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15748
15749 * config/nds32/nds32.md: Use optimize_size in the condition for
15750 alu-shift instructions.
15751
15752 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15753
15754 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
15755
15756 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15757
15758 * config/nds32/nds32.md (negsi2): Refine pattern.
15759
15760 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
15761 Chung-Ju Wu <jasonwucj@gmail.com>
15762
15763 * config/nds32/iterators.md (shift_rotate): New code iterator.
15764 (shift): New code attribute.
15765 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
15766 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
15767 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
15768 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
15769 bit-wise operations.
15770 (andsi3, *andsi3): Ditto.
15771 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
15772 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
15773 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
15774 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
15775 nds32_ior_operand, nds32_xor_operand): New predicates.
15776
15777 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15778
15779 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
15780 (addsi3, subsi3): ... this.
15781
15782 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15783
15784 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
15785
15786 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15787
15788 * config/nds32/nds32.md: Adjust indention.
15789
15790 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
15791
15792 * config/nds32/nds32.md (feature): New attribute.
15793
15794 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
15795
15796 * config/nds32/nds32.md (subtype): New attribute.
15797
15798 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
15799
15800 PR target/85203
15801 * config/arm/arm-builtins.c (arm_expand_builtin): Change
15802 expansion to perform a bitwise AND of the argument followed by a
15803 boolean negation of the result.
15804
15805 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
15806
15807 PR rtl-optimization/84878
15808 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
15809 the basic block. Assert the use reference is not artificial and that
15810 it has an associated insn.
15811
15812 2018-04-04 Michael Matz <matz@suse.de>
15813
15814 * builtins.c (compute_objsize): Pass correct operand
15815 to array_at_struct_end_p.
15816
15817 2018-04-04 Richard Biener <rguenther@suse.de>
15818
15819 PR lto/85176
15820 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
15821 from contexts for DINFO_LEVEL_TERSE and below.
15822
15823 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15824
15825 * config/nds32/nds32-doubleword.md (move_<mode>): Require
15826 resiter_operand condition.
15827 * config/nds32/nds32.md (*move<mode>): Ditto.
15828
15829 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15830 Monk Chiang <sh.chiang04@gmail.com>
15831
15832 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
15833
15834 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15835
15836 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
15837
15838 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15839 Kito Cheng <kito.cheng@gmail.com>
15840
15841 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
15842 nds32_cond_code_str, output_cond_branch,
15843 output_cond_branch_compare_zero, nds32_expand_cbranch,
15844 nds32_expand_cstore, nds32_expand_movcc,
15845 nds32_output_cbranchsi4_equality_zero,
15846 nds32_output_cbranchsi4_equality_reg,
15847 nds32_output_cbranchsi4_equality_reg_or_const_int,
15848 nds32_output_cbranchsi4_greater_less_zero: New functions.
15849 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
15850 nds32_expand_cstore, nds32_expand_movcc,
15851 nds32_output_cbranchsi4_equality_zero,
15852 nds32_output_cbranchsi4_equality_reg,
15853 nds32_output_cbranchsi4_equality_reg_or_const_int,
15854 nds32_output_cbranchsi4_greater_less_zero): Declare.
15855 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
15856 nds32_rimm11s_operand): New predicates.
15857 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
15858 * config/nds32/nds32.md: Rewrite all the branch and conditional move
15859 patterns.
15860
15861 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
15862
15863 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
15864 * config/nds32/nds32.md: Ditto.
15865 * config/nds32/pipelines.md: Ditto.
15866
15867 2018-04-04 Richard Biener <rguenther@suse.de>
15868
15869 PR tree-optimization/85168
15870 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
15871 propagating abnormals.
15872
15873 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15874
15875 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
15876
15877 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
15878 Kito Cheng <kito.cheng@gmail.com>
15879
15880 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
15881 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
15882 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
15883 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
15884 * config/nds32/nds32.md (sibcall_internal): New.
15885 (sibcall_register): Remove.
15886 (sibcall_immediate): Remove.
15887 (sibcall_value_internal): New.
15888 (sibcall_value_register): Remove.
15889 (sibcall_value_immediate): Remove.
15890 * config/nds32/predicates.md (nds32_general_register_operand): New.
15891 (nds32_call_address_operand): New.
15892
15893 2018-04-03 Jakub Jelinek <jakub@redhat.com>
15894
15895 PR rtl-optimization/85167
15896 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
15897 bb_defs if *split_p, instead preinitialize it to NULL.
15898
15899 PR tree-optimization/85156
15900 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
15901 evaluating the argument multiple times.
15902
15903 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
15904
15905 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
15906 than vector.
15907 (_mm_cvtpd_ps): Likewise.
15908 (_mm_cvttpd_epi32): Likewise.
15909 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
15910 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
15911 vector, pixel, and bool following altivec.h include.
15912
15913 2018-04-03 Martin Sebor <msebor@redhat.com>
15914
15915 * doc/extend.texi (Common Function Attributes): Clarify.
15916 (const attribute): Likewise.
15917 (pure attribute): Likewise.
15918
15919 2018-04-03 Jakub Jelinek <jakub@redhat.com>
15920
15921 PR target/85169
15922 * config/i386/i386.c (ix86_expand_vector_set): Use
15923 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
15924
15925 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
15926
15927 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
15928 instructions when changing rounding bits to preserve precision bits
15929 in the x87 control word.
15930
15931 2018-04-03 Martin Liska <mliska@suse.cz>
15932
15933 PR tree-optimization/82491
15934 * rtl.h (strip_offset_and_add): Replace += suboffset with
15935 poly_uint64 () + suboffset.
15936
15937 2018-03-29 Martin Liska <mliska@suse.cz>
15938 Martin Jambor <mjambor@suse.cz>
15939
15940 PR ipa/84947
15941 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
15942 param_type is not an integral or pointer type.
15943
15944 2018-04-03 Richard Biener <rguenther@suse.de>
15945
15946 * sese.h (recompute_all_dominators): Remove.
15947
15948 2018-04-02 Martin Sebor <msebor@redhat.com>
15949
15950 * doc/invoke.texi (-Wrestrict): Fix typos.
15951
15952 2018-04-02 Jim Wilson <jimw@sifive.com>
15953
15954 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
15955 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
15956 (<optab>di3, <optab>si3_extend): Likewise.
15957 (<optab>si3_mask, <optab>si3_mask_1): New.
15958 (<optab>di3_mask, <optab>di3_mask_1): New.
15959 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
15960 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
15961 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
15962
15963 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
15964
15965 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
15966 example.
15967
15968 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
15969
15970 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
15971 (nds32_canonicalize_comparison): New function.
15972
15973 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15974 Kito Cheng <kito.cheng@gmail.com>
15975 Kuan-Lin Chen <kuanlinchentw@gmail.com>
15976
15977 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
15978 * config/nds32/constants.md (unspec_volatile_element): Add
15979 UNSPEC_VOLATILE_RELAX_GROUP.
15980 * config/nds32/nds32-relax-opt.c: New file.
15981 * config/nds32/nds32-predicates.c
15982 (nds32_symbol_load_store_p): New function.
15983 * config/nds32/nds32-protos.h
15984 (nds32_symbol_load_store_p): Declare function.
15985 (make_pass_nds32_relax_opt): Declare new rtl pass function.
15986 * config/nds32/nds32.c
15987 (nds32_register_pass): New function to register pass.
15988 (nds32_register_passes): New function to register passes.
15989 * config/nds32/nds32.md (relax_group): New pattern.
15990 * config/nds32/nds32.opt (mrelax-hint): New option.
15991 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
15992
15993 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
15994
15995 * config/nds32/t-nds32: Modify files dependency.
15996
15997 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
15998
15999 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
16000 (PROFILE_HOOK): Define its implementation.
16001
16002 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
16003
16004 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
16005 type and 32-bit size.
16006
16007 2018-04-01 Jakub Jelinek <jakub@redhat.com>
16008
16009 PR middle-end/85090
16010 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
16011 (V_128_256): New mode iterator.
16012 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
16013 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
16014 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
16015 of V.
16016 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
16017 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
16018
16019 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
16020
16021 PR target/83315
16022 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
16023 NaN inputs correctly.
16024
16025 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
16026
16027 PR target/80546
16028 * config/rs6000/vsx.md (??r): New mode attribute.
16029 (*vsx_mov<mode>_64bit): Use it.
16030 (*vsx_mov<mode>_32bit): Likewise.
16031
16032 2018-03-30 Martin Sebor <msebor@redhat.com>
16033
16034 PR tree-optimization/84818
16035 * builtins.c (check_access): Use warning_n.
16036
16037 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16038
16039 PR target/83822
16040 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
16041 condition.
16042 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
16043 condition.
16044
16045 2018-03-30 Julia Koval <julia.koval@intel.com>
16046
16047 PR target/84413
16048 * x86-tune.def (movx, partial_reg_dependency): Enable for
16049 m_SKYLAKE_AVX512.
16050
16051 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
16052
16053 PR inline-asm/84985
16054 * lra-constraints.c (process_alt_operands): Move setting
16055 this_alternative_matches below.
16056
16057 2018-03-29 Martin Liska <mliska@suse.cz>
16058
16059 PR lto/84995.
16060 * doc/invoke.texi: Document how LTO works with debug info.
16061 Describe auto-load support of binutils. Mention 'x86-64'
16062 as valid option value of -march option.
16063
16064 2018-03-29 Jakub Jelinek <jakub@redhat.com>
16065
16066 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
16067
16068 PR c/85094
16069 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
16070 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
16071 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
16072 checking.
16073
16074 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
16075
16076 PR target/84912
16077 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
16078 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
16079 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
16080 for RS6000_BTM_POWERPC64.
16081 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
16082 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
16083 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
16084 definition.
16085 (DIVDE): Use it.
16086 (DIVDEU): Likewise.
16087
16088 2018-03-28 Carl Love <cel@us.ibm.com>
16089
16090 Revert
16091 2017-09-27 Carl Love <cel@us.ibm.com>
16092
16093 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
16094 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
16095 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
16096 fctiw instruction.
16097
16098 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16099
16100 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
16101 instead of __vector bool.
16102 (_mm_max_pu8): Likewise.
16103 (_mm_min_pi16): Likewise.
16104
16105 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
16106
16107 PR target/84912
16108 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
16109 (DIVWEUO): Likewise.
16110 (DIVDEO): Likewise.
16111 (DIVDEUO): Likewise.
16112 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
16113 DIVWEUO and DIVDEUO.
16114 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
16115 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
16116 (div_extend): Likewise.
16117 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
16118 builtin function.
16119 (__builtin_divweuo): Likewise.
16120 (__builtin_divdeo): Likewise.
16121 (__builtin_divdeuo): Likewise.
16122
16123 2018-03-28 Jakub Jelinek <jakub@redhat.com>
16124
16125 PR target/85095
16126 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
16127 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
16128
16129 PR tree-optimization/82004
16130 * gimple-match-head.c (optimize_pow_to_exp): New function.
16131 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
16132 Don't fold to exp if optimize_pow_to_exp is false.
16133
16134 2018-03-28 Martin Liska <mliska@suse.cz>
16135
16136 PR other/84819
16137 * calls.c (initialize_argument_information): Fix trailing space.
16138 * common.opt: Fix typo and provide better explanation for
16139 -fsanitize-coverage option.
16140 * config/i386/i386.opt: Fix typo.
16141
16142 2018-03-28 Jakub Jelinek <jakub@redhat.com>
16143 Martin Liska <mliska@suse.cz>
16144
16145 PR sanitizer/85081
16146 * gimplify.c (asan_poison_variable): Don't do the check for
16147 gimplify_omp_ctxp here.
16148 (gimplify_decl_expr): Do it here.
16149 (gimplify_target_expr): Likewise.
16150
16151 2018-03-28 Martin Liska <mliska@suse.cz>
16152
16153 PR target/84988
16154 * config/i386/i386.c (ix86_function_arg_advance): Do not call
16155 chkp_type_bounds_count if MPX is not enabled.
16156
16157 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
16158
16159 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
16160
16161 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
16162
16163 PR target/84914
16164 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
16165 function to create the function decl for complex long double
16166 multiply and divide for -mabi=ieeelongdouble.
16167 (init_float128_ieee): Call it.
16168
16169 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
16170
16171 PR target/85044
16172 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
16173 -fcf-protection=branch -mibt.
16174 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
16175
16176 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16177
16178 PR target/81863
16179 * config/arm/arm.c (arm_valid_symbolic_address): Handle
16180 arm_word_relocations.
16181
16182 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
16183
16184 PR target/85056
16185 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
16186 extern array declarations.
16187
16188 2018-03-27 Richard Biener <rguenther@suse.de>
16189
16190 PR middle-end/84067
16191 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
16192 explicit single_use checks.
16193
16194 2018-03-27 Richard Biener <rguenther@suse.de>
16195
16196 PR tree-optimization/85082
16197 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
16198 Valueize the VUSE.
16199
16200 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
16201
16202 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
16203 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
16204 Turn on fasynchronous-unwind-tables and funwind-tables.
16205
16206 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
16207
16208 PR target/85073
16209 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
16210 (*bmi_blsr_<mode>_ccz): Ditto.
16211
16212 2018-03-26 Tom de Vries <tom@codesourcery.com>
16213
16214 PR tree-optimization/85063
16215 * omp-general.c (offloading_function_p): New function. Factor out
16216 of ...
16217 * omp-offload.c (pass_omp_target_link::gate): ... here.
16218 * omp-general.h (offloading_function_p): Declare.
16219 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
16220 with attribute omp declare target for offloading functions.
16221
16222 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
16223
16224 PR tree-optimization/84005
16225 * tree-data-ref.h (get_base_for_alignment): Declare.
16226 * tree-data-ref.c (get_base_for_alignment_1): New function.
16227 (get_base_for_alignment): Likewise.
16228 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
16229 get_base_for_alignment to find a suitable base object, instead
16230 of always using drb->base_address.
16231
16232 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16233
16234 PR inline-asm/85022
16235 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
16236 known size by default.
16237
16238 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
16239
16240 PR inline-asm/85030
16241 * lra-constraints.c (process_alt_operands): Don't match BLKmode
16242 and non BLKmode operands.
16243
16244 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16245
16246 PR target/85026
16247 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
16248 Clean up attributes.
16249
16250 2018-03-23 Richard Biener <rguenther@suse.de>
16251
16252 PR debug/85020
16253 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
16254 we are going to emit early debug for LTO.
16255
16256 2018-03-23 Jakub Jelinek <jakub@redhat.com>
16257
16258 PR inline-asm/85034
16259 * function.c (match_asm_constraints_1): Don't optimize if input
16260 doesn't satisfy general_operand predicate for output's mode.
16261
16262 PR inline-asm/85022
16263 * alias.c (write_dependence_p): Don't require for x_canonicalized
16264 non-VOIDmode if x has VOIDmode.
16265
16266 PR sanitizer/85029
16267 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
16268 just don't try to optimize it rather than assert it never happens.
16269
16270 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16271
16272 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
16273 macro expansions for definition of ST_INTERNAL_<mode> and
16274 LD_INTERNAL_<mode> builtins.
16275 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
16276 Remove prototype.
16277 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
16278 function.
16279 (altivec_expand_st_builtin): Likewise.
16280 (altivec_expand_builtin): Remove calls to deleted functions.
16281 (rs6000_address_for_altivec): Delete this function.
16282 * config/rs6000/vector.md: Remove expands for
16283 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
16284
16285 2018-03-22 Sudakshina Das <sudi.das@arm.com>
16286
16287 PR target/84826
16288 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
16289 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
16290 re-computing once computed.
16291 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
16292 (arm_init_machine_status): Initialize
16293 machine->static_chain_stack_bytes.
16294
16295 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16296
16297 PR target/84760
16298 * doc/extend.texi: Add four new prototypes for vec_ld.
16299 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
16300 definitions for more logical presentation.
16301 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
16302 entries for V1TI variants of __builtin_altivec_ld builtin.
16303 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
16304 handling of V1TI variant of LVX icode pattern.
16305 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
16306 (rs6000_gimple_fold_builtin): Likewise.
16307 (altivec_init_builtins): Add code to define
16308 __builtin_altivec_lvx_v1ti function.
16309
16310 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16311
16312 PR inline-asm/84941
16313 * function.c (match_asm_constraints_1): Don't do the optimization
16314 if input isn't a REG, SUBREG, MEM or constant.
16315
16316 2018-03-22 Tom de Vries <tom@codesourcery.com>
16317
16318 PR tree-optimization/84956
16319 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
16320 bb_has_abnormal_pred.
16321
16322 2018-03-22 Jakub Jelinek <jakub@redhat.com>
16323
16324 PR sanitizer/85018
16325 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
16326 DECL_INITIAL (decl) to decl at the end.
16327 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
16328 adjust the comment.
16329
16330 2018-03-21 Joseph Myers <joseph@codesourcery.com>
16331
16332 * doc/extend.texi (__builtin_tgmath): Document when complex
16333 integer types are treated as _Complex _Float64.
16334
16335 2018-03-21 Tom de Vries <tom@codesourcery.com>
16336
16337 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
16338
16339 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16340
16341 PR tree-optimization/84960
16342 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
16343 if it is ENTRY block, move them into single succ of ENTRY in that case.
16344
16345 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
16346
16347 PR tree-optimization/84811
16348 * poly-int.h (poly_span_traits): Remove the T3 parameter and
16349 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
16350 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
16351 (known_subrange_p): Update accordingly. Cast each value involved
16352 in the size comparison, rather than casting the result of the
16353 subtraction.
16354
16355 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16356
16357 PR tree-optimization/84982
16358 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
16359 by flipping the least significant bit rather than all bits from
16360 bitpos to bitpos + bitsize - 1.
16361
16362 2018-03-21 Nathan Sidwell <nathan@acm.org>
16363
16364 * doc/extend.texi (Deprecated Features): Remove mention of
16365 long-deleted deprecations.
16366
16367 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16368
16369 PR jit/84288
16370 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
16371 * configure: Regenerate.
16372
16373 2018-03-21 Tom de Vries <tom@codesourcery.com>
16374
16375 PR tree-optimization/83126
16376 * tree-parloops.c (num_phis): New function.
16377 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
16378
16379 2018-03-21 Nathan Sidwell <nathan@acm.org>
16380
16381 * doc/extend.texi (Deprecated Features): Update deprecated flags,
16382 mention anon-struct/union members and trailing attributes.
16383
16384 2018-03-21 Bin Cheng <bin.cheng@arm.com>
16385
16386 PR tree-optimization/84969
16387 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
16388 builtin memset partitions if they set different rhs values.
16389
16390 2018-03-21 Jakub Jelinek <jakub@redhat.com>
16391
16392 PR rtl-optimization/84989
16393 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
16394 VEC_DUPLICATE with scalar result mode.
16395
16396 2018-03-21 Martin Liska <mliska@suse.cz>
16397
16398 PR ipa/84963
16399 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
16400 not intended return statement.
16401
16402 2018-03-21 Martin Liska <mliska@suse.cz>
16403
16404 PR target/84988
16405 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
16406 (chkp_find_bound_slots_1): Limit number of iterations.
16407
16408 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
16409
16410 PR target/84838
16411 * Minor grammar fixes for x86 options.
16412
16413 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16414
16415 PR debug/84875
16416 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
16417 holding REG_CFA_RESTORE notes, instead turn them into a USE.
16418
16419 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
16420
16421 PR target/83789
16422 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
16423 (altivec_lvx_<mode>_1op): Likewise.
16424 (altivec_stvx_<mode>_2op): Likewise.
16425 (altivec_stvx_<mode>_1op): Likewise.
16426 (altivec_lvx_<VM2:mode>): New define_expand.
16427 (altivec_stvx_<VM2:mode>): Likewise.
16428 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
16429 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16430 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
16431 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
16432 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
16433 (rs6000_gen_lvx): Likewise.
16434 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
16435 (altivec_expand_stv_builtin): Likewise.
16436 (altivec_expand_builtin): Likewise.
16437 * config/rs6000/vector.md: Likewise.
16438
16439 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16440
16441 PR target/82518
16442 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
16443 BYTES_BIG_ENDIAN.
16444
16445 2018-03-20 Richard Biener <rguenther@suse.de>
16446
16447 PR target/84986
16448 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
16449 sign-conversions as zero, fall back to standard scalar_stmt
16450 cost for the rest.
16451
16452 2018-03-20 Martin Liska <mliska@suse.cz>
16453
16454 PR ipa/84825
16455 * predict.c (rebuild_frequencies): Handle case when we have
16456 PROFILE_ABSENT, but flag_guess_branch_prob is false.
16457
16458 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16459
16460 PR target/84990
16461 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
16462 flag_section_anchors.
16463 * varasm.c (use_blocks_for_decl_p): Remove hack for
16464 dw2_force_const_mem.
16465
16466 PR target/84845
16467 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
16468 to ...
16469 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
16470 be created, use lowpart_subreg of operands[0] rather than operands[0]
16471 itself.
16472 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
16473 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
16474 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
16475 and n constraint instead of aarch64_shift_imm_di and Usd.
16476 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
16477 (*aarch64_<optab>_reg_minus<mode>3): ... this.
16478
16479 2018-03-20 Sudakshina Das <sudi.das@arm.com>
16480
16481 PR target/82989
16482 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
16483 to favor GPR over NEON registers.
16484 (<shift>di3_neon): Likewise.
16485
16486 2018-03-20 Tom de Vries <tom@codesourcery.com>
16487
16488 PR target/84952
16489 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
16490 (nvptx_process_pars): Emit bar.sync asap and alap.
16491
16492 2018-03-20 Tom de Vries <tom@codesourcery.com>
16493
16494 PR target/84954
16495 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
16496 seen_label if seen_label is already set.
16497
16498 2018-03-20 Jakub Jelinek <jakub@redhat.com>
16499
16500 PR target/84945
16501 * config/i386/i386.c (fold_builtin_cpu): For features above 31
16502 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
16503 Use 1U instead of 1. Formatting fixes.
16504
16505 PR c/84953
16506 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
16507 instead of TREE_TYPE (s1) for the return value.
16508
16509 2018-03-19 Jakub Jelinek <jakub@redhat.com>
16510
16511 PR tree-optimization/84946
16512 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
16513 bitsize + bitsize in poly_uint64 rather than poly_int64.
16514
16515 PR sanitizer/78651
16516 * dwarf2asm.c: Include fold-const.c.
16517 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
16518 of decl rather than decl itself.
16519
16520 PR rtl-optimization/84643
16521 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
16522
16523 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
16524
16525 PR sanitizer/78651
16526 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
16527 calling assemble_variable.
16528
16529 2018-03-19 Sudakshina Das <sudi.das@arm.com>
16530
16531 PR target/81647
16532 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
16533 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
16534
16535 2018-03-19 Jim Wilson <jimw@sifive.com>
16536
16537 PR bootstrap/84856
16538 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
16539 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
16540 (riscv_first_stack_step): Likewise.
16541 (riscv_option_override): Use STACK_BOUNDARY instead of
16542 MIN_STACK_BOUNDARY.
16543 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
16544 MIN_STACK_BOUNDARY.
16545 (BIGGEST_ALIGNMENT): Set to 128.
16546 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
16547 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
16548 STACK_BOUNDARY.
16549
16550 2018-03-19 Richard Biener <rguenther@suse.de>
16551
16552 PR tree-optimization/84933
16553 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
16554 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
16555
16556 2018-03-19 Richard Biener <rguenther@suse.de>
16557
16558 PR tree-optimization/84859
16559 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
16560 (cond_if_else_store_replacement): Perform sinking operation on
16561 single-store BBs regardless of MAX_STORES_TO_SINK setting.
16562 Generalize what a BB with a single eligible store is.
16563
16564 2018-03-19 Richard Biener <rguenther@suse.de>
16565
16566 PR tree-optimization/84929
16567 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
16568 chrec_is_positive against non-chrec arg.
16569
16570 2018-03-19 Tamar Christina <tamar.christina@arm.com>
16571
16572 PR target/84711
16573 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
16574
16575 2018-03-18 Martin Liska <mliska@suse.cz>
16576
16577 PR rtl-optimization/84635
16578 * regrename.c (build_def_use): Use matches_mode only when
16579 matches >= 0.
16580
16581 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
16582
16583 PR tree-optimization/84913
16584 * tree-vect-loop.c (vectorizable_reduction): Don't try to
16585 vectorize chains of COND_EXPRs.
16586
16587 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16588
16589 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
16590
16591 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16592
16593 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
16594
16595 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
16596
16597 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
16598
16599 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
16600 Kito Cheng <kito.cheng@gmail.com>
16601
16602 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
16603 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
16604 (nds32_adjust_reg_alloc_order): New function.
16605 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
16606
16607 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
16608
16609 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
16610 nds32_print_operand, nds32_print_operand_address): Use
16611 HOST_WIDE_INT_PRINT_DEC instead.
16612
16613 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
16614
16615 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
16616
16617 2018-03-17 Jakub Jelinek <jakub@redhat.com>
16618
16619 PR target/84902
16620 * config/i386/i386.c (initial_ix86_tune_features,
16621 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
16622 unsigned long long.
16623 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
16624 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
16625 rather than 1u << ix86_tune. Formatting fix.
16626 (ix86_option_override_internal): Change ix86_arch_mask from
16627 unsigned int to unsigned HOST_WIDE_INT, initialize to
16628 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
16629 (ix86_function_specific_restore): Likewise.
16630
16631 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16632
16633 PR target/84899
16634 * postreload.c (reload_combine_recognize_pattern): Perform
16635 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
16636 truncate_int_for_mode the result for the destination's mode.
16637
16638 PR c/84909
16639 * hsa-gen.c (mem_type_for_type): Fix comment typo.
16640 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
16641 Likewise.
16642 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
16643 Likewise.
16644
16645 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
16646
16647 PR target/84876
16648 * lra-assigns.c (lra_split_hard_reg_for): Don't use
16649 regno_allocno_class_array and sorted_pseudos.
16650 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
16651 insns where regno is used.
16652
16653 2018-03-16 Martin Liska <mliska@suse.cz>
16654
16655 PR ipa/84833
16656 * multiple_target.c (create_dispatcher_calls): Redirect
16657 reference in the symbol table.
16658
16659 2018-03-16 Martin Liska <mliska@suse.cz>
16660
16661 PR ipa/84722
16662 * multiple_target.c (create_dispatcher_calls): Redirect also
16663 an alias.
16664
16665 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16666
16667 PR c++/79937
16668 PR c++/82410
16669 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
16670 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
16671 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
16672
16673 2018-03-16 Julia Koval <julia.koval@intel.com>
16674
16675 * doc/invoke.texi (Skylake Server): Add CLWB.
16676 Cannonlake): Remove CLWB.
16677
16678 2018-03-16 Jakub Jelinek <jakub@redhat.com>
16679
16680 PR tree-optimization/84841
16681 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
16682 1 << 3.
16683 (FLOAT_ONE_CONST_TYPE): Define.
16684 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
16685 (sort_by_operand_rank): Put entries with higher constant_type last
16686 rather than first to match comments.
16687
16688 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
16689
16690 * config/nios2/nios2.md (movsi_internal): Fix thinko in
16691 split predicate.
16692
16693 2018-03-15 Jakub Jelinek <jakub@redhat.com>
16694
16695 PR c++/79085
16696 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
16697 check and use address of target always.
16698
16699 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
16700
16701 PR target/84574
16702 * config/i386/i386.c (indirect_thunk_needed): Update comments.
16703 (indirect_thunk_bnd_needed): Likewise.
16704 (indirect_thunks_used): Likewise.
16705 (indirect_thunks_bnd_used): Likewise.
16706 (indirect_return_needed): New.
16707 (indirect_return_bnd_needed): Likewise.
16708 (output_indirect_thunk_function): Add a bool argument for
16709 function return.
16710 (output_indirect_thunk_function): Don't generate alias for
16711 function return thunk.
16712 (ix86_code_end): Call output_indirect_thunk_function to generate
16713 function return thunks.
16714 (ix86_output_function_return): Set indirect_return_bnd_needed
16715 and indirect_return_needed instead of indirect_thunk_bnd_needed
16716 and indirect_thunk_needed.
16717
16718 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
16719
16720 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
16721 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
16722 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
16723
16724 2018-03-15 David Malcolm <dmalcolm@redhat.com>
16725 Paul Hua <paul.hua.gm@gmail.com>
16726
16727 PR c/84852
16728 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
16729
16730 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
16731
16732 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
16733 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
16734 resp. SFmode cases.
16735
16736 2018-03-15 Tamar Christina <tamar.christina@arm.com>
16737
16738 PR target/84711
16739 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
16740 instead of GET_MODE_SIZE when comparing Units.
16741
16742 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
16743
16744 PR target/68256
16745 * varasm.c (hash_section): Return an unchangeble hash value
16746 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
16747 Return !aarch64_can_use_per_function_literal_pools_p ().
16748
16749 2018-03-15 Jakub Jelinek <jakub@redhat.com>
16750
16751 PR target/84860
16752 * optabs.c (emit_conditional_move): Pass address of cmode's copy
16753 rather than address of cmode as last argument to prepare_cmp_insn.
16754
16755 2018-03-15 Julia Koval <julia.koval@intel.com>
16756
16757 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
16758 F_AVX512VNNI, F_AVX512BITALG): New.
16759
16760 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
16761
16762 PR target/83451
16763 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
16764 insn for floating-point loads and stores.
16765
16766 2018-03-14 Carl Love <cel@us.ibm.com>
16767
16768 * config/rs6000/rs6000-c.c: Add macro definitions for
16769 ALTIVEC_BUILTIN_VEC_PERMXOR.
16770 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
16771 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
16772 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
16773 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
16774 UNSPEC_VPERMXOR.
16775 * config/doc/extend.texi: Add prototypes for vec_permxor.
16776
16777 2018-03-14 David Malcolm <dmalcolm@redhat.com>
16778
16779 PR c/84852
16780 * diagnostic-show-locus.c (class layout_point): Convert m_line
16781 from int to linenum_type.
16782 (line_span::comparator): Use linenum "compare" function when
16783 comparing line numbers.
16784 (test_line_span): New function.
16785 (layout_range::contains_point): Convert param "row" from int to
16786 linenum_type.
16787 (layout_range::intersects_line_p): Likewise.
16788 (layout::will_show_line_p): Likewise.
16789 (layout::print_source_line): Likewise.
16790 (layout::should_print_annotation_line_p): Likewise.
16791 (layout::print_annotation_line): Likewise.
16792 (layout::print_leading_fixits): Likewise.
16793 (layout::annotation_line_showed_range_p): Likewise.
16794 (struct line_corrections): Likewise for field m_row.
16795 (line_corrections::line_corrections): Likewise for param "row".
16796 (layout::print_trailing_fixits): Likewise.
16797 (layout::get_state_at_point): Likewise.
16798 (layout::get_x_bound_for_row): Likewise.
16799 (layout::print_line): Likewise.
16800 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
16801 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
16802 * input.c (selftest::test_linenum_comparisons): New function.
16803 (selftest::input_c_tests): Call it.
16804 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
16805 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
16806 * selftest.h (ASSERT_GT): New macro.
16807 (ASSERT_GT_AT): New macro.
16808 (ASSERT_LT): New macro.
16809 (ASSERT_LT_AT): New macro.
16810
16811 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
16812
16813 PR rtl-optimization/84780
16814 * combine.c (distribute_links): Don't make a link based on pc_rtx.
16815
16816 2018-03-14 Martin Liska <mliska@suse.cz>
16817
16818 * tree.c (record_node_allocation_statistics): Use
16819 get_stats_node_kind.
16820 (get_stats_node_kind): New function extracted from
16821 record_node_allocation_statistics.
16822 (free_node): Use get_stats_node_kind.
16823
16824 2018-03-14 Richard Biener <rguenther@suse.de>
16825
16826 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
16827 that the value-set of ANTIC_IN doesn't grow.
16828
16829 Revert
16830 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
16831 member.
16832 (BB_VISITED_WITH_VISITED_SUCCS): New define.
16833 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
16834
16835 2018-03-14 Julia Koval <julia.koval@intel.com>
16836
16837 * config.gcc (icelake-client, icelake-server): New.
16838 (icelake): Remove.
16839 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
16840 (initial_ix86_arch_features): Ditto.
16841 (PTA_SKYLAKE): Add SGX.
16842 (PTA_ICELAKE): Remove.
16843 (PTA_ICELAKE_CLIENT): New.
16844 (PTA_ICELAKE_SERVER): New.
16845 (ix86_option_override_internal): Split up icelake on icelake client and
16846 icelake server.
16847 (get_builtin_code_for_version): Ditto.
16848 (fold_builtin_cpu): Ditto.
16849 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
16850 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
16851 * config/i386/i386.h (processor_type): Ditto.
16852 * doc/invoke.texi: Ditto.
16853
16854 2018-03-14 Jakub Jelinek <jakub@redhat.com>
16855
16856 PR sanitizer/83392
16857 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
16858 INTEGER_CST offset, add it together with bitpos / 8 and
16859 sign extend based on POINTER_SIZE.
16860
16861 PR target/84844
16862 Revert
16863 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
16864
16865 PR target/78090
16866 * config/i386/constraints.md (Yc): New register constraint.
16867 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
16868 Use Yc constraint for alternative 2 of operand 0. Remove
16869 preferred_for_speed attribute.
16870
16871 2018-03-14 Richard Biener <rguenther@suse.de>
16872
16873 PR tree-optimization/84830
16874 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
16875 with the old one to avoid oscillations.
16876
16877 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
16878
16879 PR target/83712
16880 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
16881 pseudos.
16882 (assign_by_spills): Return a flag of reload assignment failure.
16883 Do not process the reload assignment failures. Do not spill other
16884 reload pseudos if they has the same reg class. Update n if
16885 necessary.
16886 (lra_assign): Add a return arg. Set up from the result of
16887 assign_by_spills call.
16888 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
16889 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
16890 usage_insns if it is not NULL.
16891 (spill_hard_reg_in_range): New function.
16892 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
16893 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
16894 function prototypes.
16895 (lra_assign): Change prototype.
16896 * lra.c (lra): Add code to deal with fails by splitting hard reg
16897 live ranges.
16898
16899 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
16900
16901 * config/riscv/riscv.opt (mrelax): New option.
16902 * config/riscv/riscv.c (riscv_file_start): Emit ".option
16903 "norelax" when riscv_mrelax is disabled.
16904 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
16905
16906 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
16907
16908 PR target/84743
16909 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
16910 reassociation for int modes.
16911
16912 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16913
16914 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
16915 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
16916 for big-endian.
16917 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
16918 * config/aarch64/aarch64-sve.md
16919 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
16920 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
16921 (*extend<mode><Vwide>2): Rename to...
16922 (aarch64_sve_extend<mode><Vwide>2): ...this.
16923 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
16924 renaming the old pattern to...
16925 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
16926 unsigned packs.
16927 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
16928 define_expand, renaming the old pattern to...
16929 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
16930 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
16931 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
16932 account when deciding which SVE instruction the optab should use.
16933 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
16934
16935 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16936
16937 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
16938 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
16939 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
16940 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
16941 (tlsdesc_small_<mode>): Turn a define_expand and use
16942 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
16943 (tlsdesc_small_advsimd_<mode>): ...this.
16944 (tlsdesc_small_sve_<mode>): New pattern.
16945
16946 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
16947
16948 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
16949 (UNSPEC_UMUL_HIGHPART): New constants.
16950 (MUL_HIGHPART): New int iteraor.
16951 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
16952 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
16953 define_expand.
16954 (*<su>mul<mode>3_highpart): New define_insn.
16955
16956 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
16957
16958 PR lto/84805
16959 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
16960 incomplete types.
16961
16962 2018-03-13 Martin Liska <mliska@suse.cz>
16963
16964 PR ipa/84658.
16965 * (sem_item_optimizer::sem_item_optimizer): Initialize new
16966 vector.
16967 (sem_item_optimizer::~sem_item_optimizer): Release it.
16968 (sem_item_optimizer::merge_classes): Register variable aliases.
16969 (sem_item_optimizer::fixup_pt_set): New function.
16970 (sem_item_optimizer::fixup_points_to_sets): Likewise.
16971 * ipa-icf.h: Declare new variables and functions.
16972
16973 2018-03-13 Jakub Jelinek <jakub@redhat.com>
16974
16975 PR middle-end/84834
16976 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
16977 integer_pow2p@2 and test integer_pow2p in condition.
16978 (A < 0 ? C : 0): Similarly for @1.
16979
16980 PR middle-end/84831
16981 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
16982 characters starting at p contain '\0' character, don't look beyond
16983 that.
16984
16985 PR target/84827
16986 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
16987 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
16988
16989 PR target/84828
16990 * reg-stack.c (change_stack): Change update_end var from int to
16991 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
16992 also call set_block_for_insn on the newly added insns and rescan.
16993
16994 PR target/84786
16995 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
16996 on the last operand.
16997
16998 PR c++/84704
16999 * tree.c (stabilize_reference_1): Return save_expr (e) for
17000 STATEMENT_LIST even if it doesn't have side-effects.
17001
17002 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
17003
17004 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
17005
17006 2018-03-12 Renlin Li <renlin.li@arm.com>
17007
17008 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
17009 aarch64_output_scalar_simd_mov_immediate.
17010
17011 2018-03-12 Martin Sebor <msebor@redhat.com>
17012
17013 PR tree-optimization/83456
17014 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
17015 for perfectly overlapping calls to memcpy.
17016 (gimple_fold_builtin_memory_chk): Same.
17017 (gimple_fold_builtin_strcpy): Handle no-warning.
17018 (gimple_fold_builtin_stxcpy_chk): Same.
17019 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
17020
17021 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
17022
17023 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
17024 parameter. Use it for SFmode.
17025 (rs6000_function_arg_advance_1): Adjust.
17026 (rs6000_function_arg): Adjust.
17027 (rs6000_gimplify_va_arg): Pass false for that new parameter.
17028
17029 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
17030
17031 PR rtl-optimization/84169
17032 PR rtl-optimization/84780
17033 * combine.c (can_combine_p): Check for a 2-insn combination whether
17034 the destination register is used between the two insns, too.
17035
17036 2018-03-12 Richard Biener <rguenther@suse.de>
17037
17038 PR tree-optimization/84803
17039 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
17040 for refs DR analysis didn't process.
17041
17042 2018-03-12 Richard Biener <rguenther@suse.de>
17043
17044 PR tree-optimization/84777
17045 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
17046 force-vectorize loops ignore whether we are optimizing for size.
17047
17048 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
17049
17050 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
17051 (TARGET_MD_ASM_ADJUST): Define.
17052
17053 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
17054 Kito Cheng <kito.cheng@gmail.com>
17055 Chung-Ju Wu <jasonwucj@gmail.com>
17056
17057 * config/nds32/nds32.c (nds32_compute_stack_frame,
17058 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
17059 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
17060 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
17061 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
17062 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
17063 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
17064 * config/nds32/nds32.md (prologue, epilogue): Use macro
17065 NDS32_V3PUSH_AVAILABLE_P to do checking.
17066
17067 2018-03-11 Jakub Jelinek <jakub@redhat.com>
17068
17069 PR debug/58150
17070 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
17071 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
17072 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
17073 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
17074 addition of most attributes on !orig_type_die or the attribute not
17075 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
17076
17077 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
17078 Chung-Ju Wu <jasonwucj@gmail.com>
17079
17080 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
17081 __NDS32_VH__ macro.
17082 * config/nds32/nds32.opt (mvh): New option.
17083
17084 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
17085 Chung-Ju Wu <jasonwucj@gmail.com>
17086
17087 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
17088 function.
17089 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
17090 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
17091 definition.
17092
17093 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
17094 Chung-Ju Wu <jasonwucj@gmail.com>
17095
17096 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
17097 function.
17098 * config/nds32/nds32-multiple.md (strlensi): New pattern.
17099 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
17100
17101 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
17102 Kito Cheng <kito.cheng@gmail.com>
17103 Chung-Ju Wu <jasonwucj@gmail.com>
17104
17105 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
17106 UNSPEC_FFMISM and UNSPEC_FLMISM.
17107 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
17108 for ffb, ffmism and flmism.
17109 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
17110 (unspec_ffmism): Ditto.
17111 (unspec_flmism): Ditto.
17112 (nds32_expand_builtin_impl): Check if string extension is available.
17113 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
17114 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
17115
17116 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
17117
17118 Reverting patch:
17119 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
17120
17121 PR target/83712
17122 * lra-assigns.c (assign_by_spills): Return a flag of reload
17123 assignment failure. Do not process the reload assignment
17124 failures. Do not spill other reload pseudos if they has the same
17125 reg class.
17126 (lra_assign): Add a return arg. Set up from the result of
17127 assign_by_spills call.
17128 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17129 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17130 usage_insns if it is not NULL.
17131 (spill_hard_reg_in_range): New function.
17132 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17133 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17134 function prototypes.
17135 (lra_assign): Change prototype.
17136 * lra.c (lra): Add code to deal with fails by splitting hard reg
17137 live ranges.
17138
17139 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
17140
17141 PR target/84807
17142 * config/i386/i386.opt: Replace Enforcment with Enforcement.
17143
17144 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
17145
17146 PR debug/84620
17147 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
17148 (dw_val_node): Add val_symbolic_view.
17149 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
17150 (symview_upper_bound): New.
17151 (new_line_info_table): Initialize symviews_since_reset.
17152 (dwarf2out_source_line): Count symviews_since_reset and set
17153 symview_upper_bound.
17154 (dw_val_equal_p): Handle symview.
17155 (add_AT_symview): New.
17156 (print_dw_val): Handle symview.
17157 (attr_checksum, attr_checksum_ordered): Likewise.
17158 (same_dw_val_p, size_of_die): Likewise.
17159 (value_format, output_die): Likewise.
17160 (add_high_low_attributes): Use add_AT_symview for entry_view.
17161 (dwarf2out_finish): Reset symview_upper_bound, clear
17162 zero_view_p.
17163
17164 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
17165
17166 PR target/83969
17167 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
17168 Add strict argument and use it.
17169 (rs6000_split_multireg_move): Update for new strict argument.
17170 (mem_operand_gpr): Disallow all non-offsettable addresses.
17171 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
17172
17173 2018-03-09 Jakub Jelinek <jakub@redhat.com>
17174
17175 PR target/84772
17176 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
17177 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
17178 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
17179
17180 PR c++/84767
17181 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
17182 decl, use remap_type if we want to use the type.
17183
17184 2018-03-09 Martin Sebor <msebor@redhat.com>
17185
17186 PR tree-optimization/84526
17187 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
17188 Remove dead code.
17189 (builtin_access::generic_overlap): Be prepared to handle non-array
17190 base objects.
17191
17192 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
17193
17194 PR rtl-optimization/84682
17195 * lra-constraints.c (process_address_1): Check is_address flag
17196 for address constraints.
17197 (process_alt_operands): Likewise.
17198 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
17199 preprocess_constraints.
17200 * recog.h (preprocess_constraints): Add oploc parameter.
17201 Adjust callers.
17202 * recog.c (preprocess_constraints): Test address_operand for
17203 CT_ADDRESS constraints.
17204
17205 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
17206
17207 PR target/83712
17208 * lra-assigns.c (assign_by_spills): Return a flag of reload
17209 assignment failure. Do not process the reload assignment
17210 failures. Do not spill other reload pseudos if they has the same
17211 reg class.
17212 (lra_assign): Add a return arg. Set up from the result of
17213 assign_by_spills call.
17214 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
17215 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
17216 usage_insns if it is not NULL.
17217 (spill_hard_reg_in_range): New function.
17218 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
17219 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
17220 function prototypes.
17221 (lra_assign): Change prototype.
17222 * lra.c (lra): Add code to deal with fails by splitting hard reg
17223 live ranges.
17224
17225 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17226
17227 PR target/83193
17228 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
17229 Accept complain bool parameter. Only emit errors if it is true.
17230 (arm_parse_cpu_option_name): Likewise.
17231 (arm_target_thumb_only): Adjust callers of the above.
17232 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
17233 prototype to take a default true bool parameter.
17234 (arm_parse_arch_option_name): Likewise.
17235
17236 2018-03-09 David Malcolm <dmalcolm@redhat.com>
17237 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
17238
17239 PR jit/64089
17240 PR jit/84288
17241 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
17242 * configure: Regenerate.
17243 * configure.ac ("linker --version-script option"): New.
17244 ("linker soname option"): New.
17245
17246 2018-03-09 Richard Biener <rguenther@suse.de>
17247
17248 PR tree-optimization/84775
17249 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
17250 immediate uses of predicate stmts and mark them modified.
17251
17252 Revert
17253 PR tree-optimization/84178
17254 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17255 to caller.
17256 (version_loop_for_if_conversion): Delay update_ssa call.
17257 (tree_if_conversion): Delay update_ssa until after predicate
17258 insertion.
17259
17260 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
17261
17262 PR target/84763
17263 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
17264 when the function accesses prior frames.
17265
17266 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17267
17268 PR debug/84456
17269 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
17270 gen_llsym, otherwise call maybe_gen_llsym.
17271
17272 PR inline-asm/84742
17273 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
17274 has ',' character inside of it.
17275
17276 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
17277
17278 PR target/84748
17279 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
17280 as clobbering CC_REGNUM.
17281
17282 2018-03-08 Richard Biener <rguenther@suse.de>
17283
17284 PR middle-end/84552
17285 * tree-scalar-evolution.c: Include tree-into-ssa.h.
17286 (follow_copies_to_constant): Do not follow SSA names registered
17287 for update.
17288
17289 2018-03-08 Richard Biener <rguenther@suse.de>
17290
17291 PR tree-optimization/84178
17292 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
17293 to caller.
17294 (version_loop_for_if_conversion): Delay update_ssa call.
17295 (tree_if_conversion): Delay update_ssa until after predicate
17296 insertion.
17297
17298 2018-03-08 David Malcolm <dmalcolm@redhat.com>
17299
17300 PR tree-optimization/84178
17301 * tree-if-conv.c (release_bb_predicate): Remove the
17302 the assertion that the stmts have NULL use_ops.
17303 Discard the statements, asserting that they haven't
17304 yet been added to a BB.
17305
17306 2018-03-08 Richard Biener <rguenther@suse.de>
17307
17308 PR tree-optimization/84746
17309 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
17310 (phi_translate): Pass in destination ANTIC_OUT set.
17311 (phi_translate_1): Likewise. For a simplified result lookup
17312 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
17313 (phi_translate_set): Adjust.
17314 (do_pre_regular_insertion): Likewise.
17315 (do_pre_partial_partial_insertion): Likewise.
17316
17317 2018-03-08 Martin Liska <mliska@suse.cz>
17318
17319 PR gcov-profile/84735
17320 * doc/gcov.texi: Document usage of profile files.
17321 * gcov-io.h: Document changes in the format.
17322
17323 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
17324
17325 PR debug/84404
17326 PR debug/84408
17327 * dwarf2out.c (struct dw_line_info_table): Update comments for
17328 view == -1.
17329 (FORCE_RESET_NEXT_VIEW): New.
17330 (FORCE_RESETTING_VIEW_P): New.
17331 (RESETTING_VIEW_P): Check for -1 too.
17332 (ZERO_VIEW_P): Likewise.
17333 (new_line_info_table): Force-reset next view.
17334 (dwarf2out_begin_function): Likewise.
17335 (dwarf2out_source_line): Simplify zero_view_p initialization.
17336 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
17337 view directly. Omit view when omitting .loc at line 0.
17338
17339 2018-03-08 Jakub Jelinek <jakub@redhat.com>
17340
17341 PR tree-optimization/84740
17342 * tree-switch-conversion.c (process_switch): Call build_constructors
17343 only if info.phi_count is non-zero.
17344
17345 PR tree-optimization/84739
17346 * tree-tailcall.c (find_tail_calls): Check call arguments against
17347 DECL_ARGUMENTS (current_function_decl) rather than
17348 DECL_ARGUMENTS (func) when checking for tail recursion.
17349
17350 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17351
17352 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
17353 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
17354 Volker Reichelt's entry and add entries for people that perform
17355 GCC fuzzy testing and report numerous bugs.
17356
17357 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
17358
17359 PR target/82411
17360 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
17361 readonly data in sdata, if that is disabled.
17362 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
17363 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
17364 -mreadonly-in-sdata option.
17365
17366 2018-03-07 Martin Sebor <msebor@redhat.com>
17367
17368 PR tree-optimization/84468
17369 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
17370 basic block when looking for nul assignment.
17371
17372 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
17373
17374 PR target/84277
17375 * except.h (output_function_exception_table): Adjust prototype.
17376 * except.c (output_function_exception_table): Remove FNNAME parameter
17377 and add SECTION parameter. Ouput one part of the table at a time.
17378 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
17379 the first part of the exception table and emit unwind directives.
17380 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
17381 (i386_pe_seh_cold_init): Likewise.
17382 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
17383 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
17384 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
17385 (ix86_output_call_insn): Emit a nop in one more case for SEH.
17386 * config/i386/winnt.c: Include except.h.
17387 (struct seh_frame_state): Add reg_offset, after_prologue and
17388 in_cold_section fields.
17389 (i386_pe_seh_end_prologue): Set seh->after_prologue.
17390 (i386_pe_seh_cold_init): New function.
17391 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
17392 to seh->in_cold_section.
17393 (seh_emit_push): Record the offset of the push.
17394 (seh_emit_save): Record the offet of the save.
17395 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
17396 Test seh->after_prologue to disregard the epilogue.
17397 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
17398 (i386_pe_end_cold_function): New function.
17399
17400 2018-03-07 Jakub Jelinek <jakub@redhat.com>
17401
17402 PR fortran/84565
17403 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
17404 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
17405
17406 PR c++/84704
17407 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
17408 on tmp_var.
17409 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
17410 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
17411
17412 PR middle-end/84723
17413 * multiple_target.c: Include tree-inline.h and intl.h.
17414 (expand_target_clones): Diagnose and fail if node->definition and
17415 !tree_versionable_function_p (node->decl).
17416
17417 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
17418
17419 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
17420 sprint_ul.
17421 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
17422 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
17423 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
17424
17425 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17426
17427 PR target/84710
17428 * combine.c (try_combine): Use reg_or_subregno instead of handling
17429 just paradoxical SUBREGs and REGs.
17430
17431 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
17432
17433 * config/arc/arc.c (arc_finalize_pic): Remove function.
17434 (arc_must_save_register): We use single base PIC register, remove
17435 checks to save/restore the PIC register.
17436 (arc_expand_prologue): Likewise.
17437 * config/arc/arc-protos.h (arc_set_default_type_attributes):
17438 Remove.
17439 (arc_verify_short): Likewise.
17440 (arc_attr_type): Likewise.
17441 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
17442 (walk_stores): Likewise.
17443 (arc_address_cost): Make it static.
17444 (arc_verify_short): Likewise.
17445 (branch_dest): Likewise.
17446 (arc_attr_type): Likewise.
17447 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
17448 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
17449 (arc_final_prescan_insn): Remove inserting the nops due to
17450 hardware hazards. It is done in reorg step.
17451 (insn_length_variant_t): Remove.
17452 (insn_length_parameters_t): Likewise.
17453 (arc_insn_length_parameters): Likewise.
17454 (arc_get_insn_variants): Likewise.
17455 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
17456
17457 2018-03-06 Jakub Jelinek <jakub@redhat.com>
17458
17459 PR inline-asm/84683
17460 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
17461 assertion failure.
17462
17463 PR tree-optimization/84687
17464 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
17465 on new_node->decl.
17466 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
17467
17468 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
17469
17470 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
17471 Rename to ppc_speculation_barrier.
17472 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
17473 __builtin_ppc_speculation_barrier.
17474
17475 2018-03-05 Jakub Jelinek <jakub@redhat.com>
17476
17477 PR target/84700
17478 * combine.c (combine_simplify_rtx): Don't try to simplify if
17479 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
17480 are equal to x.
17481
17482 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
17483
17484 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
17485 to 32 bytes when compiling for POWER9.
17486
17487 2018-03-05 Jakub Jelinek <jakub@redhat.com>
17488
17489 PR target/84564
17490 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
17491 regparm >= 3 with no arg reg available also for calls with
17492 flag_force_indirect_call. Pass decl to ix86_function_regparm.
17493
17494 PR target/84524
17495 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
17496 orig,vex.
17497 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
17498
17499 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
17500
17501 PR target/84264
17502 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
17503
17504 2018-03-05 Richard Biener <rguenther@suse.de>
17505
17506 PR tree-optimization/84486
17507 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
17508 When inserting a __builtin_assume_aligned call set the LHS
17509 SSA name alignment info accordingly.
17510
17511 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
17512
17513 PR tree-optimization/84114
17514 * config/aarch64/aarch64.c (aarch64_reassociation_width)
17515 Avoid reassociation of FLOAT_MODE addition.
17516
17517 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
17518
17519 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
17520 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
17521 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
17522 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
17523 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
17524 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
17525 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
17526 and -mwbnoinvd.
17527 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
17528 __builtin_ia32_wbinvd): New builtins.
17529 (SPECIAL_ARGS2): New.
17530 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
17531 (SPECIAL_ARGS2): New.
17532 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
17533 (ix86_valid_target_attribute_inner_p): Ditto.
17534 (ix86_init_mmx_sse_builtins): Add special_args2.
17535 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
17536 TARGET_WBNOINVD_P): New.
17537 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
17538 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
17539 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
17540 * config/i386/immintrin.h (_wbinvd): New intrinsic.
17541 * config/i386/pconfigintrin.h: New file.
17542 * config/i386/wbnoinvdintrin.h: Ditto.
17543 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
17544 wbnoinvdintrin.h.
17545 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
17546
17547 2018-03-05 Richard Biener <rguenther@suse.de>
17548
17549 PR tree-optimization/84670
17550 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
17551 member.
17552 (BB_VISITED_WITH_VISITED_SUCCS): New define.
17553 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
17554 (compute_antic_aux): Only assert the number of values in ANTIC_IN
17555 doesn't grow if all successors (recursively) were visited at least
17556 once.
17557
17558 2018-03-05 Richard Biener <rguenther@suse.de>
17559
17560 PR tree-optimization/84650
17561 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
17562 if executed in the loop pipeline.
17563
17564 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
17565
17566 * doc/configfiles.texi (Configuration Files): Move info about
17567 conditionalizing $target-protos.h to...
17568 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
17569 differs from $target-protos.h.
17570
17571 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
17572 Chung-Ju Wu <jasonwucj@gmail.com>
17573
17574 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
17575 * config/nds32/nds32-multiple.md (setmemsi): Define.
17576 * config/nds32/nds32-memory-manipulation.c
17577 (nds32_gen_dup_4_byte_to_word_value): New.
17578 (emit_setmem_word_loop): New.
17579 (emit_setmem_byte_loop): New.
17580 (nds32_expand_setmem_loop): New.
17581 (nds32_expand_setmem_loop_v3m): New.
17582 (nds32_expand_setmem_unroll): New.
17583 (nds32_expand_setmem): New.
17584
17585 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17586 Chung-Ju Wu <jasonwucj@gmail.com>
17587
17588 * config/nds32/nds32-memory-manipulation.c
17589 (nds32_emit_load_store): New.
17590 (nds32_emit_post_inc_load_store): New.
17591 (nds32_emit_mem_move): New.
17592 (nds32_emit_mem_move_block): New.
17593 (nds32_expand_movmemsi_loop_unknown_size): New.
17594 (nds32_expand_movmemsi_loop_known_size): New.
17595 (nds32_expand_movmemsi_loop): New.
17596 (nds32_expand_movmemsi_unroll): New.
17597 (nds32_expand_movmemqi): Rename ...
17598 (nds32_expand_movmemsi): ... to this.
17599 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
17600 (movmemsi): ... to this.
17601 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
17602 (nds32_expand_movmemsi): ... to this.
17603
17604 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17605 Monk Chiang <sh.chiang04@gmail.com>
17606 Chung-Ju Wu <jasonwucj@gmail.com>
17607
17608 * config/nds32/nds32-protos.h
17609 (nds32_expand_load_multiple): New arguments.
17610 (nds32_expand_store_multiple): Ditto.
17611 (nds32_valid_multiple_load_store): Rename ...
17612 (nds32_valid_multiple_load_store_p): ... to this.
17613 * config/nds32/nds32-memory-manipulation.c
17614 (nds32_expand_load_multiple): Refine implementation.
17615 (nds32_expand_store_multiple): Ditto.
17616 * config/nds32/nds32-multiple.md
17617 (load_multiple): Update nds32_expand_load_multiple interface.
17618 (store_multiple): Update nds32_expand_store_multiple interface.
17619 * config/nds32/nds32-predicates.c
17620 (nds32_valid_multiple_load_store): Rename ...
17621 (nds32_valid_multiple_load_store_p): ... to this and refine
17622 implementation.
17623 * config/nds32/predicates.md
17624 (nds32_load_multiple_and_update_address_operation): New predicate.
17625 (nds32_store_multiple_and_update_address_operation): New predicate.
17626
17627 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
17628 Chung-Ju Wu <jasonwucj@gmail.com>
17629
17630 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
17631 (combo): New attribute.
17632 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
17633
17634 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
17635
17636 * config/nds32/nds32.opt: Change -mcmodel= default value.
17637
17638 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
17639 Monk Chiang <sh.chiang04@gmail.com>
17640 Chung-Ju Wu <jasonwucj@gmail.com>
17641
17642 * config/nds32/constants.md (unspec_element): New enum.
17643 * config/nds32/constraints.md (Umw): New constraint.
17644 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
17645 * config/nds32/nds32-intrinsic.md: Likewise.
17646 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
17647 (nds32_valid_smw_lwm_base_p): New.
17648 (nds32_output_smw_single_word): New.
17649 (nds32_output_lmw_single_word): New.
17650 (nds32_expand_unaligned_load): New.
17651 (nds32_expand_unaligned_store): New.
17652 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
17653 (nds32_output_smw_single_word): Declare.
17654 (nds32_output_lmw_single_word): Declare.
17655 (nds32_expand_unaligned_load): Declare.
17656 (nds32_expand_unaligned_store): Declare.
17657 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
17658 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
17659 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
17660 NDS32_BUILTIN_UASTORE_DW.
17661 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
17662 predicate.
17663
17664 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
17665 Kito Cheng <kito.cheng@gmail.com>
17666 Chung-Ju Wu <jasonwucj@gmail.com>
17667
17668 * config/nds32/nds32-intrinsic.c
17669 (nds32_expand_builtin_null_ftype_reg): Delete.
17670 (nds32_expand_builtin_reg_ftype_imm): Ditto.
17671 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
17672 (nds32_read_argument): New.
17673 (nds32_legitimize_target): Ditto.
17674 (nds32_legitimize_argument): Ditto.
17675 (nds32_check_constant_argument): Ditto.
17676 (nds32_expand_unop_builtin): Ditto.
17677 (nds32_expand_unopimm_builtin): Ditto.
17678 (nds32_expand_binop_builtin): Ditto.
17679 (nds32_builtin_decl_impl): Ditto.
17680 (builtin_description): Ditto.
17681 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
17682 (nds32_init_builtins_impl): Ditto.
17683 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
17684 (nds32_builtin_decl): New.
17685 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
17686 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
17687
17688 2018-03-02 Jeff Law <law@redhat.com>
17689
17690 * reorg.c (stop_search_p): Handle DEBUG_INSN.
17691 (redundant_insn, fill_simple_delay_slots): Likewise.
17692 (fill_slots_from_thread): Likewise.
17693 * resource.c (mark_referenced_resources): Likewise.
17694 (mark_set_resources, find_dead_or_set_registers): Likewise.
17695
17696 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17697
17698 * substring-locations.h (format_warning_va): Formatting fix for
17699 ATTRIBUTE_GCC_DIAG.
17700 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
17701 argument.
17702 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
17703 * substring-locations.c: Include intl.h.
17704 (format_warning_va): Turned into small wrapper around
17705 format_warning_n_va, renamed to ...
17706 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
17707 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
17708 use ngettext.
17709 (format_warning_at_substring_n): New function.
17710 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
17711 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
17712 format_warning_at_substring with just a shorter name instead of
17713 const function pointer.
17714 (fmtwarn_n): New function.
17715 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
17716 appropriate, get rid of all the fmtstr temporaries, move conditionals
17717 with G_() wrapped string literals directly into fmtwarn arguments,
17718 cast dir.len to (int), formatting fixes.
17719
17720 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
17721
17722 * doc/invoke.texi: Remove "Cilk Plus" references.
17723
17724 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17725 Richard Biener <rguenther@suse.de>
17726
17727 PR ipa/84628
17728 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
17729 for error or warning attributes if CALL_FROM_THUNK_P is set.
17730 Formatting fixes.
17731
17732 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17733
17734 PR target/56540
17735 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
17736 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
17737
17738 PR target/56540
17739 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
17740 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
17741
17742 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
17743 instead of -1U in last predictors element's probability member.
17744
17745 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
17746
17747 PR ipa/83983
17748 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
17749 arguments if they are comparable.
17750
17751 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
17752
17753 PR tree-optimization/84634
17754 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
17755 masks and masked_loop_p with a single loop_masks, making sure it's
17756 null for bb vectorization.
17757
17758 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
17759
17760 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
17761 (vect_analyze_data_ref_access): Use loop->safe_len rather than
17762 loop->force_vectorize to check whether there is no alias.
17763
17764 2018-03-02 Jakub Jelinek <jakub@redhat.com>
17765
17766 PR target/84614
17767 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
17768 prototypes.
17769 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
17770 comments.
17771 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
17772 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
17773 instead of a loop around prev_real_insn.
17774 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
17775 prev_real_insn.
17776
17777 PR inline-asm/84625
17778 * config/i386/i386.c (ix86_print_operand): Use conditional
17779 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
17780 zero vector.
17781
17782 2018-03-02 Richard Biener <rguenther@suse.de>
17783
17784 PR tree-optimization/84427
17785 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
17786 (bitmap_set_subtract_values): Rewrite to handle multiple
17787 exprs per value.
17788 (clean): Likewise.
17789 (prune_clobbered_mems): Likewise.
17790 (phi_translate): Take edge instead of pred/phiblock.
17791 (phi_translate_1): Likewise.
17792 (phi_translate_set): Likewise. Insert all translated
17793 exprs for a value into the set, keeping possibly multiple
17794 expressions per value.
17795 (compute_antic_aux): Adjust for phi_translate changes.
17796 When intersecting union the expressions and prune those
17797 not in the final value set, keeping possibly multiple
17798 expressions per value. Do not use value-insertion
17799 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
17800 all expressions. Add verification that the value-sets
17801 only shrink during iteration.
17802 (compute_partial_antic_aux): Adjust for the phi_translate changes.
17803 (do_pre_regular_insertion): Likewise.
17804 (do_pre_partial_partial_insertion): Likewise.
17805
17806 2018-03-02 Richard Biener <rguenther@suse.de>
17807
17808 PR target/82005
17809 * config/darwin.c (saved_debug_info_level): New static global.
17810 (darwin_asm_lto_start): Disable debug info generation for LTO out.
17811 (darwin_asm_lto_end): Restore debug info generation settings.
17812
17813 2018-03-01 Martin Liska <mliska@suse.cz>
17814
17815 PR sanitizer/82484
17816 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
17817 volatile arguments.
17818
17819 2018-03-01 Richard Biener <rguenther@suse.de>
17820
17821 PR debug/84645
17822 * dwarf2out.c (gen_variable_die): Properly handle late VLA
17823 type annotation with LTO when debug was disabled at compile-time.
17824
17825 2018-03-01 Matthew Fortune <mfortune@gmail.com>
17826
17827 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
17828 XINT with INTVAL.
17829 (mips_final_postscan_insn): Likewise.
17830
17831 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
17832
17833 PR rtl-optimization/84528
17834 * alias.c (init_alias_target): Add commentary.
17835 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
17836 a unique base value if the frame pointer is not eliminated
17837 to the stack pointer.
17838
17839 2018-03-01 Tom de Vries <tom@codesourcery.com>
17840
17841 PR rtl-optimization/83327
17842 * lra-int.h (hard_regs_spilled_into): Declare.
17843 * lra.c (hard_regs_spilled_into): Define.
17844 (init_reg_info): Init hard_regs_spilled_into.
17845 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
17846 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
17847 (process_bb_lives): Handle hard_regs_spilled_into.
17848 (lra_create_live_ranges_1): Before doing liveness propagation, clear
17849 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
17850
17851 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
17852
17853 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
17854 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
17855 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
17856 * config/rs6000/aix72.h: New file.
17857
17858 2018-02-28 Jakub Jelinek <jakub@redhat.com>
17859
17860 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
17861 instead of warning_at with conditional singular and plural messages
17862 where possible.
17863
17864 PR target/52991
17865 * stor-layout.c (update_alignment_for_field): For
17866 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
17867 && !DECL_PACKED (field), do the alignment update, just use
17868 only desired_align instead of MAX (type_align, desired_align)
17869 as the alignment.
17870 (place_field): Don't do known_align < desired_align handling
17871 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
17872 is non-NULL, instead do it after rli->prev_field handling and
17873 only if not within a bitfield word. For DECL_PACKED (field)
17874 use type_align of BITS_PER_UNIT.
17875
17876 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
17877
17878 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
17879 superfluous parentheses and trailing spaces.
17880
17881 2018-02-28 Richard Biener <rguenther@suse.de>
17882
17883 PR tree-optimization/84584
17884 * graphite-scop-detection.c (scop_detection::add_scop): Discard
17885 SCoPs with fake exit edge.
17886
17887 2018-02-28 Martin Liska <mliska@suse.cz>
17888
17889 PR testsuite/84597
17890 * timevar.c (timer::print): Fix format to properly print 100%
17891 values.
17892
17893 2018-02-28 Richard Biener <rguenther@suse.de>
17894
17895 PR middle-end/84607
17896 * genmatch.c (capture_info::walk_match): Do not mark
17897 captured expressions without operands as expr_p given
17898 they act more like predicates and should be subject to
17899 "lost tail" side-effect preserving.
17900
17901 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
17902
17903 PR rtl-optimization/81611
17904 * auto-inc-dec.c (attempt_change): Move dead note from
17905 mem_insn if it's the next use of regno
17906 (find_address): Take address use of reg holding
17907 non-incremented value. Add parm to limit search to the named
17908 reg only.
17909 (merge_in_block): Attempt to use a mem insn that is the next
17910 use of the original regno.
17911
17912 2018-02-27 Martin Sebor <msebor@redhat.com>
17913
17914 PR c++/83871
17915 * doc/invoke.texi (-Wmissing-attributes): New option.
17916 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
17917
17918 2018-02-27 Martin Sebor <msebor@redhat.com>
17919
17920 PR translation/84207
17921 * diagnostic-core.h (warning_n, error_n, inform_n): Change
17922 n argument to unsigned HOST_WIDE_INT.
17923 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
17924 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
17925 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
17926 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
17927
17928 2018-02-27 Richard Biener <rguenther@suse.de>
17929
17930 PR tree-optimization/84512
17931 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
17932 Do not use the estimate returned from record_stmt_cost for
17933 the scalar iteration cost but sum properly using add_stmt_cost.
17934
17935 2018-02-27 Richard Biener <rguenther@suse.de>
17936
17937 PR tree-optimization/84466
17938 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
17939 Adjust last change to less strictly validate use operands.
17940
17941 2018-02-27 Martin Liska <mliska@suse.cz>
17942
17943 PR gcov-profile/84548
17944 * gcov.c (process_file): Allow partial overlap and consider it
17945 also as group functions.
17946 (output_lines): Properly calculate range of lines for a group.
17947
17948 2018-02-27 Martin Liska <mliska@suse.cz>
17949
17950 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
17951 'ggc' suffixes. Change first column width.
17952 (timer::print): Fix formatting of the column.
17953
17954 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
17955
17956 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
17957 preserve inline entry blocks for the sake of debug inline
17958 entry point markers alone.
17959 (remove_unused_locals): Suggest in comments a better place to
17960 force the preservation of inline entry blocks that are
17961 otherwise unused, but do not preserve them.
17962
17963 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17964
17965 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
17966
17967 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
17968
17969 PR target/84039
17970 * config/i386/constraints.md (Bs): Replace
17971 ix86_indirect_branch_register with
17972 TARGET_INDIRECT_BRANCH_REGISTER.
17973 (Bw): Likewise.
17974 * config/i386/i386.md (indirect_jump): Likewise.
17975 (tablejump): Likewise.
17976 (*sibcall_memory): Likewise.
17977 (*sibcall_value_memory): Likewise.
17978 Peepholes of indirect call and jump via memory: Likewise.
17979 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
17980 (*sibcall_value_GOT_32): Likewise.
17981 * config/i386/predicates.md (indirect_branch_operand): Likewise.
17982 (GOT_memory_operand): Likewise.
17983 (call_insn_operand): Likewise.
17984 (sibcall_insn_operand): Likewise.
17985 (GOT32_symbol_operand): Likewise.
17986 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
17987
17988 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
17989
17990 PR rtl-optimization/83496
17991 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
17992 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
17993 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
17994 redundant insn, if any.
17995 (relax_delay_slots): Likewise.
17996 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
17997
17998 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
17999
18000 PR tree-optimization/83965
18001 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
18002 that grouped statements are part of a reduction chain. Return
18003 true if the statement is not marked as a reduction itself but
18004 is part of a group.
18005 (vect_recog_dot_prod_pattern): Don't check whether the statement
18006 is part of a group here.
18007 (vect_recog_sad_pattern): Likewise.
18008 (vect_recog_widen_sum_pattern): Likewise.
18009
18010 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
18011
18012 PR debug/84545
18013 * final.c (rest_of_clean_state): Also look for calls inside sequences.
18014
18015 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
18016
18017 PR target/84530
18018 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
18019 the bool argument.
18020 (ix86_output_indirect_function_return): New prototype.
18021 (ix86_split_simple_return_pop_internal): Likewise.
18022 * config/i386/i386.c (indirect_return_via_cx): New.
18023 (indirect_return_via_cx_bnd): Likewise.
18024 (indirect_thunk_name): Handle return va CX_REG.
18025 (output_indirect_thunk_function): Create alias for
18026 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
18027 (ix86_output_indirect_jmp): Remove the bool argument.
18028 (ix86_output_indirect_function_return): New function.
18029 (ix86_split_simple_return_pop_internal): Likewise.
18030 * config/i386/i386.md (*indirect_jump): Don't pass false
18031 to ix86_output_indirect_jmp.
18032 (*tablejump_1): Likewise.
18033 (simple_return_pop_internal): Change it to define_insn_and_split.
18034 Call ix86_split_simple_return_pop_internal to split it for
18035 -mfunction-return=.
18036 (simple_return_indirect_internal): Call
18037 ix86_output_indirect_function_return instead of
18038 ix86_output_indirect_jmp.
18039
18040 2018-02-26 Jakub Jelinek <jakub@redhat.com>
18041
18042 PR bootstrap/84405
18043 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
18044 memset and value initialization afterwards.
18045
18046 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
18047
18048 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
18049
18050 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
18051
18052 PR target/84521
18053 * common/config/aarch64/aarch64-common.c
18054 (aarch_option_optimization_table[]): Switch
18055 off fomit-frame-pointer
18056
18057 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
18058 Chung-Ju Wu <jasonwucj@gmail.com>
18059
18060 * config/nds32/nds32-multiple.md (load_multiple): Disallow
18061 volatile memory.
18062 (store_multiple): Ditto.
18063
18064 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
18065
18066 * config.gcc: Add --with-cpu support for nds32 target.
18067 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
18068 * config/nds32/nds32.opt: Add -mcpu= option.
18069
18070 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
18071
18072 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
18073 isel=yes): Warn for these deprecated options.
18074
18075 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
18076
18077 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
18078 ISA_2_5_MASKS_EMBEDDED.
18079
18080 2018-02-23 Jakub Jelinek <jakub@redhat.com>
18081
18082 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
18083 p->max as pointers rather than using iterative_hash_expr.
18084
18085 2018-02-23 Carl Love <cel@us.ibm.com>
18086
18087 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
18088 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
18089 BU_P8V_OVERLOAD_2.
18090 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
18091 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
18092 P8V_BUILTIN_VEC_VUNSIGNED2.
18093
18094 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
18095
18096 PR target/81572
18097 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
18098 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
18099 LRA_UNKNOWN_ALT.
18100 * lra-constraints.c (curr_insn_transform): Set up
18101 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
18102 LRA_UNKNOWN_ALT.
18103 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
18104 * lra-eliminations.c (spill_pseudos): Ditto.
18105 (process_insn_for_elimination): Ditto.
18106 * lra-lives.c (reg_early_clobber_p): Use the new macros.
18107 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
18108 LRA_NON_CLOBBERED_ALT.
18109
18110 2018-02-22 Martin Sebor <msebor@redhat.com>
18111
18112 PR tree-optimization/84480
18113 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
18114 to maybe_diag_stxncpy_trunc. Call it.
18115 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
18116 from gimple_fold_builtin_strcpy. Print inlining stack.
18117 (handle_builtin_stxncpy): Print inlining stack.
18118 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
18119
18120 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
18121
18122 PR target/84176
18123 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
18124 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
18125 and -fcheck-pointer-bounds are used together.
18126 (indirect_thunk_prefix): New enum.
18127 (indirect_thunk_need_prefix): New function.
18128 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
18129 "_nt" instead of "_bnd" for NOTRACK prefix.
18130 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
18131 (output_indirect_thunk_function): Likewise.
18132 (): Likewise.
18133 (ix86_code_end): Update output_indirect_thunk_function calls.
18134 (ix86_output_indirect_branch_via_reg): Replace
18135 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
18136 (ix86_output_indirect_branch_via_push): Likewise.
18137 (ix86_output_function_return): Likewise.
18138 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
18139 incompatible with -fcf-protection=branch and
18140 -fcheck-pointer-bounds.
18141
18142 2018-02-22 Steve Ellcey <sellcey@cavium.com>
18143
18144 PR target/83335
18145 * config/aarch64/aarch64.c (aarch64_print_address_internal):
18146 Change gcc_assert call to output_operand_lossage.
18147
18148 2018-02-22 Steve Ellcey <sellcey@cavium.com>
18149
18150 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
18151
18152 2018-02-22 DJ Delorie <dj@redhat.com>
18153 Sebastian Perta <sebastian.perta@renesas.com>
18154 Oleg Endo <olegendo@gcc.gnu.org>
18155
18156 * config/rx/rx.c (rx_rtx_costs): New function.
18157 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
18158
18159 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
18160
18161 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
18162
18163 2018-02-22 Martin Liska <mliska@suse.cz>
18164
18165 PR driver/83193
18166 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
18167 Add "native" as a possible value.
18168
18169 2018-02-22 Martin Liska <mliska@suse.cz>
18170
18171 PR driver/83193
18172 * config/i386/i386.c (ix86_option_override_internal):
18173 Add "native" as a possible value for -march and -mtune.
18174
18175 2018-02-22 Jakub Jelinek <jakub@redhat.com>
18176
18177 PR target/84502
18178 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
18179 to all type variants.
18180
18181 PR tree-optimization/84503
18182 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
18183 width as info->bitpos + info->bitsize - start.
18184 (merged_store_group::merge_overlapping): Simplify width computation.
18185 (check_no_overlap): New function.
18186 (imm_store_chain_info::try_coalesce_bswap): Compute expected
18187 start + width and last_order of the group, fail if check_no_overlap
18188 fails.
18189 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
18190 to group if check_no_overlap fails.
18191
18192 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18193
18194 * config/rs6000/altivec.md: Delete contraint arguments to
18195 define_expand, define_split, and define_peephole2, and in
18196 define_insn_and_split if always unused.
18197 * config/rs6000/darwin.md: Ditto.
18198 * config/rs6000/dfp.md: Ditto.
18199 * config/rs6000/rs6000.md: Ditto.
18200 * config/rs6000/sync.md: Ditto.
18201 * config/rs6000/vector.md: Ditto.
18202 * config/rs6000/vsx.md: Ditto.
18203
18204 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
18205
18206 * config/rs6000/altivec.md: Write output control strings as braced
18207 blocks instead of double-quoted strings.
18208 * config/rs6000/darwin.md: Ditto.
18209 * config/rs6000/rs6000.md: Ditto.
18210 * config/rs6000/vector.md: Ditto.
18211 * config/rs6000/vsx.md: Ditto.
18212
18213 2018-02-21 Jason Merrill <jason@redhat.com>
18214
18215 PR c++/84314 - ICE with templates and fastcall attribute.
18216 * attribs.c (build_type_attribute_qual_variant): Remove assert.
18217
18218 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18219
18220 * ipa-cp.c (determine_versionability): Fix comment typos.
18221
18222 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
18223
18224 PR c/84229
18225 * ipa-cp.c (determine_versionability): Do not version functions caling
18226 va_arg_pack.
18227
18228 2018-02-21 Martin Liska <mliska@suse.cz>
18229
18230 PR driver/83193
18231 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
18232 Add "native" as a possible value.
18233 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
18234 the macro when native cpu detection is available.
18235
18236 2018-02-21 Martin Liska <mliska@suse.cz>
18237
18238 PR driver/83193
18239 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
18240 Add "native" as a possible value.
18241 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
18242 when native cpu detection is available.
18243
18244 2018-02-21 Jakub Jelinek <jakub@redhat.com>
18245 Martin Sebor <msebor@redhat.com>
18246
18247 PR tree-optimization/84478
18248 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
18249 false.
18250 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
18251 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
18252 support which is conservatively correct, for 2 only stay conservative
18253 for maxlen. Formatting and comment capitalization fixes. Add STRICT
18254 argument to the 2 argument get_range_strlen, adjust 6 arg
18255 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
18256 false.
18257 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
18258 (gimple_fold_builtin_strlen): Pass true as last argument to
18259 get_range_strlen.
18260
18261 2018-02-20 Martin Sebor <msebor@redhat.com>
18262
18263 PR middle-end/84095
18264 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
18265 (builtin_memref::set_base_and_offset): Same. Handle inner references.
18266 (builtin_memref::builtin_memref): Factor out parts into
18267 set_base_and_offset and call it.
18268
18269 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18270
18271 PR middle-end/84406
18272 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
18273 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
18274 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
18275 search at the associated MODE_INT.
18276
18277 2018-02-20 Jeff Law <law@redhat.com>
18278
18279 PR middle-end/82123
18280 PR tree-optimization/81592
18281 PR middle-end/79257
18282 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
18283 for range data rather than using global data.
18284 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
18285 range data rather than using global data.
18286 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
18287 pass it to children as needed.
18288 (struct directive::fmtresult): Similarly.
18289 (struct directive::set_width): Similarly.
18290 (struct directive::set_precision): Similarly.
18291 (format_integer, format_directive, parse_directive): Similarly.
18292 (format_none): Accept unnamed vr_values parameter.
18293 (format_percent, format_floating, format_character): Similarly.
18294 (format_string, format_plain): Similarly.
18295 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
18296 the EVRP range analyzer for range data rather than using global data.
18297 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
18298 gimple-ssa-evrp-analyze.h
18299 (class sprintf_dom_walker): Add after_dom_children member function.
18300 Add evrp_range_analyzer member.
18301 (sprintf_dom_walker::before_dom_children): Call into the EVRP
18302 range analyzer as needed.
18303 (sprintf_dom_walker::after_dom_children): New member function.
18304 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
18305 if not optimizing.
18306 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
18307 (evrp_range_analyzer::pop_to_marker): Likewise.
18308
18309 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
18310
18311 PR tree-optimization/84419
18312 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
18313 with the required type if its current type is compatible but
18314 different.
18315
18316 2018-02-20 Jakub Jelinek <jakub@redhat.com>
18317
18318 PR middle-end/82004
18319 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
18320 after vectorization.
18321
18322 2018-02-20 Martin Liska <mliska@suse.cz>
18323
18324 PR driver/83193
18325 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
18326 possible values if we don't have a hint.
18327
18328 2018-02-20 Martin Liska <mliska@suse.cz>
18329
18330 PR c/84310
18331 PR target/79747
18332 * final.c (shorten_branches): Build align_tab array with one
18333 more element.
18334 * opts.c (finish_options): Add alignment option limit check.
18335 (MAX_CODE_ALIGN): Likewise.
18336 (MAX_CODE_ALIGN_VALUE): Likewise.
18337 * doc/invoke.texi: Document maximum allowed option value for
18338 all -falign-* options.
18339
18340 2018-02-19 Jakub Jelinek <jakub@redhat.com>
18341
18342 PR target/84146
18343 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
18344 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
18345 * var-tracking.c (emit_note_insn_var_location): Remove all references
18346 to NOTE_INSN_CALL_ARG_LOCATION.
18347 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
18348 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
18349 Use copy_rtx_if_shared.
18350 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
18351 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
18352 (dwarf2out_var_location): Remove handling of
18353 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
18354 on call_insn.
18355 * final.c (final_scan_insn): Remove all references to
18356 NOTE_INSN_CALL_ARG_LOCATION.
18357 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
18358 before dumping final insns.
18359 * except.c (emit_note_eh_region_end): Remove all references to
18360 NOTE_INSN_CALL_ARG_LOCATION.
18361 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
18362 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
18363 * config/arc/arc.c (hwloop_optimize): Likewise.
18364 * config/arm/arm.c (create_fix_barrier): Likewise.
18365 * config/s390/s390.c (s390_chunkify_start): Likewise.
18366 * config/sh/sh.c (find_barrier): Likewise.
18367 * config/i386/i386.c (rest_of_insert_endbranch,
18368 ix86_seh_fixup_eh_fallthru): Likewise.
18369 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
18370 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
18371 * config/frv/frv.c (frv_function_prologue): Likewise.
18372 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
18373 reg note.
18374 (note_outside_basic_block_p): Remove all references to
18375 NOTE_INSN_CALL_ARG_LOCATION.
18376 * gengtype.c (adjust_field_rtx_def): Likewise.
18377 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
18378 Likewise.
18379 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
18380 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
18381
18382 PR c++/84444
18383 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
18384 is ADDR_EXPR.
18385
18386 PR tree-optimization/84452
18387 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
18388 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
18389 is NULL.
18390
18391 2018-02-19 Martin Liska <mliska@suse.cz>
18392
18393 PR sanitizer/82183
18394 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
18395
18396 2018-02-19 Martin Liska <mliska@suse.cz>
18397 Richard Sandiford <richard.sandiford@linaro.org>
18398
18399 PR tree-optimization/82491
18400 * gimple-fold.c (get_base_constructor): Make earlier bail out
18401 to prevent ubsan.
18402
18403 2018-02-19 Carl Love <cel@us.ibm.com>
18404
18405 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
18406 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
18407 BU_P8V_OVERLOAD_1.
18408 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
18409 P8V_BUILTIN_VEC_NEG.
18410
18411 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
18412
18413 * config/rl78/rl78.md (movdf): New define expand.
18414
18415 2018-02-19 Martin Liska <mliska@suse.cz>
18416
18417 PR other/80589
18418 * doc/invoke.texi: Fix typo.
18419 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
18420
18421 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
18422
18423 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
18424 handle rs6000_single_float and rs6000_double_float specially for
18425 e500 family CPUs.
18426
18427 2018-02-16 Jeff Law <law@redhat.com>
18428
18429 * config/rx/rx.c (add_pop_cfi_notes): New function.;
18430 (pop_regs): Use it.
18431
18432 2018-02-16 Jakub Jelinek <jakub@redhat.com>
18433
18434 PR ipa/84425
18435 * ipa-inline.c (inline_small_functions): Fix a typo.
18436
18437 2018-02-16 Nathan Sidwell <nathan@acm.org>
18438
18439 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
18440
18441 2018-02-16 Carl Love <cel@us.ibm.com>
18442
18443 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
18444 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
18445 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
18446 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
18447 expansion to P8V_BUILTIN_VEC_FLOAT2.
18448
18449 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
18450
18451 PR rtl-optimization/70023
18452 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
18453 src_regno into account.
18454
18455 2018-02-16 Carl Love <cel@us.ibm.com>
18456
18457 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
18458 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
18459 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
18460 * config/rs6000/rs6000.c: Remove case statements for
18461 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
18462 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
18463 and P9V_BUILTIN_VEC_VINSERT4B.
18464 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
18465 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
18466 * config/rs6000/vsx.md:
18467 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
18468 vec_insert4b.
18469
18470 2018-02-16 Carl Love <cel@us.ibm.com>
18471
18472 * config/rs6000/altivec.h: Add builtin names vec_extract4b
18473 vec_insert4b.
18474 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
18475 definitions.
18476 * config/rs6000/rs6000-c.c: Add the definitions for
18477 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
18478 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
18479 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
18480 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
18481 definition for insert4b and define insn *insert3b_internal.
18482 * doc/extend.texi: Add documentation for vec_extract4b.
18483
18484 2018-02-16 Nathan Sidwell <nathan@acm.org>
18485
18486 * doc/extend.texi (Backwards Compatibility): Mention friend
18487 injection. Note for-scope is deprecated.
18488 * doc/invoke.texi (-ffriend-injection): Deprecate.
18489
18490 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
18491
18492 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
18493 that moved to I2, also allow destinations that are a paradoxical
18494 subreg (instead of a normal reg).
18495
18496 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
18497
18498 PR target/83831
18499 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
18500 to QImode.
18501
18502 2018-02-16 Richard Biener <rguenther@suse.de>
18503
18504 PR tree-optimization/84037
18505 PR tree-optimization/84016
18506 PR target/82862
18507 * config/i386/i386.c (ix86_builtin_vectorization_cost):
18508 Adjust vec_construct for the fact we need additional higher latency
18509 128bit inserts for AVX256 and AVX512 vector builds.
18510 (ix86_add_stmt_cost): Scale vector construction cost for
18511 elementwise loads.
18512
18513 2018-02-16 Richard Biener <rguenther@suse.de>
18514
18515 PR tree-optimization/84417
18516 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
18517 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
18518 (non_rewritable_lvalue_p): Likewise, use poly-ints.
18519
18520 2018-02-16 Martin Liska <mliska@suse.cz>
18521
18522 PR sanitizer/84307
18523 * internal-fn.def (ASAN_CHECK): Set proper flags.
18524 (ASAN_MARK): Likewise.
18525
18526 2018-02-16 Julia Koval <julia.koval@intel.com>
18527
18528 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
18529 from PTA_CANNONLAKE.
18530
18531 2018-02-16 Jakub Jelinek <jakub@redhat.com>
18532
18533 PR target/84272
18534 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
18535 Use ++iter rather than iter++ for std::list iterators.
18536 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
18537 defer deleting them until all nodes in the forest are processed. Do
18538 free even leaf nodes. Change to_process into auto_vec.
18539
18540 PR bootstrap/84405
18541 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
18542 * vec.h (vec_default_construct): Use memset instead of placement new
18543 if BROKEN_VALUE_INITIALIZATION is defined.
18544 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
18545 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
18546 is defined.
18547
18548 PR rtl-optimization/83723
18549 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
18550 * lra.c (lra_substitute_pseudo): Likewise. If true, use
18551 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
18552 recursive calls.
18553 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
18554 callers.
18555 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
18556
18557 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
18558
18559 PR rtl-optimization/81443
18560 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
18561 from inner REGs to paradoxical SUBREGs.
18562
18563 2018-02-16 Richard Biener <rguenther@suse.de>
18564
18565 PR tree-optimization/84399
18566 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
18567 For operands we can analyze at their definition make sure we can
18568 analyze them at each use as well.
18569
18570 2018-02-16 Richard Biener <rguenther@suse.de>
18571
18572 PR tree-optimization/84190
18573 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
18574 volatile accesses if the decl isn't volatile.
18575
18576 2018-02-15 Jason Merrill <jason@redhat.com>
18577
18578 PR c++/84314 - ICE with templates and fastcall attribute.
18579 * attribs.c (build_type_attribute_qual_variant): Don't clobber
18580 TYPE_CANONICAL on an existing type.
18581
18582 2018-02-15 Jakub Jelinek <jakub@redhat.com>
18583
18584 PR tree-optimization/84383
18585 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
18586 dstoff nor call operand_equal_p if dstbase is NULL.
18587
18588 PR tree-optimization/84334
18589 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
18590 also a CONSTANT_CLASS_P, punt.
18591
18592 2018-02-14 Jim Wilson <jimw@sifive.com>
18593
18594 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
18595 first SMALL_OPERAND check. New local min_second_step. Move assert
18596 to where locals are set. Add TARGET_RVC support.
18597 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
18598
18599 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
18600
18601 * doc/invoke.texi: Correct -Wformat-overflow code sample.
18602
18603 2018-02-14 Martin Sebor <msebor@redhat.com>
18604
18605 PR tree-optimization/83698
18606 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
18607 arrays constrain the offset range to their bounds.
18608 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
18609 (builtin_access::overlap): Avoid setting the size of overlap if it's
18610 already been set.
18611 (maybe_diag_overlap): Also consider arrays when deciding what values
18612 of offsets to include in diagnostics.
18613
18614 2018-02-14 Martin Sebor <msebor@redhat.com>
18615
18616 PR c/84108
18617 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
18618 that correspond to the kind of a declaration.
18619
18620 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
18621
18622 PR target/83984
18623 * config/pa/pa.md: Load address of PIC label using the linkage table
18624 if the label is nonlocal.
18625
18626 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
18627
18628 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
18629 warning message if user requests -maltivec=be.
18630 * doc/invoke.texi: Document deprecation of -maltivec=be.
18631
18632 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
18633
18634 PR target/84220
18635 * config/rs6000/rs6000-c.c: Update definitions for
18636 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
18637 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
18638
18639 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
18640
18641 PR target/84239
18642 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
18643 add _get_ssp intrinsics. Remove argument from
18644 __builtin_ia32_rdssp[d|q].
18645 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
18646 * config/i386/i386-builtin.def: Remove argument from
18647 __builtin_ia32_rdssp[d|q].
18648 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
18649 ix86_expand_special_args_builtin for _rdssp[d|q].
18650 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
18651 Clear register before usage.
18652 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
18653 Add documentation for new _get_ssp and _inc_ssp intrinsics.
18654
18655 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
18656
18657 PR tree-optimization/84357
18658 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
18659 operand 1 of an ARRAY_REF too.
18660
18661 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
18662
18663 PR target/83831
18664 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
18665 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
18666 declarations.
18667 (set_of_reg): New struct.
18668 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
18669 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
18670 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
18671 functions.
18672 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
18673 Split into bitclr, bitset, bitinvert patterns if appropriate.
18674 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
18675 use rx_fuse_in_memory_bitop.
18676 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
18677 to named insn, correct maximum insn length.
18678
18679 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
18680
18681 PR target/79242
18682 * machmode.def: Define a complex mode for PARTIAL_INT.
18683 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
18684 MODE_PARTIAL_INT.
18685 * doc/rtl.texi: Document CSPImode.
18686 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
18687 handling.
18688 (msp430_hard_regno_nregs_with_padding): Likewise.
18689
18690 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
18691
18692 PR target/84279
18693 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
18694
18695 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
18696
18697 PR rtl-optimization/84169
18698 * combine.c (try_combine): New variable split_i2i3. Set it to true if
18699 we generated a parallel as new i3 and we split that to new i2 and i3
18700 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
18701 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
18702 those to i2, not i1. Partially rewrite this scan code.
18703
18704 2018-02-13 Jakub Jelinek <jakub@redhat.com>
18705
18706 PR c/82210
18707 * stor-layout.c (place_field): For variable length fields, adjust
18708 offset_align afterwards not just based on the field's alignment,
18709 but also on the size.
18710
18711 PR middle-end/84309
18712 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
18713 of exps and logs in the use_exp2 case.
18714
18715 2018-02-13 Jeff Law <law@redhat.com>
18716
18717 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
18718 entry for "vector".
18719
18720 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
18721 ARGS as unused.
18722
18723 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
18724
18725 PR debug/84342
18726 PR debug/84319
18727 * common.opt (gas-loc-support, gas-locview-support): New.
18728 (ginline-points, ginternal-reset-location-views): New.
18729 * doc/invoke.texi: Document them. Use @itemx where intended.
18730 (gvariable-location-views): Adjust.
18731 * target.def (reset_location_view): New.
18732 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
18733 (TARGET_RESET_LOCATION_VIEW): New.
18734 * doc/tm.texi: Rebuilt.
18735 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
18736 (dwarf2out_default_as_locview_support): New.
18737 (output_asm_line_debug_info): Use option variables.
18738 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
18739 (output_loc_list): Likewise.
18740 (add_high_low_attributes): Check option variables.
18741 Don't output entry view attribute in strict mode.
18742 (gen_inlined_subroutine_die): Check option variables.
18743 (dwarf2out_inline_entry): Likewise.
18744 (init_sections_and_labels): Likewise.
18745 (dwarf2out_early_finish): Likewise.
18746 (maybe_reset_location_view): New, from...
18747 (dwarf2out_var_location): ... here. Call it.
18748 * debug.h (dwarf2out_default_as_loc_support): Declare.
18749 (dwarf2out_default_as_locview_support): Declare.
18750 * hooks.c (hook_int_rtx_insn_0): New.
18751 * hooks.h (hook_int_rtx_insn_0): Declare.
18752 * toplev.c (process_options): Take -gas-loc-support and
18753 -gas-locview-support from dwarf2out. Enable
18754 -gvariable-location-views by default only with locview
18755 assembler support. Enable -ginternal-reset-location-views by
18756 default only if the target defines the corresponding hook.
18757 Enable -ginline-points by default if location views are
18758 enabled; force it disabled if statement frontiers are
18759 disabled.
18760 * tree-inline.c (expand_call_inline): Check option variables.
18761 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
18762
18763 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
18764
18765 PR tree-optimization/84321
18766 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
18767 handling. Also check whether the anti-range contains any values
18768 that satisfy the mask; switch to a VR_RANGE if not.
18769
18770 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
18771
18772 PR sanitizer/84340
18773 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
18774
18775 2018-02-13 Martin Jambor <mjambor@suse.cz>
18776
18777 PR c++/83990
18778 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
18779 of call statements, also set location of a load to a temporary.
18780
18781 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
18782
18783 * config/rl78/rl78.c (add_vector_labels): New function.
18784 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
18785 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
18786 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
18787 which checks that no arguments are passed.
18788 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
18789 * doc/extend.texi: Documentation for the new attribute.
18790
18791 2018-02-13 Andreas Schwab <schwab@suse.de>
18792
18793 * config/riscv/linux.h (CPP_SPEC): Define.
18794
18795 2018-02-13 Jakub Jelinek <jakub@redhat.com>
18796
18797 PR target/84335
18798 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
18799 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
18800 OPTION_MASK_ISA_AES as first argument to def_builtin_const
18801 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
18802 instead of OPTION_MASK_ISA_PCLMUL as first argument to
18803 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
18804 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
18805 temporarily for AES and PCLMUL builtins.
18806
18807 PR tree-optimization/84339
18808 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
18809 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
18810 Formatting fixes.
18811
18812 PR middle-end/84309
18813 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
18814 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
18815 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
18816 inline function.
18817 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
18818 inline function.
18819 * omp-simd-clone.h: New file.
18820 * omp-simd-clone.c: Include omp-simd-clone.h.
18821 (expand_simd_clones): No longer static.
18822 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
18823 cgraph.h and omp-simd-clone.h.
18824 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
18825 (vect_recog_widen_shift_pattern): Formatting fix.
18826 (vect_pattern_recog_1): Don't check optab for calls.
18827
18828 PR target/84336
18829 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
18830 operands[2] into a REG before using gen_lowpart on it.
18831
18832 2018-02-12 Jeff Law <law@redhat.com>
18833
18834 PR target/83760
18835 * config/sh/sh.c (find_barrier): Consider a sibling call
18836 a barrier as well.
18837
18838 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
18839 successfully back substituting a reg.
18840
18841 2018-02-12 Richard Biener <rguenther@suse.de>
18842
18843 PR tree-optimization/84037
18844 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
18845 parameter, move visited init to caller.
18846 (vect_slp_analyze_operations): Separate cost from validity
18847 check, initialize visited once for all instances.
18848 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
18849 for all instances.
18850 * tree-vect-stmts.c (vect_model_simple_cost): Make early
18851 out an assert.
18852 (vect_model_promotion_demotion_cost): Likewise.
18853 (vectorizable_bswap): Guard cost modeling with !slp_node
18854 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
18855 SLP stmts.
18856 (vectorizable_call): Likewise.
18857 (vectorizable_conversion): Likewise.
18858 (vectorizable_assignment): Likewise.
18859 (vectorizable_shift): Likewise.
18860 (vectorizable_operation): Likewise.
18861 (vectorizable_store): Likewise.
18862 (vectorizable_load): Likewise.
18863 (vectorizable_condition): Likewise.
18864 (vectorizable_comparison): Likewise.
18865
18866 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
18867
18868 PR sanitizer/84307
18869 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
18870 (ASAN_MARK): Fix fnspec to account for return value, change pointer
18871 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
18872
18873 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
18874
18875 PR middle-end/83665
18876 * params.def (inline-min-speedup): Increase from 8 to 15.
18877 (max-inline-insns-auto): Decrease from 40 to 30.
18878 * ipa-split.c (consider_split): Add some buffer for function to
18879 be considered inlining candidate.
18880 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
18881 default values.
18882
18883 2018-02-12 Richard Biener <rguenther@suse.de>
18884
18885 PR tree-optimization/84037
18886 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
18887 matched stmts if we cannot swap the non-matched ones.
18888
18889 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
18890
18891 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
18892 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
18893 _mm_maskz_scalef_round_ss): New intrinsics.
18894 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
18895 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
18896 __builtin_ia32_scalefss_round): Remove.
18897 (__builtin_ia32_scalefsd_mask_round,
18898 __builtin_ia32_scalefss_mask_round): New intrinsics.
18899 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
18900 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
18901 ((match_operand:VF_128 2 "<round_nimm_predicate>"
18902 "<round_constraint>")): Changed to ...
18903 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
18904 "<round_scalar_constraint>")): ... this.
18905 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
18906 %0, %1, %2<round_op3>}"): Changed to ...
18907 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
18908 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
18909 %2<round_scalar_mask_op3>}"): ... this.
18910 * config/i386/subst.md (round_scalar_nimm_predicate): New.
18911
18912 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
18913
18914 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
18915 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
18916 (_mm_maskz_sqrt_round_ss): New intrinsics.
18917 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
18918 (__builtin_ia32_sqrtsd_mask_round)
18919 (__builtin_ia32_sqrtss_mask_round): New builtins.
18920 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
18921 (__builtin_ia32_sqrtss_round): Remove.
18922 (__builtin_ia32_sqrtsd_mask_round)
18923 (__builtin_ia32_sqrtss_mask_round): New builtins.
18924 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
18925 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
18926 ((match_operand:VF_128 1 "vector_operand"
18927 "xBm,<round_constraint>")): Changed to ...
18928 ((match_operand:VF_128 1 "vector_operand"
18929 "xBm,<round_scalar_constraint>")): ... this.
18930 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
18931 %0, %2, %<iptr>1<round_op3>}): Changed to ...
18932 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
18933 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
18934 %<iptr>1<round_scalar_mask_op3>}): ... this.
18935 ((set_attr "prefix" "<round_prefix>")): Changed to ...
18936 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
18937
18938 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
18939
18940 PR target/84266
18941 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
18942 Cast vec_cmpeq result to correct type.
18943 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
18944 Cast vec_cmpgt result to correct type.
18945
18946 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
18947
18948 * final.c (final_scan_insn_1): Renamed from...
18949 (final_scan_insn): ... this. New wrapper, to recover
18950 seen from the outermost call in recursive ones.
18951 * config/sparc/sparc.c (output_return): Drop seen from call.
18952 (output_sibcall): Likewise.
18953 * config/visium/visium.c (output_branch): Likewise.
18954
18955 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
18956
18957 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
18958 function label.
18959
18960 2018-02-10 Alan Modra <amodra@gmail.com>
18961
18962 PR target/84300
18963 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
18964 Specify LR as an input.
18965
18966 2018-02-10 Jakub Jelinek <jakub@redhat.com>
18967
18968 PR sanitizer/83987
18969 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
18970 remove_member_access_dummy_vars): New functions.
18971 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
18972 lower_omp_1, execute_lower_omp): Use them.
18973
18974 PR rtl-optimization/84308
18975 * shrink-wrap.c (spread_components): Release todo vector.
18976
18977 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
18978
18979 PR rtl-optimization/57193
18980 * ira-color.c (struct allocno_color_data): Add member
18981 conflict_allocno_hard_prefs.
18982 (update_conflict_allocno_hard_prefs): New.
18983 (bucket_allocno_compare_func): Add a preference based on
18984 conflict_allocno_hard_prefs.
18985 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
18986 (color_allocnos): Remove a dead code. Initiate
18987 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
18988
18989 2018-02-09 Jakub Jelinek <jakub@redhat.com>
18990
18991 PR target/84226
18992 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
18993 constraint from =wa to wa. Avoid a subreg on the output operand,
18994 instead use a pseudo and subreg it in a move.
18995 (p9_xxbrd_<mode>): Changed to ...
18996 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
18997 (p9_xxbrd_v2df): New expander.
18998 (p9_xxbrw_<mode>): Changed to ...
18999 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
19000 (p9_xxbrw_v4sf): New expander.
19001
19002 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
19003
19004 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
19005
19006 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
19007
19008 PR target/83926
19009 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
19010 multiply in 32-bit mode.
19011 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
19012 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
19013 mode.
19014
19015 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
19016
19017 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
19018 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
19019 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
19020 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
19021
19022 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
19023
19024 PR lto/84213
19025 * dwarf2out.c (is_trivial_indirect_ref): New function.
19026 (dwarf2out_late_global_decl): Do not generate a location
19027 attribute for variables that have a non-trivial DECL_VALUE_EXPR
19028 and that are not defined in the current unit.
19029
19030 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
19031
19032 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
19033 instead of a libcall for UNORDERED.
19034
19035 2018-02-09 Tamar Christina <tamar.christina@arm.com>
19036
19037 PR target/82641
19038 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
19039 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
19040
19041 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19042
19043 PR target/PR84295
19044 * config/s390/s390.c (s390_set_current_function): Invoke
19045 s390_indirect_branch_settings also if fndecl didn't change.
19046
19047 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
19048
19049 * config/rs6000/rs6000.md (blockage): Set length to zero.
19050
19051 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
19052
19053 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
19054
19055 2018-02-09 Jakub Jelinek <jakub@redhat.com>
19056
19057 PR sanitizer/84285
19058 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
19059 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
19060 -static-lib*san.
19061
19062 PR debug/84252
19063 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
19064 PARALLEL incoming that failed vt_get_decl_and_offset check.
19065
19066 PR middle-end/84237
19067 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
19068 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
19069 TREE_READONLY bit.
19070 (get_variable_section): For decls in named .bss* sections pass true as
19071 second argument to bss_initializer_p.
19072
19073 2018-02-09 Marek Polacek <polacek@redhat.com>
19074 Jakub Jelinek <jakub@redhat.com>
19075
19076 PR c++/83659
19077 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
19078 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
19079 Sync some changes from cxx_fold_indirect_ref.
19080
19081 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
19082
19083 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
19084 markers.
19085 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
19086 (BLOCK_INLINE_ENTRY_LABEL): New.
19087 (dwarf2out_var_location): Disregard inline entry markers.
19088 (inline_entry_data): New struct.
19089 (inline_entry_data_hasher): New hashtable type.
19090 (inline_entry_data_hasher::hash): New.
19091 (inline_entry_data_hasher::equal): New.
19092 (inline_entry_data_table): New variable.
19093 (add_high_low_attributes): Add DW_AT_entry_pc and
19094 DW_AT_GNU_entry_view attributes if a pending entry is found
19095 in inline_entry_data_table. Add old entry_pc attribute only
19096 if debug nonbinding markers are disabled.
19097 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
19098 markers are enabled.
19099 (block_within_block_p, dwarf2out_inline_entry): New.
19100 (dwarf2out_finish): Check that no entries remained in
19101 inline_entry_data_table.
19102 * final.c (reemit_insn_block_notes): Handle inline entry notes.
19103 (final_scan_insn, notice_source_line): Likewise.
19104 (rest_of_clean_state): Skip inline entry markers.
19105 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
19106 markers.
19107 * gimple.c (gimple_build_debug_inline_entry): New.
19108 * gimple.h (enum gimple_debug_subcode): Add
19109 GIMPLE_DEBUG_INLINE_ENTRY.
19110 (gimple_build_debug_inline_entry): Declare.
19111 (gimple_debug_inline_entry_p): New.
19112 (gimple_debug_nonbind_marker_p): Adjust.
19113 * insn-notes.def (INLINE_ENTRY): New.
19114 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
19115 inline entry marker notes.
19116 (print_insn): Likewise.
19117 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
19118 (INSN_DEBUG_MARKER_KIND): Likewise.
19119 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
19120 * tree-inline.c (expand_call_inline): Build and insert
19121 debug_inline_entry stmt.
19122 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
19123 inline entry blocks early, if nonbind markers are enabled.
19124 (dump_scope_block): Dump fragment info.
19125 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
19126 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
19127 (gimple_build_debug_inline_entry): New.
19128 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
19129 Enable/disable inline entry points too.
19130 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
19131 (DEBUG_INSN): Describe inline entry markers.
19132
19133 * common.opt (gvariable-location-views): New.
19134 (gvariable-location-views=incompat5): New.
19135 * config.in: Rebuilt.
19136 * configure: Rebuilt.
19137 * configure.ac: Test assembler for view support.
19138 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
19139 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
19140 * dwarf2out.c (var_loc_view): New typedef.
19141 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
19142 (dwarf2out_locviews_in_attribute): New.
19143 (dwarf2out_locviews_in_loclist): New.
19144 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
19145 (enum dw_line_info_opcode): Add LI_adv_address.
19146 (struct dw_line_info_table): Add view.
19147 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
19148 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
19149 (zero_view_p): New variable.
19150 (ZERO_VIEW_P): New macro.
19151 (output_asm_line_debug_info): New.
19152 (struct var_loc_node): Add view.
19153 (add_AT_view_list, AT_loc_list): New.
19154 (add_var_loc_to_decl): Add view param. Test it against last.
19155 (new_loc_list): Add view params. Record them.
19156 (AT_loc_list_ptr): Handle loc and view lists.
19157 (view_list_to_loc_list_val_node): New.
19158 (print_dw_val): Handle dw_val_class_view_list.
19159 (size_of_die): Likewise.
19160 (value_format): Likewise.
19161 (loc_list_has_views): New.
19162 (gen_llsym): Set vl_symbol too.
19163 (maybe_gen_llsym, skip_loc_list_entry): New.
19164 (dwarf2out_maybe_output_loclist_view_pair): New.
19165 (output_loc_list): Output view list or entries too.
19166 (output_view_list_offset): New.
19167 (output_die): Handle dw_val_class_view_list.
19168 (output_dwarf_version): New.
19169 (output_compilation_unit_header): Use it.
19170 (output_skeleton_debug_sections): Likewise.
19171 (output_rnglists, output_line_info): Likewise.
19172 (output_pubnames, output_aranges): Update version comments.
19173 (output_one_line_info_table): Output view numbers in asm comments.
19174 (dw_loc_list): Determine current endview, pass it to new_loc_list.
19175 Call maybe_gen_llsym.
19176 (loc_list_from_tree_1): Adjust.
19177 (add_AT_location_description): Create view list attribute if
19178 needed, check it's absent otherwise.
19179 (convert_cfa_to_fb_loc_list): Adjust.
19180 (maybe_emit_file): Call output_asm_line_debug_info for test.
19181 (dwarf2out_var_location): Reset views as needed. Precompute
19182 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
19183 attribute. Set view.
19184 (new_line_info_table): Reset next view.
19185 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
19186 (dwarf2out_source_line): Likewise. Output view resets and labels to
19187 the assembler, or select appropriate line info opcodes.
19188 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
19189 (optimize_string_length): Catch it. Adjust.
19190 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
19191 dw_val_class_view_list, and remove it if no longer needed.
19192 (hash_loc_list): Hash view numbers.
19193 (loc_list_hasher::equal): Compare them.
19194 (optimize_location_lists): Check whether a view list symbol is
19195 needed, and whether the locview attribute is present, and
19196 whether they match. Remove the locview attribute if no longer
19197 needed.
19198 (index_location_lists): Call skip_loc_list_entry for test.
19199 (dwarf2out_finish): Call output_asm_line_debug_info for test.
19200 Use output_dwarf_version.
19201 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
19202 (struct dw_val_node): Add val_view_list.
19203 * final.c (SEEN_NEXT_VIEW): New.
19204 (set_next_view_needed): New.
19205 (clear_next_view_needed): New.
19206 (maybe_output_next_view): New.
19207 (final_start_function): Rename to...
19208 (final_start_function_1): ... this. Take pointer to FIRST,
19209 add SEEN parameter. Emit param bindings in the initial view.
19210 (final_start_function): Reintroduce SEEN-less interface.
19211 (final): Rename to...
19212 (final_1): ... this. Take SEEN parameter. Output final pending
19213 next view at the end.
19214 (final): Reintroduce seen-less interface.
19215 (final_scan_insn): Output pending next view before switching
19216 sections or ending a block. Mark the next view as needed when
19217 outputting variable locations. Notify debug backend of section
19218 changes, and of location view changes.
19219 (rest_of_handle_final): Adjust.
19220 * toplev.c (process_options): Autodetect value for debug variable
19221 location views option. Warn on incompat5 without -gdwarf-5.
19222 * doc/invoke.texi (gvariable-location-views): New.
19223 (gvariable-location-views=incompat5): New.
19224 (gno-variable-location-views): New.
19225
19226 2018-02-08 David Malcolm <dmalcolm@redhat.com>
19227
19228 PR tree-optimization/84136
19229 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
19230 that the result of find_edge is non-NULL.
19231
19232 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19233
19234 PR target/83008
19235 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
19236 storing integer register in SImode. Fix cost of 256 and 512
19237 byte aligned SSE register store.
19238
19239 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
19240
19241 * config/i386/i386.c (ix86_multiplication_cost): Fix
19242 multiplication cost for TARGET_AVX512DQ.
19243
19244 2018-02-08 Marek Polacek <polacek@redhat.com>
19245
19246 PR tree-optimization/84238
19247 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
19248 get_range_strlen.
19249
19250 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19251
19252 PR tree-optimization/84265
19253 * tree-vect-stmts.c (vectorizable_store): Don't treat
19254 VMAT_CONTIGUOUS accesses as grouped.
19255 (vectorizable_load): Likewise.
19256
19257 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19258
19259 PR tree-optimization/81635
19260 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
19261 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
19262 (test_round_for_mask): New functions.
19263 (wide_int_cc_tests): Call test_round_for_mask.
19264 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
19265 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
19266 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
19267 range returned by get_range_info.
19268
19269 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
19270
19271 PR ipa/81360
19272 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
19273 * symtab.c: Include builtins.h
19274 (symtab_node::output_to_lto_symbol_table_p): Move here
19275 from lto-streamer-out.c:output_symbol_p.
19276 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
19277 (output_symbol_p): Move all logic to symtab.c
19278 (produce_symtab): Update.
19279
19280 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
19281
19282 * config/s390/s390-opts.h (enum indirect_branch): Define.
19283 * config/s390/s390-protos.h (s390_return_addr_from_memory)
19284 (s390_indirect_branch_via_thunk)
19285 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
19286 (enum s390_indirect_branch_type): Define.
19287 * config/s390/s390.c (struct s390_frame_layout, struct
19288 machine_function): Remove.
19289 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
19290 (indirect_branch_table_label_no, indirect_branch_table_name):
19291 Define variables.
19292 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
19293 (enum s390_indirect_branch_option): Define.
19294 (s390_return_addr_from_memory): New function.
19295 (s390_handle_string_attribute): New function.
19296 (s390_attribute_table): Add new attribute handler.
19297 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
19298 (s390_indirect_branch_via_thunk): New function.
19299 (s390_indirect_branch_via_inline_thunk): New function.
19300 (s390_function_ok_for_sibcall): When jumping via thunk disallow
19301 sibling call optimization for non z10 compiles.
19302 (s390_emit_call): Force indirect branch target to be a single
19303 register. Add r1 clobber for non-z10 compiles.
19304 (s390_emit_epilogue): Emit return jump via return_use expander.
19305 (s390_reorg): Handle JUMP_INSNs as execute targets.
19306 (s390_option_override_internal): Perform validity checks for the
19307 new command line options.
19308 (s390_indirect_branch_attrvalue): New function.
19309 (s390_indirect_branch_settings): New function.
19310 (s390_set_current_function): Invoke s390_indirect_branch_settings.
19311 (s390_output_indirect_thunk_function): New function.
19312 (s390_code_end): Implement target hook.
19313 (s390_case_values_threshold): Implement target hook.
19314 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
19315 macros.
19316 * config/s390/s390.h (struct s390_frame_layout)
19317 (struct machine_function): Move here from s390.c.
19318 (TARGET_INDIRECT_BRANCH_NOBP_RET)
19319 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
19320 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
19321 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
19322 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
19323 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
19324 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
19325 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
19326 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
19327 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
19328 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
19329 (mnemonic attribute): Add values which aren't recognized
19330 automatically.
19331 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
19332 pattern for branch conversion. Fix mnemonic attribute.
19333 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
19334 indirect branch via thunk if requested.
19335 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
19336 ("*indirect_jump"): Disable for branch conversion using out of
19337 line thunks.
19338 ("indirect_jump_via_thunk<mode>_z10")
19339 ("indirect_jump_via_thunk<mode>")
19340 ("indirect_jump_via_inlinethunk<mode>_z10")
19341 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
19342 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
19343 ("casesi_jump_via_inlinethunk<mode>_z10")
19344 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
19345 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
19346 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
19347 ("*indirect2_jump"): Disable for branch conversion.
19348 ("casesi_jump"): Turn into expander and expand patterns for branch
19349 conversion.
19350 ("return_use"): New expander.
19351 ("*return"): Emit return via thunk and rename it to ...
19352 ("*return<mode>"): ... this one.
19353 * config/s390/s390.opt: Add new options and and enum for the
19354 option values.
19355
19356 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
19357
19358 * lra-constraints.c (match_reload): Unconditionally use
19359 gen_lowpart_SUBREG, rather than selecting between that
19360 and equivalent gen_rtx_SUBREG code.
19361
19362 2018-02-08 Richard Biener <rguenther@suse.de>
19363
19364 PR tree-optimization/84233
19365 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
19366 changed flag instead of boguously re-using phi_inserted.
19367
19368 2018-02-08 Martin Jambor <mjambor@suse.cz>
19369
19370 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
19371 static local variables.
19372
19373 2018-02-08 Richard Biener <rguenther@suse.de>
19374
19375 PR tree-optimization/84278
19376 * tree-vect-stmts.c (vectorizable_store): When looking for
19377 smaller vector types to perform grouped strided loads/stores
19378 make sure the mode is supported by the target.
19379 (vectorizable_load): Likewise.
19380
19381 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19382
19383 * config/aarch64/aarch64.c (aarch64_components_for_bb):
19384 Increase LDP/STP opportunities by adding adjacent callee-saves.
19385
19386 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
19387
19388 PR rtl-optimization/84068
19389 PR rtl-optimization/83459
19390 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
19391
19392 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
19393
19394 PR tree-optimization/84224
19395 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
19396 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
19397 non-zero arguments.
19398
19399 2018-02-07 Iain Sandoe <iain@codesourcery.com>
19400
19401 PR target/84113
19402 * config/rs6000/altivec.md (*restore_world): Remove LR use.
19403 * config/rs6000/predicates.md (restore_world_operation): Adjust op
19404 count, remove one USE.
19405
19406 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
19407
19408 * doc/install.texi (Configuration): Document the
19409 --with-long-double-format={ibm,ieee} PowerPC configuration
19410 options.
19411
19412 PR target/84154
19413 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
19414 Convert from define_expand to be define_insn_and_split. Rework
19415 float/double/_Float128 conversions to QI/HI/SImode to work with
19416 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
19417 conversions to QI/HImode types did a store and then a load to
19418 truncate the value. For conversions to VSX registers, don't split
19419 the insn, instead emit the code directly. Use the code iterator
19420 any_fix to combine signed and unsigned conversions.
19421 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
19422 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
19423 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19424 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
19425 (fix_<mode>di2_hw): Likewise.
19426 (fixuns_<mode>di2_hw): Likewise.
19427 (fix_<mode>si2_hw): Likewise.
19428 (fixuns_<mode>si2_hw): Likewise.
19429 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
19430 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
19431 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
19432 fix<uns>_trunc<SFDF:mode>si2_p8.
19433 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
19434 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
19435 (fix<uns>_<mode>_mem): Likewise.
19436 (fctiw<u>z_<mode>_mem): Likewise.
19437 (fix<uns>_<mode>_mem): Likewise.
19438 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
19439 the register allocator from doing a direct move to the GPRs to do
19440 a store, and instead use the ISA 3.0 store byte/half-word from
19441 vector register instruction. For IEEE 128-bit floating point,
19442 also optimize stores of 32-bit ints.
19443 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
19444
19445 2018-02-07 Alan Hayward <alan.hayward@arm.com>
19446
19447 * genextract.c (push_pathstr_operand): New function to support
19448 [a-zA-Z].
19449 (walk_rtx): Call push_pathstr_operand.
19450 (print_path): Support [a-zA-Z].
19451
19452 2018-02-07 Richard Biener <rguenther@suse.de>
19453
19454 PR tree-optimization/84037
19455 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
19456 (cse_and_gimplify_to_preheader): Declare.
19457 (vect_get_place_in_interleaving_chain): Likewise.
19458 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
19459 ivexpr_map.
19460 (_loop_vec_info::~_loop_vec_info): Delete it.
19461 (cse_and_gimplify_to_preheader): New function.
19462 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
19463 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
19464 (vectorizable_load): Likewise. For grouped stores always base
19465 the IV on the first element.
19466 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
19467 condition before gimplifying.
19468
19469 2018-02-07 Jakub Jelinek <jakub@redhat.com>
19470
19471 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
19472 *DIV_EXPR and *MOD_EXPR.
19473
19474 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
19475
19476 PR target/84248
19477 * config/i386/i386.c (ix86_option_override_internal): Mask out
19478 the CF_SET bit when checking -fcf-protection.
19479
19480 2018-02-07 Tom de Vries <tom@codesourcery.com>
19481
19482 PR libgomp/84217
19483 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
19484 enough.
19485
19486 2018-02-07 Richard Biener <rguenther@suse.de>
19487
19488 PR tree-optimization/84204
19489 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
19490 this place.
19491
19492 PR tree-optimization/84205
19493 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
19494 special-case isl_ast_op_zdiv_r.
19495
19496 PR tree-optimization/84223
19497 * graphite-scop-detection.c (gather_bbs::before_dom_children):
19498 Only add conditions from within the region.
19499 (gather_bbs::after_dom_children): Adjust.
19500
19501 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
19502
19503 PR target/84209
19504 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
19505 * config/avr/avr.md: Only post-reload split REG-REG moves if
19506 either register is GENERAL_REG_P.
19507
19508 2018-02-07 Jakub Jelinek <jakub@redhat.com>
19509
19510 PR tree-optimization/84235
19511 * tree-ssa-scopedtables.c
19512 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
19513 if the subtraction is performed in floating point type where NaNs are
19514 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
19515 build 1. Formatting fix.
19516
19517 2018-02-06 Jakub Jelinek <jakub@redhat.com>
19518
19519 PR target/84146
19520 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
19521 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
19522 and skip it regardless of bb boundaries. Use CALL_P macro,
19523 don't test INSN_P (insn) together with CALL_P or JUMP_P check
19524 unnecessarily, formatting fix.
19525
19526 2018-02-06 Michael Collison <michael.collison@arm.com>
19527
19528 * config/arm/thumb2.md:
19529 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
19530 (*thumb_mov_notscc): Ditto.
19531
19532 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
19533
19534 PR target/84154
19535 * config/rs6000/rs6000.md (su code attribute): Use "u" for
19536 unsigned_fix, not "s".
19537
19538 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19539
19540 * configure.ac (gcc_fn_eh_frame_ro): New function.
19541 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
19542 correct .eh_frame permissions.
19543 * configure: Regenerate.
19544
19545 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
19546
19547 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
19548 irrelevant options.
19549
19550 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
19551
19552 * config/rs6000/rs6000.c (rs6000_option_override_internal):
19553 Display warning message for -mno-speculate-indirect-jumps.
19554
19555 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
19556
19557 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
19558 Undocumented.
19559 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
19560
19561 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
19562
19563 PR tree-optimization/84225
19564 * tree-eh.c (find_trapping_overflow): Only call
19565 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
19566
19567 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19568
19569 PR target/84145
19570 * config/i386/i386.c: Reimplement the check of possible options
19571 -mibt/-mshstk conbination. Change error messages.
19572 * doc/invoke.texi: Fix a typo: remove extra '='.
19573
19574 2018-02-06 Marek Polacek <polacek@redhat.com>
19575
19576 PR tree-optimization/84228
19577 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
19578
19579 2018-02-06 Tamar Christina <tamar.christina@arm.com>
19580
19581 PR target/82641
19582 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
19583 emitted arch directives.
19584 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
19585 __ARM_FEATURE_COPROC before changing architectures.
19586
19587 2018-02-06 Richard Biener <rguenther@suse.de>
19588
19589 * config/i386/i386.c (print_reg): Fix typo.
19590 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
19591
19592 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
19593
19594 * configure: Regenerate.
19595
19596 2018-02-05 Martin Sebor <msebor@redhat.com>
19597
19598 PR tree-optimization/83369
19599 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
19600 inlining context.
19601
19602 2018-02-05 Martin Liska <mliska@suse.cz>
19603
19604 * doc/invoke.texi: Cherry-pick upstream r323995.
19605
19606 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
19607
19608 * ira.c (ira_init_register_move_cost): Adjust comment.
19609
19610 2018-02-05 Martin Liska <mliska@suse.cz>
19611
19612 PR gcov-profile/84137
19613 * doc/gcov.texi: Fix typo in documentation.
19614
19615 2018-02-05 Martin Liska <mliska@suse.cz>
19616
19617 PR gcov-profile/83879
19618 * doc/gcov.texi: Document necessity of --dynamic-list-data when
19619 using dlopen functionality.
19620
19621 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
19622
19623 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
19624 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
19625 _mm_maskz_range_ss, _mm_mask_range_round_ss,
19626 _mm_maskz_range_round_ss): New intrinsics.
19627 (__builtin_ia32_rangesd128_round)
19628 (__builtin_ia32_rangess128_round): Remove.
19629 (__builtin_ia32_rangesd128_mask_round,
19630 __builtin_ia32_rangess128_mask_round): New builtins.
19631 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
19632 __builtin_ia32_rangess128_round): Remove.
19633 (__builtin_ia32_rangesd128_mask_round,
19634 __builtin_ia32_rangess128_mask_round): New builtins.
19635 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
19636 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
19637 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
19638 "<round_saeonly_constraint>")): Changed to ...
19639 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
19640 "<round_saeonly_scalar_constraint>")): ... this.
19641 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
19642 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
19643 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
19644 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
19645 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
19646
19647 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
19648
19649 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
19650 options.
19651 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
19652 Remove all values except native, 8540 and 8548.
19653
19654 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
19655
19656 * config/i386/i386.c (ix86_output_function_return): Pass
19657 INVALID_REGNUM, instead of -1, as invalid register number to
19658 indirect_thunk_name and output_indirect_thunk.
19659
19660 2018-02-02 Julia Koval <julia.koval@intel.com>
19661
19662 * config.gcc: Add -march=icelake.
19663 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
19664 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
19665 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
19666 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
19667 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
19668 (processor_target_table): Add icelake.
19669 (ix86_option_override_internal): Handle new PTAs.
19670 (get_builtin_code_for_version): Handle icelake.
19671 (M_INTEL_COREI7_ICELAKE): New.
19672 (fold_builtin_cpu): Handle icelake.
19673 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
19674 * doc/invoke.texi: Add -march=icelake.
19675
19676 2018-02-02 Julia Koval <julia.koval@intel.com>
19677
19678 * config/i386/i386.c (ix86_option_override_internal): Change flags type
19679 to wide_int_bitmask.
19680 * wide-int-bitmask.h: New.
19681
19682 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
19683
19684 PR target/84066
19685 * config/i386/i386.md: Replace Pmode with word_mode in
19686 builtin_setjmp_setup and builtin_longjmp to support x32.
19687
19688 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
19689
19690 PR target/56010
19691 PR target/83743
19692 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
19693 #include "opts.h".
19694 (rs6000_supported_cpu_names): New static variable.
19695 (linux_cpu_translation_table): Likewise.
19696 (elf_platform) <cpu>: Define new static variable and use it.
19697 Translate kernel AT_PLATFORM name to canonical name if needed.
19698 Error if platform name is unknown.
19699
19700 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
19701
19702 PR target/84089
19703 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
19704
19705 2018-02-01 Jeff Law <law@redhat.com>
19706
19707 PR target/84128
19708 * config/i386/i386.c (release_scratch_register_on_entry): Add new
19709 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
19710 the scratch if RELEASE_VIA_POP is false.
19711 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
19712 If we have to save a temporary register, decrement SIZE appropriately.
19713 Pass new arguments to release_scratch_register_on_entry.
19714 (ix86_adjust_stack_and_probe): Likewise.
19715 (ix86_emit_probe_stack_range): Pass new arguments to
19716 release_scratch_register_on_entry.
19717
19718 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
19719
19720 PR rtl-optimization/84157
19721 * combine.c (change_zero_ext): Use REG_P predicate in
19722 front of HARD_REGISTER_P predicate.
19723
19724 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
19725
19726 * config/avr/avr.c (avr_option_override): Move disabling of
19727 -fdelete-null-pointer-checks to...
19728 * common/config/avr/avr-common.c (avr_option_optimization_table):
19729 ...here.
19730
19731 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19732
19733 PR tree-optimization/81635
19734 * tree-data-ref.c (split_constant_offset_1): For types that
19735 wrap on overflow, try to use range info to prove that wrapping
19736 cannot occur.
19737
19738 2018-02-01 Renlin Li <renlin.li@arm.com>
19739
19740 PR target/83370
19741 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
19742 TAILCALL_ADDR_REGS.
19743 (aarch64_register_move_cost): Likewise.
19744 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
19745 TAILCALL_ADDR_REGS.
19746 (REG_CLASS_NAMES): Likewise.
19747 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
19748 TAILCALL_ADDR_REGS. Remove IP registers.
19749 * config/aarch64/aarch64.md (Ucs): Update register constraint.
19750
19751 2018-02-01 Richard Biener <rguenther@suse.de>
19752
19753 * domwalk.h (dom_walker::dom_walker): Add additional constructor
19754 for specifying RPO order and allow NULL for that.
19755 * domwalk.c (dom_walker::dom_walker): Likewise.
19756 (dom_walker::walk): Handle NULL RPO order.
19757 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
19758 in RPO order.
19759 (rewrite_update_dom_walker): Likewise.
19760 (mark_def_dom_walker): Likewise.
19761
19762 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19763
19764 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
19765 (aarch64_maybe_expand_sve_subreg_move): Declare.
19766 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
19767 * config/aarch64/predicates.md (aarch64_any_register_operand): New
19768 predicate.
19769 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
19770 that are semantically a reverse operation.
19771 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
19772 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
19773 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
19774 functions.
19775 (aarch64_can_change_mode_class): For big-endian, forbid changes
19776 between two SVE modes if they have different element sizes.
19777
19778 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19779
19780 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
19781 the TImode handling for big-endian targets.
19782
19783 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19784
19785 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
19786 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
19787 not just bytes.
19788 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
19789 Remove BSWAP handing for big-endian targets and use the form of
19790 LD1RQ appropariate for the mode.
19791
19792 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19793
19794 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
19795 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
19796 duplicated element.
19797
19798 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
19799
19800 PR tearget/83845
19801 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
19802 check for operands that need to go through aarch64_sve_reload_be.
19803
19804 2018-02-01 Jakub Jelinek <jakub@redhat.com>
19805
19806 PR tree-optimization/81661
19807 PR tree-optimization/84117
19808 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
19809 * tree-eh.c: Include gimplify.h.
19810 (find_trapping_overflow, replace_trapping_overflow,
19811 rewrite_to_non_trapping_overflow): New functions.
19812 * tree-vect-loop.c: Include tree-eh.h.
19813 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
19814 * tree-data-ref.c: Include tree-eh.h.
19815 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
19816
19817 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
19818
19819 PR rtl-optimization/84123
19820 * combine.c (change_zero_ext): Check if hard register satisfies
19821 can_change_dest_mode before calling gen_lowpart_SUBREG.
19822
19823 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
19824
19825 PR target/82444
19826 * ira.c (ira_init_register_move_cost): Remove assert.
19827
19828 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19829
19830 PR rtl-optimization/84071
19831 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
19832 * doc/tm.texi: Regenerate.
19833
19834 2018-01-31 Richard Biener <rguenther@suse.de>
19835
19836 PR tree-optimization/84132
19837 * tree-data-ref.c (analyze_miv_subscript): Properly
19838 check whether evolution_function_is_affine_multivariate_p
19839 before calling gcd_of_steps_may_divide_p.
19840
19841 2018-01-31 Julia Koval <julia.koval@intel.com>
19842
19843 PR target/83618
19844 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
19845 * config/i386/i386.md (rdpid_rex64) New.
19846 (rdpid): Make 32bit only.
19847
19848 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
19849
19850 PR lto/84105
19851 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
19852 an IDENTIFIER_NODE for FUNCTION_TYPE's.
19853
19854 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19855
19856 Revert
19857 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
19858
19859 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
19860
19861 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
19862
19863 PR rtl-optimization/84071
19864 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
19865 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
19866
19867 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
19868
19869 * config/arc/arc.c (arc_handle_aux_attribute): New function.
19870 (arc_attribute_table): Add 'aux' attribute.
19871 (arc_in_small_data_p): Consider aux like variables.
19872 (arc_is_aux_reg_p): New function.
19873 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
19874 (arc_get_aux_arg): New function.
19875 (prepare_move_operands): Handle aux-register access.
19876 (arc_handle_aux_attribute): New function.
19877 * doc/extend.texi (ARC Variable attributes): Add subsection.
19878
19879 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
19880
19881 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
19882 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
19883 (arc_attribute_table): Add 'uncached' attribute.
19884 (arc_print_operand): Print '.di' flag for uncached memory
19885 accesses.
19886 (arc_in_small_data_p): Do not consider for small data the uncached
19887 types.
19888 (arc_is_uncached_mem_p): New function.
19889 * config/arc/predicates.md (compact_store_memory_operand): Check
19890 for uncached memory accesses.
19891 (nonvol_nonimm_operand): Likewise.
19892 * doc/extend.texi (ARC Type Attribute): New subsection.
19893
19894 2018-01-31 Jakub Jelinek <jakub@redhat.com>
19895
19896 PR c/84100
19897 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
19898 falign-loops=): Add Optimization flag.
19899
19900 2018-01-30 Jeff Law <law@redhat.com>
19901
19902 PR target/84064
19903 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
19904 INT_REGISTERS_SAVED. Check it prior to calling
19905 get_scratch_register_on_entry.
19906 (ix86_adjust_stack_and_probe): Similarly.
19907 (ix86_emit_probe_stack_range): Similarly.
19908 (ix86_expand_prologue): Corresponding changes.
19909
19910 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19911
19912 PR target/40411
19913 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
19914 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
19915
19916 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
19917
19918 PR target/84112
19919 * lra-constraints.c (curr_insn_transform): Process AND in the
19920 address.
19921
19922 2018-01-30 Jakub Jelinek <jakub@redhat.com>
19923
19924 PR rtl-optimization/83986
19925 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
19926 dependence against last_pending_memory_flush in addition to
19927 pending_jump_insns.
19928
19929 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
19930
19931 PR tree-optimization/81611
19932 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
19933 copies.
19934
19935 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
19936
19937 PR target/83758
19938 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
19939 a reg rtx.
19940
19941 2018-01-30 Richard Biener <rguenther@suse.de>
19942 Jakub Jelinek <jakub@redhat.com>
19943
19944 PR tree-optimization/84111
19945 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
19946 inner loops added during recursion, as they don't have up-to-date
19947 SSA form.
19948
19949 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
19950
19951 PR ipa/81360
19952 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
19953 (can_inline_edge_by_limits_p): ... here.
19954 (can_early_inline_edge_p, check_callers,
19955 update_caller_keys, update_callee_keys, recursive_inlining,
19956 add_new_edges_to_heap, speculation_useful_p,
19957 inline_small_functions,
19958 inline_small_functions, flatten_function,
19959 inline_to_all_callers_1): Update.
19960
19961 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
19962
19963 * profile-count.c (profile_count::combine_with_ipa_count): Handle
19964 zeros correctly.
19965
19966 2018-01-30 Richard Biener <rguenther@suse.de>
19967
19968 PR tree-optimization/83008
19969 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
19970 invariant and constant vector uses in stmts when they need
19971 more than one stmt.
19972
19973 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19974
19975 PR bootstrap/84017
19976 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
19977 * configure: Regenerate.
19978
19979 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19980
19981 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
19982 pattern.
19983 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
19984 Use gen_rtx_REG rather than gen_lowpart.
19985
19986 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19987
19988 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
19989 rather than 0 when creating partial subregs.
19990
19991 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
19992
19993 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
19994 of usage.
19995
19996 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
19997
19998 PR target/81550
19999 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
20000 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
20001 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
20002 flags. This restores the settings used before the 2017-07-24.
20003 Turning off pre increment/decrement/modify allows IVOPTS to
20004 optimize DF/SF loops where the index is an int.
20005
20006 2018-01-29 Richard Biener <rguenther@suse.de>
20007 Kelvin Nilsen <kelvin@gcc.gnu.org>
20008
20009 PR bootstrap/80867
20010 * tree-vect-stmts.c (vectorizable_call): Don't call
20011 targetm.vectorize_builtin_md_vectorized_function if callee is
20012 NULL.
20013
20014 2018-01-22 Carl Love <cel@us.ibm.com>
20015
20016 * doc/extend.tex: Fix typo in second arg in
20017 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
20018
20019 2018-01-29 Richard Biener <rguenther@suse.de>
20020
20021 PR tree-optimization/84086
20022 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
20023 (flush_ssaname_freelist): When SSA names were released reset
20024 the SCEV hash table.
20025
20026 2018-01-29 Richard Biener <rguenther@suse.de>
20027
20028 PR tree-optimization/84057
20029 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
20030 removed paths when removing edges.
20031
20032 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
20033
20034 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
20035 -mfunction-return=@var{choice}.
20036
20037 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
20038
20039 PR diagnostic/84034
20040 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
20041 Handle CR like TAB.
20042 (layout::print_source_line): Likewise.
20043 (test_get_line_width_without_trailing_whitespace): Add test cases.
20044
20045 2018-01-27 Jakub Jelinek <jakub@redhat.com>
20046
20047 PR middle-end/84040
20048 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
20049 debug insns.
20050
20051 2018-01-26 Jim Wilson <jimw@sifive.com>
20052
20053 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
20054
20055 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
20056 specified.
20057
20058 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20059
20060 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
20061 and CMP + SUB-immediate -> SUBS.
20062
20063 2018-01-26 Martin Sebor <msebor@redhat.com>
20064
20065 PR tree-optimization/83896
20066 * tree-ssa-strlen.c (get_string_len): Rename...
20067 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
20068 Avoid assuming length is constant.
20069 (handle_char_store): Use HOST_WIDE_INT for string length.
20070
20071 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
20072
20073 PR target/81763
20074 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
20075 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
20076
20077 2018-01-26 Richard Biener <rguenther@suse.de>
20078
20079 PR rtl-optimization/84003
20080 * dse.c (record_store): Only record redundant stores when
20081 the earlier store aliases at least all accesses the later one does.
20082
20083 2018-01-26 Jakub Jelinek <jakub@redhat.com>
20084
20085 PR rtl-optimization/83985
20086 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
20087 REG_CFA_RESTORE insns.
20088 (delete_unmarked_insns): Don't ignore separate shrink wrapping
20089 REG_CFA_RESTORE insns here.
20090
20091 PR c/83989
20092 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
20093 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
20094
20095 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20096
20097 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
20098 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
20099 (arc_init): Likewise.
20100 (arc_override_options): Likewise.
20101 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
20102 value.
20103 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
20104 support.
20105 * config/arc/arc.h (TARGET_DBNZ): Define.
20106 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
20107 properly set the tune attribute.
20108 (dbnz): Use TARGET_DBNZ guard.
20109 * config/arc/arc.opt (mtune): Add core3 option.
20110
20111 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20112
20113 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
20114 recognize new pic like addresses.
20115 (arc_delegitimize_address): Clean up.
20116
20117 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20118
20119 * config/arc/arc-arches.def: Option mrf16 valid for all
20120 architectures.
20121 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
20122 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
20123 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
20124 * config/arc/arc-tables.opt: Regenerate.
20125 * config/arc/arc.c (arc_conditional_register_usage): Handle
20126 reduced register file case.
20127 (arc_file_start): Set must have build attributes.
20128 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
20129 mrf16 option value.
20130 * config/arc/arc.opt (mrf16): Add new option.
20131 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
20132 * config/arc/genmultilib.awk: Handle new mrf16 option.
20133 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
20134 * config/arc/t-multilib: Regenerate.
20135 * doc/invoke.texi (ARC Options): Document mrf16 option.
20136
20137 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20138
20139 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
20140 * config/arc/arc.c (arc_handle_secure_attribute): New function.
20141 (arc_attribute_table): Add 'secure_call' attribute.
20142 (arc_print_operand): Print secure call operand.
20143 (arc_function_ok_for_sibcall): Don't optimize tail calls when
20144 secure.
20145 (arc_is_secure_call_p): New function. * config/arc/arc.md
20146 (call_i): Add support for sjli instruction.
20147 (call_value_i): Likewise.
20148 * config/arc/constraints.md (Csc): New constraint.
20149
20150 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
20151 John Eric Martin <John.Martin@emmicro-us.com>
20152
20153 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
20154 * config/arc/arc.c (_arc_jli_section): New struct.
20155 (arc_jli_section): New type.
20156 (rc_jli_sections): New static variable.
20157 (arc_handle_jli_attribute): New function.
20158 (arc_attribute_table): Add jli_always and jli_fixed attribute.
20159 (arc_file_end): New function.
20160 (TARGET_ASM_FILE_END): Define.
20161 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
20162 (arc_add_jli_section): New function.
20163 (jli_call_scan): Likewise.
20164 (arc_reorg): Call jli_call_scan.
20165 (arc_output_addsi): Remove 'S' from printing asm operand.
20166 (arc_is_jli_call_p): New function.
20167 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
20168 operand.
20169 (movhi_insn): Likewise.
20170 (movsi_insn): Likewise.
20171 (movsi_set_cc_insn): Likewise.
20172 (loadqi_update): Likewise.
20173 (load_zeroextendqisi_update): Likewise.
20174 (load_signextendqisi_update): Likewise.
20175 (loadhi_update): Likewise.
20176 (load_zeroextendhisi_update): Likewise.
20177 (load_signextendhisi_update): Likewise.
20178 (loadsi_update): Likewise.
20179 (loadsf_update): Likewise.
20180 (movsicc_insn): Likewise.
20181 (bset_insn): Likewise.
20182 (bxor_insn): Likewise.
20183 (bclr_insn): Likewise.
20184 (bmsk_insn): Likewise.
20185 (bicsi3_insn): Likewise.
20186 (cmpsi_cc_c_insn): Likewise.
20187 (movsi_ne): Likewise.
20188 (movsi_cond_exec): Likewise.
20189 (clrsbsi2): Likewise.
20190 (norm_f): Likewise.
20191 (normw): Likewise.
20192 (swap): Likewise.
20193 (divaw): Likewise.
20194 (flag): Likewise.
20195 (sr): Likewise.
20196 (kflag): Likewise.
20197 (ffs): Likewise.
20198 (ffs_f): Likewise.
20199 (fls): Likewise.
20200 (call_i): Remove 'S' asm letter, add jli instruction.
20201 (call_value_i): Likewise.
20202 * config/arc/arc.op (mjli-always): New option.
20203 * config/arc/constraints.md (Cji): New constraint.
20204 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
20205 operand.
20206 (subsf3_fpx): Likewise.
20207 (mulsf3_fpx): Likewise.
20208 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
20209 asm operand.
20210 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
20211 function attrbutes.
20212 * doc/invoke.texi (ARC): Document mjli-always option.
20213
20214 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
20215
20216 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
20217 avoid addition with 0 and use incw and decw where possible.
20218
20219 2018-01-26 Richard Biener <rguenther@suse.de>
20220
20221 PR tree-optimization/81082
20222 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
20223 association if it requires casting to unsigned.
20224 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
20225 from fold_plusminus_mult_expr to catch important cases late when
20226 range info is available.
20227
20228 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
20229
20230 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
20231 * configure.ac (hidden_linkonce): New test.
20232 * configure: Regenerate.
20233 * config.in: Regenerate.
20234
20235 2018-01-26 Julia Koval <julia.koval@intel.com>
20236
20237 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
20238 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
20239 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
20240 _mm_mask_bitshuffle_epi64_mask): Fix type.
20241 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
20242 USI_FTYPE_V4DI_V4DI_USI): Remove.
20243 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
20244 __builtin_ia32_vpshufbitqmb256_mask,
20245 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
20246 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
20247 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
20248
20249 2018-01-26 Alan Modra <amodra@gmail.com>
20250
20251 PR target/84033
20252 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
20253 UNSPEC_VBPERMQ. Sort other unspecs.
20254
20255 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
20256
20257 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
20258
20259 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
20260
20261 PR middle-end/83055
20262 * predict.c (drop_profile): Do not push/pop cfun; update also
20263 node->count.
20264 (handle_missing_profiles): Fix logic looking for zero profiles.
20265
20266 2018-01-25 Jakub Jelinek <jakub@redhat.com>
20267
20268 PR middle-end/83977
20269 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
20270 on functions with #pragma omp declare simd or functions with simd
20271 attribute.
20272 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
20273 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
20274 Remove trailing \n from warning_at calls.
20275
20276 2018-01-25 Tom de Vries <tom@codesourcery.com>
20277
20278 PR target/84028
20279 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20280 for neutered workers.
20281
20282 2018-01-24 Joseph Myers <joseph@codesourcery.com>
20283
20284 PR target/68467
20285 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
20286 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
20287
20288 2018-01-24 Jeff Law <law@redhat.com>
20289
20290 PR target/83994
20291 * i386.c (get_probe_interval): Move to earlier point.
20292 (ix86_compute_frame_layout): If -fstack-clash-protection and
20293 the frame is larger than the probe interval, then use pushes
20294 to save registers rather than reg->mem moves.
20295 (ix86_expand_prologue): Remove conditional for int_registers_saved
20296 assertion.
20297
20298 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
20299
20300 PR target/84014
20301 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
20302 min/max for never referenced object.
20303
20304 2018-01-24 Jakub Jelinek <jakub@redhat.com>
20305
20306 PR middle-end/83977
20307 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
20308 here.
20309 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
20310 attributes from DECL_ATTRIBUTES (decl) without affecting
20311 DECL_ATTRIBUTES (current_function_decl).
20312 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
20313 functions with non-NULL DECL_ABSTRACT_ORIGIN.
20314
20315 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
20316
20317 PR tree-optimization/83979
20318 * fold-const.c (fold_comparison): Use constant_boolean_node
20319 instead of boolean_{true,false}_node.
20320
20321 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
20322
20323 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
20324 with zero counts.
20325
20326 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20327
20328 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20329 Simplify the clause that sets the length attribute.
20330 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20331 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
20332 clause that sets the length attribute.
20333 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20334
20335 2018-01-24 Tom de Vries <tom@codesourcery.com>
20336
20337 PR target/83589
20338 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
20339 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
20340 Add strict parameter.
20341 (prevent_branch_around_nothing): Insert dummy insn between branch to
20342 label and label with no ptx insn inbetween.
20343 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
20344
20345 2018-01-24 Tom de Vries <tom@codesourcery.com>
20346
20347 PR target/81352
20348 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
20349 for neutered threads in warp.
20350 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
20351
20352 2018-01-24 Richard Biener <rguenther@suse.de>
20353
20354 PR tree-optimization/83176
20355 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
20356 operands.
20357
20358 2018-01-24 Richard Biener <rguenther@suse.de>
20359
20360 PR tree-optimization/82819
20361 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
20362 code generating pluses that are no-ops in the target precision.
20363
20364 2018-01-24 Richard Biener <rguenther@suse.de>
20365
20366 PR middle-end/84000
20367 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
20368
20369 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20370
20371 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
20372 to merge probabilities.
20373 * predict.c (probably_never_executed): Also mark as cold functions
20374 with global 0 profile and guessed local profile.
20375 * profile-count.c (profile_probability::combine_with_count): New
20376 member function.
20377 * profile-count.h (profile_probability::operator*,
20378 profile_probability::operator*=, profile_probability::operator/,
20379 profile_probability::operator/=): Reduce precision to adjusted
20380 and set value to guessed on contradictory divisions.
20381 (profile_probability::combine_with_freq): Remove.
20382 (profile_probability::combine_wiht_count): Declare.
20383 (profile_count::force_nonzero):: Set to adjusted.
20384 (profile_count::probability_in):: Set quality to adjusted.
20385 * tree-ssa-tail-merge.c (replace_block_by): Use
20386 combine_with_count.
20387
20388 2018-01-23 Andrew Waterman <andrew@sifive.com>
20389 Jim Wilson <jimw@sifive.com>
20390
20391 * config/riscv/riscv.c (riscv_stack_boundary): New.
20392 (riscv_option_override): Set riscv_stack_boundary. Handle
20393 riscv_preferred_stack_boundary_arg.
20394 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
20395 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
20396 (STACK_BOUNDARY): Set to riscv_stack_boundary.
20397 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
20398 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
20399 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
20400
20401 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
20402
20403 PR target/83905
20404 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
20405 of struct ix86_frame.
20406 (ix86_expand_epilogue): Likewise. Add a local variable for
20407 the reg_save_offset field in struct ix86_frame.
20408
20409 2018-01-23 Bin Cheng <bin.cheng@arm.com>
20410
20411 PR tree-optimization/82604
20412 * tree-loop-distribution.c (enum partition_kind): New enum item
20413 PKIND_PARTIAL_MEMSET.
20414 (partition_builtin_p): Support above new enum item.
20415 (generate_code_for_partition): Ditto.
20416 (compute_access_range): Differentiate cases that equality can be
20417 proven at all loops, the innermost loops or no loops.
20418 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
20419 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
20420 (finalize_partitions, distribute_loop): Don't fuse partition of
20421 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
20422 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
20423 parloop is enabled.
20424
20425 2018-01-23 Martin Liska <mliska@suse.cz>
20426
20427 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
20428 order to ignore the predictor.
20429 (PRED_POLYMORPHIC_CALL): Likewise.
20430 (PRED_RECURSIVE_CALL): Likewise.
20431
20432 2018-01-23 Martin Liska <mliska@suse.cz>
20433
20434 * tree-profile.c (tree_profiling): Print function header to
20435 aware reader which function we are working on.
20436 * value-prof.c (gimple_find_values_to_profile): Do not print
20437 not interesting value histograms.
20438
20439 2018-01-23 Martin Liska <mliska@suse.cz>
20440
20441 * profile-count.h (enum profile_quality): Add
20442 profile_uninitialized as the first value. Do not number values
20443 as they are zero based.
20444 (profile_count::verify): Update sanity check.
20445 (profile_probability::verify): Likewise.
20446
20447 2018-01-23 Nathan Sidwell <nathan@acm.org>
20448
20449 * doc/invoke.texi (ffor-scope): Deprecate.
20450
20451 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20452
20453 PR tree-optimization/83510
20454 * domwalk.c (set_all_edges_as_executable): New function.
20455 (dom_walker::dom_walker): Convert bool param
20456 "skip_unreachable_blocks" to enum reachability. Move setup of
20457 edge flags to set_all_edges_as_executable and only do it when
20458 reachability is REACHABLE_BLOCKS.
20459 * domwalk.h (enum dom_walker::reachability): New enum.
20460 (dom_walker::dom_walker): Convert bool param
20461 "skip_unreachable_blocks" to enum reachability.
20462 (set_all_edges_as_executable): New decl.
20463 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
20464 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
20465 "reachability".
20466 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
20467 but converting true to REACHABLE_BLOCKS.
20468 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
20469 * tree-vrp.c
20470 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
20471 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
20472 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
20473 REACHABLE_BLOCKS.
20474 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
20475 if check_all_array_refs will be called.
20476
20477 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20478
20479 * tree.c (selftest::test_location_wrappers): Add more test
20480 coverage.
20481
20482 2018-01-23 David Malcolm <dmalcolm@redhat.com>
20483
20484 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
20485 (selftest::test_bit_in_range): Likewise.
20486
20487 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
20488
20489 PR testsuite/83888
20490 * doc/sourcebuild.texi (vect_float): Say that the selector
20491 only describes the situation when -funsafe-math-optimizations is on.
20492 (vect_float_strict): Document.
20493
20494 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
20495
20496 PR tree-optimization/83965
20497 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
20498 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
20499 instead of checking only for a reduction.
20500 (vect_recog_widen_sum_pattern): Likewise.
20501
20502 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20503
20504 * predict.c (probably_never_executed): Only use precise profile info.
20505 (compute_function_frequency): Skip after inlining hack since we now
20506 have quality checking.
20507
20508 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
20509
20510 * profile-count.h (profile_probability::very_unlikely,
20511 profile_probability::unlikely, profile_probability::even): Set
20512 precision to guessed.
20513
20514 2018-01-23 Richard Biener <rguenther@suse.de>
20515
20516 PR tree-optimization/83963
20517 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
20518 Properly terminate dominator walk when crossing the exit edge not
20519 when visiting its source block.
20520
20521 2018-01-23 Jakub Jelinek <jakub@redhat.com>
20522
20523 PR c++/83918
20524 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
20525 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
20526
20527 2018-01-22 Jakub Jelinek <jakub@redhat.com>
20528
20529 PR tree-optimization/83957
20530 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
20531 semicolon after for body surrounded by braces.
20532
20533 PR tree-optimization/83081
20534 * profile-count.h (profile_probability::split): New method.
20535 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
20536 Use profile_probability::split.
20537 (do_compare_rtx_and_jump): Fix adjustment of probabilities
20538 when splitting a single conditional jump into 2.
20539
20540 2018-01-22 David Malcolm <dmalcolm@redhat.com>
20541
20542 PR tree-optimization/69452
20543 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
20544 decl.
20545
20546 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20547
20548 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
20549 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
20550 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
20551
20552 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20553
20554 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
20555 declaration.
20556 * config/rl78/rl78.md (movdi): New define_expand.
20557 * config/rl78/rl78.c (rl78_split_movdi): New function.
20558
20559 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
20560
20561 PR target/83862
20562 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
20563 no longer used.
20564 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
20565 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
20566 128-bit to produce an UNSPEC move to get the double word with the
20567 signbit and then a shift directly to do signbit.
20568 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
20569 implementation with a new version that just does either a direct
20570 move or a regular move. Move memory interface to separate insns.
20571 Move insns so they are next to the expander.
20572 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
20573 with signbit move. Split big and little endian case.
20574 (signbit<mode>2_dm_mem_le): Likewise.
20575 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
20576 (signbit<mode>2_dm2): Likewise.
20577
20578 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20579
20580 * config/rl78/rl78.md (anddi3): New define_expand.
20581
20582 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20583
20584 * config/rl78/rl78.md (umindi3): New define_expand.
20585
20586 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20587
20588 * config/rl78/rl78.md (smindi3): New define_expand.
20589
20590 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20591
20592 * config/rl78/rl78.md (smaxdi3): New define_expand.
20593
20594 2018-01-22 Carl Love <cel@us.ibm.com>
20595
20596 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
20597 LVX_V1TI): Add macro expansion.
20598 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
20599 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
20600 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
20601 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
20602 Change check to determine if the instruction is a byte reversing
20603 entry. Fix typo in comment.
20604 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
20605 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
20606 Add def_builtin calls for new builtins.
20607 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
20608 Add define_insn expansion.
20609
20610 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20611
20612 * config/rl78/rl78.md (umaxdi3): New define_expand.
20613
20614 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
20615
20616 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
20617 for non-QImode registers.
20618
20619 2018-01-22 Richard Biener <rguenther@suse.de>
20620
20621 PR tree-optimization/83963
20622 * graphite-scop-detection.c (scop_detection::get_sese): Delay
20623 including the loop exit block.
20624 (scop_detection::merge_sese): Likewise.
20625 (scop_detection::add_scop): Do it here instead.
20626
20627 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20628
20629 * doc/sourcebuild.texi (arm_softfloat): Document.
20630
20631 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
20632
20633 PR gcc/77734
20634 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
20635 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
20636 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
20637
20638 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20639 David Edelsohn <dje.gcc@gmail.com>
20640
20641 PR target/83946
20642 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20643 Change "crset eq" to "crset 2".
20644 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20645 (*call_indirect_aix<mode>_nospec): Likewise.
20646 (*call_value_indirect_aix<mode>_nospec): Likewise.
20647 (*call_indirect_elfv2<mode>_nospec): Likewise.
20648 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
20649 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
20650 change assembly output from . to $.
20651 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20652 (indirect_jump<mode>_nospec): Change assembly output from . to $.
20653 (*tablejump<mode>_internal1_nospec): Likewise.
20654
20655 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
20656
20657 PR target/80870
20658 * config/sh/sh_optimize_sett_clrt.cc:
20659 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
20660
20661 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
20662
20663 PR tree-optimization/83940
20664 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
20665 offset_dt to vect_constant_def rather than vect_unknown_def_type.
20666 (vect_check_load_store_mask): Add a mask_dt_out parameter and
20667 use it to pass back the definition type.
20668 (vect_check_store_rhs): Likewise rhs_dt_out.
20669 (vect_build_gather_load_calls): Add a mask_dt argument and use
20670 it instead of a call to vect_is_simple_use.
20671 (vectorizable_store): Update calls to vect_check_load_store_mask
20672 and vect_check_store_rhs. Use the dt returned by the latter instead
20673 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
20674 instead of calls to vect_is_simple_use. Pass the scalar rather
20675 than the vector operand to vect_is_simple_use when handling
20676 second and subsequent copies of an rhs value.
20677 (vectorizable_load): Update calls to vect_check_load_store_mask
20678 and vect_build_gather_load_calls. Use the cached mask_dt and
20679 gs_info.offset_dt instead of calls to vect_is_simple_use.
20680
20681 2018-01-20 Jakub Jelinek <jakub@redhat.com>
20682
20683 PR middle-end/83945
20684 * tree-emutls.c: Include gimplify.h.
20685 (lower_emutls_2): New function.
20686 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
20687 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
20688 it before further processing.
20689
20690 PR target/83930
20691 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
20692 UINTVAL (trueop1) instead of INTVAL (op1).
20693
20694 2018-01-19 Jakub Jelinek <jakub@redhat.com>
20695
20696 PR debug/81570
20697 PR debug/83728
20698 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
20699 INCOMING_FRAME_SP_OFFSET if not defined.
20700 (scan_trace): Add ENTRY argument. If true and
20701 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
20702 emit a note to adjust the CFA offset.
20703 (create_cfi_notes): Adjust scan_trace callers.
20704 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
20705 INCOMING_FRAME_SP_OFFSET in the CIE.
20706 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
20707 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
20708 Likewise.
20709 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
20710 * doc/tm.texi: Regenerated.
20711
20712 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
20713
20714 PR rtl-optimization/83147
20715 * lra-constraints.c (remove_inheritance_pseudos): Use
20716 lra_substitute_pseudo_within_insn.
20717
20718 2018-01-19 Tom de Vries <tom@codesourcery.com>
20719 Cesar Philippidis <cesar@codesourcery.com>
20720
20721 PR target/83920
20722 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
20723
20724 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
20725
20726 PR target/83790
20727 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
20728 spaces for function labels.
20729
20730 2018-01-19 Martin Liska <mliska@suse.cz>
20731
20732 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
20733 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
20734 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
20735 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
20736 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
20737 (PRED_CONST_RETURN): Change from 69 to 65.
20738 (PRED_NULL_RETURN): Change from 91 to 71.
20739 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
20740 (PRED_LOOP_GUARD): Change from 66 to 73.
20741
20742 2018-01-19 Martin Liska <mliska@suse.cz>
20743
20744 * predict.c (predict_insn_def): Add new assert.
20745 (struct branch_predictor): Change type to signed integer.
20746 (test_prediction_value_range): Amend test to cover
20747 PROB_UNINITIALIZED.
20748 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
20749 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
20750 (PRED_LOOP_ITERATIONS_MAX): Likewise.
20751 (PRED_LOOP_IV_COMPARE): Likewise.
20752 * predict.h (PROB_UNINITIALIZED): Define new constant.
20753
20754 2018-01-19 Martin Liska <mliska@suse.cz>
20755
20756 * predict.c (dump_prediction): Add new format for
20757 analyze_brprob.py script which is enabled with -details
20758 suboption.
20759 * profile-count.h (precise_p): New function.
20760
20761 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
20762
20763 PR tree-optimization/83922
20764 * tree-vect-loop.c (vect_verify_full_masking): Return false if
20765 there are no statements that need masking.
20766 (vect_active_double_reduction_p): New function.
20767 (vect_analyze_loop_operations): Use it when handling phis that
20768 are not in the loop header.
20769
20770 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
20771
20772 PR tree-optimization/83914
20773 * tree-vect-loop.c (vectorizable_induction): Don't convert
20774 init_expr or apply the peeling adjustment for inductions
20775 that are nested within the vectorized loop.
20776
20777 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20778
20779 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
20780 instead of NEG.
20781
20782 2018-01-18 Jakub Jelinek <jakub@redhat.com>
20783
20784 PR sanitizer/81715
20785 PR testsuite/83882
20786 * function.h (gimplify_parameters): Add gimple_seq * argument.
20787 * function.c: Include gimple.h and options.h.
20788 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
20789 for the added local temporaries if needed.
20790 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
20791 if there are any parameter cleanups, wrap whole body into a
20792 try/finally with the cleanups.
20793
20794 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
20795
20796 PR target/82964
20797 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
20798 Use GET_MODE_CLASS for scalar floating point.
20799
20800 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
20801
20802 PR ipa/82256
20803 patch by PaX Team
20804 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
20805 Fix call of call_cgraph_insertion_hooks.
20806
20807 2018-01-18 Martin Sebor <msebor@redhat.com>
20808
20809 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
20810
20811 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
20812
20813 PR ipa/83619
20814 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
20815 frequencies.
20816
20817 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
20818
20819 PR other/70268
20820 * common.opt: (-ffile-prefix-map): New option.
20821 * opts.c (common_handle_option): Defer it.
20822 * opts-global.c (handle_common_deferred_options): Handle it.
20823 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
20824 * file-prefix-map.h: New file.
20825 (remap_debug_filename, add_debug_prefix_map): ...here.
20826 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
20827 * final.c (debug_prefix_map, add_debug_prefix_map
20828 remap_debug_filename): Move to...
20829 * file-prefix-map.c: New file.
20830 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
20831 generalize, get rid of alloca(), use strrchr() instead of strchr().
20832 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
20833 Implement in terms of add_prefix_map().
20834 (remap_macro_filename, remap_debug_filename): Implement in term of
20835 remap_filename().
20836 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
20837 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
20838 * dbxout.c: Include file-prefix-map.h.
20839 * varasm.c: Likewise.
20840 * vmsdbgout.c: Likewise.
20841 * xcoffout.c: Likewise.
20842 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
20843 * doc/cppopts.texi (-fmacro-prefix-map): Document.
20844 * doc/invoke.texi (-ffile-prefix-map): Document.
20845 (-fdebug-prefix-map): Update description.
20846
20847 2018-01-18 Martin Liska <mliska@suse.cz>
20848
20849 * config/i386/i386.c (indirect_thunk_name): Document that also
20850 lfence is emitted.
20851 (output_indirect_thunk): Document why both instructions
20852 (pause and lfence) are generated.
20853
20854 2018-01-18 Richard Biener <rguenther@suse.de>
20855
20856 PR tree-optimization/83887
20857 * graphite-scop-detection.c
20858 (scop_detection::get_nearest_dom_with_single_entry): Remove.
20859 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
20860 (scop_detection::merge_sese): Re-implement with a flood-fill
20861 algorithm that properly finds a SESE region if it exists.
20862
20863 2018-01-18 Jakub Jelinek <jakub@redhat.com>
20864
20865 PR c/61240
20866 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
20867 pointer_diff optimizations use view_convert instead of convert.
20868
20869 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
20870
20871 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
20872 Generate different code for -mno-speculate-indirect-jumps.
20873 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
20874 (*call_indirect_aix<mode>): Disable for
20875 -mno-speculate-indirect-jumps.
20876 (*call_indirect_aix<mode>_nospec): New define_insn.
20877 (*call_value_indirect_aix<mode>): Disable for
20878 -mno-speculate-indirect-jumps.
20879 (*call_value_indirect_aix<mode>_nospec): New define_insn.
20880 (*sibcall_nonlocal_sysv<mode>): Generate different code for
20881 -mno-speculate-indirect-jumps.
20882 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
20883
20884 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
20885
20886 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
20887 long double type, set the flags for noting the default long double
20888 type, even if we don't pass or return a long double type.
20889
20890 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
20891
20892 PR ipa/83051
20893 * ipa-inline.c (flatten_function): Do not overwrite final inlining
20894 failure.
20895
20896 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
20897
20898 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
20899 support for merge[hl].
20900 (fold_mergehl_helper): New helper function.
20901 (tree-vector-builder.h): New #include for tree_vector_builder usage.
20902 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
20903 (altivec_vmrglw_direct): Add xxmrglw insn.
20904
20905 2018-01-17 Andrew Waterman <andrew@sifive.com>
20906
20907 * config/riscv/riscv.c (riscv_conditional_register_usage): If
20908 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
20909
20910 2018-01-17 David Malcolm <dmalcolm@redhat.com>
20911
20912 PR lto/83121
20913 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
20914 call the lto_location_cache before reading the
20915 DECL_SOURCE_LOCATION of the types.
20916
20917 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
20918 Richard Sandiford <richard.sandiford@linaro.org>
20919
20920 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
20921 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
20922 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
20923 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
20924 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
20925 Add declaration.
20926 * config/aarch64/constraints.md (aarch64_movti_operand):
20927 Limit immediates.
20928 * config/aarch64/predicates.md (Uti): Add new constraint.
20929
20930 2018-01-17 Carl Love <cel@us.ibm.com>
20931
20932 * config/rs6000/vsx.md (define_expand xl_len_r,
20933 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
20934 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
20935 lxvll.
20936 (define_expand, define_insn): Move the shift left from the
20937 define_insn to the define_expand for lxvl and stxvl instructions.
20938 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
20939 and XL_LEN_R definitions to PURE.
20940
20941 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
20942
20943 * config/i386/i386.c (indirect_thunk_name): Declare regno
20944 as unsigned int. Compare regno with INVALID_REGNUM.
20945 (output_indirect_thunk): Ditto.
20946 (output_indirect_thunk_function): Ditto.
20947 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
20948 in the call to output_indirect_thunk_function.
20949
20950 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
20951
20952 PR middle-end/83884
20953 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
20954 rather than the size of inner_type to determine the stack slot size
20955 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
20956
20957 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
20958
20959 PR target/83546
20960 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
20961 to PTA_SILVERMONT.
20962
20963 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
20964
20965 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
20966 endian Linux systems to optionally enable multilibs for selecting
20967 the long double type if the user configured an explicit type.
20968 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
20969 have no long double multilibs if not defined.
20970 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
20971 warn if the user used -mabi={ieee,ibm}longdouble and we built
20972 multilibs for long double.
20973 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
20974 appropriate multilib option.
20975 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
20976 multilib options.
20977 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
20978 for building long double multilibs.
20979 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
20980
20981 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
20982
20983 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
20984 copies.
20985
20986 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
20987 64 bits.
20988 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
20989 128 bits.
20990
20991 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
20992 variables.
20993
20994 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
20995 return value.
20996
20997 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
20998
20999 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
21000 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
21001
21002 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
21003
21004 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
21005 different rtl trees depending on TARGET_64BIT.
21006 (rs6000_gen_lvx): Likewise.
21007
21008 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
21009
21010 * config/visium/visium.md (nop): Tweak comment.
21011 (hazard_nop): Likewise.
21012
21013 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
21014
21015 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
21016 -mspeculate-indirect-jumps.
21017 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
21018 for -mno-speculate-indirect-jumps.
21019 (*call_indirect_elfv2<mode>_nospec): New define_insn.
21020 (*call_value_indirect_elfv2<mode>): Disable for
21021 -mno-speculate-indirect-jumps.
21022 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
21023 (indirect_jump): Emit different RTL for
21024 -mno-speculate-indirect-jumps.
21025 (*indirect_jump<mode>): Disable for
21026 -mno-speculate-indirect-jumps.
21027 (*indirect_jump<mode>_nospec): New define_insn.
21028 (tablejump): Emit different RTL for
21029 -mno-speculate-indirect-jumps.
21030 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
21031 (tablejumpsi_nospec): New define_expand.
21032 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
21033 (tablejumpdi_nospec): New define_expand.
21034 (*tablejump<mode>_internal1): Disable for
21035 -mno-speculate-indirect-jumps.
21036 (*tablejump<mode>_internal1_nospec): New define_insn.
21037 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
21038 option.
21039
21040 2018-01-16 Artyom Skrobov tyomitch@gmail.com
21041
21042 * caller-save.c (insert_save): Drop unnecessary parameter. All
21043 callers updated.
21044
21045 2018-01-16 Jakub Jelinek <jakub@redhat.com>
21046 Richard Biener <rguenth@suse.de>
21047
21048 PR libgomp/83590
21049 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
21050 return early, inline manually is_gimple_sizepos. Make sure if we
21051 call gimplify_expr we don't end up with a gimple constant.
21052 * tree.c (variably_modified_type_p): Don't return true for
21053 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
21054 * gimplify.h (is_gimple_sizepos): Remove.
21055
21056 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21057
21058 PR tree-optimization/83857
21059 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
21060 vectorizable_live_operation for pure SLP statements.
21061 (vectorizable_live_operation): Handle PHIs.
21062
21063 2018-01-16 Richard Biener <rguenther@suse.de>
21064
21065 PR tree-optimization/83867
21066 * tree-vect-stmts.c (vect_transform_stmt): Precompute
21067 nested_in_vect_loop_p since the scalar stmt may get invalidated.
21068
21069 2018-01-16 Jakub Jelinek <jakub@redhat.com>
21070
21071 PR c/83844
21072 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
21073 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
21074 If off is not INTEGER_CST, issue a may not be aligned warning
21075 rather than isn't aligned. Use isn%'t rather than isn't.
21076 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
21077 into MULT_EXPR.
21078 <case MULT_EXPR>: Improve the case when bottom and one of the
21079 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
21080 operand, in that case check if the other operand is multiple of
21081 bottom divided by the INTEGER_CST operand.
21082
21083 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21084
21085 PR target/83858
21086 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
21087 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
21088 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
21089 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
21090 * config/pa/pa.c (pa_function_arg_advance): Likewise.
21091 (pa_function_arg, pa_arg_partial_bytes): Likewise.
21092 (pa_function_arg_size): New function.
21093
21094 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21095
21096 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
21097 in a separate statement.
21098
21099 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
21100
21101 PR tree-optimization/83847
21102 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
21103 group gathers and scatters.
21104
21105 2018-01-16 Jakub Jelinek <jakub@redhat.com>
21106
21107 PR rtl-optimization/86620
21108 * params.def (max-sched-ready-insns): Bump minimum value to 1.
21109
21110 PR rtl-optimization/83213
21111 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
21112 to last if both are JUMP_INSNs.
21113
21114 PR tree-optimization/83843
21115 * gimple-ssa-store-merging.c
21116 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
21117 store_immediate_info for bswap/nop orig_stores.
21118
21119 2018-01-15 Andrew Waterman <andrew@sifive.com>
21120
21121 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
21122 !TARGET_MUL.
21123 <UDIV>: Increase cost if !TARGET_DIV.
21124
21125 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
21126
21127 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
21128 (define_attr "cr_logical_3op"): New.
21129 (cceq_ior_compare): Adjust.
21130 (cceq_ior_compare_complement): Adjust.
21131 (*cceq_rev_compare): Adjust.
21132 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
21133 (is_cracked_insn): Adjust.
21134 (insn_must_be_first_in_group): Adjust.
21135 * config/rs6000/40x.md: Adjust.
21136 * config/rs6000/440.md: Adjust.
21137 * config/rs6000/476.md: Adjust.
21138 * config/rs6000/601.md: Adjust.
21139 * config/rs6000/603.md: Adjust.
21140 * config/rs6000/6xx.md: Adjust.
21141 * config/rs6000/7450.md: Adjust.
21142 * config/rs6000/7xx.md: Adjust.
21143 * config/rs6000/8540.md: Adjust.
21144 * config/rs6000/cell.md: Adjust.
21145 * config/rs6000/e300c2c3.md: Adjust.
21146 * config/rs6000/e500mc.md: Adjust.
21147 * config/rs6000/e500mc64.md: Adjust.
21148 * config/rs6000/e5500.md: Adjust.
21149 * config/rs6000/e6500.md: Adjust.
21150 * config/rs6000/mpc.md: Adjust.
21151 * config/rs6000/power4.md: Adjust.
21152 * config/rs6000/power5.md: Adjust.
21153 * config/rs6000/power6.md: Adjust.
21154 * config/rs6000/power7.md: Adjust.
21155 * config/rs6000/power8.md: Adjust.
21156 * config/rs6000/power9.md: Adjust.
21157 * config/rs6000/rs64.md: Adjust.
21158 * config/rs6000/titan.md: Adjust.
21159
21160 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21161
21162 * config/i386/predicates.md (indirect_branch_operand): Rewrite
21163 ix86_indirect_branch_register logic.
21164
21165 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21166
21167 * config/i386/constraints.md (Bs): Update
21168 ix86_indirect_branch_register check. Don't check
21169 ix86_indirect_branch_register with GOT_memory_operand.
21170 (Bw): Likewise.
21171 * config/i386/predicates.md (GOT_memory_operand): Don't check
21172 ix86_indirect_branch_register here.
21173 (GOT32_symbol_operand): Likewise.
21174
21175 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21176
21177 * config/i386/predicates.md (constant_call_address_operand):
21178 Rewrite ix86_indirect_branch_register logic.
21179 (sibcall_insn_operand): Likewise.
21180
21181 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21182
21183 * config/i386/constraints.md (Bs): Replace
21184 ix86_indirect_branch_thunk_register with
21185 ix86_indirect_branch_register.
21186 (Bw): Likewise.
21187 * config/i386/i386.md (indirect_jump): Likewise.
21188 (tablejump): Likewise.
21189 (*sibcall_memory): Likewise.
21190 (*sibcall_value_memory): Likewise.
21191 Peepholes of indirect call and jump via memory: Likewise.
21192 * config/i386/i386.opt: Likewise.
21193 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21194 (GOT_memory_operand): Likewise.
21195 (call_insn_operand): Likewise.
21196 (sibcall_insn_operand): Likewise.
21197 (GOT32_symbol_operand): Likewise.
21198
21199 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21200
21201 PR middle-end/83837
21202 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
21203 type rather than type addr's type points to.
21204 (expand_omp_atomic_mutex): Likewise.
21205 (expand_omp_atomic): Likewise.
21206
21207 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
21208
21209 PR target/83839
21210 * config/i386/i386.c (output_indirect_thunk_function): Use
21211 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
21212 for __x86_return_thunk.
21213
21214 2018-01-15 Richard Biener <rguenther@suse.de>
21215
21216 PR middle-end/83850
21217 * expmed.c (extract_bit_field_1): Fix typo.
21218
21219 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21220
21221 PR target/83687
21222 * config/arm/iterators.md (VF): New mode iterator.
21223 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
21224 Remove integer-related logic from pattern.
21225 (neon_vabd<mode>_3): Likewise.
21226
21227 2018-01-15 Jakub Jelinek <jakub@redhat.com>
21228
21229 PR middle-end/82694
21230 * common.opt (fstrict-overflow): No longer an alias.
21231 (fwrapv-pointer): New option.
21232 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
21233 also for pointer types based on flag_wrapv_pointer.
21234 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
21235 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
21236 opts->x_flag_wrapv got set.
21237 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
21238 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
21239 POINTER_TYPE_OVERFLOW_UNDEFINED.
21240 * match.pd: Likewise in address comparison pattern.
21241 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
21242
21243 2018-01-15 Richard Biener <rguenther@suse.de>
21244
21245 PR lto/83804
21246 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
21247 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
21248 Reset type names to their identifier if their TYPE_DECL doesn't
21249 have linkage (and thus is used for ODR and devirt).
21250 (save_debug_info_for_decl): Remove.
21251 (save_debug_info_for_type): Likewise.
21252 (add_tree_to_fld_list): Adjust.
21253 * tree-pretty-print.c (dump_generic_node): Make dumping of
21254 type names more robust.
21255
21256 2018-01-15 Richard Biener <rguenther@suse.de>
21257
21258 * BASE-VER: Bump to 8.0.1.
21259
21260 2018-01-14 Martin Sebor <msebor@redhat.com>
21261
21262 PR other/83508
21263 * builtins.c (check_access): Avoid warning when the no-warning bit
21264 is set.
21265
21266 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
21267
21268 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
21269 * ira-color (allocno_hard_regs_compare): Likewise.
21270
21271 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
21272
21273 PR target/83013
21274 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
21275 Use .pushsection/.popsection.
21276
21277 2018-01-14 Martin Sebor <msebor@redhat.com>
21278
21279 PR c++/81327
21280 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
21281
21282 2018-01-14 Jakub Jelinek <jakub@redhat.com>
21283
21284 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
21285 entry from extra_headers.
21286 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
21287 extra_headers, make the list bitwise identical to the i?86-*-* one.
21288
21289 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21290
21291 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
21292 -mcmodel=large with -mindirect-branch=thunk,
21293 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
21294 -mfunction-return=thunk-extern.
21295 * doc/invoke.texi: Document -mcmodel=large is incompatible with
21296 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
21297 -mfunction-return=thunk and -mfunction-return=thunk-extern.
21298
21299 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21300
21301 * config/i386/i386.c (print_reg): Print the name of the full
21302 integer register without '%'.
21303 (ix86_print_operand): Handle 'V'.
21304 * doc/extend.texi: Document 'V' modifier.
21305
21306 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21307
21308 * config/i386/constraints.md (Bs): Disallow memory operand for
21309 -mindirect-branch-register.
21310 (Bw): Likewise.
21311 * config/i386/predicates.md (indirect_branch_operand): Likewise.
21312 (GOT_memory_operand): Likewise.
21313 (call_insn_operand): Likewise.
21314 (sibcall_insn_operand): Likewise.
21315 (GOT32_symbol_operand): Likewise.
21316 * config/i386/i386.md (indirect_jump): Call convert_memory_address
21317 for -mindirect-branch-register.
21318 (tablejump): Likewise.
21319 (*sibcall_memory): Likewise.
21320 (*sibcall_value_memory): Likewise.
21321 Disallow peepholes of indirect call and jump via memory for
21322 -mindirect-branch-register.
21323 (*call_pop): Replace m with Bw.
21324 (*call_value_pop): Likewise.
21325 (*sibcall_pop_memory): Replace m with Bs.
21326 * config/i386/i386.opt (mindirect-branch-register): New option.
21327 * doc/invoke.texi: Document -mindirect-branch-register option.
21328
21329 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21330
21331 * config/i386/i386-protos.h (ix86_output_function_return): New.
21332 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
21333 set function_return_type.
21334 (indirect_thunk_name): Add ret_p to indicate thunk for function
21335 return.
21336 (output_indirect_thunk_function): Pass false to
21337 indirect_thunk_name.
21338 (ix86_output_indirect_branch_via_reg): Likewise.
21339 (ix86_output_indirect_branch_via_push): Likewise.
21340 (output_indirect_thunk_function): Create alias for function
21341 return thunk if regno < 0.
21342 (ix86_output_function_return): New function.
21343 (ix86_handle_fndecl_attribute): Handle function_return.
21344 (ix86_attribute_table): Add function_return.
21345 * config/i386/i386.h (machine_function): Add
21346 function_return_type.
21347 * config/i386/i386.md (simple_return_internal): Use
21348 ix86_output_function_return.
21349 (simple_return_internal_long): Likewise.
21350 * config/i386/i386.opt (mfunction-return=): New option.
21351 (indirect_branch): Mention -mfunction-return=.
21352 * doc/extend.texi: Document function_return function attribute.
21353 * doc/invoke.texi: Document -mfunction-return= option.
21354
21355 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
21356
21357 * config/i386/i386-opts.h (indirect_branch): New.
21358 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
21359 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
21360 with local indirect jump when converting indirect call and jump.
21361 (ix86_set_indirect_branch_type): New.
21362 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
21363 (indirectlabelno): New.
21364 (indirect_thunk_needed): Likewise.
21365 (indirect_thunk_bnd_needed): Likewise.
21366 (indirect_thunks_used): Likewise.
21367 (indirect_thunks_bnd_used): Likewise.
21368 (INDIRECT_LABEL): Likewise.
21369 (indirect_thunk_name): Likewise.
21370 (output_indirect_thunk): Likewise.
21371 (output_indirect_thunk_function): Likewise.
21372 (ix86_output_indirect_branch_via_reg): Likewise.
21373 (ix86_output_indirect_branch_via_push): Likewise.
21374 (ix86_output_indirect_branch): Likewise.
21375 (ix86_output_indirect_jmp): Likewise.
21376 (ix86_code_end): Call output_indirect_thunk_function if needed.
21377 (ix86_output_call_insn): Call ix86_output_indirect_branch if
21378 needed.
21379 (ix86_handle_fndecl_attribute): Handle indirect_branch.
21380 (ix86_attribute_table): Add indirect_branch.
21381 * config/i386/i386.h (machine_function): Add indirect_branch_type
21382 and has_local_indirect_jump.
21383 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
21384 to true.
21385 (tablejump): Likewise.
21386 (*indirect_jump): Use ix86_output_indirect_jmp.
21387 (*tablejump_1): Likewise.
21388 (simple_return_indirect_internal): Likewise.
21389 * config/i386/i386.opt (mindirect-branch=): New option.
21390 (indirect_branch): New.
21391 (keep): Likewise.
21392 (thunk): Likewise.
21393 (thunk-inline): Likewise.
21394 (thunk-extern): Likewise.
21395 * doc/extend.texi: Document indirect_branch function attribute.
21396 * doc/invoke.texi: Document -mindirect-branch= option.
21397
21398 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
21399
21400 PR ipa/83051
21401 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
21402
21403 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
21404
21405 * ipa-inline.c (want_inline_small_function_p): Return false if
21406 inlining has already failed with CIF_FINAL_ERROR.
21407 (update_caller_keys): Call want_inline_small_function_p before
21408 can_inline_edge_p.
21409 (update_callee_keys): Likewise.
21410
21411 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
21412
21413 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
21414 New function.
21415 (rs6000_quadword_masked_address_p): Likewise.
21416 (quad_aligned_load_p): Likewise.
21417 (quad_aligned_store_p): Likewise.
21418 (const_load_sequence_p): Add comment to describe the outer-most loop.
21419 (mimic_memory_attributes_and_flags): New function.
21420 (rs6000_gen_stvx): Likewise.
21421 (replace_swapped_aligned_store): Likewise.
21422 (rs6000_gen_lvx): Likewise.
21423 (replace_swapped_aligned_load): Likewise.
21424 (replace_swapped_load_constant): Capitalize argument name in
21425 comment describing this function.
21426 (rs6000_analyze_swaps): Add a third pass to search for vector loads
21427 and stores that access quad-word aligned addresses and replace
21428 with stvx or lvx instructions when appropriate.
21429 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
21430 New function prototype.
21431 (rs6000_quadword_masked_address_p): Likewise.
21432 (rs6000_gen_lvx): Likewise.
21433 (rs6000_gen_stvx): Likewise.
21434 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
21435 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
21436 when memory address is aligned.
21437 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
21438 this split to select lvx instruction when memory address is aligned.
21439 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
21440 instruction when memory address is aligned.
21441 (*vsx_le_perm_load_v16qi): Likewise.
21442 (four unnamed splitters): Modify to select the stvx instruction
21443 when memory is aligned.
21444
21445 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
21446
21447 * predict.c (determine_unlikely_bbs): Handle correctly BBs
21448 which appears in the queue multiple times.
21449
21450 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21451 Alan Hayward <alan.hayward@arm.com>
21452 David Sherwood <david.sherwood@arm.com>
21453
21454 * tree-vectorizer.h (vec_lower_bound): New structure.
21455 (_loop_vec_info): Add check_nonzero and lower_bounds.
21456 (LOOP_VINFO_CHECK_NONZERO): New macro.
21457 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
21458 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
21459 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
21460 fields. Make seg_len the distance travelled, not including the
21461 access size.
21462 (dr_direction_indicator): Declare.
21463 (dr_zero_step_indicator): Likewise.
21464 (dr_known_forward_stride_p): Likewise.
21465 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
21466 tree-ssanames.h.
21467 (runtime_alias_check_p): Allow runtime alias checks with
21468 variable strides.
21469 (operator ==): Compare access_size and align.
21470 (prune_runtime_alias_test_list): Rework for new distinction between
21471 the access_size and seg_len.
21472 (create_intersect_range_checks_index): Likewise. Cope with polynomial
21473 segment lengths.
21474 (get_segment_min_max): New function.
21475 (create_intersect_range_checks): Use it.
21476 (dr_step_indicator): New function.
21477 (dr_direction_indicator): Likewise.
21478 (dr_zero_step_indicator): Likewise.
21479 (dr_known_forward_stride_p): Likewise.
21480 * tree-loop-distribution.c (data_ref_segment_size): Return
21481 DR_STEP * (niters - 1).
21482 (compute_alias_check_pairs): Update call to the dr_with_seg_len
21483 constructor.
21484 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
21485 (vect_preserves_scalar_order_p): New function, split out from...
21486 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
21487 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
21488 (vect_vfa_access_size): New function.
21489 (vect_vfa_align): Likewise.
21490 (vect_compile_time_alias): Take access_size_a and access_b arguments.
21491 (dump_lower_bound): New function.
21492 (vect_check_lower_bound): Likewise.
21493 (vect_small_gap_p): Likewise.
21494 (vectorizable_with_step_bound_p): Likewise.
21495 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
21496 depencies if the vectorization factor is 1. Convert the checks
21497 for nonzero steps into checks on the bounds of DR_STEP. Try using
21498 a bunds check for variable steps if the minimum required step is
21499 relatively small. Update calls to the dr_with_seg_len
21500 constructor and to vect_compile_time_alias.
21501 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
21502 function.
21503 (vect_loop_versioning): Call it.
21504 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
21505 when retrying.
21506 (vect_estimate_min_profitable_iters): Account for any bounds checks.
21507
21508 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21509 Alan Hayward <alan.hayward@arm.com>
21510 David Sherwood <david.sherwood@arm.com>
21511
21512 * doc/sourcebuild.texi (vect_scatter_store): Document.
21513 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
21514 optabs.
21515 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
21516 Document.
21517 * genopinit.c (main): Add supports_vec_scatter_store and
21518 supports_vec_scatter_store_cached to target_optabs.
21519 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
21520 IFN_MASK_SCATTER_STORE.
21521 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
21522 functions.
21523 * internal-fn.h (internal_store_fn_p): Declare.
21524 (internal_fn_stored_value_index): Likewise.
21525 * internal-fn.c (scatter_store_direct): New macro.
21526 (expand_scatter_store_optab_fn): New function.
21527 (direct_scatter_store_optab_supported_p): New macro.
21528 (internal_store_fn_p): New function.
21529 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
21530 IFN_MASK_SCATTER_STORE.
21531 (internal_fn_mask_index): Likewise.
21532 (internal_fn_stored_value_index): New function.
21533 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
21534 for scatter stores.
21535 * optabs-query.h (supports_vec_scatter_store_p): Declare.
21536 * optabs-query.c (supports_vec_scatter_store_p): New function.
21537 * tree-vectorizer.h (vect_get_store_rhs): Declare.
21538 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
21539 true for scatter stores.
21540 (vect_gather_scatter_fn_p): Handle scatter stores too.
21541 (vect_check_gather_scatter): Consider using scatter stores if
21542 supports_vec_scatter_store_p.
21543 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
21544 scatter stores too.
21545 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
21546 internal_fn_stored_value_index.
21547 (check_load_store_masking): Handle scatter stores too.
21548 (vect_get_store_rhs): Make public.
21549 (vectorizable_call): Use internal_store_fn_p.
21550 (vectorizable_store): Handle scatter store internal functions.
21551 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
21552 when deciding whether the end of the group has been reached.
21553 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
21554 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
21555 (mask_scatter_store<mode>): New insns.
21556
21557 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21558 Alan Hayward <alan.hayward@arm.com>
21559 David Sherwood <david.sherwood@arm.com>
21560
21561 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
21562 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
21563 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
21564 function.
21565 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
21566 Use vect_truncate_gather_scatter_offset if we can't treat the
21567 operation as a normal gather load or scatter store.
21568 (get_group_load_store_type): Take the gather_scatter_info
21569 as argument. Try using a gather load or scatter store for
21570 single-element groups.
21571 (get_load_store_type): Update calls to get_group_load_store_type
21572 and vect_use_strided_gather_scatters_p.
21573
21574 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21575 Alan Hayward <alan.hayward@arm.com>
21576 David Sherwood <david.sherwood@arm.com>
21577
21578 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
21579 optional tree argument.
21580 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
21581 null target hooks.
21582 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
21583 but continue to use the current value as a fallback.
21584 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
21585 to compare the updates.
21586 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
21587 (get_load_store_type): Use it when handling a strided access.
21588 (vect_get_strided_load_store_ops): New function.
21589 (vect_get_data_ptr_increment): Likewise.
21590 (vectorizable_load): Handle strided gather loads. Always pass
21591 a step to vect_create_data_ref_ptr and bump_vector_ptr.
21592
21593 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21594 Alan Hayward <alan.hayward@arm.com>
21595 David Sherwood <david.sherwood@arm.com>
21596
21597 * doc/md.texi (gather_load@var{m}): Document.
21598 (mask_gather_load@var{m}): Likewise.
21599 * genopinit.c (main): Add supports_vec_gather_load and
21600 supports_vec_gather_load_cached to target_optabs.
21601 * optabs-tree.c (init_tree_optimization_optabs): Use
21602 ggc_cleared_alloc to allocate target_optabs.
21603 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
21604 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
21605 functions.
21606 * internal-fn.h (internal_load_fn_p): Declare.
21607 (internal_gather_scatter_fn_p): Likewise.
21608 (internal_fn_mask_index): Likewise.
21609 (internal_gather_scatter_fn_supported_p): Likewise.
21610 * internal-fn.c (gather_load_direct): New macro.
21611 (expand_gather_load_optab_fn): New function.
21612 (direct_gather_load_optab_supported_p): New macro.
21613 (direct_internal_fn_optab): New function.
21614 (internal_load_fn_p): Likewise.
21615 (internal_gather_scatter_fn_p): Likewise.
21616 (internal_fn_mask_index): Likewise.
21617 (internal_gather_scatter_fn_supported_p): Likewise.
21618 * optabs-query.c (supports_at_least_one_mode_p): New function.
21619 (supports_vec_gather_load_p): Likewise.
21620 * optabs-query.h (supports_vec_gather_load_p): Declare.
21621 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
21622 and memory_type field.
21623 (NUM_PATTERNS): Bump to 15.
21624 * tree-vect-data-refs.c: Include internal-fn.h.
21625 (vect_gather_scatter_fn_p): New function.
21626 (vect_describe_gather_scatter_call): Likewise.
21627 (vect_check_gather_scatter): Try using internal functions for
21628 gather loads. Recognize existing calls to a gather load function.
21629 (vect_analyze_data_refs): Consider using gather loads if
21630 supports_vec_gather_load_p.
21631 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
21632 (vect_get_gather_scatter_offset_type): Likewise.
21633 (vect_convert_mask_for_vectype): Likewise.
21634 (vect_add_conversion_to_patterm): Likewise.
21635 (vect_try_gather_scatter_pattern): Likewise.
21636 (vect_recog_gather_scatter_pattern): New pattern recognizer.
21637 (vect_vect_recog_func_ptrs): Add it.
21638 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
21639 internal_fn_mask_index and internal_gather_scatter_fn_p.
21640 (check_load_store_masking): Take the gather_scatter_info as an
21641 argument and handle gather loads.
21642 (vect_get_gather_scatter_ops): New function.
21643 (vectorizable_call): Check internal_load_fn_p.
21644 (vectorizable_load): Likewise. Handle gather load internal
21645 functions.
21646 (vectorizable_store): Update call to check_load_store_masking.
21647 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
21648 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
21649 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
21650 (aarch64_gather_scale_operand_d): New predicates.
21651 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
21652 (mask_gather_load<mode>): New insns.
21653
21654 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21655 Alan Hayward <alan.hayward@arm.com>
21656 David Sherwood <david.sherwood@arm.com>
21657
21658 * optabs.def (fold_left_plus_optab): New optab.
21659 * doc/md.texi (fold_left_plus_@var{m}): Document.
21660 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
21661 * internal-fn.c (fold_left_direct): Define.
21662 (expand_fold_left_optab_fn): Likewise.
21663 (direct_fold_left_optab_supported_p): Likewise.
21664 * fold-const-call.c (fold_const_fold_left): New function.
21665 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
21666 * tree-parloops.c (valid_reduction_p): New function.
21667 (gather_scalar_reductions): Use it.
21668 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
21669 (vect_finish_replace_stmt): Declare.
21670 * tree-vect-loop.c (fold_left_reduction_fn): New function.
21671 (needs_fold_left_reduction_p): New function, split out from...
21672 (vect_is_simple_reduction): ...here. Accept reductions that
21673 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
21674 (vect_force_simple_reduction): Also store the reduction type in
21675 the assignment's STMT_VINFO_REDUC_TYPE.
21676 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
21677 (merge_with_identity): New function.
21678 (vect_expand_fold_left): Likewise.
21679 (vectorize_fold_left_reduction): Likewise.
21680 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
21681 scalar phi in place for it. Check for target support and reject
21682 cases that would reassociate the operation. Defer the transform
21683 phase to vectorize_fold_left_reduction.
21684 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
21685 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
21686 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
21687
21688 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21689
21690 * tree-if-conv.c (predicate_mem_writes): Remove redundant
21691 call to ifc_temp_var.
21692
21693 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21694 Alan Hayward <alan.hayward@arm.com>
21695 David Sherwood <david.sherwood@arm.com>
21696
21697 * target.def (legitimize_address_displacement): Take the original
21698 offset as a poly_int.
21699 * targhooks.h (default_legitimize_address_displacement): Update
21700 accordingly.
21701 * targhooks.c (default_legitimize_address_displacement): Likewise.
21702 * doc/tm.texi: Regenerate.
21703 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
21704 as an argument, moving assert of ad->disp == ad->disp_term to...
21705 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
21706 Try calling targetm.legitimize_address_displacement before expanding
21707 the address rather than afterwards, and adjust for the new interface.
21708 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
21709 Match the new hook interface. Handle SVE addresses.
21710 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
21711 new hook interface.
21712
21713 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21714
21715 * Makefile.in (OBJS): Add early-remat.o.
21716 * target.def (select_early_remat_modes): New hook.
21717 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
21718 * doc/tm.texi: Regenerate.
21719 * targhooks.h (default_select_early_remat_modes): Declare.
21720 * targhooks.c (default_select_early_remat_modes): New function.
21721 * timevar.def (TV_EARLY_REMAT): New timevar.
21722 * passes.def (pass_early_remat): New pass.
21723 * tree-pass.h (make_pass_early_remat): Declare.
21724 * early-remat.c: New file.
21725 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
21726 function.
21727 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
21728
21729 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21730 Alan Hayward <alan.hayward@arm.com>
21731 David Sherwood <david.sherwood@arm.com>
21732
21733 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
21734 vfm1 with a bound_epilog parameter.
21735 (vect_do_peeling): Update calls accordingly, and move the prologue
21736 call earlier in the function. Treat the base bound_epilog as 0 for
21737 fully-masked loops and retain vf - 1 for other loops. Add 1 to
21738 this base when peeling for gaps.
21739 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
21740 with fully-masked loops.
21741 (vect_estimate_min_profitable_iters): Handle the single peeled
21742 iteration in that case.
21743
21744 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21745 Alan Hayward <alan.hayward@arm.com>
21746 David Sherwood <david.sherwood@arm.com>
21747
21748 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
21749 single-element interleaving even if the size is not a power of 2.
21750 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
21751 accesses for single-element interleaving if the group size is
21752 not a power of 2.
21753
21754 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21755 Alan Hayward <alan.hayward@arm.com>
21756 David Sherwood <david.sherwood@arm.com>
21757
21758 * doc/md.texi (fold_extract_last_@var{m}): Document.
21759 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
21760 * optabs.def (fold_extract_last_optab): New optab.
21761 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
21762 * internal-fn.c (fold_extract_direct): New macro.
21763 (expand_fold_extract_optab_fn): Likewise.
21764 (direct_fold_extract_optab_supported_p): Likewise.
21765 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
21766 * tree-vect-loop.c (vect_model_reduction_cost): Handle
21767 EXTRACT_LAST_REDUCTION.
21768 (get_initial_def_for_reduction): Do not create an initial vector
21769 for EXTRACT_LAST_REDUCTION reductions.
21770 (vectorizable_reduction): Leave the scalar phi in place for
21771 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
21772 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
21773 epilogue code for EXTRACT_LAST_REDUCTION and defer the
21774 transform phase to vectorizable_condition.
21775 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
21776 split out from...
21777 (vect_finish_stmt_generation): ...here.
21778 (vect_finish_replace_stmt): New function.
21779 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
21780 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
21781 pattern.
21782 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
21783
21784 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21785 Alan Hayward <alan.hayward@arm.com>
21786 David Sherwood <david.sherwood@arm.com>
21787
21788 * doc/md.texi (extract_last_@var{m}): Document.
21789 * optabs.def (extract_last_optab): New optab.
21790 * internal-fn.def (EXTRACT_LAST): New internal function.
21791 * internal-fn.c (cond_unary_direct): New macro.
21792 (expand_cond_unary_optab_fn): Likewise.
21793 (direct_cond_unary_optab_supported_p): Likewise.
21794 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
21795 loops using EXTRACT_LAST.
21796 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
21797 (extract_last_<mode>): ...this optab.
21798 (vec_extract<mode><Vel>): Update accordingly.
21799
21800 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21801 Alan Hayward <alan.hayward@arm.com>
21802 David Sherwood <david.sherwood@arm.com>
21803
21804 * target.def (empty_mask_is_expensive): New hook.
21805 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
21806 * doc/tm.texi: Regenerate.
21807 * targhooks.h (default_empty_mask_is_expensive): Declare.
21808 * targhooks.c (default_empty_mask_is_expensive): New function.
21809 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
21810 if the target says that empty masks are expensive.
21811 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
21812 New function.
21813 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
21814
21815 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21816 Alan Hayward <alan.hayward@arm.com>
21817 David Sherwood <david.sherwood@arm.com>
21818
21819 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
21820 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
21821 (vect_use_loop_mask_for_alignment_p): New function.
21822 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
21823 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
21824 niters_skip argument. Make sure that the first niters_skip elements
21825 of the first iteration are inactive.
21826 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
21827 Update call to vect_set_loop_masks_directly.
21828 (get_misalign_in_elems): New function, split out from...
21829 (vect_gen_prolog_loop_niters): ...here.
21830 (vect_update_init_of_dr): Take a code argument that specifies whether
21831 the adjustment should be added or subtracted.
21832 (vect_update_init_of_drs): Likewise.
21833 (vect_prepare_for_masked_peels): New function.
21834 (vect_do_peeling): Skip prologue peeling if we're using a mask
21835 instead. Update call to vect_update_inits_of_drs.
21836 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
21837 mask_skip_niters.
21838 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
21839 alignment. Do not include the number of peeled iterations in
21840 the minimum threshold in that case.
21841 (vectorizable_induction): Adjust the start value down by
21842 LOOP_VINFO_MASK_SKIP_NITERS iterations.
21843 (vect_transform_loop): Call vect_prepare_for_masked_peels.
21844 Take the number of skipped iterations into account when calculating
21845 the loop bounds.
21846 * tree-vect-stmts.c (vect_gen_while_not): New function.
21847
21848 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21849 Alan Hayward <alan.hayward@arm.com>
21850 David Sherwood <david.sherwood@arm.com>
21851
21852 * doc/sourcebuild.texi (vect_fully_masked): Document.
21853 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
21854 default value to 0.
21855 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
21856 split out from...
21857 (vect_analyze_loop_2): ...here. Don't check the vectorization
21858 factor against the number of loop iterations if the loop is
21859 fully-masked.
21860
21861 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21862 Alan Hayward <alan.hayward@arm.com>
21863 David Sherwood <david.sherwood@arm.com>
21864
21865 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
21866 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
21867 (dump_groups): Update accordingly.
21868 (iv_use::mem_type): New member variable.
21869 (address_p): New function.
21870 (record_use): Add a mem_type argument and initialize the new
21871 mem_type field.
21872 (record_group_use): Add a mem_type argument. Use address_p.
21873 Remove obsolete null checks of base_object. Update call to record_use.
21874 (find_interesting_uses_op): Update call to record_group_use.
21875 (find_interesting_uses_cond): Likewise.
21876 (find_interesting_uses_address): Likewise.
21877 (get_mem_type_for_internal_fn): New function.
21878 (find_address_like_use): Likewise.
21879 (find_interesting_uses_stmt): Try find_address_like_use before
21880 calling find_interesting_uses_op.
21881 (addr_offset_valid_p): Use the iv mem_type field as the type
21882 of the addressed memory.
21883 (add_autoinc_candidates): Likewise.
21884 (get_address_cost): Likewise.
21885 (split_small_address_groups_p): Use address_p.
21886 (split_address_groups): Likewise.
21887 (add_iv_candidate_for_use): Likewise.
21888 (autoinc_possible_for_pair): Likewise.
21889 (rewrite_groups): Likewise.
21890 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
21891 (determine_group_iv_cost): Update after split of USE_ADDRESS.
21892 (get_alias_ptr_type_for_ptr_address): New function.
21893 (rewrite_use_address): Rewrite address uses in calls that were
21894 identified by find_address_like_use.
21895
21896 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21897 Alan Hayward <alan.hayward@arm.com>
21898 David Sherwood <david.sherwood@arm.com>
21899
21900 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
21901 TARGET_MEM_REFs.
21902 * gimple-expr.h (is_gimple_addressable: Likewise.
21903 * gimple-expr.c (is_gimple_address): Likewise.
21904 * internal-fn.c (expand_call_mem_ref): New function.
21905 (expand_mask_load_optab_fn): Use it.
21906 (expand_mask_store_optab_fn): Likewise.
21907
21908 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21909 Alan Hayward <alan.hayward@arm.com>
21910 David Sherwood <david.sherwood@arm.com>
21911
21912 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
21913 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
21914 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
21915 (cond_umax@var{mode}): Document.
21916 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
21917 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
21918 (cond_umin_optab, cond_umax_optab): New optabs.
21919 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
21920 (COND_IOR, COND_XOR): New internal functions.
21921 * internal-fn.h (get_conditional_internal_fn): Declare.
21922 * internal-fn.c (cond_binary_direct): New macro.
21923 (expand_cond_binary_optab_fn): Likewise.
21924 (direct_cond_binary_optab_supported_p): Likewise.
21925 (get_conditional_internal_fn): New function.
21926 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
21927 Cope with reduction statements that are vectorized as calls rather
21928 than assignments.
21929 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
21930 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
21931 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
21932 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
21933 (UNSPEC_COND_EOR): New unspecs.
21934 (optab): Add mappings for them.
21935 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
21936 (sve_int_op, sve_fp_op): New int attributes.
21937
21938 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
21939 Alan Hayward <alan.hayward@arm.com>
21940 David Sherwood <david.sherwood@arm.com>
21941
21942 * optabs.def (while_ult_optab): New optab.
21943 * doc/md.texi (while_ult@var{m}@var{n}): Document.
21944 * internal-fn.def (WHILE_ULT): New internal function.
21945 * internal-fn.h (direct_internal_fn_supported_p): New override
21946 that takes two types as argument.
21947 * internal-fn.c (while_direct): New macro.
21948 (expand_while_optab_fn): New function.
21949 (convert_optab_supported_p): Likewise.
21950 (direct_while_optab_supported_p): New macro.
21951 * wide-int.h (wi::udiv_ceil): New function.
21952 * tree-vectorizer.h (rgroup_masks): New structure.
21953 (vec_loop_masks): New typedef.
21954 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
21955 and fully_masked_p.
21956 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
21957 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
21958 (vect_max_vf): New function.
21959 (slpeel_make_loop_iterate_ntimes): Delete.
21960 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
21961 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
21962 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
21963 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
21964 internal-fn.h, stor-layout.h and optabs-query.h.
21965 (vect_set_loop_mask): New function.
21966 (add_preheader_seq): Likewise.
21967 (add_header_seq): Likewise.
21968 (interleave_supported_p): Likewise.
21969 (vect_maybe_permute_loop_masks): Likewise.
21970 (vect_set_loop_masks_directly): Likewise.
21971 (vect_set_loop_condition_masked): Likewise.
21972 (vect_set_loop_condition_unmasked): New function, split out from
21973 slpeel_make_loop_iterate_ntimes.
21974 (slpeel_make_loop_iterate_ntimes): Rename to..
21975 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
21976 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
21977 (vect_do_peeling): Update call accordingly.
21978 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
21979 loops.
21980 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
21981 mask_compare_type, can_fully_mask_p and fully_masked_p.
21982 (release_vec_loop_masks): New function.
21983 (_loop_vec_info): Use it to free the loop masks.
21984 (can_produce_all_loop_masks_p): New function.
21985 (vect_get_max_nscalars_per_iter): Likewise.
21986 (vect_verify_full_masking): Likewise.
21987 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
21988 retries, and free the mask rgroups before retrying. Check loop-wide
21989 reasons for disallowing fully-masked loops. Make the final decision
21990 about whether use a fully-masked loop or not.
21991 (vect_estimate_min_profitable_iters): Do not assume that peeling
21992 for the number of iterations will be needed for fully-masked loops.
21993 (vectorizable_reduction): Disable fully-masked loops.
21994 (vectorizable_live_operation): Likewise.
21995 (vect_halve_mask_nunits): New function.
21996 (vect_double_mask_nunits): Likewise.
21997 (vect_record_loop_mask): Likewise.
21998 (vect_get_loop_mask): Likewise.
21999 (vect_transform_loop): Handle the case in which the final loop
22000 iteration might handle a partial vector. Call vect_set_loop_condition
22001 instead of slpeel_make_loop_iterate_ntimes.
22002 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
22003 (check_load_store_masking): New function.
22004 (prepare_load_store_mask): Likewise.
22005 (vectorizable_store): Handle fully-masked loops.
22006 (vectorizable_load): Likewise.
22007 (supportable_widening_operation): Use vect_halve_mask_nunits for
22008 booleans.
22009 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
22010 (vect_gen_while): New function.
22011 * config/aarch64/aarch64.md (umax<mode>3): New expander.
22012 (aarch64_uqdec<mode>): New insn.
22013
22014 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22015 Alan Hayward <alan.hayward@arm.com>
22016 David Sherwood <david.sherwood@arm.com>
22017
22018 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
22019 (reduc_xor_scal_optab): New optabs.
22020 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
22021 (reduc_xor_scal_@var{m}): Document.
22022 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
22023 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
22024 internal functions.
22025 * fold-const-call.c (fold_const_call): Handle them.
22026 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
22027 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
22028 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
22029 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
22030 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
22031 (UNSPEC_XORV): New unspecs.
22032 (optab): Add entries for them.
22033 (BITWISEV): New int iterator.
22034 (bit_reduc_op): New int attributes.
22035
22036 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22037 Alan Hayward <alan.hayward@arm.com>
22038 David Sherwood <david.sherwood@arm.com>
22039
22040 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
22041 * internal-fn.def (VEC_SHL_INSERT): New internal function.
22042 * optabs.def (vec_shl_insert_optab): New optab.
22043 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
22044 (duplicate_and_interleave): Likewise.
22045 * tree-vect-loop.c: Include internal-fn.h.
22046 (neutral_op_for_slp_reduction): New function, split out from
22047 get_initial_defs_for_reduction.
22048 (get_initial_def_for_reduction): Handle option 2 for variable-length
22049 vectors by loading the neutral value into a vector and then shifting
22050 the initial value into element 0.
22051 (get_initial_defs_for_reduction): Replace the code argument with
22052 the neutral value calculated by neutral_op_for_slp_reduction.
22053 Use gimple_build_vector for constant-length vectors.
22054 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
22055 but the first group_size elements have a neutral value.
22056 Use duplicate_and_interleave otherwise.
22057 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
22058 Update call to get_initial_defs_for_reduction. Handle SLP
22059 reductions for variable-length vectors by creating one vector
22060 result for each scalar result, with the elements associated
22061 with other scalar results stubbed out with the neutral value.
22062 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
22063 Require IFN_VEC_SHL_INSERT for double reductions on
22064 variable-length vectors, or SLP reductions that have
22065 a neutral value. Require can_duplicate_and_interleave_p
22066 support for variable-length unchained SLP reductions if there
22067 is no neutral value, such as for MIN/MAX reductions. Also require
22068 the number of vector elements to be a multiple of the number of
22069 SLP statements when doing variable-length unchained SLP reductions.
22070 Update call to vect_create_epilog_for_reduction.
22071 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
22072 and remove initial values.
22073 (duplicate_and_interleave): Make public.
22074 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
22075 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
22076
22077 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22078 Alan Hayward <alan.hayward@arm.com>
22079 David Sherwood <david.sherwood@arm.com>
22080
22081 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
22082 (can_duplicate_and_interleave_p): New function.
22083 (vect_get_and_check_slp_defs): Take the vector of statements
22084 rather than just the current one. Remove excess parentheses.
22085 Restriction rejectinon of vect_constant_def and vect_external_def
22086 for variable-length vectors to boolean types, or types for which
22087 can_duplicate_and_interleave_p is false.
22088 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
22089 (duplicate_and_interleave): New function.
22090 (vect_get_constant_vectors): Use gimple_build_vector for
22091 constant-length vectors and suitable variable-length constant
22092 vectors. Use duplicate_and_interleave for other variable-length
22093 vectors. Don't defer the update when inserting new statements.
22094
22095 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22096 Alan Hayward <alan.hayward@arm.com>
22097 David Sherwood <david.sherwood@arm.com>
22098
22099 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
22100 min_profitable_iters doesn't go negative.
22101
22102 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22103 Alan Hayward <alan.hayward@arm.com>
22104 David Sherwood <david.sherwood@arm.com>
22105
22106 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
22107 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
22108 * optabs.def (vec_mask_load_lanes_optab): New optab.
22109 (vec_mask_store_lanes_optab): Likewise.
22110 * internal-fn.def (MASK_LOAD_LANES): New internal function.
22111 (MASK_STORE_LANES): Likewise.
22112 * internal-fn.c (mask_load_lanes_direct): New macro.
22113 (mask_store_lanes_direct): Likewise.
22114 (expand_mask_load_optab_fn): Handle masked operations.
22115 (expand_mask_load_lanes_optab_fn): New macro.
22116 (expand_mask_store_optab_fn): Handle masked operations.
22117 (expand_mask_store_lanes_optab_fn): New macro.
22118 (direct_mask_load_lanes_optab_supported_p): Likewise.
22119 (direct_mask_store_lanes_optab_supported_p): Likewise.
22120 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
22121 parameter.
22122 (vect_load_lanes_supported): Likewise.
22123 * tree-vect-data-refs.c (strip_conversion): New function.
22124 (can_group_stmts_p): Likewise.
22125 (vect_analyze_data_ref_accesses): Use it instead of checking
22126 for a pair of assignments.
22127 (vect_store_lanes_supported): Take a masked_p parameter.
22128 (vect_load_lanes_supported): Likewise.
22129 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
22130 vect_store_lanes_supported and vect_load_lanes_supported.
22131 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
22132 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
22133 parameter. Don't allow gaps for masked accesses.
22134 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
22135 and vect_load_lanes_supported.
22136 (get_load_store_type): Take a masked_p parameter and update
22137 call to get_group_load_store_type.
22138 (vectorizable_store): Update call to get_load_store_type.
22139 Handle IFN_MASK_STORE_LANES.
22140 (vectorizable_load): Update call to get_load_store_type.
22141 Handle IFN_MASK_LOAD_LANES.
22142
22143 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22144 Alan Hayward <alan.hayward@arm.com>
22145 David Sherwood <david.sherwood@arm.com>
22146
22147 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
22148 modes for SVE.
22149 * config/aarch64/aarch64-protos.h
22150 (aarch64_sve_struct_memory_operand_p): Declare.
22151 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
22152 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
22153 (VPRED, vpred): Handle SVE structure modes.
22154 * config/aarch64/constraints.md (Utx): New constraint.
22155 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
22156 (aarch64_sve_struct_nonimmediate_operand): New predicates.
22157 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
22158 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
22159 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
22160 structure modes. Split into pieces after RA.
22161 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
22162 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
22163 New patterns.
22164 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
22165 SVE structure modes.
22166 (aarch64_classify_address): Likewise.
22167 (sizetochar): Move earlier in file.
22168 (aarch64_print_operand): Handle SVE register lists.
22169 (aarch64_array_mode): New function.
22170 (aarch64_sve_struct_memory_operand_p): Likewise.
22171 (TARGET_ARRAY_MODE): Redefine.
22172
22173 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22174 Alan Hayward <alan.hayward@arm.com>
22175 David Sherwood <david.sherwood@arm.com>
22176
22177 * target.def (array_mode): New target hook.
22178 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
22179 * doc/tm.texi: Regenerate.
22180 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
22181 * hooks.c (hook_optmode_mode_uhwi_none): New function.
22182 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
22183 targetm.array_mode.
22184 * stor-layout.c (mode_for_array): Likewise. Support polynomial
22185 type sizes.
22186
22187 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22188 Alan Hayward <alan.hayward@arm.com>
22189 David Sherwood <david.sherwood@arm.com>
22190
22191 * fold-const.c (fold_binary_loc): Check the argument types
22192 rather than the result type when testing for a vector operation.
22193
22194 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22195
22196 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
22197 * doc/tm.texi: Regenerate.
22198
22199 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22200 Alan Hayward <alan.hayward@arm.com>
22201 David Sherwood <david.sherwood@arm.com>
22202
22203 * doc/invoke.texi (-msve-vector-bits=): Document new option.
22204 (sve): Document new AArch64 extension.
22205 * doc/md.texi (w): Extend the description of the AArch64
22206 constraint to include SVE vectors.
22207 (Upl, Upa): Document new AArch64 predicate constraints.
22208 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
22209 enum.
22210 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
22211 (msve-vector-bits=): New option.
22212 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
22213 SVE when these are disabled.
22214 (sve): New extension.
22215 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
22216 modes. Adjust their number of units based on aarch64_sve_vg.
22217 (MAX_BITSIZE_MODE_ANY_MODE): Define.
22218 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
22219 aarch64_addr_query_type.
22220 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
22221 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
22222 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
22223 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
22224 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
22225 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
22226 (aarch64_simd_imm_zero_p): Delete.
22227 (aarch64_check_zero_based_sve_index_immediate): Declare.
22228 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22229 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22230 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22231 (aarch64_sve_float_mul_immediate_p): Likewise.
22232 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22233 rather than an rtx.
22234 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
22235 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
22236 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
22237 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
22238 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
22239 (aarch64_regmode_natural_size): Likewise.
22240 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
22241 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
22242 left one place.
22243 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
22244 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
22245 for VG and the SVE predicate registers.
22246 (V_ALIASES): Add a "z"-prefixed alias.
22247 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
22248 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
22249 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
22250 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
22251 (REG_CLASS_NAMES): Add entries for them.
22252 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
22253 and the predicate registers.
22254 (aarch64_sve_vg): Declare.
22255 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
22256 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
22257 (REGMODE_NATURAL_SIZE): Define.
22258 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
22259 SVE macros.
22260 * config/aarch64/aarch64.c: Include cfgrtl.h.
22261 (simd_immediate_info): Add a constructor for series vectors,
22262 and an associated step field.
22263 (aarch64_sve_vg): New variable.
22264 (aarch64_dbx_register_number): Handle VG and the predicate registers.
22265 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
22266 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
22267 (VEC_ANY_DATA, VEC_STRUCT): New constants.
22268 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
22269 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
22270 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
22271 (aarch64_get_mask_mode): New functions.
22272 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
22273 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
22274 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
22275 predicate modes and predicate registers. Explicitly restrict
22276 GPRs to modes of 16 bytes or smaller. Only allow FP registers
22277 to store a vector mode if it is recognized by
22278 aarch64_classify_vector_mode.
22279 (aarch64_regmode_natural_size): New function.
22280 (aarch64_hard_regno_caller_save_mode): Return the original mode
22281 for predicates.
22282 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
22283 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
22284 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
22285 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
22286 functions.
22287 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
22288 does not overlap dest if the function is frame-related. Handle
22289 SVE constants.
22290 (aarch64_split_add_offset): New function.
22291 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
22292 them aarch64_add_offset.
22293 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
22294 and update call to aarch64_sub_sp.
22295 (aarch64_add_cfa_expression): New function.
22296 (aarch64_expand_prologue): Pass extra temporary registers to the
22297 functions above. Handle the case in which we need to emit new
22298 DW_CFA_expressions for registers that were originally saved
22299 relative to the stack pointer, but now have to be expressed
22300 relative to the frame pointer.
22301 (aarch64_output_mi_thunk): Pass extra temporary registers to the
22302 functions above.
22303 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
22304 IP0 and IP1 values for SVE frames.
22305 (aarch64_expand_vec_series): New function.
22306 (aarch64_expand_sve_widened_duplicate): Likewise.
22307 (aarch64_expand_sve_const_vector): Likewise.
22308 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
22309 Handle SVE constants. Use emit_move_insn to move a force_const_mem
22310 into the register, rather than emitting a SET directly.
22311 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
22312 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
22313 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
22314 (offset_9bit_signed_scaled_p): New functions.
22315 (aarch64_replicate_bitmask_imm): New function.
22316 (aarch64_bitmask_imm): Use it.
22317 (aarch64_cannot_force_const_mem): Reject expressions involving
22318 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
22319 (aarch64_classify_index): Handle SVE indices, by requiring
22320 a plain register index with a scale that matches the element size.
22321 (aarch64_classify_address): Handle SVE addresses. Assert that
22322 the mode of the address is VOIDmode or an integer mode.
22323 Update call to aarch64_classify_symbol.
22324 (aarch64_classify_symbolic_expression): Update call to
22325 aarch64_classify_symbol.
22326 (aarch64_const_vec_all_in_range_p): New function.
22327 (aarch64_print_vector_float_operand): Likewise.
22328 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
22329 "vN" for FP registers with SVE modes. Handle (const ...) vectors
22330 and the FP immediates 1.0 and 0.5.
22331 (aarch64_print_address_internal): Handle SVE addresses.
22332 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
22333 (aarch64_regno_regclass): Handle predicate registers.
22334 (aarch64_secondary_reload): Handle big-endian reloads of SVE
22335 data modes.
22336 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
22337 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
22338 (aarch64_convert_sve_vector_bits): New function.
22339 (aarch64_override_options): Use it to handle -msve-vector-bits=.
22340 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
22341 rather than an rtx.
22342 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
22343 Handle SVE vector and predicate modes. Accept VL-based constants
22344 that need only one temporary register, and VL offsets that require
22345 no temporary registers.
22346 (aarch64_conditional_register_usage): Mark the predicate registers
22347 as fixed if SVE isn't available.
22348 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
22349 Return true for SVE vector and predicate modes.
22350 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
22351 rather than an unsigned int. Handle SVE modes.
22352 (aarch64_preferred_simd_mode): Update call accordingly. Handle
22353 SVE modes.
22354 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
22355 if SVE is enabled.
22356 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
22357 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
22358 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
22359 (aarch64_sve_float_mul_immediate_p): New functions.
22360 (aarch64_sve_valid_immediate): New function.
22361 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
22362 Explicitly reject structure modes. Check for INDEX constants.
22363 Handle PTRUE and PFALSE constants.
22364 (aarch64_check_zero_based_sve_index_immediate): New function.
22365 (aarch64_simd_imm_zero_p): Delete.
22366 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
22367 vector modes. Accept constants in the range of CNT[BHWD].
22368 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
22369 ask for an Advanced SIMD mode.
22370 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
22371 (aarch64_simd_vector_alignment): Handle SVE predicates.
22372 (aarch64_vectorize_preferred_vector_alignment): New function.
22373 (aarch64_simd_vector_alignment_reachable): Use it instead of
22374 the vector size.
22375 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
22376 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
22377 functions.
22378 (MAX_VECT_LEN): Delete.
22379 (expand_vec_perm_d): Add a vec_flags field.
22380 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
22381 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
22382 (aarch64_evpc_ext): Don't apply a big-endian lane correction
22383 for SVE modes.
22384 (aarch64_evpc_rev): Rename to...
22385 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
22386 (aarch64_evpc_rev_global): New function.
22387 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
22388 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
22389 MAX_VECT_LEN.
22390 (aarch64_evpc_sve_tbl): New function.
22391 (aarch64_expand_vec_perm_const_1): Update after rename of
22392 aarch64_evpc_rev. Handle SVE permutes too, trying
22393 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
22394 than aarch64_evpc_tbl.
22395 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
22396 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
22397 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
22398 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
22399 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
22400 (aarch64_expand_sve_vcond): New functions.
22401 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
22402 of aarch64_vector_mode_p.
22403 (aarch64_dwarf_poly_indeterminate_value): New function.
22404 (aarch64_compute_pressure_classes): Likewise.
22405 (aarch64_can_change_mode_class): Likewise.
22406 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
22407 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
22408 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
22409 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
22410 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
22411 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
22412 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
22413 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
22414 constraints.
22415 (Dn, Dl, Dr): Accept const as well as const_vector.
22416 (Dz): Likewise. Compare against CONST0_RTX.
22417 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
22418 of "vector" where appropriate.
22419 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
22420 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
22421 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
22422 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
22423 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
22424 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
22425 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
22426 (v_int_equiv): Extend to SVE modes.
22427 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
22428 mode attributes.
22429 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
22430 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
22431 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
22432 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
22433 (SVE_COND_FP_CMP): New int iterators.
22434 (perm_hilo): Handle the new unpack unspecs.
22435 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
22436 attributes.
22437 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
22438 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
22439 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
22440 (aarch64_equality_operator, aarch64_constant_vector_operand)
22441 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
22442 (aarch64_sve_nonimmediate_operand): Likewise.
22443 (aarch64_sve_general_operand): Likewise.
22444 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
22445 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
22446 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
22447 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
22448 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
22449 (aarch64_sve_float_arith_immediate): Likewise.
22450 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
22451 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
22452 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
22453 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
22454 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
22455 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
22456 (aarch64_sve_float_arith_operand): Likewise.
22457 (aarch64_sve_float_arith_with_sub_operand): Likewise.
22458 (aarch64_sve_float_mul_operand): Likewise.
22459 (aarch64_sve_vec_perm_operand): Likewise.
22460 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
22461 (aarch64_mov_operand): Accept const_poly_int and const_vector.
22462 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
22463 as well as const_vector.
22464 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
22465 in file. Use CONST0_RTX and CONSTM1_RTX.
22466 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
22467 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
22468 Use aarch64_simd_imm_zero.
22469 * config/aarch64/aarch64-sve.md: New file.
22470 * config/aarch64/aarch64.md: Include it.
22471 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
22472 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
22473 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
22474 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
22475 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
22476 (sve): New attribute.
22477 (enabled): Disable instructions with the sve attribute unless
22478 TARGET_SVE.
22479 (movqi, movhi): Pass CONST_POLY_INT operaneds through
22480 aarch64_expand_mov_immediate.
22481 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
22482 CNT[BHSD] immediates.
22483 (movti): Split CONST_POLY_INT moves into two halves.
22484 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
22485 Split additions that need a temporary here if the destination
22486 is the stack pointer.
22487 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
22488 (*add<mode>3_poly_1): New instruction.
22489 (set_clobber_cc): New expander.
22490
22491 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22492
22493 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
22494 parameter and use it instead of GET_MODE_SIZE (innermode). Use
22495 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
22496 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
22497 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
22498 Change innermode from fixed_mode_size to machine_mode.
22499 (simplify_subreg): Update call accordingly. Handle a constant-sized
22500 subreg of a variable-length CONST_VECTOR.
22501
22502 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
22503 Alan Hayward <alan.hayward@arm.com>
22504 David Sherwood <david.sherwood@arm.com>
22505
22506 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
22507 (add_offset_to_base): New function, split out from...
22508 (create_mem_ref): ...here. When handling a scale other than 1,
22509 check first whether the address is valid without the offset.
22510 Add it into the base if so, leaving the index and scale as-is.
22511
22512 2018-01-12 Jakub Jelinek <jakub@redhat.com>
22513
22514 PR c++/83778
22515 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
22516 fold_for_warn before checking if arg2 is INTEGER_CST.
22517
22518 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
22519
22520 * config/rs6000/predicates.md (load_multiple_operation): Delete.
22521 (store_multiple_operation): Delete.
22522 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
22523 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
22524 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
22525 guarded by TARGET_STRING.
22526 (rs6000_output_load_multiple): Delete.
22527 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
22528 OPTION_MASK_STRING / TARGET_STRING handling.
22529 (print_operand) <'N', 'O'>: Add comment that these are unused now.
22530 (const rs6000_opt_masks) <"string">: Change mask to 0.
22531 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
22532 (MASK_STRING): Delete.
22533 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
22534 parts. Simplify.
22535 (load_multiple): Delete.
22536 (*ldmsi8): Delete.
22537 (*ldmsi7): Delete.
22538 (*ldmsi6): Delete.
22539 (*ldmsi5): Delete.
22540 (*ldmsi4): Delete.
22541 (*ldmsi3): Delete.
22542 (store_multiple): Delete.
22543 (*stmsi8): Delete.
22544 (*stmsi7): Delete.
22545 (*stmsi6): Delete.
22546 (*stmsi5): Delete.
22547 (*stmsi4): Delete.
22548 (*stmsi3): Delete.
22549 (movmemsi_8reg): Delete.
22550 (corresponding unnamed define_insn): Delete.
22551 (movmemsi_6reg): Delete.
22552 (corresponding unnamed define_insn): Delete.
22553 (movmemsi_4reg): Delete.
22554 (corresponding unnamed define_insn): Delete.
22555 (movmemsi_2reg): Delete.
22556 (corresponding unnamed define_insn): Delete.
22557 (movmemsi_1reg): Delete.
22558 (corresponding unnamed define_insn): Delete.
22559 * config/rs6000/rs6000.opt (mno-string): New.
22560 (mstring): Replace by deprecation warning stub.
22561 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
22562
22563 2018-01-12 Jakub Jelinek <jakub@redhat.com>
22564
22565 * regrename.c (regrename_do_replace): If replacing the same
22566 reg multiple times, try to reuse last created gen_raw_REG.
22567
22568 PR debug/81155
22569 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
22570 main to workaround a bug in GDB.
22571
22572 2018-01-12 Tom de Vries <tom@codesourcery.com>
22573
22574 PR target/83737
22575 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
22576
22577 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
22578
22579 PR rtl-optimization/80481
22580 * ira-color.c (get_cap_member): New function.
22581 (allocnos_conflict_by_live_ranges_p): Use it.
22582 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
22583 (setup_slot_coalesced_allocno_live_ranges): Ditto.
22584
22585 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
22586
22587 PR target/83628
22588 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
22589 (*saddl_se_1): Ditto.
22590 (*ssubsi_1): Ditto.
22591 (*ssubl_se_1): Ditto.
22592
22593 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22594
22595 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
22596 rather than wi::to_widest for DR_INITs.
22597 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
22598 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
22599 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
22600 INTEGER_CSTs.
22601 (vect_analyze_group_access_1): Note that here.
22602
22603 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22604
22605 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
22606 polynomial type sizes.
22607
22608 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
22609
22610 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
22611 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
22612 (gimple_add_tmp_var): Likewise.
22613
22614 2018-01-12 Martin Liska <mliska@suse.cz>
22615
22616 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
22617 (gimple_alloc_sizes): Likewise.
22618 (dump_gimple_statistics): Use PRIu64 in printf format.
22619 * gimple.h: Change uint64_t to int.
22620
22621 2018-01-12 Martin Liska <mliska@suse.cz>
22622
22623 * tree-core.h: Use uint64_t instead of int.
22624 * tree.c (tree_node_counts): Likewise.
22625 (tree_node_sizes): Likewise.
22626 (dump_tree_statistics): Use PRIu64 in printf format.
22627
22628 2018-01-12 Martin Liska <mliska@suse.cz>
22629
22630 * Makefile.in: As qsort_chk is implemented in vec.c, add
22631 vec.o to linkage of gencfn-macros.
22632 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
22633 passing the info to record_node_allocation_statistics.
22634 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
22635 and pass the info.
22636 * ggc-common.c (struct ggc_usage): Add operator== and use
22637 it in operator< and compare function.
22638 * mem-stats.h (struct mem_usage): Likewise.
22639 * vec.c (struct vec_usage): Remove operator< and compare
22640 function. Can be simply inherited.
22641
22642 2018-01-12 Martin Jambor <mjambor@suse.cz>
22643
22644 PR target/81616
22645 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
22646 * tree-ssa-math-opts.c: Include domwalk.h.
22647 (convert_mult_to_fma_1): New function.
22648 (fma_transformation_info): New type.
22649 (fma_deferring_state): Likewise.
22650 (cancel_fma_deferring): New function.
22651 (result_of_phi): Likewise.
22652 (last_fma_candidate_feeds_initial_phi): Likewise.
22653 (convert_mult_to_fma): Added deferring logic, split actual
22654 transformation to convert_mult_to_fma_1.
22655 (math_opts_dom_walker): New type.
22656 (math_opts_dom_walker::after_dom_children): New method, body moved
22657 here from pass_optimize_widening_mul::execute, added deferring logic
22658 bits.
22659 (pass_optimize_widening_mul::execute): Moved most of code to
22660 math_opts_dom_walker::after_dom_children.
22661 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
22662 * config/i386/i386.c (ix86_option_override_internal): Added
22663 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
22664
22665 2018-01-12 Richard Biener <rguenther@suse.de>
22666
22667 PR debug/83157
22668 * dwarf2out.c (gen_variable_die): Do not reset old_die for
22669 inline instance vars.
22670
22671 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
22672
22673 PR target/81819
22674 * config/rx/rx.c (rx_is_restricted_memory_address):
22675 Handle SUBREG case.
22676
22677 2018-01-12 Richard Biener <rguenther@suse.de>
22678
22679 PR tree-optimization/80846
22680 * target.def (split_reduction): New target hook.
22681 * targhooks.c (default_split_reduction): New function.
22682 * targhooks.h (default_split_reduction): Declare.
22683 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
22684 target requests first reduce vectors by combining low and high
22685 parts.
22686 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
22687 (get_vectype_for_scalar_type_and_size): Export.
22688 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
22689 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
22690 * doc/tm.texi: Regenerate.
22691 * config/i386/i386.c (ix86_split_reduction): Implement
22692 TARGET_VECTORIZE_SPLIT_REDUCTION.
22693
22694 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
22695
22696 PR target/83368
22697 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
22698 in PIC mode except for TARGET_VXWORKS_RTP.
22699 * config/sparc/sparc.c: Include cfgrtl.h.
22700 (TARGET_INIT_PIC_REG): Define.
22701 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
22702 (sparc_pic_register_p): New predicate.
22703 (sparc_legitimate_address_p): Use it.
22704 (sparc_legitimize_pic_address): Likewise.
22705 (sparc_delegitimize_address): Likewise.
22706 (sparc_mode_dependent_address_p): Likewise.
22707 (gen_load_pcrel_sym): Remove 4th parameter.
22708 (load_got_register): Adjust call to above. Remove obsolete stuff.
22709 (sparc_expand_prologue): Do not call load_got_register here.
22710 (sparc_flat_expand_prologue): Likewise.
22711 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
22712 (sparc_use_pseudo_pic_reg): New function.
22713 (sparc_init_pic_reg): Likewise.
22714 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
22715 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
22716
22717 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
22718
22719 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
22720 Add item for branch_cost.
22721
22722 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
22723
22724 PR rtl-optimization/83565
22725 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
22726 not extend the result to a larger mode for rotate operations.
22727 (num_sign_bit_copies1): Likewise.
22728
22729 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
22730
22731 PR target/40411
22732 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
22733 -symbolic.
22734 Use values-Xc.o for -pedantic.
22735 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
22736
22737 2018-01-12 Martin Liska <mliska@suse.cz>
22738
22739 PR ipa/83054
22740 * ipa-devirt.c (final_warning_record::grow_type_warnings):
22741 New function.
22742 (possible_polymorphic_call_targets): Use it.
22743 (ipa_devirt): Likewise.
22744
22745 2018-01-12 Martin Liska <mliska@suse.cz>
22746
22747 * profile-count.h (enum profile_quality): Use 0 as invalid
22748 enum value of profile_quality.
22749
22750 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
22751
22752 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
22753 -mext-string options.
22754
22755 2018-01-12 Richard Biener <rguenther@suse.de>
22756
22757 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
22758 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
22759 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
22760 Likewise.
22761 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
22762
22763 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
22764
22765 * configure.ac (--with-long-double-format): Add support for the
22766 configuration option to change the default long double format on
22767 PowerPC systems.
22768 * config.gcc (powerpc*-linux*-*): Likewise.
22769 * configure: Regenerate.
22770 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
22771 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
22772 used without modification.
22773
22774 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
22775
22776 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
22777 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
22778 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
22779 MISC_BUILTIN_SPEC_BARRIER.
22780 (rs6000_init_builtins): Likewise.
22781 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
22782 enum value.
22783 (speculation_barrier): New define_insn.
22784 * doc/extend.texi: Document __builtin_speculation_barrier.
22785
22786 2018-01-11 Jakub Jelinek <jakub@redhat.com>
22787
22788 PR target/83203
22789 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
22790 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
22791 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
22792 iterators.
22793 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
22794 integral modes instead of "ss" and "sd".
22795 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
22796 vectors with 32-bit and 64-bit elements.
22797 (vecdupssescalarmodesuffix): New mode attribute.
22798 (vec_dup<mode>): Use it.
22799
22800 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
22801
22802 PR target/83330
22803 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
22804 frame if argument is passed on stack.
22805
22806 2018-01-11 Jakub Jelinek <jakub@redhat.com>
22807
22808 PR target/82682
22809 * ree.c (combine_reaching_defs): Optimize also
22810 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
22811 reg2=any_extend(exp); reg1=reg2;, formatting fix.
22812
22813 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
22814
22815 PR middle-end/83189
22816 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
22817
22818 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
22819
22820 PR middle-end/83718
22821 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
22822 after they are computed.
22823
22824 2018-01-11 Bin Cheng <bin.cheng@arm.com>
22825
22826 PR tree-optimization/83695
22827 * gimple-loop-linterchange.cc
22828 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
22829 reset cached scev information after interchange.
22830 (pass_linterchange::execute): Remove call to scev_reset_htab.
22831
22832 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22833
22834 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
22835 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
22836 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
22837 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
22838 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
22839 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
22840 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
22841 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
22842 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
22843 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
22844 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
22845 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
22846 (V_lane_reg): Likewise.
22847 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
22848 New define_expand.
22849 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
22850 (vfmal_lane_low<mode>_intrinsic,
22851 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
22852 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
22853 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
22854 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
22855 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
22856 vfmsl_lane_high<mode>_intrinsic): New define_insns.
22857
22858 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22859
22860 * config/arm/arm-cpus.in (fp16fml): New feature.
22861 (ALL_SIMD): Add fp16fml.
22862 (armv8.2-a): Add fp16fml as an option.
22863 (armv8.3-a): Likewise.
22864 (armv8.4-a): Add fp16fml as part of fp16.
22865 * config/arm/arm.h (TARGET_FP16FML): Define.
22866 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
22867 when appropriate.
22868 * config/arm/arm-modes.def (V2HF): Define.
22869 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
22870 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
22871 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
22872 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
22873 vfmsl_low, vfmsl_high): New set of builtins.
22874 * config/arm/iterators.md (PLUSMINUS): New code iterator.
22875 (vfml_op): New code attribute.
22876 (VFMLHALVES): New int iterator.
22877 (VFML, VFMLSEL): New mode attributes.
22878 (V_reg): Define mapping for V2HF.
22879 (V_hi, V_lo): New mode attributes.
22880 (VF_constraint): Likewise.
22881 (vfml_half, vfml_half_selector): New int attributes.
22882 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
22883 define_expand.
22884 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
22885 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
22886 New define_insn.
22887 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
22888 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
22889 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
22890 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
22891 documentation.
22892 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
22893 Document new effective target and option set.
22894
22895 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22896
22897 * config/arm/arm-cpus.in (armv8_4): New feature.
22898 (ARMv8_4a): New fgroup.
22899 (armv8.4-a): New arch.
22900 * config/arm/arm-tables.opt: Regenerate.
22901 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
22902 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
22903 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
22904 Add matching rules for -march=armv8.4-a and extensions.
22905 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
22906
22907 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
22908
22909 PR target/81821
22910 * config/rx/rx.md (BW): New mode attribute.
22911 (sync_lock_test_and_setsi): Add mode suffix to insn output.
22912
22913 2018-01-11 Richard Biener <rguenther@suse.de>
22914
22915 PR tree-optimization/83435
22916 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
22917 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
22918 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
22919
22920 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22921 Alan Hayward <alan.hayward@arm.com>
22922 David Sherwood <david.sherwood@arm.com>
22923
22924 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
22925 field.
22926 (aarch64_classify_address): Initialize it. Track polynomial offsets.
22927 (aarch64_print_address_internal): Use it to check for a zero offset.
22928
22929 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
22930 Alan Hayward <alan.hayward@arm.com>
22931 David Sherwood <david.sherwood@arm.com>
22932
22933 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
22934 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
22935 Return a poly_int64 rather than a HOST_WIDE_INT.
22936 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
22937 rather than a HOST_WIDE_INT.
22938 * config/aarch64/aarch64.h (aarch64_frame): Protect with
22939 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
22940 hard_fp_offset, frame_size, initial_adjust, callee_offset and
22941 final_offset from HOST_WIDE_INT to poly_int64.
22942 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
22943 to_constant when getting the number of units in an Advanced SIMD
22944 mode.
22945 (aarch64_builtin_vectorized_function): Check for a constant number
22946 of units.
22947 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
22948 GET_MODE_SIZE.
22949 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
22950 attribute instead of GET_MODE_NUNITS.
22951 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
22952 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
22953 GET_MODE_SIZE for fixed-size registers.
22954 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
22955 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
22956 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
22957 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
22958 (aarch64_print_operand, aarch64_print_address_internal)
22959 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
22960 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
22961 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
22962 Handle polynomial GET_MODE_SIZE.
22963 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
22964 wider than SImode without modification.
22965 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
22966 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
22967 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
22968 passing and returning SVE modes.
22969 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
22970 rather than GEN_INT.
22971 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
22972 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
22973 (aarch64_allocate_and_probe_stack_space): Likewise.
22974 (aarch64_layout_frame): Cope with polynomial offsets.
22975 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
22976 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
22977 polynomial offsets.
22978 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
22979 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
22980 poly_int64 rather than a HOST_WIDE_INT.
22981 (aarch64_get_separate_components, aarch64_process_components)
22982 (aarch64_expand_prologue, aarch64_expand_epilogue)
22983 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
22984 (aarch64_anchor_offset): New function, split out from...
22985 (aarch64_legitimize_address): ...here.
22986 (aarch64_builtin_vectorization_cost): Handle polynomial
22987 TYPE_VECTOR_SUBPARTS.
22988 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
22989 GET_MODE_NUNITS.
22990 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
22991 number of elements from the PARALLEL rather than the mode.
22992 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
22993 rather than GET_MODE_BITSIZE.
22994 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
22995 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
22996 (aarch64_expand_vec_perm_const_1): Handle polynomial
22997 d->perm.length () and d->perm elements.
22998 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
22999 Apply to_constant to d->perm elements.
23000 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
23001 polynomial CONST_VECTOR_NUNITS.
23002 (aarch64_move_pointer): Take amount as a poly_int64 rather
23003 than an int.
23004 (aarch64_progress_pointer): Avoid temporary variable.
23005 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
23006 the mode attribute instead of GET_MODE.
23007
23008 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23009 Alan Hayward <alan.hayward@arm.com>
23010 David Sherwood <david.sherwood@arm.com>
23011
23012 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
23013 x exists before using it.
23014 (aarch64_add_constant_internal): Rename to...
23015 (aarch64_add_offset_1): ...this. Replace regnum with separate
23016 src and dest rtxes. Handle the case in which they're different,
23017 including when the offset is zero. Replace scratchreg with an rtx.
23018 Use 2 additions if there is no spare register into which we can
23019 move a 16-bit constant.
23020 (aarch64_add_constant): Delete.
23021 (aarch64_add_offset): Replace reg with separate src and dest
23022 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
23023 Use aarch64_add_offset_1.
23024 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
23025 an rtx rather than an int. Take the delta as a poly_int64
23026 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
23027 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
23028 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
23029 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
23030 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
23031 and aarch64_add_sp.
23032 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
23033 aarch64_add_constant.
23034
23035 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23036
23037 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
23038 Use scalar_float_mode.
23039
23040 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23041
23042 * config/aarch64/aarch64-simd.md
23043 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
23044 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
23045 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
23046 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
23047 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
23048 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
23049 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
23050 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
23051 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
23052 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
23053
23054 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23055
23056 PR target/83514
23057 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
23058 targ_options->x_arm_arch_string is non NULL.
23059
23060 2018-01-11 Tamar Christina <tamar.christina@arm.com>
23061
23062 * config/aarch64/aarch64.h
23063 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
23064
23065 2018-01-11 Sudakshina Das <sudi.das@arm.com>
23066
23067 PR target/82096
23068 * expmed.c (emit_store_flag_force): Swap if const op0
23069 and change VOIDmode to mode of op0.
23070
23071 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
23072
23073 PR rtl-optimization/83761
23074 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
23075 than bytes to mode_for_size.
23076
23077 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23078
23079 PR middle-end/83189
23080 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
23081 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
23082 profile.
23083
23084 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23085
23086 PR middle-end/83575
23087 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
23088 when in layout mode.
23089 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
23090 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
23091 partition fixup.
23092
23093 2018-01-10 Michael Collison <michael.collison@arm.com>
23094
23095 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
23096 * config/aarch64/aarch64-option-extension.def: Add
23097 AARCH64_OPT_EXTENSION of 'fp16fml'.
23098 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23099 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
23100 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
23101 * config/aarch64/constraints.md (Ui7): New constraint.
23102 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
23103 (VFMLA_SEL_W): Ditto.
23104 (f16quad): Ditto.
23105 (f16mac1): Ditto.
23106 (VFMLA16_LOW): New int iterator.
23107 (VFMLA16_HIGH): Ditto.
23108 (UNSPEC_FMLAL): New unspec.
23109 (UNSPEC_FMLSL): Ditto.
23110 (UNSPEC_FMLAL2): Ditto.
23111 (UNSPEC_FMLSL2): Ditto.
23112 (f16mac): New code attribute.
23113 * config/aarch64/aarch64-simd-builtins.def
23114 (aarch64_fmlal_lowv2sf): Ditto.
23115 (aarch64_fmlsl_lowv2sf): Ditto.
23116 (aarch64_fmlalq_lowv4sf): Ditto.
23117 (aarch64_fmlslq_lowv4sf): Ditto.
23118 (aarch64_fmlal_highv2sf): Ditto.
23119 (aarch64_fmlsl_highv2sf): Ditto.
23120 (aarch64_fmlalq_highv4sf): Ditto.
23121 (aarch64_fmlslq_highv4sf): Ditto.
23122 (aarch64_fmlal_lane_lowv2sf): Ditto.
23123 (aarch64_fmlsl_lane_lowv2sf): Ditto.
23124 (aarch64_fmlal_laneq_lowv2sf): Ditto.
23125 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
23126 (aarch64_fmlalq_lane_lowv4sf): Ditto.
23127 (aarch64_fmlsl_lane_lowv4sf): Ditto.
23128 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
23129 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
23130 (aarch64_fmlal_lane_highv2sf): Ditto.
23131 (aarch64_fmlsl_lane_highv2sf): Ditto.
23132 (aarch64_fmlal_laneq_highv2sf): Ditto.
23133 (aarch64_fmlsl_laneq_highv2sf): Ditto.
23134 (aarch64_fmlalq_lane_highv4sf): Ditto.
23135 (aarch64_fmlsl_lane_highv4sf): Ditto.
23136 (aarch64_fmlalq_laneq_highv4sf): Ditto.
23137 (aarch64_fmlsl_laneq_highv4sf): Ditto.
23138 * config/aarch64/aarch64-simd.md:
23139 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
23140 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
23141 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
23142 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
23143 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
23144 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
23145 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
23146 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
23147 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
23148 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
23149 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
23150 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
23151 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
23152 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
23153 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
23154 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
23155 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
23156 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
23157 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
23158 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
23159 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
23160 (vfmlsl_low_u32): Ditto.
23161 (vfmlalq_low_u32): Ditto.
23162 (vfmlslq_low_u32): Ditto.
23163 (vfmlal_high_u32): Ditto.
23164 (vfmlsl_high_u32): Ditto.
23165 (vfmlalq_high_u32): Ditto.
23166 (vfmlslq_high_u32): Ditto.
23167 (vfmlal_lane_low_u32): Ditto.
23168 (vfmlsl_lane_low_u32): Ditto.
23169 (vfmlal_laneq_low_u32): Ditto.
23170 (vfmlsl_laneq_low_u32): Ditto.
23171 (vfmlalq_lane_low_u32): Ditto.
23172 (vfmlslq_lane_low_u32): Ditto.
23173 (vfmlalq_laneq_low_u32): Ditto.
23174 (vfmlslq_laneq_low_u32): Ditto.
23175 (vfmlal_lane_high_u32): Ditto.
23176 (vfmlsl_lane_high_u32): Ditto.
23177 (vfmlal_laneq_high_u32): Ditto.
23178 (vfmlsl_laneq_high_u32): Ditto.
23179 (vfmlalq_lane_high_u32): Ditto.
23180 (vfmlslq_lane_high_u32): Ditto.
23181 (vfmlalq_laneq_high_u32): Ditto.
23182 (vfmlslq_laneq_high_u32): Ditto.
23183 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
23184 (AARCH64_FL_FOR_ARCH8_4): New.
23185 (AARCH64_ISA_F16FML): New ISA flag.
23186 (TARGET_F16FML): New feature flag for fp16fml.
23187 (doc/invoke.texi): Document new fp16fml option.
23188
23189 2018-01-10 Michael Collison <michael.collison@arm.com>
23190
23191 * config/aarch64/aarch64-builtins.c:
23192 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
23193 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23194 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
23195 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
23196 (AARCH64_ISA_SHA3): New ISA flag.
23197 (TARGET_SHA3): New feature flag for sha3.
23198 * config/aarch64/iterators.md (sha512_op): New int attribute.
23199 (CRYPTO_SHA512): New int iterator.
23200 (UNSPEC_SHA512H): New unspec.
23201 (UNSPEC_SHA512H2): Ditto.
23202 (UNSPEC_SHA512SU0): Ditto.
23203 (UNSPEC_SHA512SU1): Ditto.
23204 * config/aarch64/aarch64-simd-builtins.def
23205 (aarch64_crypto_sha512hqv2di): New builtin.
23206 (aarch64_crypto_sha512h2qv2di): Ditto.
23207 (aarch64_crypto_sha512su0qv2di): Ditto.
23208 (aarch64_crypto_sha512su1qv2di): Ditto.
23209 (aarch64_eor3qv8hi): Ditto.
23210 (aarch64_rax1qv2di): Ditto.
23211 (aarch64_xarqv2di): Ditto.
23212 (aarch64_bcaxqv8hi): Ditto.
23213 * config/aarch64/aarch64-simd.md:
23214 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
23215 (aarch64_crypto_sha512su0qv2di): Ditto.
23216 (aarch64_crypto_sha512su1qv2di): Ditto.
23217 (aarch64_eor3qv8hi): Ditto.
23218 (aarch64_rax1qv2di): Ditto.
23219 (aarch64_xarqv2di): Ditto.
23220 (aarch64_bcaxqv8hi): Ditto.
23221 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
23222 (vsha512h2q_u64): Ditto.
23223 (vsha512su0q_u64): Ditto.
23224 (vsha512su1q_u64): Ditto.
23225 (veor3q_u16): Ditto.
23226 (vrax1q_u64): Ditto.
23227 (vxarq_u64): Ditto.
23228 (vbcaxq_u16): Ditto.
23229 * config/arm/types.md (crypto_sha512): New type attribute.
23230 (crypto_sha3): Ditto.
23231 (doc/invoke.texi): Document new sha3 option.
23232
23233 2018-01-10 Michael Collison <michael.collison@arm.com>
23234
23235 * config/aarch64/aarch64-builtins.c:
23236 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
23237 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23238 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
23239 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
23240 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
23241 (AARCH64_ISA_SM4): New ISA flag.
23242 (TARGET_SM4): New feature flag for sm4.
23243 * config/aarch64/aarch64-simd-builtins.def
23244 (aarch64_sm3ss1qv4si): Ditto.
23245 (aarch64_sm3tt1aq4si): Ditto.
23246 (aarch64_sm3tt1bq4si): Ditto.
23247 (aarch64_sm3tt2aq4si): Ditto.
23248 (aarch64_sm3tt2bq4si): Ditto.
23249 (aarch64_sm3partw1qv4si): Ditto.
23250 (aarch64_sm3partw2qv4si): Ditto.
23251 (aarch64_sm4eqv4si): Ditto.
23252 (aarch64_sm4ekeyqv4si): Ditto.
23253 * config/aarch64/aarch64-simd.md:
23254 (aarch64_sm3ss1qv4si): Ditto.
23255 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
23256 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
23257 (aarch64_sm4eqv4si): Ditto.
23258 (aarch64_sm4ekeyqv4si): Ditto.
23259 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
23260 (sm3part_op): Ditto.
23261 (CRYPTO_SM3TT): Ditto.
23262 (CRYPTO_SM3PART): Ditto.
23263 (UNSPEC_SM3SS1): New unspec.
23264 (UNSPEC_SM3TT1A): Ditto.
23265 (UNSPEC_SM3TT1B): Ditto.
23266 (UNSPEC_SM3TT2A): Ditto.
23267 (UNSPEC_SM3TT2B): Ditto.
23268 (UNSPEC_SM3PARTW1): Ditto.
23269 (UNSPEC_SM3PARTW2): Ditto.
23270 (UNSPEC_SM4E): Ditto.
23271 (UNSPEC_SM4EKEY): Ditto.
23272 * config/aarch64/constraints.md (Ui2): New constraint.
23273 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
23274 * config/arm/types.md (crypto_sm3): New type attribute.
23275 (crypto_sm4): Ditto.
23276 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
23277 (vsm3tt1aq_u32): Ditto.
23278 (vsm3tt1bq_u32): Ditto.
23279 (vsm3tt2aq_u32): Ditto.
23280 (vsm3tt2bq_u32): Ditto.
23281 (vsm3partw1q_u32): Ditto.
23282 (vsm3partw2q_u32): Ditto.
23283 (vsm4eq_u32): Ditto.
23284 (vsm4ekeyq_u32): Ditto.
23285 (doc/invoke.texi): Document new sm4 option.
23286
23287 2018-01-10 Michael Collison <michael.collison@arm.com>
23288
23289 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
23290 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
23291 (AARCH64_FL_FOR_ARCH8_4): New.
23292 (AARCH64_FL_V8_4): New flag.
23293 (doc/invoke.texi): Document new armv8.4-a option.
23294
23295 2018-01-10 Michael Collison <michael.collison@arm.com>
23296
23297 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
23298 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
23299 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
23300 * config/aarch64/aarch64-option-extension.def: Add
23301 AARCH64_OPT_EXTENSION of 'sha2'.
23302 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
23303 (crypto): Disable sha2 and aes if crypto disabled.
23304 (crypto): Enable aes and sha2 if enabled.
23305 (simd): Disable sha2 and aes if simd disabled.
23306 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
23307 New flags.
23308 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
23309 (TARGET_SHA2): New feature flag for sha2.
23310 (TARGET_AES): New feature flag for aes.
23311 * config/aarch64/aarch64-simd.md:
23312 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
23313 conditional on TARGET_AES.
23314 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
23315 (aarch64_crypto_sha1hsi): Make pattern conditional
23316 on TARGET_SHA2.
23317 (aarch64_crypto_sha1hv4si): Ditto.
23318 (aarch64_be_crypto_sha1hv4si): Ditto.
23319 (aarch64_crypto_sha1su1v4si): Ditto.
23320 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
23321 (aarch64_crypto_sha1su0v4si): Ditto.
23322 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
23323 (aarch64_crypto_sha256su0v4si): Ditto.
23324 (aarch64_crypto_sha256su1v4si): Ditto.
23325 (doc/invoke.texi): Document new aes and sha2 options.
23326
23327 2018-01-10 Martin Sebor <msebor@redhat.com>
23328
23329 PR tree-optimization/83781
23330 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
23331 as string arrays.
23332
23333 2018-01-11 Martin Sebor <msebor@gmail.com>
23334 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
23335
23336 PR tree-optimization/83501
23337 PR tree-optimization/81703
23338
23339 * tree-ssa-strlen.c (get_string_cst): Rename...
23340 (get_string_len): ...to this. Handle global constants.
23341 (handle_char_store): Adjust.
23342
23343 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
23344 Jim Wilson <jimw@sifive.com>
23345
23346 * config/riscv/riscv-protos.h (riscv_output_return): New.
23347 * config/riscv/riscv.c (struct machine_function): New naked_p field.
23348 (riscv_attribute_table, riscv_output_return),
23349 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
23350 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
23351 (riscv_compute_frame_info): Only compute frame->mask if not a naked
23352 function.
23353 (riscv_expand_prologue): Add early return for naked function.
23354 (riscv_expand_epilogue): Likewise.
23355 (riscv_function_ok_for_sibcall): Return false for naked function.
23356 (riscv_set_current_function): New.
23357 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
23358 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
23359 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
23360 * doc/extend.texi (RISC-V Function Attributes): New.
23361
23362 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
23363
23364 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
23365 check for 128-bit long double before checking TCmode.
23366 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
23367 128-bit long doubles before checking TFmode or TCmode.
23368 (FLOAT128_IBM_P): Likewise.
23369
23370 2018-01-10 Martin Sebor <msebor@redhat.com>
23371
23372 PR tree-optimization/83671
23373 * builtins.c (c_strlen): Unconditionally return zero for the empty
23374 string.
23375 Use -Warray-bounds for warnings.
23376 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
23377 for non-constant array indices with COMPONENT_REF, arrays of
23378 arrays, and pointers to arrays.
23379 (gimple_fold_builtin_strlen): Determine and set length range for
23380 non-constant character arrays.
23381
23382 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
23383
23384 PR middle-end/81897
23385 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
23386 empty blocks.
23387
23388 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
23389
23390 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
23391
23392 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23393
23394 PR target/83399
23395 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
23396 VECTOR_MEM_ALTIVEC_OR_VSX_P.
23397 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
23398 indexed_or_indirect_operand predicate.
23399 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
23400 (*vsx_le_perm_load_v8hi): Likewise.
23401 (*vsx_le_perm_load_v16qi): Likewise.
23402 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
23403 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
23404 (*vsx_le_perm_store_v8hi): Likewise.
23405 (*vsx_le_perm_store_v16qi): Likewise.
23406 (eight unnamed splitters): Likewise.
23407
23408 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
23409
23410 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
23411 * config/rs6000/emmintrin.h: Likewise.
23412 * config/rs6000/mmintrin.h: Likewise.
23413 * config/rs6000/xmmintrin.h: Likewise.
23414
23415 2018-01-10 David Malcolm <dmalcolm@redhat.com>
23416
23417 PR c++/43486
23418 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
23419 "public_flag".
23420 * tree.c (tree_nop_conversion): Return true for location wrapper
23421 nodes.
23422 (maybe_wrap_with_location): New function.
23423 (selftest::check_strip_nops): New function.
23424 (selftest::test_location_wrappers): New function.
23425 (selftest::tree_c_tests): Call it.
23426 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
23427 (maybe_wrap_with_location): New decl.
23428 (EXPR_LOCATION_WRAPPER_P): New macro.
23429 (location_wrapper_p): New inline function.
23430 (tree_strip_any_location_wrapper): New inline function.
23431
23432 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
23433
23434 PR target/83735
23435 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
23436 stack_realign_offset for the largest alignment of stack slot
23437 actually used.
23438 (ix86_find_max_used_stack_alignment): New function.
23439 (ix86_finalize_stack_frame_flags): Use it. Set
23440 max_used_stack_alignment if we don't realign stack.
23441 * config/i386/i386.h (machine_function): Add
23442 max_used_stack_alignment.
23443
23444 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
23445
23446 * config/arm/arm.opt (-mbranch-cost): New option.
23447 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
23448 account.
23449
23450 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
23451
23452 PR target/83629
23453 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
23454 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
23455
23456 2018-01-10 Richard Biener <rguenther@suse.de>
23457
23458 PR debug/83765
23459 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
23460 early out so it also covers the case where we have a non-NULL
23461 origin.
23462
23463 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
23464
23465 PR tree-optimization/83753
23466 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
23467 for non-strided grouped accesses if the number of elements is 1.
23468
23469 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
23470
23471 PR target/81616
23472 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
23473 * i386.h (TARGET_USE_GATHER): Define.
23474 * x86-tune.def (X86_TUNE_USE_GATHER): New.
23475
23476 2018-01-10 Martin Liska <mliska@suse.cz>
23477
23478 PR bootstrap/82831
23479 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
23480 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
23481 partitioning.
23482 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
23483 CLEANUP_NO_PARTITIONING is not set.
23484
23485 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
23486
23487 * doc/rtl.texi: Remove documentation of (const ...) wrappers
23488 for vectors, as a partial revert of r254296.
23489 * rtl.h (const_vec_p): Delete.
23490 (const_vec_duplicate_p): Don't test for vector CONSTs.
23491 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
23492 * expmed.c (make_tree): Likewise.
23493
23494 Revert:
23495 * common.md (E, F): Use CONSTANT_P instead of checking for
23496 CONST_VECTOR.
23497 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
23498 checking for CONST_VECTOR.
23499
23500 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23501
23502 PR middle-end/83575
23503 * predict.c (force_edge_cold): Handle in more sane way edges
23504 with no prediction.
23505
23506 2018-01-09 Carl Love <cel@us.ibm.com>
23507
23508 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
23509 V4SI, V4SF types.
23510 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
23511 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
23512 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
23513 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
23514 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
23515 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
23516 * config/rs6000/rs6000-protos.h: Add extern defition for
23517 rs6000_generate_float2_double_code.
23518 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
23519 function.
23520 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
23521 (float2_v2df): Add define_expand.
23522
23523 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
23524
23525 PR target/83628
23526 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
23527 op_mode in the force_to_mode call.
23528
23529 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23530
23531 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
23532 instead of checking each element individually.
23533 (aarch64_evpc_uzp): Likewise.
23534 (aarch64_evpc_zip): Likewise.
23535 (aarch64_evpc_ext): Likewise.
23536 (aarch64_evpc_rev): Likewise.
23537 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
23538 instead of checking each element individually. Return true without
23539 generating rtl if
23540 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
23541 whether all selected elements come from the same input, instead of
23542 checking each element individually. Remove calls to gen_rtx_REG,
23543 start_sequence and end_sequence and instead assert that no rtl is
23544 generated.
23545
23546 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23547
23548 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
23549 order of HIGH and CONST checks.
23550
23551 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
23552
23553 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
23554 if the destination isn't an SSA_NAME.
23555
23556 2018-01-09 Richard Biener <rguenther@suse.de>
23557
23558 PR tree-optimization/83668
23559 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
23560 move prologue...
23561 (canonicalize_loop_form): ... here, renamed from ...
23562 (canonicalize_loop_closed_ssa_form): ... this and amended to
23563 swap successor edges for loop exit blocks to make us use
23564 the RPO order we need for initial schedule generation.
23565
23566 2018-01-09 Joseph Myers <joseph@codesourcery.com>
23567
23568 PR tree-optimization/64811
23569 * match.pd: When optimizing comparisons with Inf, avoid
23570 introducing or losing exceptions from comparisons with NaN.
23571
23572 2018-01-09 Martin Liska <mliska@suse.cz>
23573
23574 PR sanitizer/82517
23575 * asan.c (shadow_mem_size): Add gcc_assert.
23576
23577 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
23578
23579 Don't save registers in main().
23580
23581 PR target/83738
23582 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
23583 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
23584 * config/avr/avr.c (avr_set_current_function): Don't error if
23585 naked, OS_task or OS_main are specified at the same time.
23586 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
23587 OS_main.
23588 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
23589 attribute.
23590 * common/config/avr/avr-common.c (avr_option_optimization_table):
23591 Switch on -mmain-is-OS_task for optimizing compilations.
23592
23593 2018-01-09 Richard Biener <rguenther@suse.de>
23594
23595 PR tree-optimization/83572
23596 * graphite.c: Include cfganal.h.
23597 (graphite_transform_loops): Connect infinite loops to exit
23598 and remove fake edges at the end.
23599
23600 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23601
23602 * ipa-inline.c (edge_badness): Revert accidental checkin.
23603
23604 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
23605
23606 PR ipa/80763
23607 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
23608 symbols; not inline clones.
23609
23610 2018-01-09 Jakub Jelinek <jakub@redhat.com>
23611
23612 PR target/83507
23613 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
23614 hard registers. Formatting fixes.
23615
23616 PR preprocessor/83722
23617 * gcc.c (try_generate_repro): Pass
23618 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
23619 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
23620 do_report_bug.
23621
23622 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
23623 Kito Cheng <kito.cheng@gmail.com>
23624
23625 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
23626 (riscv_leaf_function_p): Delete.
23627 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
23628
23629 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23630
23631 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
23632 function.
23633 (do_ifelse): New function.
23634 (do_isel): New function.
23635 (do_sub3): New function.
23636 (do_add3): New function.
23637 (do_load_mask_compare): New function.
23638 (do_overlap_load_compare): New function.
23639 (expand_compare_loop): New function.
23640 (expand_block_compare): Call expand_compare_loop() when appropriate.
23641 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
23642 option description.
23643 (-mblock-compare-inline-loop-limit): New option.
23644
23645 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
23646
23647 PR target/83677
23648 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
23649 Reverse order of second and third operands in first alternative.
23650 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
23651 of first and second elements in UNSPEC_VPERMR vector.
23652 (altivec_expand_vec_perm_le): Likewise.
23653
23654 2018-01-08 Jeff Law <law@redhat.com>
23655
23656 PR rtl-optimizatin/81308
23657 * tree-switch-conversion.c (cfg_altered): New file scoped static.
23658 (process_switch): If group_case_labels makes a change, then set
23659 cfg_altered.
23660 (pass_convert_switch::execute): If a switch is converted, then
23661 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
23662
23663 PR rtl-optimization/81308
23664 * recog.c (split_all_insns): Conditionally cleanup the CFG after
23665 splitting insns.
23666
23667 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
23668
23669 PR target/83663 - Revert r255946
23670 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
23671 generation for cases where splatting a value is not useful.
23672 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
23673 across a vec_duplicate and a paradoxical subreg forming a vector
23674 mode to a vec_concat.
23675
23676 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23677
23678 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
23679 -march=armv8.3-a variants.
23680 * config/arm/t-multilib: Likewise.
23681 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
23682
23683 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
23684
23685 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
23686 to generate rtl.
23687 (cceq_ior_compare_complement): Give it a name so I can use it, and
23688 change boolean_or_operator predicate to boolean_operator so it can
23689 be used to generate a crand.
23690 (eqne): New code iterator.
23691 (bd/bd_neg): New code_attrs.
23692 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
23693 a single define_insn.
23694 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
23695 decrement (bdnzt/bdnzf/bdzt/bdzf).
23696 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
23697 with the new names of the branch decrement patterns, and added the
23698 names of the branch decrement conditional patterns.
23699
23700 2018-01-08 Richard Biener <rguenther@suse.de>
23701
23702 PR tree-optimization/83563
23703 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
23704 cache.
23705
23706 2018-01-08 Richard Biener <rguenther@suse.de>
23707
23708 PR middle-end/83713
23709 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
23710
23711 2018-01-08 Richard Biener <rguenther@suse.de>
23712
23713 PR tree-optimization/83685
23714 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
23715 references to abnormals.
23716
23717 2018-01-08 Richard Biener <rguenther@suse.de>
23718
23719 PR lto/83719
23720 * dwarf2out.c (output_indirect_strings): Handle empty
23721 skeleton_debug_str_hash.
23722 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
23723
23724 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
23725
23726 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
23727 (emit_store_direct): Likewise.
23728 (arc_trampoline_adjust_address): Likewise.
23729 (arc_asm_trampoline_template): New function.
23730 (arc_initialize_trampoline): Use asm_trampoline_template.
23731 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
23732 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
23733 * config/arc/arc.md (flush_icache): Delete pattern.
23734
23735 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
23736
23737 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
23738 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
23739 munaligned-access.
23740
23741 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
23742
23743 PR target/83681
23744 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
23745 by not USED_FOR_TARGET.
23746 (make_pass_resolve_sw_modes): Likewise.
23747
23748 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
23749
23750 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
23751 USED_FOR_TARGET.
23752
23753 2018-01-08 Richard Biener <rguenther@suse.de>
23754
23755 PR middle-end/83580
23756 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
23757
23758 2018-01-08 Richard Biener <rguenther@suse.de>
23759
23760 PR middle-end/83517
23761 * match.pd ((t * 2) / 2) -> t): Add missing :c.
23762
23763 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
23764
23765 PR middle-end/81897
23766 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
23767 basic blocks with a small number of successors.
23768 (convert_control_dep_chain_into_preds): Improve handling of
23769 forwarder blocks.
23770 (dump_predicates): Split apart into...
23771 (dump_pred_chain): ...here...
23772 (dump_pred_info): ...and here.
23773 (can_one_predicate_be_invalidated_p): Add debugging printfs.
23774 (can_chain_union_be_invalidated_p): Improve check for invalidation
23775 of paths.
23776 (uninit_uses_cannot_happen): Avoid unnecessary if
23777 convert_control_dep_chain_into_preds yielded nothing.
23778
23779 2018-01-06 Martin Sebor <msebor@redhat.com>
23780
23781 PR tree-optimization/83640
23782 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
23783 subtracting negative offset from size.
23784 (builtin_access::overlap): Adjust offset bounds of the access to fall
23785 within the size of the object if possible.
23786
23787 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
23788
23789 PR rtl-optimization/83699
23790 * expmed.c (extract_bit_field_1): Restrict the vector usage of
23791 extract_bit_field_as_subreg to cases in which the extracted
23792 value is also a vector.
23793
23794 * lra-constraints.c (process_alt_operands): Test for the equivalence
23795 substitutions when detecting a possible reload cycle.
23796
23797 2018-01-06 Jakub Jelinek <jakub@redhat.com>
23798
23799 PR debug/83480
23800 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
23801 by default if flag_selective_schedling{,2}. Formatting fixes.
23802
23803 PR rtl-optimization/83682
23804 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
23805 if it has non-VECTOR_MODE element mode.
23806 (vec_duplicate_p): Likewise.
23807
23808 PR middle-end/83694
23809 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
23810 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
23811
23812 2018-01-05 Jakub Jelinek <jakub@redhat.com>
23813
23814 PR target/83604
23815 * config/i386/i386-builtin.def
23816 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
23817 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
23818 Require also OPTION_MASK_ISA_AVX512F in addition to
23819 OPTION_MASK_ISA_GFNI.
23820 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
23821 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
23822 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
23823 to OPTION_MASK_ISA_GFNI.
23824 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
23825 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
23826 OPTION_MASK_ISA_AVX512BW.
23827 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
23828 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
23829 addition to OPTION_MASK_ISA_GFNI.
23830 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
23831 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
23832 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
23833 to OPTION_MASK_ISA_GFNI.
23834 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
23835 a requirement for all ISAs rather than any of them with a few
23836 exceptions.
23837 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
23838 processing.
23839 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
23840 bitmasks to be enabled with 3 exceptions, instead of requiring any
23841 enabled ISA with lots of exceptions.
23842 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
23843 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
23844 Change avx512bw in isa attribute to avx512f.
23845 * config/i386/sgxintrin.h: Add license boilerplate.
23846 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
23847 to __AVX512F__ and __AVX512VL to __AVX512VL__.
23848 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
23849 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
23850 defined.
23851 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
23852 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
23853 temporarily sse2 rather than sse if not enabled already.
23854
23855 PR target/83604
23856 * config/i386/sse.md (VI248_VLBW): Rename to ...
23857 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
23858 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
23859 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
23860 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
23861 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
23862 mode iterator instead of VI248_VLBW.
23863
23864 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
23865
23866 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
23867 (record_modified): Skip clobbers; add debug output.
23868 (param_change_prob): Use sreal frequencies.
23869
23870 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
23871
23872 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
23873 punt for user-aligned variables.
23874
23875 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
23876
23877 * tree-chrec.c (chrec_contains_symbols): Return true for
23878 POLY_INT_CST.
23879
23880 2018-01-05 Sudakshina Das <sudi.das@arm.com>
23881
23882 PR target/82439
23883 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
23884 of (x|y) == x for BICS pattern.
23885
23886 2018-01-05 Jakub Jelinek <jakub@redhat.com>
23887
23888 PR tree-optimization/83605
23889 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
23890 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
23891 can throw.
23892
23893 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
23894
23895 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
23896 * config/epiphany/rtems.h: New file.
23897
23898 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23899 Uros Bizjak <ubizjak@gmail.com>
23900
23901 PR target/83554
23902 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
23903 QIreg_operand instead of register_operand predicate.
23904 * config/i386/i386.c (ix86_rop_should_change_byte_p,
23905 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
23906 comments instead of -fmitigate[-_]rop.
23907
23908 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
23909
23910 PR bootstrap/81926
23911 * cgraphunit.c (symbol_table::compile): Switch to text_section
23912 before calling assembly_start debug hook.
23913 * run-rtl-passes.c (run_rtl_passes): Likewise.
23914 Include output.h.
23915
23916 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23917
23918 * tree-vrp.c (extract_range_from_binary_expr_1): Check
23919 range_int_cst_p rather than !symbolic_range_p before calling
23920 extract_range_from_multiplicative_op_1.
23921
23922 2018-01-04 Jeff Law <law@redhat.com>
23923
23924 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
23925 redundant test in assertion.
23926
23927 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23928
23929 * doc/rtl.texi: Document machine_mode wrapper classes.
23930
23931 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23932
23933 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
23934 using tree_to_uhwi.
23935
23936 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23937
23938 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
23939 the VEC_PERM_EXPR fold to fail.
23940
23941 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23942
23943 PR debug/83585
23944 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
23945 to switched_sections.
23946
23947 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23948
23949 PR target/83680
23950 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
23951 test for d.testing.
23952
23953 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
23954
23955 PR target/83387
23956 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
23957 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
23958
23959 2018-01-04 Jakub Jelinek <jakub@redhat.com>
23960
23961 PR debug/83666
23962 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
23963 is BLKmode and bitpos not zero or mode change is needed.
23964
23965 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
23966
23967 PR target/83675
23968 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
23969 TARGET_VIS2.
23970
23971 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
23972
23973 PR target/83628
23974 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
23975 instead of MULT rtx. Update all corresponding splitters.
23976 (*saddl_se): Ditto.
23977 (*ssub<modesuffix>): Ditto.
23978 (*ssubl_se): Ditto.
23979 (*cmp_sadd_di): Update split patterns.
23980 (*cmp_sadd_si): Ditto.
23981 (*cmp_sadd_sidi): Ditto.
23982 (*cmp_ssub_di): Ditto.
23983 (*cmp_ssub_si): Ditto.
23984 (*cmp_ssub_sidi): Ditto.
23985 * config/alpha/predicates.md (const23_operand): New predicate.
23986 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
23987 Look for ASHIFT, not MULT inner operand.
23988 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
23989
23990 2018-01-04 Martin Liska <mliska@suse.cz>
23991
23992 PR gcov-profile/83669
23993 * gcov.c (output_intermediate_file): Add version to intermediate
23994 gcov file.
23995 * doc/gcov.texi: Document new field 'version' in intermediate
23996 file format. Fix location of '-k' option of gcov command.
23997
23998 2018-01-04 Martin Liska <mliska@suse.cz>
23999
24000 PR ipa/82352
24001 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
24002
24003 2018-01-04 Jakub Jelinek <jakub@redhat.com>
24004
24005 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
24006
24007 2018-01-03 Martin Sebor <msebor@redhat.com>
24008
24009 PR tree-optimization/83655
24010 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
24011 checking calls with invalid arguments.
24012
24013 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24014
24015 * tree-vect-stmts.c (vect_get_store_rhs): New function.
24016 (vectorizable_mask_load_store): Delete.
24017 (vectorizable_call): Return false for masked loads and stores.
24018 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
24019 instead of gimple_assign_rhs1.
24020 (vectorizable_load): Handle IFN_MASK_LOAD.
24021 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
24022
24023 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24024
24025 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
24026 split out from..,
24027 (vectorizable_mask_load_store): ...here.
24028 (vectorizable_load): ...and here.
24029
24030 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24031
24032 * tree-vect-stmts.c (vect_build_all_ones_mask)
24033 (vect_build_zero_merge_argument): New functions, split out from...
24034 (vectorizable_load): ...here.
24035
24036 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24037
24038 * tree-vect-stmts.c (vect_check_store_rhs): New function,
24039 split out from...
24040 (vectorizable_mask_load_store): ...here.
24041 (vectorizable_store): ...and here.
24042
24043 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24044
24045 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
24046 split out from...
24047 (vectorizable_mask_load_store): ...here.
24048
24049 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24050
24051 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
24052 (vect_model_store_cost): Take a vec_load_store_type instead of a
24053 vect_def_type.
24054 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
24055 (vect_model_store_cost): Take a vec_load_store_type instead of a
24056 vect_def_type.
24057 (vectorizable_mask_load_store): Update accordingly.
24058 (vectorizable_store): Likewise.
24059 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
24060
24061 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24062
24063 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
24064 IFN_MASK_LOAD calls here rather than...
24065 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
24066
24067 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24068 Alan Hayward <alan.hayward@arm.com>
24069 David Sherwood <david.sherwood@arm.com>
24070
24071 * expmed.c (extract_bit_field_1): For vector extracts,
24072 fall back to extract_bit_field_as_subreg if vec_extract
24073 isn't available.
24074
24075 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24076 Alan Hayward <alan.hayward@arm.com>
24077 David Sherwood <david.sherwood@arm.com>
24078
24079 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
24080 they are variable or constant sized.
24081 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
24082 slots for constant-sized data.
24083
24084 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24085 Alan Hayward <alan.hayward@arm.com>
24086 David Sherwood <david.sherwood@arm.com>
24087
24088 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
24089 handling COND_EXPRs with boolean comparisons, try to find a better
24090 basis for the mask type than the boolean itself.
24091
24092 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24093
24094 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
24095 is calculated and how it can be overridden.
24096 * genmodes.c (max_bitsize_mode_any_mode): New variable.
24097 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
24098 if defined.
24099 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
24100 if nonzero.
24101
24102 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24103 Alan Hayward <alan.hayward@arm.com>
24104 David Sherwood <david.sherwood@arm.com>
24105
24106 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
24107 Remove the mode argument.
24108 (aarch64_simd_valid_immediate): Remove the mode and inverse
24109 arguments.
24110 * config/aarch64/iterators.md (bitsize): New iterator.
24111 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
24112 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
24113 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
24114 aarch64_simd_valid_immediate.
24115 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
24116 (aarch64_reg_or_bic_imm): Likewise.
24117 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
24118 with an insn_type enum and msl with a modifier_type enum.
24119 Replace element_width with a scalar_mode. Change the shift
24120 to unsigned int. Add constructors for scalar_float_mode and
24121 scalar_int_mode elements.
24122 (aarch64_vect_float_const_representable_p): Delete.
24123 (aarch64_can_const_movi_rtx_p)
24124 (aarch64_simd_scalar_immediate_valid_for_move)
24125 (aarch64_simd_make_constant): Update call to
24126 aarch64_simd_valid_immediate.
24127 (aarch64_advsimd_valid_immediate_hs): New function.
24128 (aarch64_advsimd_valid_immediate): Likewise.
24129 (aarch64_simd_valid_immediate): Remove mode and inverse
24130 arguments. Rewrite to use the above. Use const_vec_duplicate_p
24131 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
24132 and aarch64_float_const_representable_p on the result.
24133 (aarch64_output_simd_mov_immediate): Remove mode argument.
24134 Update call to aarch64_simd_valid_immediate and use of
24135 simd_immediate_info.
24136 (aarch64_output_scalar_simd_mov_immediate): Update call
24137 accordingly.
24138
24139 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24140 Alan Hayward <alan.hayward@arm.com>
24141 David Sherwood <david.sherwood@arm.com>
24142
24143 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
24144 (mode_nunits): Likewise CONST_MODE_NUNITS.
24145 * machmode.def (ADJUST_NUNITS): Document.
24146 * genmodes.c (mode_data::need_nunits_adj): New field.
24147 (blank_mode): Update accordingly.
24148 (adj_nunits): New variable.
24149 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
24150 parameter.
24151 (emit_mode_size_inline): Set need_bytesize_adj for all modes
24152 listed in adj_nunits.
24153 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
24154 listed in adj_nunits. Don't emit case statements for such modes.
24155 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
24156 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
24157 nothing if adj_nunits is nonnull.
24158 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
24159 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
24160 (emit_mode_fbit): Update use of print_maybe_const_decl.
24161 (emit_move_size): Likewise. Treat the array as non-const
24162 if adj_nunits.
24163 (emit_mode_adjustments): Handle adj_nunits.
24164
24165 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24166
24167 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
24168 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
24169 (VECTOR_MODES): Use it.
24170 (make_vector_modes): Take the prefix as an argument.
24171
24172 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24173 Alan Hayward <alan.hayward@arm.com>
24174 David Sherwood <david.sherwood@arm.com>
24175
24176 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
24177 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
24178 for MODE_VECTOR_BOOL.
24179 * machmode.def (VECTOR_BOOL_MODE): Document.
24180 * genmodes.c (VECTOR_BOOL_MODE): New macro.
24181 (make_vector_bool_mode): New function.
24182 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
24183 MODE_VECTOR_BOOL.
24184 * lto-streamer-in.c (lto_input_mode_table): Likewise.
24185 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
24186 Likewise.
24187 * stor-layout.c (int_mode_for_mode): Likewise.
24188 * tree.c (build_vector_type_for_mode): Likewise.
24189 * varasm.c (output_constant_pool_2): Likewise.
24190 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
24191 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
24192 for MODE_VECTOR_BOOL.
24193 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
24194 of mode class checks.
24195 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
24196 instead of a list of mode class checks.
24197 (expand_vector_scalar_condition): Likewise.
24198 (type_for_widest_vector_mode): Handle BImode as an inner mode.
24199
24200 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24201 Alan Hayward <alan.hayward@arm.com>
24202 David Sherwood <david.sherwood@arm.com>
24203
24204 * machmode.h (mode_size): Change from unsigned short to
24205 poly_uint16_pod.
24206 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
24207 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24208 or if measurement_type is not polynomial.
24209 (fixed_size_mode::includes_p): Check for constant-sized modes.
24210 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
24211 return a poly_uint16 rather than an unsigned short.
24212 (emit_mode_size): Change the type of mode_size from unsigned short
24213 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
24214 (emit_mode_adjustments): Cope with polynomial vector sizes.
24215 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24216 for GET_MODE_SIZE.
24217 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24218 for GET_MODE_SIZE.
24219 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
24220 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
24221 * caller-save.c (setup_save_areas): Likewise.
24222 (replace_reg_with_saved_mem): Likewise.
24223 * calls.c (emit_library_call_value_1): Likewise.
24224 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
24225 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
24226 (gen_lowpart_for_combine): Likewise.
24227 * convert.c (convert_to_integer_1): Likewise.
24228 * cse.c (equiv_constant, cse_insn): Likewise.
24229 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
24230 (cselib_subst_to_values): Likewise.
24231 * dce.c (word_dce_process_block): Likewise.
24232 * df-problems.c (df_word_lr_mark_ref): Likewise.
24233 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
24234 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
24235 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
24236 (rtl_for_decl_location): Likewise.
24237 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
24238 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
24239 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
24240 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
24241 (expand_expr_real_1): Likewise.
24242 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
24243 (pad_below): Likewise.
24244 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24245 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
24246 * ira.c (get_subreg_tracking_sizes): Likewise.
24247 * ira-build.c (ira_create_allocno_objects): Likewise.
24248 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
24249 (ira_sort_regnos_for_alter_reg): Likewise.
24250 * ira-costs.c (record_operand_costs): Likewise.
24251 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
24252 (resolve_simple_move): Likewise.
24253 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
24254 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
24255 (lra_constraints): Likewise.
24256 (CONST_POOL_OK_P): Reject variable-sized modes.
24257 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
24258 (add_pseudo_to_slot, lra_spill): Likewise.
24259 * omp-low.c (omp_clause_aligned_alignment): Likewise.
24260 * optabs-query.c (get_best_extraction_insn): Likewise.
24261 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
24262 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
24263 (expand_mult_highpart, valid_multiword_target_p): Likewise.
24264 * recog.c (offsettable_address_addr_space_p): Likewise.
24265 * regcprop.c (maybe_mode_change): Likewise.
24266 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
24267 * regrename.c (build_def_use): Likewise.
24268 * regstat.c (dump_reg_info): Likewise.
24269 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
24270 (find_reloads, find_reloads_subreg_address): Likewise.
24271 * reload1.c (eliminate_regs_1): Likewise.
24272 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
24273 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
24274 (simplify_binary_operation_1, simplify_subreg): Likewise.
24275 * targhooks.c (default_function_arg_padding): Likewise.
24276 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
24277 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
24278 (verify_gimple_assign_ternary): Likewise.
24279 * tree-inline.c (estimate_move_cost): Likewise.
24280 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24281 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
24282 (get_address_cost_ainc): Likewise.
24283 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
24284 (vect_supportable_dr_alignment): Likewise.
24285 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24286 (vectorizable_reduction): Likewise.
24287 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
24288 (vectorizable_operation, vectorizable_load): Likewise.
24289 * tree.c (build_same_sized_truth_vector_type): Likewise.
24290 * valtrack.c (cleanup_auto_inc_dec): Likewise.
24291 * var-tracking.c (emit_note_insn_var_location): Likewise.
24292 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
24293 (ADDR_VEC_ALIGN): Likewise.
24294
24295 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24296 Alan Hayward <alan.hayward@arm.com>
24297 David Sherwood <david.sherwood@arm.com>
24298
24299 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
24300 unsigned short.
24301 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
24302 or if measurement_type is polynomial.
24303 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
24304 * combine.c (make_extraction): Likewise.
24305 * dse.c (find_shift_sequence): Likewise.
24306 * dwarf2out.c (mem_loc_descriptor): Likewise.
24307 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
24308 (extract_bit_field, extract_low_bits): Likewise.
24309 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
24310 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
24311 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
24312 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
24313 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
24314 * reload.c (find_reloads): Likewise.
24315 * reload1.c (alter_reg): Likewise.
24316 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
24317 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
24318 * tree-if-conv.c (predicate_mem_writes): Likewise.
24319 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
24320 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
24321 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
24322 * valtrack.c (dead_debug_insert_temp): Likewise.
24323 * varasm.c (mergeable_constant_section): Likewise.
24324 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
24325
24326 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24327 Alan Hayward <alan.hayward@arm.com>
24328 David Sherwood <david.sherwood@arm.com>
24329
24330 * expr.c (expand_assignment): Cope with polynomial mode sizes
24331 when assigning to a CONCAT.
24332
24333 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24334 Alan Hayward <alan.hayward@arm.com>
24335 David Sherwood <david.sherwood@arm.com>
24336
24337 * machmode.h (mode_precision): Change from unsigned short to
24338 poly_uint16_pod.
24339 (mode_to_precision): Return a poly_uint16 rather than an unsigned
24340 short.
24341 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
24342 or if measurement_type is not polynomial.
24343 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
24344 in which the mode is already known to be a scalar_int_mode.
24345 * genmodes.c (emit_mode_precision): Change the type of mode_precision
24346 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
24347 initializer.
24348 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24349 for GET_MODE_PRECISION.
24350 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24351 for GET_MODE_PRECISION.
24352 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
24353 as polynomial.
24354 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
24355 (expand_field_assignment, make_extraction): Likewise.
24356 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
24357 (get_last_value): Likewise.
24358 * convert.c (convert_to_integer_1): Likewise.
24359 * cse.c (cse_insn): Likewise.
24360 * expr.c (expand_expr_real_1): Likewise.
24361 * lra-constraints.c (simplify_operand_subreg): Likewise.
24362 * optabs-query.c (can_atomic_load_p): Likewise.
24363 * optabs.c (expand_atomic_load): Likewise.
24364 (expand_atomic_store): Likewise.
24365 * ree.c (combine_reaching_defs): Likewise.
24366 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
24367 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
24368 * tree.h (type_has_mode_precision_p): Likewise.
24369 * ubsan.c (instrument_si_overflow): Likewise.
24370
24371 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24372 Alan Hayward <alan.hayward@arm.com>
24373 David Sherwood <david.sherwood@arm.com>
24374
24375 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
24376 polynomial numbers of units.
24377 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
24378 (valid_vector_subparts_p): New function.
24379 (build_vector_type): Remove temporary shim and take the number
24380 of units as a poly_uint64 rather than an int.
24381 (build_opaque_vector_type): Take the number of units as a
24382 poly_uint64 rather than an int.
24383 * tree.c (build_vector_from_ctor): Handle polynomial
24384 TYPE_VECTOR_SUBPARTS.
24385 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
24386 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
24387 (build_vector_from_val): If the number of units is variable,
24388 use build_vec_duplicate_cst for constant operands and
24389 VEC_DUPLICATE_EXPR otherwise.
24390 (make_vector_type): Remove temporary is_constant ().
24391 (build_vector_type, build_opaque_vector_type): Take the number of
24392 units as a poly_uint64 rather than an int.
24393 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
24394 VECTOR_CST_NELTS.
24395 * cfgexpand.c (expand_debug_expr): Likewise.
24396 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
24397 (store_constructor, expand_expr_real_1): Likewise.
24398 (const_scalar_mask_from_tree): Likewise.
24399 * fold-const-call.c (fold_const_reduction): Likewise.
24400 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
24401 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
24402 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
24403 (fold_relational_const): Likewise.
24404 (native_interpret_vector): Likewise. Change the size from an
24405 int to an unsigned int.
24406 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
24407 TYPE_VECTOR_SUBPARTS.
24408 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
24409 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
24410 duplicating a non-constant operand into a variable-length vector.
24411 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
24412 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
24413 * ipa-icf.c (sem_variable::equals): Likewise.
24414 * match.pd: Likewise.
24415 * omp-simd-clone.c (simd_clone_subparts): Likewise.
24416 * print-tree.c (print_node): Likewise.
24417 * stor-layout.c (layout_type): Likewise.
24418 * targhooks.c (default_builtin_vectorization_cost): Likewise.
24419 * tree-cfg.c (verify_gimple_comparison): Likewise.
24420 (verify_gimple_assign_binary): Likewise.
24421 (verify_gimple_assign_ternary): Likewise.
24422 (verify_gimple_assign_single): Likewise.
24423 * tree-pretty-print.c (dump_generic_node): Likewise.
24424 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
24425 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
24426 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
24427 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
24428 (vect_shift_permute_load_chain): Likewise.
24429 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
24430 (expand_vector_condition, optimize_vector_constructor): Likewise.
24431 (lower_vec_perm, get_compute_type): Likewise.
24432 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
24433 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
24434 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
24435 (vect_recog_mask_conversion_pattern): Likewise.
24436 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
24437 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
24438 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
24439 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
24440 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
24441 (vectorizable_shift, vectorizable_operation, vectorizable_store)
24442 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
24443 (supportable_widening_operation): Likewise.
24444 (supportable_narrowing_operation): Likewise.
24445 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
24446 Likewise.
24447 * varasm.c (output_constant): Likewise.
24448
24449 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24450 Alan Hayward <alan.hayward@arm.com>
24451 David Sherwood <david.sherwood@arm.com>
24452
24453 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
24454 so that both the length == 3 and length != 3 cases set up their
24455 own permute vectors. Add comments explaining why we know the
24456 number of elements is constant.
24457 (vect_permute_load_chain): Likewise.
24458
24459 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24460 Alan Hayward <alan.hayward@arm.com>
24461 David Sherwood <david.sherwood@arm.com>
24462
24463 * machmode.h (mode_nunits): Change from unsigned char to
24464 poly_uint16_pod.
24465 (ONLY_FIXED_SIZE_MODES): New macro.
24466 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
24467 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
24468 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
24469 New typedefs.
24470 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
24471 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
24472 or if measurement_type is not polynomial.
24473 * genmodes.c (ZERO_COEFFS): New macro.
24474 (emit_mode_nunits_inline): Make mode_nunits_inline return a
24475 poly_uint16.
24476 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
24477 Use ZERO_COEFFS when emitting initializers.
24478 * data-streamer.h (bp_pack_poly_value): New function.
24479 (bp_unpack_poly_value): Likewise.
24480 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
24481 for GET_MODE_NUNITS.
24482 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
24483 for GET_MODE_NUNITS.
24484 * tree.c (make_vector_type): Remove temporary shim and make
24485 the real function take the number of units as a poly_uint64
24486 rather than an int.
24487 (build_vector_type_for_mode): Handle polynomial nunits.
24488 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
24489 * emit-rtl.c (const_vec_series_p_1): Likewise.
24490 (gen_rtx_CONST_VECTOR): Likewise.
24491 * fold-const.c (test_vec_duplicate_folding): Likewise.
24492 * genrecog.c (validate_pattern): Likewise.
24493 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
24494 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
24495 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
24496 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
24497 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
24498 * rtlanal.c (subreg_get_info): Likewise.
24499 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
24500 (vect_grouped_load_supported): Likewise.
24501 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
24502 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
24503 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
24504 (simplify_const_unary_operation, simplify_binary_operation_1)
24505 (simplify_const_binary_operation, simplify_ternary_operation)
24506 (test_vector_ops_duplicate, test_vector_ops): Likewise.
24507 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
24508 instead of CONST_VECTOR_NUNITS.
24509 * varasm.c (output_constant_pool_2): Likewise.
24510 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
24511 explicit-encoded elements in the XVEC for variable-length vectors.
24512
24513 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24514
24515 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
24516
24517 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24518 Alan Hayward <alan.hayward@arm.com>
24519 David Sherwood <david.sherwood@arm.com>
24520
24521 * coretypes.h (fixed_size_mode): Declare.
24522 (fixed_size_mode_pod): New typedef.
24523 * builtins.h (target_builtins::x_apply_args_mode)
24524 (target_builtins::x_apply_result_mode): Change type to
24525 fixed_size_mode_pod.
24526 * builtins.c (apply_args_size, apply_result_size, result_vector)
24527 (expand_builtin_apply_args_1, expand_builtin_apply)
24528 (expand_builtin_return): Update accordingly.
24529
24530 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24531
24532 * cse.c (hash_rtx_cb): Hash only the encoded elements.
24533 * cselib.c (cselib_hash_rtx): Likewise.
24534 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
24535 CONST_VECTOR encoding.
24536
24537 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24538 Jeff Law <law@redhat.com>
24539
24540 PR target/83641
24541 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
24542 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
24543 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
24544 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
24545
24546 PR target/83641
24547 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
24548 explicitly probe *sp in a noreturn function if there were any callee
24549 register saves or frame pointer is needed.
24550
24551 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24552
24553 PR debug/83621
24554 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
24555 BLKmode for ternary, binary or unary expressions.
24556
24557 PR debug/83645
24558 * var-tracking.c (delete_vta_debug_insn): New inline function.
24559 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
24560 insns from get_insns () to NULL instead of each bb separately.
24561 Use delete_vta_debug_insn. No longer static.
24562 (vt_debug_insns_local, variable_tracking_main_1): Adjust
24563 delete_vta_debug_insns callers.
24564 * rtl.h (delete_vta_debug_insns): Declare.
24565 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
24566 instead of variable_tracking_main.
24567
24568 2018-01-03 Martin Sebor <msebor@redhat.com>
24569
24570 PR tree-optimization/83603
24571 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
24572 arguments past the endof the argument list in functions declared
24573 without a prototype.
24574 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
24575 Avoid checking when arguments are null.
24576
24577 2018-01-03 Martin Sebor <msebor@redhat.com>
24578
24579 PR c/83559
24580 * doc/extend.texi (attribute const): Fix a typo.
24581 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
24582 issuing -Wsuggest-attribute for void functions.
24583
24584 2018-01-03 Martin Sebor <msebor@redhat.com>
24585
24586 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
24587 offset_int::from instead of wide_int::to_shwi.
24588 (maybe_diag_overlap): Remove assertion.
24589 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
24590 * gimple-ssa-sprintf.c (format_directive): Same.
24591 (parse_directive): Same.
24592 (sprintf_dom_walker::compute_format_length): Same.
24593 (try_substitute_return_value): Same.
24594
24595 2018-01-03 Jeff Law <law@redhat.com>
24596
24597 PR middle-end/83654
24598 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
24599 non-constant residual for zero at runtime and avoid probing in
24600 that case. Reorganize code for trailing problem to mirror handling
24601 of the residual.
24602
24603 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
24604
24605 PR tree-optimization/83501
24606 * tree-ssa-strlen.c (get_string_cst): New.
24607 (handle_char_store): Call get_string_cst.
24608
24609 2018-01-03 Martin Liska <mliska@suse.cz>
24610
24611 PR tree-optimization/83593
24612 * tree-ssa-strlen.c: Include tree-cfg.h.
24613 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
24614 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
24615 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
24616 to false.
24617 (strlen_dom_walker::before_dom_children): Call
24618 gimple_purge_dead_eh_edges. Dump tranformation with details
24619 dump flags.
24620 (strlen_dom_walker::before_dom_children): Update call by adding
24621 new argument cleanup_eh.
24622 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
24623
24624 2018-01-03 Martin Liska <mliska@suse.cz>
24625
24626 PR ipa/83549
24627 * cif-code.def (VARIADIC_THUNK): New enum value.
24628 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
24629 thunks.
24630
24631 2018-01-03 Jan Beulich <jbeulich@suse.com>
24632
24633 * sse.md (mov<mode>_internal): Tighten condition for when to use
24634 vmovdqu<ssescalarsize> for TI and OI modes.
24635
24636 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24637
24638 Update copyright years.
24639
24640 2018-01-03 Martin Liska <mliska@suse.cz>
24641
24642 PR ipa/83594
24643 * ipa-visibility.c (function_and_variable_visibility): Skip
24644 functions with noipa attribure.
24645
24646 2018-01-03 Jakub Jelinek <jakub@redhat.com>
24647
24648 * gcc.c (process_command): Update copyright notice dates.
24649 * gcov-dump.c (print_version): Ditto.
24650 * gcov.c (print_version): Ditto.
24651 * gcov-tool.c (print_version): Ditto.
24652 * gengtype.c (create_file): Ditto.
24653 * doc/cpp.texi: Bump @copying's copyright year.
24654 * doc/cppinternals.texi: Ditto.
24655 * doc/gcc.texi: Ditto.
24656 * doc/gccint.texi: Ditto.
24657 * doc/gcov.texi: Ditto.
24658 * doc/install.texi: Ditto.
24659 * doc/invoke.texi: Ditto.
24660
24661 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24662
24663 * vector-builder.h (vector_builder::m_full_nelts): Change from
24664 unsigned int to poly_uint64.
24665 (vector_builder::full_nelts): Update prototype accordingly.
24666 (vector_builder::new_vector): Likewise.
24667 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
24668 (vector_builder::operator ==): Likewise.
24669 (vector_builder::finalize): Likewise.
24670 * int-vector-builder.h (int_vector_builder::int_vector_builder):
24671 Take the number of elements as a poly_uint64 rather than an
24672 unsigned int.
24673 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
24674 from unsigned int to poly_uint64.
24675 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
24676 (vec_perm_indices::new_vector): Likewise.
24677 (vec_perm_indices::length): Likewise.
24678 (vec_perm_indices::nelts_per_input): Likewise.
24679 (vec_perm_indices::input_nelts): Likewise.
24680 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
24681 number of elements per input as a poly_uint64 rather than an
24682 unsigned int. Use the original encoding for variable-length
24683 vectors, rather than clamping each individual element.
24684 For the second and subsequent elements in each pattern,
24685 clamp the step and base before clamping their sum.
24686 (vec_perm_indices::series_p): Handle polynomial element counts.
24687 (vec_perm_indices::all_in_range_p): Likewise.
24688 (vec_perm_indices_to_tree): Likewise.
24689 (vec_perm_indices_to_rtx): Likewise.
24690 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
24691 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
24692 (tree_vector_builder::new_binary_operation): Handle polynomial
24693 element counts. Return false if we need to know the number
24694 of elements at compile time.
24695 * fold-const.c (fold_vec_perm): Punt if the number of elements
24696 isn't known at compile time.
24697
24698 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24699
24700 * vec-perm-indices.h (vec_perm_builder): Change element type
24701 from HOST_WIDE_INT to poly_int64.
24702 (vec_perm_indices::element_type): Update accordingly.
24703 (vec_perm_indices::clamp): Handle polynomial element_types.
24704 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
24705 (vec_perm_indices::all_in_range_p): Likewise.
24706 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
24707 than shwi trees.
24708 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
24709 polynomial vec_perm_indices element types.
24710 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
24711 * fold-const.c (fold_vec_perm): Likewise.
24712 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
24713 * tree-vect-generic.c (lower_vec_perm): Likewise.
24714 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
24715 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
24716 element type to HOST_WIDE_INT.
24717
24718 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24719 Alan Hayward <alan.hayward@arm.com>
24720 David Sherwood <david.sherwood@arm.com>
24721
24722 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
24723 rather than an int. Use plus_constant.
24724 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
24725 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
24726
24727 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24728 Alan Hayward <alan.hayward@arm.com>
24729 David Sherwood <david.sherwood@arm.com>
24730
24731 * calls.c (emit_call_1, expand_call): Change struct_value_size from
24732 a HOST_WIDE_INT to a poly_int64.
24733
24734 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24735 Alan Hayward <alan.hayward@arm.com>
24736 David Sherwood <david.sherwood@arm.com>
24737
24738 * calls.c (load_register_parameters): Cope with polynomial
24739 mode sizes. Require a constant size for BLKmode parameters
24740 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
24741 forces a parameter to be padded at the lsb end in order to
24742 fill a complete number of words, require the parameter size
24743 to be ordered wrt UNITS_PER_WORD.
24744
24745 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24746 Alan Hayward <alan.hayward@arm.com>
24747 David Sherwood <david.sherwood@arm.com>
24748
24749 * reload1.c (spill_stack_slot_width): Change element type
24750 from unsigned int to poly_uint64_pod.
24751 (alter_reg): Treat mode sizes as polynomial.
24752
24753 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24754 Alan Hayward <alan.hayward@arm.com>
24755 David Sherwood <david.sherwood@arm.com>
24756
24757 * reload.c (complex_word_subreg_p): New function.
24758 (reload_inner_reg_of_subreg, push_reload): Use it.
24759
24760 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24761 Alan Hayward <alan.hayward@arm.com>
24762 David Sherwood <david.sherwood@arm.com>
24763
24764 * lra-constraints.c (process_alt_operands): Reject matched
24765 operands whose sizes aren't ordered.
24766 (match_reload): Refer to this check here.
24767
24768 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24769 Alan Hayward <alan.hayward@arm.com>
24770 David Sherwood <david.sherwood@arm.com>
24771
24772 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
24773 that the mode size is in the set {1, 2, 4, 8, 16}.
24774
24775 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24776 Alan Hayward <alan.hayward@arm.com>
24777 David Sherwood <david.sherwood@arm.com>
24778
24779 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
24780 Use plus_constant instead of gen_rtx_PLUS.
24781
24782 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24783 Alan Hayward <alan.hayward@arm.com>
24784 David Sherwood <david.sherwood@arm.com>
24785
24786 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
24787 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
24788 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
24789 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
24790 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
24791 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
24792 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
24793 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
24794 * config/i386/i386.c (ix86_push_rounding): ...this new function.
24795 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
24796 a poly_int64.
24797 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
24798 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
24799 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
24800 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
24801 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
24802 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
24803 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
24804 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
24805 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
24806 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
24807 function.
24808 * expr.c (emit_move_resolve_push): Treat the input and result
24809 of PUSH_ROUNDING as a poly_int64.
24810 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
24811 (emit_push_insn): Likewise.
24812 * lra-eliminations.c (mark_not_eliminable): Likewise.
24813 * recog.c (push_operand): Likewise.
24814 * reload1.c (elimination_effects): Likewise.
24815 * rtlanal.c (nonzero_bits1): Likewise.
24816 * calls.c (store_one_arg): Likewise. Require the padding to be
24817 known at compile time.
24818
24819 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24820 Alan Hayward <alan.hayward@arm.com>
24821 David Sherwood <david.sherwood@arm.com>
24822
24823 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
24824 Use plus_constant instead of gen_rtx_PLUS.
24825
24826 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24827 Alan Hayward <alan.hayward@arm.com>
24828 David Sherwood <david.sherwood@arm.com>
24829
24830 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
24831 rather than an int.
24832
24833 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24834 Alan Hayward <alan.hayward@arm.com>
24835 David Sherwood <david.sherwood@arm.com>
24836
24837 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
24838 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
24839 via stack temporaries. Treat the mode size as polynomial too.
24840
24841 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24842 Alan Hayward <alan.hayward@arm.com>
24843 David Sherwood <david.sherwood@arm.com>
24844
24845 * expr.c (expand_expr_real_2): When handling conversions involving
24846 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
24847 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
24848 as a poly_uint64 too.
24849
24850 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24851 Alan Hayward <alan.hayward@arm.com>
24852 David Sherwood <david.sherwood@arm.com>
24853
24854 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
24855
24856 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24857 Alan Hayward <alan.hayward@arm.com>
24858 David Sherwood <david.sherwood@arm.com>
24859
24860 * combine.c (can_change_dest_mode): Handle polynomial
24861 REGMODE_NATURAL_SIZE.
24862 * expmed.c (store_bit_field_1): Likewise.
24863 * expr.c (store_constructor): Likewise.
24864 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
24865 and polynomial REGMODE_NATURAL_SIZE.
24866 (gen_lowpart_common): Likewise.
24867 * reginfo.c (record_subregs_of_mode): Likewise.
24868 * rtlanal.c (read_modify_subreg_p): Likewise.
24869
24870 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24871 Alan Hayward <alan.hayward@arm.com>
24872 David Sherwood <david.sherwood@arm.com>
24873
24874 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
24875 numbers of elements.
24876
24877 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24878 Alan Hayward <alan.hayward@arm.com>
24879 David Sherwood <david.sherwood@arm.com>
24880
24881 * match.pd: Cope with polynomial numbers of vector elements.
24882
24883 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24884 Alan Hayward <alan.hayward@arm.com>
24885 David Sherwood <david.sherwood@arm.com>
24886
24887 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
24888 in a POINTER_PLUS_EXPR.
24889
24890 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24891 Alan Hayward <alan.hayward@arm.com>
24892 David Sherwood <david.sherwood@arm.com>
24893
24894 * omp-simd-clone.c (simd_clone_subparts): New function.
24895 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
24896 (ipa_simd_modify_function_body): Likewise.
24897
24898 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24899 Alan Hayward <alan.hayward@arm.com>
24900 David Sherwood <david.sherwood@arm.com>
24901
24902 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
24903 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
24904 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
24905 (expand_vector_condition, vector_element): Likewise.
24906 (subparts_gt): New function.
24907 (get_compute_type): Use subparts_gt.
24908 (count_type_subparts): Delete.
24909 (expand_vector_operations_1): Use subparts_gt instead of
24910 count_type_subparts.
24911
24912 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24913 Alan Hayward <alan.hayward@arm.com>
24914 David Sherwood <david.sherwood@arm.com>
24915
24916 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
24917 (vect_compile_time_alias): ...this new function. Do the calculation
24918 on poly_ints rather than trees.
24919 (vect_prune_runtime_alias_test_list): Update call accordingly.
24920
24921 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24922 Alan Hayward <alan.hayward@arm.com>
24923 David Sherwood <david.sherwood@arm.com>
24924
24925 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
24926 numbers of units.
24927 (vect_schedule_slp_instance): Likewise.
24928
24929 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24930 Alan Hayward <alan.hayward@arm.com>
24931 David Sherwood <david.sherwood@arm.com>
24932
24933 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
24934 constant and extern definitions for variable-length vectors.
24935 (vect_get_constant_vectors): Note that the number of units
24936 is known to be constant.
24937
24938 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24939 Alan Hayward <alan.hayward@arm.com>
24940 David Sherwood <david.sherwood@arm.com>
24941
24942 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
24943 of units as polynomial. Choose between WIDE and NARROW based
24944 on multiple_p.
24945
24946 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24947 Alan Hayward <alan.hayward@arm.com>
24948 David Sherwood <david.sherwood@arm.com>
24949
24950 * tree-vect-stmts.c (simd_clone_subparts): New function.
24951 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
24952
24953 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24954 Alan Hayward <alan.hayward@arm.com>
24955 David Sherwood <david.sherwood@arm.com>
24956
24957 * tree-vect-stmts.c (vectorizable_call): Treat the number of
24958 vectors as polynomial. Use build_index_vector for
24959 IFN_GOMP_SIMD_LANE.
24960
24961 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24962 Alan Hayward <alan.hayward@arm.com>
24963 David Sherwood <david.sherwood@arm.com>
24964
24965 * tree-vect-stmts.c (get_load_store_type): Treat the number of
24966 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
24967 for variable-length vectors.
24968 (vectorizable_mask_load_store): Treat the number of units as
24969 polynomial, asserting that it is constant if the condition has
24970 already been enforced.
24971 (vectorizable_store, vectorizable_load): Likewise.
24972
24973 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24974 Alan Hayward <alan.hayward@arm.com>
24975 David Sherwood <david.sherwood@arm.com>
24976
24977 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
24978 of units as polynomial. Punt if we can't tell at compile time
24979 which vector contains the final result.
24980
24981 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24982 Alan Hayward <alan.hayward@arm.com>
24983 David Sherwood <david.sherwood@arm.com>
24984
24985 * tree-vect-loop.c (vectorizable_induction): Treat the number
24986 of units as polynomial. Punt on SLP inductions. Use an integer
24987 VEC_SERIES_EXPR for variable-length integer reductions. Use a
24988 cast of such a series for variable-length floating-point
24989 reductions.
24990
24991 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
24992 Alan Hayward <alan.hayward@arm.com>
24993 David Sherwood <david.sherwood@arm.com>
24994
24995 * tree.h (build_index_vector): Declare.
24996 * tree.c (build_index_vector): New function.
24997 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
24998 of units as polynomial, forcibly converting it to a constant if
24999 vectorizable_reduction has already enforced the condition.
25000 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
25001 to create a {1,2,3,...} vector.
25002 (vectorizable_reduction): Treat the number of units as polynomial.
25003 Choose vectype_in based on the largest scalar element size rather
25004 than the smallest number of units. Enforce the restrictions
25005 relied on above.
25006
25007 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25008 Alan Hayward <alan.hayward@arm.com>
25009 David Sherwood <david.sherwood@arm.com>
25010
25011 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
25012 number of units as polynomial.
25013
25014 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25015 Alan Hayward <alan.hayward@arm.com>
25016 David Sherwood <david.sherwood@arm.com>
25017
25018 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
25019 * target.def (autovectorize_vector_sizes): Return the vector sizes
25020 by pointer, using vector_sizes rather than a bitmask.
25021 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
25022 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
25023 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
25024 Likewise.
25025 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
25026 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
25027 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
25028 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
25029 * omp-general.c (omp_max_vf): Likewise.
25030 * omp-low.c (omp_clause_aligned_alignment): Likewise.
25031 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
25032 * tree-vect-loop.c (vect_analyze_loop): Likewise.
25033 * tree-vect-slp.c (vect_slp_bb): Likewise.
25034 * doc/tm.texi: Regenerate.
25035 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
25036 to a poly_uint64.
25037 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
25038 the vector size as a poly_uint64 rather than an unsigned int.
25039 (current_vector_size): Change from an unsigned int to a poly_uint64.
25040 (get_vectype_for_scalar_type): Update accordingly.
25041 * tree.h (build_truth_vector_type): Take the size and number of
25042 units as a poly_uint64 rather than an unsigned int.
25043 (build_vector_type): Add a temporary overload that takes
25044 the number of units as a poly_uint64 rather than an unsigned int.
25045 * tree.c (make_vector_type): Likewise.
25046 (build_truth_vector_type): Take the number of units as a poly_uint64
25047 rather than an unsigned int.
25048
25049 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25050 Alan Hayward <alan.hayward@arm.com>
25051 David Sherwood <david.sherwood@arm.com>
25052
25053 * target.def (get_mask_mode): Take the number of units and length
25054 as poly_uint64s rather than unsigned ints.
25055 * targhooks.h (default_get_mask_mode): Update accordingly.
25056 * targhooks.c (default_get_mask_mode): Likewise.
25057 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
25058 * doc/tm.texi: Regenerate.
25059
25060 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25061 Alan Hayward <alan.hayward@arm.com>
25062 David Sherwood <david.sherwood@arm.com>
25063
25064 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
25065 * omp-general.c (omp_max_vf): Likewise.
25066 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
25067 (expand_omp_simd): Handle polynomial safelen.
25068 * omp-low.c (omplow_simd_context): Add a default constructor.
25069 (omplow_simd_context::max_vf): Change from int to poly_uint64.
25070 (lower_rec_simd_input_clauses): Update accordingly.
25071 (lower_rec_input_clauses): Likewise.
25072
25073 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25074 Alan Hayward <alan.hayward@arm.com>
25075 David Sherwood <david.sherwood@arm.com>
25076
25077 * tree-vectorizer.h (vect_nunits_for_cost): New function.
25078 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
25079 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
25080 (vect_analyze_slp_cost): Likewise.
25081 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
25082 (vect_model_load_cost): Likewise.
25083
25084 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25085 Alan Hayward <alan.hayward@arm.com>
25086 David Sherwood <david.sherwood@arm.com>
25087
25088 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
25089 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
25090 from an unsigned int * to a poly_uint64_pod *.
25091 (calculate_unrolling_factor): New function.
25092 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
25093
25094 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25095 Alan Hayward <alan.hayward@arm.com>
25096 David Sherwood <david.sherwood@arm.com>
25097
25098 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
25099 from an unsigned int to a poly_uint64.
25100 (_loop_vec_info::slp_unrolling_factor): Likewise.
25101 (_loop_vec_info::vectorization_factor): Change from an int
25102 to a poly_uint64.
25103 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
25104 (vect_get_num_vectors): New function.
25105 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
25106 (vect_get_num_copies): Use vect_get_num_vectors.
25107 (vect_analyze_data_ref_dependences): Change max_vf from an int *
25108 to an unsigned int *.
25109 (vect_analyze_data_refs): Change min_vf from an int * to a
25110 poly_uint64 *.
25111 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
25112 than an unsigned HOST_WIDE_INT.
25113 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
25114 (vect_analyze_data_ref_dependence): Change max_vf from an int *
25115 to an unsigned int *.
25116 (vect_analyze_data_ref_dependences): Likewise.
25117 (vect_compute_data_ref_alignment): Handle polynomial vf.
25118 (vect_enhance_data_refs_alignment): Likewise.
25119 (vect_prune_runtime_alias_test_list): Likewise.
25120 (vect_shift_permute_load_chain): Likewise.
25121 (vect_supportable_dr_alignment): Likewise.
25122 (dependence_distance_ge_vf): Take the vectorization factor as a
25123 poly_uint64 rather than an unsigned HOST_WIDE_INT.
25124 (vect_analyze_data_refs): Change min_vf from an int * to a
25125 poly_uint64 *.
25126 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
25127 vfm1 as a poly_uint64 rather than an int. Make the same change
25128 for the returned bound_scalar.
25129 (vect_gen_vector_loop_niters): Handle polynomial vf.
25130 (vect_do_peeling): Likewise. Update call to
25131 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
25132 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
25133 be constant.
25134 * tree-vect-loop.c (vect_determine_vectorization_factor)
25135 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
25136 (vect_get_known_peeling_cost): Likewise.
25137 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
25138 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
25139 (vect_transform_loop): Likewise. Use the lowest possible VF when
25140 updating the upper bounds of the loop.
25141 (vect_min_worthwhile_factor): Make static. Return an unsigned int
25142 rather than an int.
25143 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
25144 polynomial unroll factors.
25145 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
25146 (vect_make_slp_decision): Likewise.
25147 (vect_supported_load_permutation_p): Likewise, and polynomial
25148 vf too.
25149 (vect_analyze_slp_cost): Handle polynomial vf.
25150 (vect_slp_analyze_node_operations): Likewise.
25151 (vect_slp_analyze_bb_1): Likewise.
25152 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
25153 than an unsigned HOST_WIDE_INT.
25154 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
25155 (vectorizable_load): Handle polynomial vf.
25156 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
25157 a poly_uint64.
25158 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
25159
25160 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25161 Alan Hayward <alan.hayward@arm.com>
25162 David Sherwood <david.sherwood@arm.com>
25163
25164 * match.pd: Handle bit operations involving three constants
25165 and try to fold one pair.
25166
25167 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
25168
25169 * tree-vect-loop-manip.c: Include gimple-fold.h.
25170 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
25171 niters_maybe_zero parameters. Handle other cases besides a step of 1.
25172 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
25173 Add a path that uses a step of VF instead of 1, but disable it
25174 for now.
25175 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
25176 and niters_no_overflow parameters. Update calls to
25177 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
25178 Create a new SSA name if the latter choses to use a ste other
25179 than zero, and return it via niters_vector_mult_vf_var.
25180 * tree-vect-loop.c (vect_transform_loop): Update calls to
25181 vect_do_peeling, vect_gen_vector_loop_niters and
25182 slpeel_make_loop_iterate_ntimes.
25183 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
25184 (vect_gen_vector_loop_niters): Update declarations after above changes.
25185
25186 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
25187
25188 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
25189 128-bit round to integer instructions.
25190 (ceil<mode>2): Likewise.
25191 (btrunc<mode>2): Likewise.
25192 (round<mode>2): Likewise.
25193
25194 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25195
25196 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
25197 unaligned VSX load/store on P8/P9.
25198 (expand_block_clear): Allow the use of unaligned VSX
25199 load/store on P8/P9.
25200
25201 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25202
25203 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
25204 New function.
25205 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
25206 swap associated with both a load and a store.
25207
25208 2018-01-02 Andrew Waterman <andrew@sifive.com>
25209
25210 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
25211 * config/riscv/riscv.md (clear_cache): Use it.
25212
25213 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
25214
25215 * web.c: Remove out-of-date comment.
25216
25217 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25218
25219 * expr.c (fixup_args_size_notes): Check that any existing
25220 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
25221 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
25222 (emit_single_push_insn): ...here.
25223
25224 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25225
25226 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
25227 (const_vector_encoded_nelts): New function.
25228 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
25229 (const_vector_int_elt, const_vector_elt): Declare.
25230 * emit-rtl.c (const_vector_int_elt_1): New function.
25231 (const_vector_elt): Likewise.
25232 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
25233 of CONST_VECTOR_ELT.
25234
25235 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25236
25237 * expr.c: Include rtx-vector-builder.h.
25238 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
25239 directly on the tree encoding.
25240 (const_vector_from_tree): Likewise.
25241 * optabs.c: Include rtx-vector-builder.h.
25242 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
25243 sequence of "u" values.
25244 * vec-perm-indices.c: Include rtx-vector-builder.h.
25245 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
25246 directly on the vec_perm_indices encoding.
25247
25248 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25249
25250 * doc/rtl.texi (const_vector): Describe new encoding scheme.
25251 * Makefile.in (OBJS): Add rtx-vector-builder.o.
25252 * rtx-vector-builder.h: New file.
25253 * rtx-vector-builder.c: Likewise.
25254 * rtl.h (rtx_def::u2): Add a const_vector field.
25255 (CONST_VECTOR_NPATTERNS): New macro.
25256 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
25257 (CONST_VECTOR_DUPLICATE_P): Likewise.
25258 (CONST_VECTOR_STEPPED_P): Likewise.
25259 (CONST_VECTOR_ENCODED_ELT): Likewise.
25260 (const_vec_duplicate_p): Check for a duplicated vector encoding.
25261 (unwrap_const_vec_duplicate): Likewise.
25262 (const_vec_series_p): Check for a non-duplicated vector encoding.
25263 Say that the function only returns true for integer vectors.
25264 * emit-rtl.c: Include rtx-vector-builder.h.
25265 (gen_const_vec_duplicate_1): Delete.
25266 (gen_const_vector): Call gen_const_vec_duplicate instead of
25267 gen_const_vec_duplicate_1.
25268 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
25269 (gen_const_vec_duplicate): Use rtx_vector_builder.
25270 (gen_const_vec_series): Likewise.
25271 (gen_rtx_CONST_VECTOR): Likewise.
25272 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
25273 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25274 Build a new vector rather than modifying a CONST_VECTOR in-place.
25275 (handle_special_swappables): Update call accordingly.
25276 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
25277 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
25278 Build a new vector rather than modifying a CONST_VECTOR in-place.
25279 (handle_special_swappables): Update call accordingly.
25280
25281 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25282
25283 * simplify-rtx.c (simplify_const_binary_operation): Use
25284 CONST_VECTOR_ELT instead of XVECEXP.
25285
25286 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25287
25288 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
25289 the selector elements to be different from the data elements
25290 if the selector is a VECTOR_CST.
25291 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
25292 ssizetype for the selector.
25293
25294 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25295
25296 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
25297 before testing each element individually.
25298 * tree-vect-generic.c (lower_vec_perm): Likewise.
25299
25300 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25301
25302 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
25303 * selftest-run-tests.c (selftest::run_tests): Call it.
25304 * vector-builder.h (vector_builder::operator ==): New function.
25305 (vector_builder::operator !=): Likewise.
25306 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
25307 (vec_perm_indices::all_from_input_p): New function.
25308 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
25309 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
25310 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
25311 instead of reading the VECTOR_CST directly. Detect whether both
25312 vector inputs are the same before constructing the vec_perm_indices,
25313 and update the number of inputs argument accordingly. Use the
25314 utility functions added above. Only construct sel2 if we need to.
25315
25316 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25317
25318 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
25319 the broadcast of the low byte.
25320 (expand_mult_highpart): Use an explicit encoding for the permutes.
25321 * optabs-query.c (can_mult_highpart_p): Likewise.
25322 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
25323 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25324 (vectorizable_bswap): Likewise.
25325 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
25326 explicit encoding for the power-of-2 permutes.
25327 (vect_permute_store_chain): Likewise.
25328 (vect_grouped_load_supported): Likewise.
25329 (vect_permute_load_chain): Likewise.
25330
25331 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25332
25333 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
25334 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
25335 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
25336 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
25337 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
25338 (vect_gen_perm_mask_any): Likewise.
25339
25340 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25341
25342 * int-vector-builder.h: New file.
25343 * vec-perm-indices.h: Include int-vector-builder.h.
25344 (vec_perm_indices): Redefine as an int_vector_builder.
25345 (auto_vec_perm_indices): Delete.
25346 (vec_perm_builder): Redefine as a stand-alone class.
25347 (vec_perm_indices::vec_perm_indices): New function.
25348 (vec_perm_indices::clamp): Likewise.
25349 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
25350 (vec_perm_indices::new_vector): New function.
25351 (vec_perm_indices::new_expanded_vector): Update for new
25352 vec_perm_indices class.
25353 (vec_perm_indices::rotate_inputs): New function.
25354 (vec_perm_indices::all_in_range_p): Operate directly on the
25355 encoded form, without computing elided elements.
25356 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
25357 encoding. Update for new vec_perm_indices class.
25358 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
25359 the given vec_perm_builder.
25360 (expand_vec_perm_var): Update vec_perm_builder constructor.
25361 (expand_mult_highpart): Use vec_perm_builder instead of
25362 auto_vec_perm_indices.
25363 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
25364 vec_perm_indices instead of auto_vec_perm_indices. Use a single
25365 or double series encoding as appropriate.
25366 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
25367 vec_perm_indices instead of auto_vec_perm_indices.
25368 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25369 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25370 (vect_permute_store_chain): Likewise.
25371 (vect_grouped_load_supported): Likewise.
25372 (vect_permute_load_chain): Likewise.
25373 (vect_shift_permute_load_chain): Likewise.
25374 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25375 (vect_transform_slp_perm_load): Likewise.
25376 (vect_schedule_slp_instance): Likewise.
25377 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25378 (vectorizable_mask_load_store): Likewise.
25379 (vectorizable_bswap): Likewise.
25380 (vectorizable_store): Likewise.
25381 (vectorizable_load): Likewise.
25382 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
25383 vec_perm_indices instead of auto_vec_perm_indices. Use
25384 tree_to_vec_perm_builder to read the vector from a tree.
25385 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
25386 vec_perm_builder instead of a vec_perm_indices.
25387 (have_whole_vector_shift): Use vec_perm_builder and
25388 vec_perm_indices instead of auto_vec_perm_indices. Leave the
25389 truncation to calc_vec_perm_mask_for_shift.
25390 (vect_create_epilog_for_reduction): Likewise.
25391 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
25392 from auto_vec_perm_indices to vec_perm_indices.
25393 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25394 instead of changing individual elements.
25395 (aarch64_vectorize_vec_perm_const): Use new_vector to install
25396 the vector in d.perm.
25397 * config/arm/arm.c (expand_vec_perm_d::perm): Change
25398 from auto_vec_perm_indices to vec_perm_indices.
25399 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
25400 instead of changing individual elements.
25401 (arm_vectorize_vec_perm_const): Use new_vector to install
25402 the vector in d.perm.
25403 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
25404 Update vec_perm_builder constructor.
25405 (rs6000_expand_interleave): Likewise.
25406 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
25407 (rs6000_expand_interleave): Likewise.
25408
25409 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25410
25411 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
25412 to qimode could truncate the indices.
25413 * optabs.c (expand_vec_perm_var): Likewise.
25414
25415 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25416
25417 * Makefile.in (OBJS): Add vec-perm-indices.o.
25418 * vec-perm-indices.h: New file.
25419 * vec-perm-indices.c: Likewise.
25420 * target.h (vec_perm_indices): Replace with a forward class
25421 declaration.
25422 (auto_vec_perm_indices): Move to vec-perm-indices.h.
25423 * optabs.h: Include vec-perm-indices.h.
25424 (expand_vec_perm): Delete.
25425 (selector_fits_mode_p, expand_vec_perm_var): Declare.
25426 (expand_vec_perm_const): Declare.
25427 * target.def (vec_perm_const_ok): Replace with...
25428 (vec_perm_const): ...this new hook.
25429 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
25430 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
25431 * doc/tm.texi: Regenerate.
25432 * optabs.def (vec_perm_const): Delete.
25433 * doc/md.texi (vec_perm_const): Likewise.
25434 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
25435 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
25436 expand_vec_perm for constant permutation vectors. Assert that
25437 the mode of variable permutation vectors is the integer equivalent
25438 of the mode that is being permuted.
25439 * optabs-query.h (selector_fits_mode_p): Declare.
25440 * optabs-query.c: Include vec-perm-indices.h.
25441 (selector_fits_mode_p): New function.
25442 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
25443 is defined, instead of checking whether the vec_perm_const_optab
25444 exists. Use targetm.vectorize.vec_perm_const instead of
25445 targetm.vectorize.vec_perm_const_ok. Check whether the indices
25446 fit in the vector mode before using a variable permute.
25447 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
25448 vec_perm_indices instead of an rtx.
25449 (expand_vec_perm): Replace with...
25450 (expand_vec_perm_const): ...this new function. Take the selector
25451 as a vec_perm_indices rather than an rtx. Also take the mode of
25452 the selector. Update call to shift_amt_for_vec_perm_mask.
25453 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
25454 Use vec_perm_indices::new_expanded_vector to expand the original
25455 selector into bytes. Check whether the indices fit in the vector
25456 mode before using a variable permute.
25457 (expand_vec_perm_var): Make global.
25458 (expand_mult_highpart): Use expand_vec_perm_const.
25459 * fold-const.c: Includes vec-perm-indices.h.
25460 * tree-ssa-forwprop.c: Likewise.
25461 * tree-vect-data-refs.c: Likewise.
25462 * tree-vect-generic.c: Likewise.
25463 * tree-vect-loop.c: Likewise.
25464 * tree-vect-slp.c: Likewise.
25465 * tree-vect-stmts.c: Likewise.
25466 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
25467 Delete.
25468 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
25469 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
25470 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
25471 (aarch64_vectorize_vec_perm_const): ...this new function.
25472 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25473 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25474 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
25475 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
25476 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25477 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25478 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
25479 into...
25480 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
25481 check for NEON modes.
25482 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
25483 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
25484 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
25485 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
25486 into...
25487 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
25488 the old VEC_PERM_CONST conditions.
25489 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
25490 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
25491 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
25492 (ia64_vectorize_vec_perm_const_ok): Merge into...
25493 (ia64_vectorize_vec_perm_const): ...this new function.
25494 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
25495 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
25496 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
25497 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
25498 * config/mips/mips.c (mips_expand_vec_perm_const)
25499 (mips_vectorize_vec_perm_const_ok): Merge into...
25500 (mips_vectorize_vec_perm_const): ...this new function.
25501 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
25502 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
25503 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
25504 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
25505 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
25506 (rs6000_expand_vec_perm_const): Delete.
25507 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
25508 Delete.
25509 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25510 (altivec_expand_vec_perm_const_le): Take each operand individually.
25511 Operate on constant selectors rather than rtxes.
25512 (altivec_expand_vec_perm_const): Likewise. Update call to
25513 altivec_expand_vec_perm_const_le.
25514 (rs6000_expand_vec_perm_const): Delete.
25515 (rs6000_vectorize_vec_perm_const_ok): Delete.
25516 (rs6000_vectorize_vec_perm_const): New function.
25517 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
25518 an element count and rtx array.
25519 (rs6000_expand_extract_even): Update call accordingly.
25520 (rs6000_expand_interleave): Likewise.
25521 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
25522 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
25523 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
25524 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
25525 (rs6000_expand_vec_perm_const): Delete.
25526 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
25527 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25528 (altivec_expand_vec_perm_const_le): Take each operand individually.
25529 Operate on constant selectors rather than rtxes.
25530 (altivec_expand_vec_perm_const): Likewise. Update call to
25531 altivec_expand_vec_perm_const_le.
25532 (rs6000_expand_vec_perm_const): Delete.
25533 (rs6000_vectorize_vec_perm_const_ok): Delete.
25534 (rs6000_vectorize_vec_perm_const): New function. Remove stray
25535 reference to the SPE evmerge intructions.
25536 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
25537 an element count and rtx array.
25538 (rs6000_expand_extract_even): Update call accordingly.
25539 (rs6000_expand_interleave): Likewise.
25540 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
25541 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
25542 new function.
25543 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
25544
25545 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25546
25547 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
25548 vector mode and that that mode matches the mode of the data
25549 being permuted.
25550 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
25551 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
25552 directly using expand_vec_perm_1 when forcing selectors into
25553 registers.
25554 (expand_vec_perm_var): New function, split out from expand_vec_perm.
25555
25556 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25557
25558 * optabs-query.h (can_vec_perm_p): Delete.
25559 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
25560 * optabs-query.c (can_vec_perm_p): Split into...
25561 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
25562 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
25563 particular selector is valid.
25564 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
25565 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
25566 (vect_grouped_load_supported): Likewise.
25567 (vect_shift_permute_load_chain): Likewise.
25568 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
25569 (vect_transform_slp_perm_load): Likewise.
25570 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
25571 (vectorizable_bswap): Likewise.
25572 (vect_gen_perm_mask_checked): Likewise.
25573 * fold-const.c (fold_ternary_loc): Likewise. Don't take
25574 implementations of variable permutation vectors into account
25575 when deciding which selector to use.
25576 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
25577 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
25578 with a false third argument.
25579 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
25580 to test whether the constant selector is valid and can_vec_perm_var_p
25581 to test whether a variable selector is valid.
25582
25583 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25584
25585 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
25586 * optabs-query.c (can_vec_perm_p): Likewise.
25587 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
25588 instead of vec_perm_indices.
25589 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
25590 (vect_gen_perm_mask_checked): Likewise,
25591 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
25592 (vect_gen_perm_mask_checked): Likewise,
25593
25594 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
25595
25596 * optabs-query.h (qimode_for_vec_perm): Declare.
25597 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
25598 (qimode_for_vec_perm): ...this new function.
25599 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
25600
25601 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25602
25603 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
25604 does not have a conditional at the top.
25605
25606 2018-01-02 Richard Biener <rguenther@suse.de>
25607
25608 * ipa-inline.c (big_speedup_p): Fix expression.
25609
25610 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
25611
25612 PR target/81616
25613 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
25614 for generic 4->6.
25615
25616 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
25617
25618 PR target/81616
25619 Generic tuning.
25620 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
25621 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
25622 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
25623 cond_taken_branch_cost 3->4.
25624
25625 2018-01-01 Jakub Jelinek <jakub@redhat.com>
25626
25627 PR tree-optimization/83581
25628 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
25629 TODO_cleanup_cfg if any changes have been made.
25630
25631 PR middle-end/83608
25632 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
25633 convert_modes if target mode has the right side, but different mode
25634 class.
25635
25636 PR middle-end/83609
25637 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
25638 last argument when extracting from CONCAT. If either from_real or
25639 from_imag is NULL, use expansion through memory. If result is not
25640 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
25641 the parts directly to inner mode, if even that fails, use expansion
25642 through memory.
25643
25644 PR middle-end/83623
25645 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
25646 check for bswap in mode rather than HImode and use that in expand_unop
25647 too.
25648 \f
25649 Copyright (C) 2018 Free Software Foundation, Inc.
25650
25651 Copying and distribution of this file, with or without modification,
25652 are permitted in any medium without royalty provided the copyright
25653 notice and this notice are preserved.