1 2018-11-29 Alan Modra <amodra@gmail.com>
3 * config.in (HAVE_AS_PLTSEQ): Add.
4 * config/rs6000/predicates.md (indirect_call_operand): New.
5 * config/rs6000/rs6000-protos.h (rs6000_pltseq_template),
6 (rs6000_sibcall_sysv): Declare.
7 * config/rs6000/rs6000.c (init_cumulative_args): Set cookie
8 CALL_LONG for -fno-plt.
9 (print_operand <T, z, 0>): Handle UNSPEC_PLTSEQ.
10 (rs6000_indirect_call_template_1): Emit .reloc directives for
12 (rs6000_pltseq_template): New function.
13 (rs6000_longcall_ref): Add arg parameter. Use PLT16 insns if
14 relocs supported by assembler. Move SYMBOL_REF test to callers.
15 (rs6000_call_aix): Adjust rs6000_longcall_ref call. Package
16 insns in UNSPEC_PLTSEQ, preserving original func_desc.
17 (rs6000_call_sysv): Likewise.
18 (rs6000_sibcall_sysv): New function.
19 * config/rs6000/rs6000.h (HAVE_AS_PLTSEQ): Provide default.
20 * config/rs6000/rs6000.md (UNSPEC_PLTSEQ, UNSPEC_PLT16_HA,
21 UNSPEC_PLT16_LO): New.
22 (pltseq_tocsave, pltseq_plt16_ha, pltseq_plt16_lo, pltseq_mtctr): New.
23 (call_indirect_nonlocal_sysv): Don't differentiate zero from non-zero
24 cookie in constraints. Test explicitly for flags in length attr.
25 Handle unspec operand 1.
26 (call_value_indirect_nonlocal_sysv): Likewise.
27 (call_indirect_aix, call_value_indirect_aix): Handle unspec operand 1.
28 (call_indirect_elfv2, call_value_indirect_elfv2): Likewise.
29 (sibcall, sibcall_value): Use rs6000_sibcall_sysv.
30 (sibcall_indirect_nonlocal_sysv): New pattern.
31 (sibcall_value_indirect_nonlocal_sysv): Likewise.
32 (sibcall_nonlocal_sysv, sibcall_value_nonlocal_sysv): Remove indirect
34 * configure.ac: Check for gas plt sequence marker support.
35 * configure: Regenerate.
37 2018-11-29 Alan Modra <amodra@gmail.com>
39 * config/rs6000/predicates.md (unspec_tls): New.
40 * config/rs6000/rs6000-protos.h (rs6000_call_template),
41 (rs6000_sibcall_template): Update prototype.
42 (rs6000_longcall_ref): Delete.
43 (rs6000_call_sysv): Declare.
44 * config/rs6000/rs6000.c (edit_tls_call_insn): New function.
45 (global_tlsarg): New variable.
46 (rs6000_legitimize_tls_address): Rewrite __tls_get_addr call
48 (print_operand): Extract UNSPEC_TLSGD address operand.
49 (rs6000_call_template, rs6000_sibcall_template): Remove arg
50 parameter, extract from second call operand instead.
51 (rs6000_longcall_ref): Make static, localize vars.
52 (rs6000_call_aix): Rename parameter to reflect new usage. Take
53 tlsarg from global_tlsarg. Don't create unused rtl or nop insns.
54 (rs6000_sibcall_aix): Rename parameter to reflect new usage. Take
55 tlsarg from global_tlsarg.
56 (rs6000_call_sysv): New function.
57 * config/rs6000/rs6000.md: Adjust rs6000_call_template and
58 rs6000_sibcall_template throughout.
59 (tls_gd_aix, tls_gd_sysv, tls_gd_call_aix, tls_gd_call_sysv): Delete.
60 (tls_ld_aix, tls_ld_sysv, tls_ld_call_aix, tls_ld_call_sysv): Delete.
61 (tls_gdld_nomark): New insn.
62 (tls_gd): Swap operand order. Simplify mode selection.
63 (tls_gd_high, tls_gd_low): Swap operand order.
64 (tls_ld): Remove const_int 0 vector element from UNSPEC_TLSLD.
65 Simplify mode selection.
66 (tls_ld_high, tls_ld_low): Similarly adjust UNSPEC_TLSLD.
67 (call, call_value): Don't assert for second call operand.
70 2018-11-29 Alan Modra <amodra@gmail.com>
72 * config/rs6000/darwin.md (call_indirect_nonlocal_darwin64),
73 (call_nonlocal_darwin64, call_value_indirect_nonlocal_darwin64),
74 (call_value_nonlocal_darwin64): Remove constraints from second call
75 arg, the rounded_stack_size_rtx arg.
76 * config/rs6000/rs6000.md (tls_gd_aix, tls_gd_sysv, tls_gd_call_aix),
77 (tls_gd_call_sysv, tls_ld_aix, tls_ld_sysv, tls_ld_call_aix),
78 (tls_ld_call_sysv, call_local32, call_local64, call_value_local32),
79 (call_value_local64, call_indirect_nonlocal_sysv),
80 (call_nonlocal_sysv, call_nonlocal_sysv_secure),
81 (call_value_indirect_nonlocal_sysv, call_value_nonlocal_sysv),
82 (call_value_nonlocal_sysv_secure, call_local_aix),
83 (call_value_local_aix, call_nonlocal_aix, call_value_nonlocal_aix),
84 (call_indirect_aix, call_value_indirect_aix, call_indirect_elfv2),
85 (call_value_indirect_elfv2, sibcall_local32, sibcall_local64),
86 (sibcall_value_local32, sibcall_value_local64, sibcall_aix),
87 (sibcall_value_aix): Likewise.
89 2018-11-29 Alan Modra <amodra@gmail.com>
91 * config/rs6000/rs6000.md (TLSmode): Delete mode iterator. Replace
92 with P throughout except for call mems which should use SI.
93 (tls_abi_suffix, tls_sysv_suffix, tls_insn_suffix): Delete mode
94 attributes. Replace with bits, mode and ptrload respectively.
96 2018-11-29 Alan Modra <amodra@gmail.com>
98 * config/rs6000/rs6000-protos.h (rs6000_indirect_call_template),
99 (rs6000_indirect_sibcall_template): Declare.
100 * config/rs6000/rs6000.c (rs6000_indirect_call_template_1),
101 (rs6000_indirect_call_template, rs6000_indirect_sibcall_template):
103 * config/rs6000/rs6000.md (call_indirect_nonlocal_sysv),
104 (call_value_indirect_nonlocal_sysv, sibcall_nonlocal_sysv),
105 (call_indirect_aix, call_value_indirect_aix): Use
106 rs6000_indirect_call_template and rs6000_indirect_sibcall_template.
107 call_indirect_elfv2, call_value_indirect_elfv2): Likewise, and
108 handle both speculation and non-speculation cases.
109 (call_indirect_aix_nospec, call_value_indirect_aix_nospec): Delete.
110 (call_indirect_elfv2_nospec, call_value_indirect_elfv2_nospec): Delete.
112 2018-11-29 Alan Modra <amodra@gmail.com>
114 * config/rs6000/rs6000-protos.h (rs6000_call_template): Declare.
115 (rs6000_sibcall_template): Declare.
116 (macho_call_template): Rename from output_call.
117 * config/rs6000/rs6000.c (rs6000_call_template_1): New function.
118 (rs6000_call_template, rs6000_sibcall_template): Likewise.
119 (macho_call_template): Rename from output_call.
120 * config/rs6000/rs6000.md (tls_gd_aix, tls_gd_sysv),
121 (tls_gd_call_aix, tls_gd_call_sysv, tls_ld_aix, tls_ld_sysv),
122 (tls_ld_call_aix, tls_ld_call_sysv, call_nonlocal_sysv),
123 (call_nonlocal_sysv_secure, call_value_nonlocal_sysv),
124 (call_value_nonlocal_sysv_secure, call_nonlocal_aix),
125 (call_value_nonlocal_aix): Use rs6000_call_template and update
126 occurrences of output_call to macho_call_template.
127 (sibcall_nonlocal_sysv, sibcall_value_nonlocal_sysv, sibcall_aix),
128 (sibcall_value_aix): Use rs6000_sibcall_template.
130 2018-11-28 Aaron Sawdey <acsawdey@linux.ibm.com>
132 * config/rs6000/rs6000-string.c (expand_block_clear): Change how
133 we determine if unaligned vsx is ok.
135 2018-11-28 Jan Hubicka <jh@suse.cz>
137 * predict.c (determine_unlikely_bbs): Forward declare; also determine
138 edges that are always known to be taken because it is only likely
139 edge out of given BB.
140 (tree_estimate_probability): Call before profile guessing to get bit
143 2018-11-28 Jan Hubicka <jh@suse.cz>
145 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Handle
146 profile_probability::always better.
148 2018-11-28 Jan Hubicka <jh@suse.cz>
150 * profile-count.h (profile_count::split): Give better result when
151 splitting profile_probability::always.
153 2018-11-28 Vladimir Makarov <vmakarov@redhat.com>
156 * ira-costs.c (scan_one_insn): Process subregs when updating costs
157 for pseudos and allocnos from insn.
159 2018-11-28 David Edelsohn <dje.gcc@gmail.com>
161 * config/rs6000/aix72.h: Update to match aix71.h changes.
163 2018-11-28 Jeff Law <law@redhat.com>
165 * config/h8300/constraints.md: Add "C" constraint for call insns.
166 * config/h8300/h8300.md (call, call_value): Turn into a define_expand
167 and define_insn pair. Move invalid call targets into a register in
168 the expander and fix constraints in the matching pattern.
169 * config/h8300/predicates.md (call_expander_operand): Renamed from
170 call_insn_operand. Reject things we shouldn't be trying to handle.
171 (call_insn_operand): New predicate for use by the call/call_value
173 (small_call_insn_operand): Update appropriately.
175 2018-11-28 Sam Tebbs <sam.tebbs@arm.com>
177 * config/aarch64/aarch64.c (aarch64_process_target_attr): Replace
178 calls to strtok with strtok_r.
180 2018-11-28 Richard Biener <rguenther@suse.de>
182 PR tree-optimization/88223
183 * tree-ssa-sccvn.c (vn_reference_lookup_3): When skipping
184 over a stored-same value may-alias store make sure to consider
185 partial overlaps which are valid when TBAA reasonings do not
186 apply and byte-granular overlaps are possible at all.
188 2018-11-28 Richard Biener <rguenther@suse.de>
190 PR tree-optimization/88217
191 * vr-values.c (vr_values::extract_range_from_phi_node): Make
192 sure to handle results > +INF and < -INF correctly when
193 trying to drop down to +INF - 1 or -INF + 1.
195 2018-11-28 Alan Modra <amodra@gmail.com>
197 * xcoffout.c (do_block): Signed/unsigned warning fix.
199 2018-11-28 Richard Biener <rguenther@suse.de>
201 PR tree-optimization/79351
202 * tree-ssa-sccvn.c (vn_reference_lookup_3): For assignments from
203 empty CONSTRUCTORs ensure the store is at a constant position.
205 2018-11-28 Richard Biener <rguenther@suse.de>
207 PR tree-optimization/88229
208 * tree-ssa.c (non_rewritable_mem_ref_base): Check DECL_SIZE_UNIT
209 is an INTEGER_CST before accessing it so.
211 2018-11-28 Sam Tebbs <sam.tebbs@arm.com>
213 * doc/invoke.texi (-mtune=): Remove obsolete CPU names.
215 2018-11-28 Jakub Jelinek <jakub@redhat.com>
218 * config/i386/i386.c (ix86_expand_sse_movcc): Handle DFmode and
219 SFmode using sse4_1_blendvs[sd] with TARGET_SSE4_1. Formatting fixes.
220 * config/i386/sse.md (sse4_1_blendv<ssemodesuffix>): New pattern.
222 2018-11-27 Martin Liska <mliska@suse.cz>
224 * gcov.c (generate_results): Append current_working_directory
227 2018-11-27 Mike Gulick <mgulick@mathworks.com>
229 PR preprocessor/83173
230 * input.c (dump_location_info): Dump reason and included_from
231 fields from line_map_ordinary struct. Fix indentation when
233 * diagnostic-show-locus.c (num_digits, num_digits): Move to
234 diagnostic.c to allow it to be utilized by input.c.
235 * diagnostic.c (num_digits, selftest::test_num_digits): Moved
237 (selftest::diagnostic_c_tests): Run selftest::test_num_digits.
238 * diagnostic.h (num_digits): Add extern definition.
240 2018-11-27 Fredrik Noring <noring@nocrew.org>
242 * config/mips/mips.c (mips_reorg_process_insns)
243 (mips_option_override): Handle `-mfix-r5900'.
244 * config/mips/mips.h (ASM_SPEC): Add `mfix-r5900' and
246 * config/mips/mips.opt (mfix-r5900): New option.
247 * doc/invoke.texi: Document the `r5900' processor name, and
248 `-mfix-r5900' and `-mno-fix-r5900' options.
250 2018-11-27 Jakub Jelinek <jakub@redhat.com>
253 * config/rs6000/rs6000.c (print_operand) <case 'D'>: Use
254 output_operand_lossage instead of gcc_assert.
255 <case 't'>: Likewise.
256 <case 'z'>: Likewise.
257 <case 'V'>: Use output_operand_lossage instead of gcc_unreachable.
259 2018-11-27 Jeff Law <law@redhat.com>
261 * config/riscv/riscv (riscv_block_mvoe_straight): Use RETURN_BEGIN
262 in call to move_by_pieces.
264 * config/sh/sh-mem.c (expand_block_move): Use RETURN_BEGIN in call
267 * config/lm32/lm32.c (lm32_block_move_inline): Use RETURN_BEGIN in
268 call to move_by_pieces.
270 * config/mips/mips.c (mips_block_move_straight): Use RETURN_BEGIN
271 in call to move_by_pieces.
273 * config/microblaze/microblaze.c (microblaze_block_move_straight): Use
274 RETURN_BEGIN in call to move_by_pieces.
275 (microblaze_expand_block_move): Likewise.
277 2018-11-27 Tamar Christina <tamar.christina@arm.com>
279 * doc/md.texi (xorsign): Document it.
281 2018-11-27 Martin Liska <mliska@suse.cz>
283 * config/gnu-user.h (TARGET_F951_OPTIONS): New.
284 * gcc.c (find_fortran_preinclude_file): New function
285 to handle Fortran pre-include.
287 2018-11-27 Martin Liska <mliska@suse.cz>
289 * asan.c (asan_emit_stack_protection): Use new enum values
290 instead of int constants.
291 * builtins.c (expand_builtin_memory_copy_args): Replace int
292 type with memop_ret enum type.
293 (expand_builtin_mempcpy_args): Likewise.
294 (expand_builtin_memcpy): Use new enum values
295 instead of int constants. Likewise.
296 (expand_builtin_mempcpy): Likewise.
297 (expand_movstr): Likewise.
298 (expand_builtin_strcpy_args): Likewise.
299 (expand_builtin_stpcpy_1): Likewise.
300 (expand_builtin_strncpy): Likewise.
301 (expand_builtin_memset_args): Likewise.
302 * expr.c (move_by_pieces_d::finish_endp): Rename to ...
303 (move_by_pieces_d::finish_retmode): ... this.
304 (move_by_pieces): Change last argument type to memop_ret.
305 (store_by_pieces): Use new enum values
306 instead of int constants.
307 (emit_block_move_hints): Likewise.
308 (emit_push_insn): Likewise.
309 (store_expr): Likewise.
310 * expr.h (store_by_pieces): Change int to newly added enum
312 * rtl.h (enum memop_ret): Define.
313 (move_by_pieces): Use the enum type.
315 2018-11-27 Alan Modra <amodra@gmail.com>
317 * config/rs6000/aix71.h (ASM_SPEC): Don't select default -maix64
319 (ASM_CPU_SPEC): Do so here. Rewrite using if .. else if .. specs
320 form. Error on missing -mcpu case.
321 * config/rs6000/driver-rs6000.c (asm_names <_AIX>): Update NULL case.
322 (asm_names <!_AIX>): Add missing cpus. Update NULL case. Apply
323 PR63177 fix for -mcpu=power8 and -mcpu=powerpc64le.
324 * config/rs6000/rs6000.h (ASM_CPU_SPEC): Rewrite using if ..
325 else if .. specs form. Error on missing -mcpu case. Don't output
326 duplicate -maltivec. Apply PR63177 fix for -mcpu=powerpc64le.
328 2018-11-26 David Malcolm <dmalcolm@redhat.com>
330 * dump-context.h (dump_context::dump_loc): Convert 1st param from
331 dump_flags_t to const dump_metadata_t &. Convert 2nd param from
332 const dump_location_t & to const dump_user_location_t &.
333 (dump_context::dump_loc_immediate): Convert 2nd param from
334 const dump_location_t & to const dump_user_location_t &.
335 (dump_context::dump_gimple_stmt): Convert 1st param from
336 dump_flags_t to const dump_metadata_t &.
337 (dump_context::void dump_gimple_stmt_loc): Likewise; convert
338 2nd param from const dump_location_t & to
339 const dump_user_location_t &.
340 (dump_context::dump_gimple_expr): Convert 1st param from
341 dump_flags_t to const dump_metadata_t &.
342 (dump_context::dump_gimple_expr_loc): Likewise; convert
343 2nd param from const dump_location_t & to
344 const dump_user_location_t &.
345 (dump_context::dump_generic_expr): Convert 1st param from
346 dump_flags_t to const dump_metadata_t &.
347 (dump_context::dump_generic_expr_loc): Likewise; convert
348 2nd param from const dump_location_t & to
349 const dump_user_location_t &.
350 (dump_context::dump_printf_va): Convert 1st param from
351 dump_flags_t to const dump_metadata_t &.
352 (dump_context::dump_printf_loc_va): Likewise; convert
353 2nd param from const dump_location_t & to
354 const dump_user_location_t &.
355 (dump_context::dump_dec): Convert 1st param from
356 dump_flags_t to const dump_metadata_t &.
357 (dump_context::dump_symtab_node): Likewise.
358 (dump_context::begin_scope): Split out 2nd param into
359 user and impl locations.
360 (dump_context::ensure_pending_optinfo): Add metadata param.
361 (dump_context::begin_next_optinfo): Replace dump_location_t param
362 with metadata and user location.
363 * dumpfile.c (dump_context::dump_loc): Convert 1st param from
364 dump_flags_t to const dump_metadata_t &. Convert 2nd param from
365 const dump_location_t & to const dump_user_location_t &.
366 (dump_context::dump_loc_immediate): Convert 2nd param from
367 const dump_location_t & to const dump_user_location_t &.
368 (dump_context::dump_gimple_stmt): Convert 1st param from
369 dump_flags_t to const dump_metadata_t &.
370 (dump_context::void dump_gimple_stmt_loc): Likewise; convert
371 2nd param from const dump_location_t & to
372 const dump_user_location_t &.
373 (dump_context::dump_gimple_expr): Convert 1st param from
374 dump_flags_t to const dump_metadata_t &.
375 (dump_context::dump_gimple_expr_loc): Likewise; convert
376 2nd param from const dump_location_t & to
377 const dump_user_location_t &.
378 (dump_context::dump_generic_expr): Convert 1st param from
379 dump_flags_t to const dump_metadata_t &.
380 (dump_context::dump_generic_expr_loc): Likewise; convert
381 2nd param from const dump_location_t & to
382 const dump_user_location_t &.
383 (dump_context::dump_printf_va): Convert 1st param from
384 dump_flags_t to const dump_metadata_t &.
385 (dump_context::dump_printf_loc_va): Likewise; convert
386 2nd param from const dump_location_t & to
387 const dump_user_location_t &.
388 (dump_context::dump_dec): Convert 1st param from
389 dump_flags_t to const dump_metadata_t &.
390 (dump_context::dump_symtab_node): Likewise.
391 (dump_context::begin_scope): Split out 2nd param into
392 user and impl locations.
393 (dump_context::ensure_pending_optinfo): Add metadata param.
394 (dump_context::begin_next_optinfo): Replace dump_location_t param
395 with metadata and user location.
396 (dump_gimple_stmt): Convert 1st param from dump_flags_t to
397 const dump_metadata_t &.
398 (dump_gimple_stmt_loc): Likewise; convert 2nd param from
399 const dump_location_t & to const dump_user_location_t &.
400 (dump_gimple_expr): Convert 1st param from dump_flags_t to
401 const dump_metadata_t &.
402 (dump_gimple_expr_loc): Likewise; convert 2nd param from
403 const dump_location_t & to const dump_user_location_t &.
404 (dump_generic_expr): Convert 1st param from dump_flags_t to
405 const dump_metadata_t &.
406 (dump_generic_expr_loc): Likewise; convert 2nd param from
407 const dump_location_t & to const dump_user_location_t &.
408 (dump_printf): Convert 1st param from dump_flags_t to
409 const dump_metadata_t &.
410 (dump_printf_loc): Likewise; convert 2nd param from
411 const dump_location_t & to const dump_user_location_t &.
412 (dump_dec): Convert 1st param from dump_flags_t to
413 const dump_metadata_t &.
414 (dump_symtab_node): Likewise.
415 (dump_begin_scope): Split out 2nd param into user and impl
417 (selftest::assert_impl_location_eq): New function.
418 (ASSERT_IMPL_LOCATION_EQ): New macro.
419 (selftest::test_impl_location): Update to use
420 ASSERT_IMPL_LOCATION_EQ.
421 (selftest::test_capture_of_dump_calls): Convert "loc" to
422 dump_user_location_t. Add ASSERT_IMPL_LOCATION_EQ throughout,
423 verifying line numbers of dump emissions.
424 * dumpfile.h (class dump_metadata_t): New class.
425 (dump_printf): Convert 1st param from dump_flags_t to
426 const dump_metadata_t &.
427 (dump_printf_loc): Likewise; convert 2nd param from
428 const dump_location_t & to const dump_user_location_t &.
429 (dump_generic_expr_loc): Likewise.
430 (dump_generic_expr): Convert 1st param from dump_flags_t to
431 const dump_metadata_t &.
432 (dump_gimple_stmt_loc): Likewise; convert 2nd param from
433 const dump_location_t & to const dump_user_location_t &.
434 (dump_gimple_stmt): Convert 1st param from dump_flags_t to
435 const dump_metadata_t &.
436 (dump_gimple_expr_loc): Likewise; convert 2nd param from
437 const dump_location_t & to const dump_user_location_t &.
438 (dump_gimple_expr): Convert 1st param from dump_flags_t to
439 const dump_metadata_t &.
440 (dump_symtab_node): Likewise.
441 (dump_dec): Likewise.
442 (dump_begin_scope): Split out 2nd param into user and impl
444 (auto_dump_scope::auto_dump_scope): Split "loc" param into a user
445 location and impl_location, and capture the impl_location.
446 (AUTO_DUMP_SCOPE): Rename param from LOC to USER_LOC.
447 * loop-unroll.c (report_unroll): Update for changes to
448 dump_printf_loc and dump_printf.
449 * opt-problem.cc (opt_problem::opt_problem): Update for change to
451 * optinfo-emit-json.cc
452 (selftest::test_building_json_from_dump_calls): Convert "loc" from
453 dump_location_t to dump_user_location_t.
454 * optinfo.cc (optinfo::emit_for_opt_problem): Update for change in
456 * profile.c (compute_branch_probabilities): Update for change to
458 * selftest.h (ASSERT_STR_CONTAINS_AT): New macro.
459 * tree-vect-slp.c (vect_print_slp_tree): Update for change to
462 2018-11-27 Jozef Lawrynowicz <jozef.l@mittosystems.com>
464 * doc/sourcebuild.texi: Document check_effective_target_int_eq_float
465 and check_effective_target_ptr_eq_long.
467 2018-11-27 Alan Modra <amodra@gmail.com>
469 * config.gcc (powerpc*-*-freebsd*, powerpc-*-netbsd*),
470 (powerpc-*-eabisimaltivec*, powerpc-*-eabisim*, powerpc-*-elf*),
471 (powerpc-*-eabialtivec*, powerpc-*-eabi*, powerpc-*-rtems*),
472 (powerpc-wrs-vxworks*, powerpc-*-lynxos*, powerpcle-*-elf*),
473 (powerpcle-*-eabisim*, powerpcle-*-eabi*): Add gnu-user.h to tm_file.
474 * config/rs6000/freebsd.h (CPLUSPLUS_CPP_SPEC),
475 (LINK_GCC_C_SEQUENCE_SPEC): Undef.
476 (ASM_APP_ON, ASM_APP_OFF): Don't define.
477 * config/rs6000/freebsd64.h (ASM_APP_ON, ASM_APP_OFF): Don't define.
478 * config/rs6000/lynx.h (ASM_APP_ON, ASM_APP_OFF): Don't define.
479 * config/rs6000/linux64.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
480 * config/rs6000/netbsd.h (CPLUSPLUS_CPP_SPEC),
481 (LINK_GCC_C_SEQUENCE_SPEC): Undef.
482 * config/rs6000/rtems.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
483 * config/rs6000/sysv4.h (GNU_USER_TARGET_CRTI): Redefine.
484 (GNU_USER_TARGET_CRTN): Redefine.
485 (CC1_SPEC): Use GNU_USER_TARGET_CC1_SPEC.
486 (LIB_LINUX_SPEC): Use GNU_USER_TARGET_LIB_SPEC.
487 (CRTOFFLOADBEGIN, CRTOFFLOADEND): Don't define.
488 (STARTFILE_LINUX_SPEC): Define as GNU_USER_TARGET_STARTFILE_SPEC.
489 (ENDFILE_LINUX_SPEC): Define as GNU_USER_TARGET_ENDFILE_SPEC.
490 (UCLIBC_DYNAMIC_LINKER, CHOOSE_DYNAMIC_LINKER): Don't define.
491 (LINK_EH_SPEC): Don't define.
493 2018-11-26 Martin Sebor <msebor@redhat.com>
496 * expr.c (string_constant): Handle top-level decls of all character
497 types and subobjects of narrow character type.
499 2018-11-27 Alan Modra <amodra@gmail.com>
501 * config.gcc (powerpc*-*-linux*): Add linux.h to tm_file.
502 * config/rs6000/linux.h (TARGET_OS_CPP_BUILTINS): Use
503 GNU_USER_TARGET_OS_CPP_BUILTINS.
504 (RS6000_ABI_NAME): Define.
505 * config/rs6000/linux64.h (TARGET_OS_CPP_BUILTINS): Use
506 GNU_USER_TARGET_OS_CPP_BUILTINS.
507 (MUSL_DYNAMIC_LINKER32): Undef before defining.
508 (UCLIBC_DYNAMIC_LINKER32, UCLIBC_DYNAMIC_LINKER64): Don't define.
509 (CHOOSE_DYNAMIC_LINKER): Don't define.
510 (GNU_USER_DYNAMIC_LINKER32, GNU_USER_DYNAMIC_LINKER64): Don't define.
511 * config/rs6000/sysv4.h (MUSL_DYNAMIC_LINKER): Undef before defining.
512 (CHOOSE_DYNAMIC_LINKER, GNU_USER_DYNAMIC_LINKER): Only define when
514 (CPP_OS_LINUX_SPEC): Remove defines and asserts handled by
515 TARGET_OS_CPP_BUILTINS.
517 2018-11-26 Jozef Lawrynowicz <jozef.l@mittosystems.com>
519 * doc/sourcebuild.texi: Document check_effective_target_newlib_nano_io.
521 2018-11-26 Jakub Jelinek <jakub@redhat.com>
524 * config/i386/i386.c (def_builtin2): If tcode == VOID_FTYPE_UINT64
525 and !TARGET_64BIT, return NULL_TREE.
528 * dwarf2out.c (secname_for_decl): For functions with
529 DECL_SECTION_NAME if in_cold_section_p, try to return
530 current_function_section's name if it is a named section.
532 2018-11-26 Maya Rashish <coypu@sdf.org>
535 * config.host (*-*-netbsd*): Use x-netbsd and host-netbsd.o.
536 * config/x-netbsd: New file.
537 * config/host-netbsd: New file.
539 2018-11-26 Jozef Lawrynowicz <jozef.l@mittosystems.com>
541 * doc/sourcebuild.texi: Document check_effective_target_size20plus.
542 Clarify documentation for check_effective_target_size32plus.
544 2018-11-26 Aaron Sawdey <acsawdey@linux.ibm.com>
546 * config/rs6000/rs6000-string.c (expand_cmp_vec_sequence): Rename
547 and modify expand_strncmp_vec_sequence.
548 (emit_final_compare_vec): Rename and modify emit_final_str_compare_vec.
549 (generate_6432_conversion): New function.
550 (expand_block_compare): Add support for vsx.
551 (expand_block_compare_gpr): New function.
552 * config/rs6000/rs6000.opt (rs6000_block_compare_inline_limit): Increase
553 default limit to 63 because of more compact vsx code.
555 2018-11-26 Uros Bizjak <ubizjak@gmail.com>
558 * config/i386/i386.c (dbx_register_map): Use IGNORED_DWARF_REGNUM
559 for registers for which no debug information can be generated.
560 Use INVALID_REGNUM for non-existent registers.
561 (dbx64_register_map): Ditto.
562 (svr4_dbx_register_map): Ditto.
564 2018-11-26 Sandra Loosemore <sandra@codesourcery.com>
567 * doc/invoke.texi (Optimize Options): Document that
568 -freorder-functions also uses "hot" and "cold" attributes.
570 2018-11-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
572 * config/aarch64/aarch64.c (aarch64_mangle_type): Fix typo in comment.
574 2018-11-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
576 * config/arm/types.md (mul64): Rename to...
577 (widen_mul64): ... This.
578 * config/arm/arm-generic.md: Rename mul64 to widen_mul64.
579 * config/arm/cortex-a15.md: Likewise.
580 * config/arm/cortex-a5.md: Likewise.
581 * config/arm/cortex-a53.md: Likewise.
582 * config/arm/cortex-a57.md: Likewise.
583 * config/arm/cortex-a7.md: Likewise.
584 * config/arm/cortex-m4.md: Likewise.
585 * config/arm/exynos-m1.md: Likewise.
586 * config/arm/marvell-pj4.md: Likewise.
587 * config/arm/xgene1.md: Likewise.
589 2018-11-26 Richard Biener <rguenther@suse.de>
591 PR tree-optimization/88182
592 * tree-vect-loop.c (vectorizable_reduction): Pick up single
593 correct reduc_def_info.
594 * tree-vect-slp.c (vect_analyze_slp_instance): Set
595 STMT_VINFO_REDUC_DEF of the first stmt.
597 2018-11-26 Andreas Krebbel <krebbel@linux.ibm.com>
599 * doc/invoke.texi: Document z14/arch12 -march option.
601 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
604 * doc/invoke.texi (Option Summary): Add -Wmissing-noreturn.
605 (Warning Options): Likewise.
607 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
610 * doc/extend.texi (Common Function Attributes): Clarify that
611 functions with "const" attribute can read const global variables.
613 2018-11-26 Alan Modra <amodra@gmail.com>
615 * config/rs6000/rs6000.h (OBJECT_PEF, TARGET_MACOS): Delete.
617 2018-11-25 Sandra Loosemore <sandra@codesourcery.com>
620 * doc/extend.texi (Common Variable Attributes): Use preferred
621 placement of type attributes in examples, plus whitespace fixes.
622 (Type Attributes): Clarify why placement of attributes
623 immediately after struct/union/enum keyword is preferred.
624 (Common Type Attributes): Use preferred placement of type
625 attributes in examples, plus more whitespace fixes.
627 2018-11-25 Paul Koning <ni1d@arrl.net>
629 * config/pdp11/pdp11.h (TARGET_HAS_NO_HW_DIVIDE): Define.
631 2018-11-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
633 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
634 attributes): Document d_runtime.
636 2018-11-25 Vladimir Makarov <vmakarov@redhat.com>
639 * ira-costs.c (record_operand_costs): Initiate register move cost
642 2018-11-23 Jeff Law <law@redhat.com>
644 PR rtl-optimization/87468
645 * tree-ssa-threadupdate.c (create_block_for_threading): Clear
646 EDGE_IGNORE on all outgoing edges of the duplicate block.
648 2018-11-23 Vladimir Makarov <vmakarov@redhat.com>
651 * ira-costs.c (record_operand_costs): Use bigger hard reg class if
652 its mode does not fit to the original class.
654 2018-11-23 Martin Sebor <msebor@redhat.com>
656 PR tree-optimization/87756
657 * expr.c (string_constant): Synthesize a string literal from
658 the address of a constant character.
659 * tree.c (build_string_literal): Add an argument.
660 * tree.h (build_string_literal): Same.
662 2018-11-23 Christoph Muellner <christoph.muellner@theobroma-systems.com>
664 * config/aarch64/aarch64-cores.def: Define emag.
665 * config/aarch64/aarch64-tune.md: Regenerated with emag.
666 * config/aarch64/aarch64.c (emag_tunings): New struct.
667 * doc/invoke.texi: Document mtune value.
669 2018-11-23 Christoph Muellner <christoph.muellner@theobroma-systems.com>
670 Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
672 * config/aarch64/aarch64.c (xgene1_tunings): Optimize Xgene1 tunings
674 * config/aarch64/aarch64.c (xgene1_tunings): Add Xgene1 specific
676 * config/aarch64/aarch64.c (xgene1_addrcost_table): Correct the post-modify
678 * config/arm/aarch-cost-tables.h (xgene1_extra_costs): Update the cost table
681 2018-11-23 Richard Biener <rguenther@suse.de>
683 PR tree-optimization/88149
684 * tree-vect-slp.c (vect_slp_analyze_node_operations): Detect
685 the case where there are two different def types for the
686 same operand at different operand position in the same stmt.
688 2018-23-11 Mihail Ionescu <mihail.ionescu@arm.com>
690 * config/arm/arm.c (arm_expand_compare_and_swap): Simplify and call
691 gen_atomic_compare_swap_1.
692 (arm_evpc_neon_vuzp): Likewise gen_neon_vuzp_internal.
693 (arm_evpc_neon_vtrn): Likewise gen_neon_vtrn_internal.
694 (arm_evpc_neon_vext): Likewise gen_neon_vext_internal.
695 (arm_evpc_neon_vzip): Likewise gen_neon_vzip_internal.
696 (arm_evpc_neon_vrev): Replace the function pointer and simplify the mode
698 * config/arm/arm.md (neon_vext<mode>),
699 (neon_vrev64<mode>, neon_vrev32<mode>),
700 (neon_vrev16<mode>, neon_vtrn<mode>_internal),
701 (neon_vzip<mode>_internal, neon_vuzp<mode>_internal): Add an '@'character
702 before the pattern name.
703 * config/arm/sync.md:
704 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1),
705 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): Likewise.
707 2018-11-23 Jakub Jelinek <jakub@redhat.com>
709 PR tree-optimization/86614
710 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return early
711 if TREE_NO_WARNING is set on ref.ref.
713 2018-11-22 Sandra Loosemore <sandra@codesourcery.com>
714 Alan Coopersmith <alan.coopersmith@oracle.com>
717 * doc/extend.texi (Designated Inits): Clarify handling of multiple
718 initializers for unions.
720 2018-11-22 Jan Hubicka <jh@suse.cz>
723 * ipa-devirt.c (type_variants_equivalent_p): Drop warn and warned
724 parameters; do not warn here.
725 (odr_subtypes_equivalent_p): Likewise.
726 (warn_odr): Fix typo.
727 (warn_types_mismatch): Do not output confused warnings on integer types.
728 (odr_types_equivalent_p): Update.
730 2018-11-22 Jakub Jelinek <jakub@redhat.com>
732 * config/i386/i386.c (ix86_option_override_internal,
733 ix86_can_inline_p, classify_argument, construct_container,
734 ix86_expand_prologue, ix86_expand_split_stack_prologue,
735 ix86_expand_carry_flag_compare, expand_set_or_movmem_via_loop,
736 expand_setmem_epilogue_via_loop, promote_duplicated_reg,
737 ix86_expand_set_or_movmem, ix86_init_builtins_va_builtins_abi):
740 * config/i386/i386.c (ix86_option_override_internal): For
741 stack_protector_guard related options, use opts_set->x_ instead
742 of global_options_set. and prefix options with opts->x_ . Move
743 defaults for offset and reg into else block.
745 2018-11-22 Uros Bizjak <ubizjak@gmail.com>
747 * config/i386/i386.c (ix86_check_avx_upper_register):
748 Return true for all SSE registers with mode bitsize > 128.
750 2018-11-22 Vladimir Makarov <vmakarov@redhat.com>
752 PR rtl-optimization/87718
753 * ira-costs.c: Remove trailing white-spaces.
754 (record_operand_costs): Add a special treatment for moves
755 involving a hard register.
757 2018-11-22 Uros Bizjak <ubizjak@gmail.com>
759 * config/i386/i386.c (ix86_avx_emit_vzeroupper): Remove.
760 (ix86_emit_mode_set) <case AVX_U128>: Emit vzeroupper here.
762 2018-11-22 Martin Liska <mliska@suse.cz>
764 * common/config/i386/i386-common.c (processor_names): Add
765 static assert and add missing "znver2".
766 (ix86_get_valid_option_values): Add checking assert for null
767 values and add "native" value if feasible.
768 * config/i386/i386.h: Do not declare size of processor_names.
769 * common/config/i386/i386-common.c:
770 * config/i386/i386.c: Add static assert for size
771 of processor_cost_table.
773 2018-11-22 Thomas Preud'homme <thomas.preudhomme@linaro.org>
775 * target-insns.def (stack_protect_combined_set): Define new standard
777 (stack_protect_combined_test): Likewise.
778 * cfgexpand.c (stack_protect_prologue): Try new
779 stack_protect_combined_set pattern first.
780 * function.c (stack_protect_epilogue): Try new
781 stack_protect_combined_test pattern first.
782 * config/arm/arm.c (require_pic_register): Add pic_reg and compute_now
783 parameters to control which register to use as PIC register and force
784 reloading PIC register respectively. Insert in the stream of insns if
786 (legitimize_pic_address): Expose above new parameters in prototype and
787 adapt recursive calls accordingly. Use pic_reg if non null instead of
789 (arm_load_pic_register): Add pic_reg parameter and use it if non null.
790 (arm_legitimize_address): Adapt to new legitimize_pic_address
792 (thumb_legitimize_address): Likewise.
793 (arm_emit_call_insn): Adapt to require_pic_register prototype change.
794 (arm_expand_prologue): Adapt to arm_load_pic_register prototype change.
795 (thumb1_expand_prologue): Likewise.
796 * config/arm/arm-protos.h (legitimize_pic_address): Adapt to prototype
798 (arm_load_pic_register): Likewise.
799 * config/arm/predicated.md (guard_addr_operand): New predicate.
800 (guard_operand): New predicate.
801 * config/arm/arm.md (movsi expander): Adapt to legitimize_pic_address
803 (builtin_setjmp_receiver expander): Adapt to thumb1_expand_prologue
805 (stack_protect_combined_set): New expander..
806 (stack_protect_combined_set_insn): New insn_and_split pattern.
807 (stack_protect_set_insn): New insn pattern.
808 (stack_protect_combined_test): New expander.
809 (stack_protect_combined_test_insn): New insn_and_split pattern.
810 (arm_stack_protect_test_insn): New insn pattern.
811 * config/arm/thumb1.md (thumb1_stack_protect_test_insn): New insn pattern.
812 * config/arm/unspecs.md (UNSPEC_SP_SET): New unspec.
813 (UNSPEC_SP_TEST): Likewise.
814 * doc/md.texi (stack_protect_combined_set): Document new standard
816 (stack_protect_set): Clarify that the operand for guard's address is
818 (stack_protect_combined_test): Document new standard pattern name.
819 (stack_protect_test): Clarify that the operand for guard's address is
822 2018-11-22 Richard Biener <rguenther@suse.de>
824 PR tree-optimization/88148
825 * tree-ssa-loop-niter.c (simplify_replace_tree): Get optional
826 valueization callback parameter and handle it.
827 * tree-ssa-loop-niter.h (simplify_replace_tree): Export.
828 * tree-ssa-sccvn.c (process_bb): Eliminate in loop niter trees.
830 2018-11-22 Richard Biener <rguenther@suse.de>
834 * lto-streamer-out.c (lto_is_streamable): Allow CALL_EXPRs
835 which can appear in size expressions.
836 * tree-streamer-in.c (unpack_ts_base_value_fields): Stream
837 CALL_EXPR_BY_DESCRIPTOR.
838 (streamer_read_tree_bitfields): Stream CALL_EXPR_IFN.
839 * tree-streamer-out.c (pack_ts_base_value_fields): Stream
840 CALL_EXPR_BY_DESCRIPTOR.
841 (streamer_write_tree_bitfields): Stream CALL_EXPR_IFN.
845 * tree.c (free_lang_data_in_one_sizepos): Free non-gimple-val
848 2018-11-22 Richard Biener <rguenther@suse.de>
850 PR tree-optimization/88069
851 * tree-ssa-sccvn.c (visit_phi): Tweak previous fix to not
852 apply to default defs.
854 2018-11-22 Jakub Jelinek <jakub@redhat.com>
858 * config/i386/i386.c (ix86_option_override_internal): Default
859 ix86_stack_protector_guard to SSP_TLS only if TARGET_THREAD_SSP_OFFSET
861 * config/i386/i386.md (stack_protect_set, stack_protect_set_<mode>,
862 stack_protect_test, stack_protect_test_<mode>): Use empty condition
863 instead of TARGET_SSP_TLS_GUARD.
865 2018-11-22 Martin Liska <mliska@suse.cz>
868 * config/i386/i386.c (ix86_option_override_internal): Error
869 about usage -mabi=ms and -fsanitize=thread.
871 2018-11-21 Uros Bizjak <ubizjak@gmail.com>
874 2013-10-26 Vladimir Makarov <vmakarov@redhat.com>
877 2013-10-25 Vladimir Makarov <vmakarov@redhat.com>
879 * lra-spills.c (lra_final_code_change): Remove useless move insns.
881 2018-11-21 Uros Bizjak <ubizjak@gmail.com>
884 * function.c (expand_function_end): Do not emit extra blockage insn.
886 2018-11-21 Lokesh Janghel <lokeshjanghel91@gmail.com>
889 * config/i386/i386.c (function_value_ms_64): Return AX_REG instead
890 of FIRST_SSE_REG for 4 or 8 byte modes.
892 2018-11-21 Jan Hubicka <jh@suse.cz>
895 * tree.c (fld_decl_context): Break out from ...
896 (free_lang_data_in_decl): ... here; free TREE_PUBLIC, TREE_PRIVATE
897 DECL_ARTIFICIAL of TYPE_DECL; do not free TREE_TYPE of TYPE_DECL.
898 (fld_incomplete_type_of): Build copy of TYP_DECL.
899 * ipa-devirt.c (free_enum_values): Rename to ...
900 (free_odr_warning_data): ... this one; free also duplicated TYPE_DECLs
901 and TREE_TYPEs of TYPE_DECLs.
902 (get_odr_type): Initialize odr_vtable_hash if needed.
904 2018-11-21 Alexandre Oliva <oliva@adacore.com>
906 * final.c (compute_discriminator): Declare. Renamed from...
907 (maybe_set_discriminator): ... this. Set and return a local.
908 (override_discriminator): New.
909 (final_scan_insn_1): Set it.
910 (notice_source_line): Adjust. Always set discriminator.
912 2018-11-21 Jakub Jelinek <jakub@redhat.com>
915 * config/aarch64/atomics.md (@aarch64_compare_and_swap<mode>): Use
916 rIJ constraint for aarch64_plus_operand rather than rn.
918 2018-11-21 Renlin Li <renlin.li@arm.com>
921 * explow.h (get_dynamic_stack_size): Declare it as external.
922 * explow.c (record_new_stack_level): Remove function static attribute.
923 * function.c (assign_stack_local_1): Dynamically align the stack slot
924 addr for parameter copy on the stack.
926 2018-11-21 Richard Biener <rguenther@suse.de>
929 * bitmap.c (bitmap_last_set_bit): Refactor to avoid warning.
930 * Makefile.in (bitmap.o-warn): Remove again.
932 2018-11-20 Jeff Law <law@redhat.com>
934 PR tree-optimization/88069
935 * tree-ssa-dom.c (record_equivalences_from_phis): Propagate away
936 degenerate virtual PHIs.
938 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
941 * config/i386/sse.md (sse4_1_<code>v8qiv8hi2<mask_name>): Replace
942 nonimmediate_operand with register_operand.
943 (avx2_<code>v8qiv8si2<mask_name>): Likewise.
944 (sse4_1_<code>v4qiv4si2<mask_name>): Likewise.
945 (sse4_1_<code>v4hiv4si2<mask_name>): Likewise.
946 (sse4_1_<code>v2qiv2di2<mask_name>): Likewise.
947 (avx512f_<code>v8qiv8di2<mask_name>): Likewise.
948 (avx2_<code>v4qiv4di2<mask_name>): Likewise.
949 (avx2_<code>v4hiv4di2<mask_name>): Likewise.
950 (sse4_1_<code>v2hiv2di2<mask_name>): Likewise.
951 (sse4_1_<code>v2siv2di2<mask_name>): Likewise.
952 (*sse4_1_<code>v8qiv8hi2<mask_name>_1): New pattern.
953 (*sse4_1_<code>v8qiv8hi2<mask_name>_2): Likewise.
954 (*avx2_<code>v8qiv8si2<mask_name>_1): Likewise.
955 (*avx2_<code>v8qiv8si2<mask_name>_2): Likewise.
956 (*sse4_1_<code>v4qiv4si2<mask_name>_1): Likewise.
957 (*sse4_1_<code>v4qiv4si2<mask_name>_2): Likewise.
958 (*sse4_1_<code>v4hiv4si2<mask_name>_1): Likewise.
959 (*sse4_1_<code>v4hiv4si2<mask_name>_2): Likewise.
960 (*avx512f_<code>v8qiv8di2<mask_name>_1): Likewise.
961 (*avx512f_<code>v8qiv8di2<mask_name>_2): Likewise.
962 (*avx2_<code>v4qiv4di2<mask_name>_1): Likewise.
963 (*avx2_<code>v4qiv4di2<mask_name>_2): Likewise.
964 (*avx2_<code>v4hiv4di2<mask_name>_1): Likewise.
965 (*avx2_<code>v4hiv4di2<mask_name>_2): Likewise.
966 (*sse4_1_<code>v2hiv2di2<mask_name>_1): Likewise.
967 (*sse4_1_<code>v2hiv2di2<mask_name>_2): Likewise.
968 (*sse4_1_<code>v2siv2di2<mask_name>_1): Likewise.
969 (*sse4_1_<code>v2siv2di2<mask_name>_2): Likewise.
971 2018-11-21 H.J. Lu <hongjiu.lu@intel.com>
973 * read-rtl.c (apply_subst_iterator): Handle define_split and
974 define_insn_and_split.
976 2018-11-21 Jakub Jelinek <jakub@redhat.com>
978 PR rtl-optimization/87817
979 * config/i386/i386.md (bmi2_bzhi_<mode>3, *bmi2_bzhi_<mode>3,
980 *bmi2_bzhi_<mode>3_1, *bmi2_bzhi_<mode>3_1_ccz): Use IF_THEN_ELSE
981 in the pattern to avoid triggering UB when operands[2] is zero.
982 (tbm_bextri_<mode>): New expander. Renamed the old define_insn to ...
983 (*tbm_bextri_<mode>): ... this.
985 2018-11-21 Tom de Vries <tdevries@suse.de>
988 * params.def (HOT_BB_COUNT_FRACTION): Terminate help message with
991 2018-11-21 Andreas Krebbel <krebbel@linux.ibm.com>
993 * configure.ac: Add check for Binutils to determine whether vector
994 load/store alignments hints are being supported.
995 * config.in: Regenerate.
996 * configure: Regenerate.
997 * config/s390/s390.c (print_operand): Support new output
999 * config/s390/s390.md ("movti"): Append alignment hint output
1000 using the new output modifier 'A'.
1001 * config/s390/vector.md ("mov<mode>", "*vec_tf_to_v1tf")
1002 ("*vec_ti_to_v1ti"): Likewise.
1004 2018-11-20 Martin Sebor <msebor@redhat.com>
1006 * doc/extend.texi (Other Builtins): Add __builtin_has_attribute.
1008 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1011 * ipa-devirt.c (odr_types_equivalent_p): Use operand_equal_p to
1012 compare ENUM values.
1014 2018-11-20 Jakub Jelinek <jakub@redhat.com>
1016 PR tree-optimization/87895
1017 * omp-simd-clone.c (ipa_simd_modify_function_body): When removing
1018 or replacing GIMPLE_RETURN, set EDGE_FALLTHRU on the edge to EXIT.
1019 (simd_clone_adjust): Don't set EDGE_FALLTHRU here. In a loop that
1020 redirects edges to EXIT to edges to incr_bb, iterate while EXIT
1021 has any preds and always use EDGE_PRED (, 0).
1023 2018-11-20 Uros Bizjak <ubizjak@gmail.com>
1026 * mode-switching.c (create_pre_exit): After reload, always split the
1027 fallthrough edge to the exit block.
1029 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1031 * ipa-devirt.c (add_type_duplicate): Do not ICE on incomplete enums.
1032 * tree.c (build_array_type_1): Forward declare.
1033 (fld_type_variant_equal_p): Add INNER_TYPE parameter.
1034 (fld_type_variant): Likewise.
1035 (fld_simplified_types): New hash.
1036 (fld_process_array_type): New function.
1037 (fld_incomplete_type_of): Handle array and enumeration types.
1038 (fld_simplified_type): Handle simplification of arrays.
1039 (free_lang_data): Allocate and free simplified types hash.
1041 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1044 * ipa-devirt.c (odr_subtypes_equivalent_p): Report ODR violation
1045 when sybtype already violates ODR.
1046 (get_odr_type): Do not ICE when insert is false and type duplicate
1047 is not registered yet.
1048 (register_odr_type): Be sure to register subtypes first.
1050 2018-11-20 Andreas Krebbel <krebbel@linux.ibm.com>
1052 * config/s390/s390.md ("clztidi2"): Swap the RTX's written to the
1053 DImode parts of the target operand.
1055 2018-11-20 Nathan Sidwell <nathan@acm.org>
1058 * Makefile.in (bitmap.o-warn): Use -Wno-error=array-bounds.
1060 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1063 * tree.c (free_lang_data_in_cgraph): Add argument fld; break out
1065 (free_lang_data) ... here; update call of free_lang_data_in_cgraph.
1067 2018-11-20 Jan Hubicka <hubicka@ucw.cz>
1070 * ipa-fnsummary.c (pass_ipa_fnsummary): Do not remove functions
1071 * ipa.c (possible_inline_candidate_p): Break out from ..
1072 (process_references): ... here ; drop before_inlining_p;
1073 cleanup handling of alises.
1074 (walk_polymorphic_call_targets): Likewise.
1075 (symbol_table::remove_unreachable_nodes): Likewise.
1076 * passes.c (pass_data_ipa_remove_symbols): New structure.
1077 (pass_ipa_remove_symbols): New pass.
1078 (make_pass_ipa_remove_symbols): New function.
1079 * tree-pass.h (make_pass_ipa_remove_symbols): Declare.
1080 * passes.def (pass_ipa_remove_symbols): Schedule after early passes.
1082 2018-11-20 Richard Biener <rguenther@suse.de>
1084 * tree-vect-stmts.c (vectorizable_condition): Do not get
1085 at else_clause vect def for EXTRACT_LAST_REDUCTION. Remove
1086 pointless vect_is_simple_use calls.
1088 2018-11-20 Richard Biener <rguenther@suse.de>
1090 PR tree-optimization/88074
1091 * tree-ssa-dom.c (pass_dominator::execute): Do not walk
1094 2018-11-20 Richard Biener <rguenther@suse.de>
1096 PR tree-optimization/88069
1097 * tree-ssa-sccvn.c (visit_phi): Do not value-number to unvisited
1098 virtual PHI arguments.
1100 2018-11-20 Ilya Leoshkevich <iii@linux.ibm.com>
1103 * config/s390/s390.md: Skip LT(G) peephole when literal pool is
1105 * rtl.h (contains_constant_pool_address_p): New function.
1106 * rtlanal.c (contains_constant_pool_address_p): Likewise.
1108 2018-11-20 Richard Biener <rguenther@suse.de>
1111 * alias.c (component_uses_parent_alias_set_from): Remove
1112 alias-set zero and TYPE_TYPELESS_STORAGE case both already
1113 handled in other ways.
1115 2018-11-20 Richard Biener <rguenther@suse.de>
1117 PR tree-optimization/88087
1118 * tree-ssa-pre.c (create_expression_by_pieces): Re-materialize
1120 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Remember
1123 2018-11-20 Richard Biener <rguenther@suse.de>
1126 * tree-data-ref.c (lambda_matrix_right_hermite): Use abs_hwi.
1128 2018-11-19 Jan Hubicka <hubicka@ucw.cz>
1131 * ipa-devirt.c (free_enum_values): Do not ICE on ODR vilations.
1133 2018-11-19 Sandra Loosemore <sandra@codesourcery.com>
1136 * doc/invoke.texi (Link Options): Mention shared libraries
1137 in documentation for the -l option. Simplify discussion and
1138 point to the system linker documentation for details.
1140 2018-11-19 Peter Bergner <bergner@linux.ibm.com>
1142 PR rtl-optimization/88033
1143 * ira-lives.c (non_conflicting_reg_copy_p): Skip copies from a register
1144 to itself. Use HARD_REGISTER_NUM_P.
1146 2018-11-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
1148 * tree-vect-loop.c (vect_transform_loop): Disable further unrolling
1149 of the loop if vf is non-constant.
1151 2018-11-19 David Malcolm <dmalcolm@redhat.com>
1153 PR tree-optimization/87025
1154 * dumpfile.c (dump_context::begin_scope): Call end_any_optinfo
1155 immediately after creating the scope optinfo.
1156 (selftest::test_pr87025): New function.
1157 (selftest::dumpfile_c_tests): Call it.
1158 * optinfo-emit-json.cc (optrecord_json_writer::pop_scope): Assert
1159 that we're not popping the top-level records array.
1160 * optinfo.cc (optinfo::handle_dump_file_kind): Assert that we're
1161 not changing the kind of a "scope" optinfo.
1163 2018-11-19 David Malcolm <dmalcolm@redhat.com>
1165 PR tree-optimization/87025
1166 * dump-context.h: Include "optinfo.h".
1167 (class optrecord_json_writer): New forward decl.
1168 (dump_context::forcibly_enable_optinfo_p): Delete.
1169 (dump_context::optinfo_enabled_p): New member function.
1170 (dump_context::optimization_records_enabled_p): New member
1172 (dump_context::set_json_writer): New member function.
1173 (dump_context::emit_optinfo): New member function.
1174 (dump_context::m_forcibly_enable_optinfo): Delete.
1175 (dump_context::m_json_writer): New member data.
1176 * dumpfile.c (dump_context::set_json_writer): New member function.
1177 (dump_context::finish_any_json_writer): New member function.
1178 (dump_context::end_scope): Replace call to
1179 optimization_records_maybe_pop_dump_scope with call to
1180 m_json_writer->pop_scope.
1181 (dump_context::optinfo_enabled_p): New member function.
1182 (dump_context::end_any_optinfo): Replace call to optinfo::emit with call
1183 to dump_context::emit_optinfo.
1184 (dump_context::emit_optinfo): New member function.
1185 (temp_dump_context::temp_dump_context): Replace
1186 m_forcibly_enable_optinfo with call to set_json_writer.
1187 (temp_dump_context::~temp_dump_context): Clean up any json writer.
1188 * optinfo-emit-json.cc (class optrecord_json_writer): Move to
1190 (the_json_writer): Delete.
1191 (optimization_records_start): Delete.
1192 (optimization_records_finish): Delete.
1193 (optimization_records_enabled_p): Delete, in favor of
1194 dump_context::optimization_records_enabled_p.
1195 (optimization_records_maybe_record_optinfo): Delete.
1196 (optimization_records_maybe_pop_dump_scope): Delete.
1197 * optinfo-emit-json.h: Include "json.h". Delete forward
1199 (optimization_records_start): Delete.
1200 (optimization_records_finish): Delete.
1201 (optimization_records_enabled_p): Delete.
1202 (optimization_records_maybe_record_optinfo): Delete.
1203 (optimization_records_maybe_pop_dump_scope): Delete.
1204 (class optrecord_json_writer): Move here from
1205 optinfo-emit-json.cc.
1206 * optinfo.cc (optinfo::emit_for_opt_problem): Replace call
1207 to optinfo::emit with call to dump_context::emit_optinfo.
1208 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
1209 (optinfo_enabled_p): Delete, in favor of
1210 dump_context::optinfo_enabled_p.
1211 (optinfo_wants_inlining_info_p): Update for conversion o
1212 optimization_records_enabled_p to a member function of
1214 * optinfo.h (optinfo_enabled_p): Delete, in favor of
1215 dump_context::optinfo_enabled_p.
1216 (optinfo::emit): Delete, in favor of dump_context::emit_optinfo.
1217 * toplev.c: Include "dump-context.h".
1218 (compile_file): Replace call to optimization_records_finish with
1219 dump_context::finish_any_json_writer.
1220 (do_compile): Replace call to optimization_records_start with
1221 conditionally creating a optrecord_json_writer for the
1224 2018-11-19 Jakub Jelinek <jakub@redhat.com>
1226 PR tree-optimization/88071
1227 * tree-vect-loop.c (vectorize_fold_left_reduction): Pass true instead
1228 of false as last argument to gsi_remove.
1229 * tree-vect-stmts.c (vect_finish_replace_stmt): Pass true instead of
1230 false as last argument to gsi_replace.
1233 * omp-expand.c: Don't include debug.h.
1234 (adjust_context_and_scope): Add REGION argument. Find DECL_CONTEXT
1235 from innermost outer parallel, task, teams or target that has a
1236 child_fn set, or, if there is no such outer region, use
1237 current_function_decl. Do the DECL_CONTEXT adjustment regardless of
1238 whether a suitable BLOCK is found or not.
1239 (expand_parallel_call, expand_teams_call): Don't call
1240 adjust_context_and_scope here.
1241 (grid_expand_target_grid_body): Revert 2017-01-25 changes.
1242 (expand_omp_taskreg, expand_omp_target): Likewise. Call
1243 adjust_context_and_scope.
1244 * dwarf2out.c (dwarf2out_early_global_decl): For
1245 decl_function_context recurse instead of calling dwarf2out_decl.
1247 2018-11-20 Eric Botcazou <ebotcazou@adacore.com>
1249 PR rtl-optimization/85925
1250 * rtl.h (word_register_operation_p): New predicate.
1251 * combine.c (record_dead_and_set_regs_1): Only apply specific handling
1252 for WORD_REGISTER_OPERATIONS targets to word_register_operation_p RTX.
1253 * rtlanal.c (nonzero_bits1): Likewise. Adjust couple of comments.
1254 (num_sign_bit_copies1): Likewise.
1256 2018-11-19 Richard Biener <rguenther@suse.de>
1259 * tree.c (free_lang_data_in_one_sizepos): Free non-gimple-val
1262 2018-11-19 Eric Botcazou <ebotcazou@adacore.com>
1264 * compare-elim.c (struct comparison): Add not_in_a field.
1265 (is_not): New static function.
1266 (strip_not): Likewise.
1267 (conforming_compare): Handle a NOT in the first operand.
1268 (can_eliminate_compare): Likewise.
1269 (find_comparison_dom_walker::before_dom_children): Likewise.
1270 (try_eliminate_compare): Likewise.
1271 * config/visium/visium.md (negsi2_insn_set_carry): Turn into...
1272 (neg<mode>2_insn_set_carry): ...this and add missing NEG operation.
1274 2018-11-19 Jonathan Wakely <jwakely@redhat.com>
1276 * doc/extend.texi (Common Type Attributes): Fix typos.
1278 2018-11-19 Martin Liska <mliska@suse.cz>
1280 PR gcov-profile/88045
1281 * coverage.c (coverage_begin_function): Add assert.
1283 2018-11-18 Sandra Loosemore <sandra@codesourcery.com>
1286 * doc/extend.texi (Common Function Attributes): Document that
1287 no_instrument_function applies to -p and -pg, too.
1288 * doc/invoke.texi (Instrumentation Options): Add cross-references
1289 to docs for -p, -pg, and -finstrument-functions.
1291 2018-11-18 Alan Modra <amodra@gmail.com>
1293 * config/gnu-user.h (GNU_USER_TARGET_CRTI): Define.
1294 (GNU_USER_TARGET_STARTFILE_SPEC): Use it here.
1295 (GNU_USER_TARGET_CRTN): Define.
1296 (GNU_USER_TARGET_ENDFILE_SPEC): Use it here.
1297 (GNU_USER_TARGET_LINK_GCC_C_SEQUENCE_SPEC): Define.
1299 2018-11-18 Alan Modra <amodra@gmail.com>
1301 * config/gnu-user.h (GNU_USER_TARGET_STARTFILE_SPEC): Delete
1302 !HAVE_LD_PIE variant.
1303 (GNU_USER_TARGET_ENDFILE_SPEC): Likewise.
1305 2018-11-17 Nick Clifton <nickc@redhat.com>
1306 Sandra Loosemore <sandra@codesourcery.com>
1309 * doc/invoke.texi (Overall Options): Document that --help and
1310 --help=<value> options cannot be combined.
1312 2018-11-17 Jim Wilson <jimw@sifive.com>
1314 * config/riscv/riscv.c (epilogue_cfa_sp_offset): New.
1315 (riscv_restore_reg): If restoring HARD_FRAME_POINTER_REGNUM, and
1316 epilogue_cfa_sp_offset set, then add REG_CFA_DEF_CFA regnote.
1317 (riscv_expand_epilogue): Initialize epilogue_cfa_sp_offset. Set it
1318 to step2 if frame_pointer_needed and step1 is 0.
1320 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
1323 * doc/cpp.texi (System Headers): Add note about implicit
1324 extern "C" block on targets that define SYSTEM_IMPLICIT_EXTERN_C.
1326 2018-11-17 Sandra Loosemore <sandra@codesourcery.com>
1329 * doc/invoke.texi (C++ Dialect Options): Clarify usage of
1330 -fno-implicit-templates.
1332 2018-11-17 Jakub Jelinek <jakub@redhat.com>
1334 PR tree-optimization/87546
1335 * tree-vect-patterns.c (vect_look_through_possible_promotion): Add
1336 min_precision variable, initially set it to orig_precision, only does
1337 something if op_type's precision is <= min_precision and update
1338 min_precision whenever calling set_op.
1340 2018-11-16 Jan Hubicka <hubicka@ucw.cz>
1343 * ipa-devirt.c (warn_odr): Look for main variant to get TYPE_DECL.
1345 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
1347 * doc/invoke.texi (Option Summary): Fix whitespace and line
1348 breaks in @gccoptlist environments.
1349 (Warning Options): Likewise.
1350 (Optimize Options): Likewise.
1351 (PowerPC SPE Options): Likewise.
1352 (RS/6000 and PowerPC Options): Likewise.
1354 2018-11-16 Jeff Law <law@redhat.com>
1356 * config/mn10300/mn10300.md (adddi3_degenerate): Remove bogus
1359 2018-11-16 Sandra Loosemore <sandra@codesourcery.com>
1362 * doc/invoke.texi (Optimize Options): Update options enabled by
1363 fprofile-generate, -fprofile-use, and -fauto-profile.
1365 2018-11-16 Jozef Lawrynowicz <jozef.l@mittosystems.com>
1368 * target-def.h: Initialize TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP.
1369 Add them to the TARGET_ASM_{,UN}ALIGNED_INT_OP structs.
1370 * target.def: Enumerate TARGET_ASM_{,UN}ALIGNED_P{S,D,T}I_OP in
1372 * target.h: Add psi, pdi, pti to struct asm_int_op definition.
1373 * targhooks.c (default_print_patchable_function_entry): Assert
1374 asm_int_op does not return a NULL string.
1375 * varasm.c (integer_asm_op): Return the op for a partial int type
1376 when the requested size does not correspond to an integer type.
1377 * config/msp430/msp430.c: Initialize TARGET_ASM_{,UN}ALIGNED_PSI_OP.
1378 * doc/tm.texi: Regenerate.
1380 2018-11-12 Jason Merrill <jason@redhat.com>
1382 * gimplify.c (gimplify_case_label_expr): Handle hot/cold attributes.
1384 2018-11-16 Michael Meissner <meissner@linux.ibm.com>
1386 * config/rs6000/constraints.md (wF constraint): Remove power9
1387 fusion documentation. Just document wF's use for power8 fusion.
1388 * config/rs6000/predicates.md (p9_fusion_reg_operand): Delete.
1389 (fusion_gpr_addis): Delete power9 fusion support. Change power8
1390 fusion support to require the upper 12 bits to be all 0's or all
1392 (fusion_gpr_mem_load): Add comment.
1393 (fusion_addis_mem_combo_load): Remove power9 fusion support.
1394 (fusion_addis_mem_combo_store): Delete.
1395 (fusion_offsettable_mem_operand): Delete.
1396 * config/rs6000/rs6000-cpus.def (ISA_2_7_MASKS_SERVER): Do not set
1398 (ISA_3_0_MASKS_SERVER): Delete power9 fusion.
1399 (POWERPC_MASKS): Delete power9 fusion.
1400 * config/rs6000/rs6000-protos.h (emit_fusion_load_store): Delete.
1401 (fusion_p9_p): Delete.
1402 (expand_fusion_p9_load): Delete.
1403 (expand_fusion_p9_store): Delete.
1404 (emit_fusion_p9_load): Delete.
1405 (emit_fusion_p9_store): Delete.
1406 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Delete power9
1408 (rs6000_option_override_internal): Set power8 fusion based on
1409 whether we are tuning for power8. Delete power9 fusion support.
1410 (rs6000_opt_masks): Delete -mpower9-fusion switch.
1411 (emit_fusion_load): Rename emit_fusion_load_store to
1412 emit_fusion_load, and drop fusion store support. Update callers.
1413 (emit_fusion_load_store): Likewise.
1414 (emit_fusion_gpr_load): Likewise.
1415 (fusion_p9_p): Delete.
1416 (expand_fusion_p9_load): Delete.
1417 (expand_fusion_p9_store): Delete.
1418 (emit_fusion_p9_load): Delete.
1419 (emit_fusion_p9_store): Delete.
1420 * config/rs6000/rs6000.md (UNSPEC_FUSION_P9): Delete.
1421 (GPR_FUSION): Delete.
1422 (FPR_FUSION): Delete.
1423 (power9 fusion peephole2s): Delete.
1424 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Delete.
1425 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Delete.
1426 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_load): Delete.
1427 (fusion_vsx_<P:mode>_<FPR_FUSION:mode>_store): Delete.
1428 (fusion_p9_<mode>_constant): Delete.
1429 * config/rs6000/rs6000.opt (-mpower9-fusion): Delete undocumented
1430 power9 fusion switch.
1431 * doc/md.texi (PowerPC constraints): Update wF constraint
1432 documentation for power8 fusion only.
1434 2018-11-16 Jakub Jelinek <jakub@redhat.com>
1436 PR rtl-optimization/87475
1437 * cfgrtl.c (patch_jump_insn): Allow redirection failure for
1438 CROSSING_JUMP_P insns.
1439 (cfg_layout_redirect_edge_and_branch): Don't ICE if ret is NULL.
1441 2018-11-16 Uros Bizjak <ubizjak@gmail.com>
1444 * config/i386/i386.md (floatunsdidf2): Allow only 64bit AVX512F targets.
1445 * config/i386/sse.md (UNSPEC_MOVDI_TO_SSE): New UNSPEC.
1446 (movdi_to_sse): Rewrite using UNSPEC_MOVDI_TO_SSE unspec.
1448 2018-11-16 Jakub Jelinek <jakub@redhat.com>
1451 * optabs.c (expand_binop): For op0_mode use GET_MODE (op0), unless it
1452 is VOIDmode, in which case use int_mode. Similarly for op1_mode.
1454 2018-11-16 Eric Botcazou <ebotcazou@adacore.com>
1456 * tree-switch-conversion.h (switch_decision_tree::emit_case_nodes): Add
1457 location_t parameter.
1458 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
1459 (switch_decision_tree::do_jump_if_equal): Likewise.
1460 * tree-switch-conversion.c (switch_decision_tree::emit): Pass location
1461 of switch statement to emit_case_nodes.
1462 (switch_decision_tree::emit_cmp_and_jump_insns): Add LOC parameter and
1463 set it on the newly built GIMPLE comparison statement.
1464 (switch_decision_tree::do_jump_if_equal): Likewise.
1465 (switch_decision_tree::emit_case_nodes): Add LOC parameter and pass it
1466 in calls to do_jump_if_equal as well as recursive calls.
1468 2018-11-16 Andi Kleen <ak@linux.intel.com>
1470 * config/i386/i386.md: Allow memory operands to ptwrite.
1472 2018-11-16 Michael Matz <matz@suse.de>
1474 * system.h (PRsa): New macro.
1475 (SIZE_AMOUNT): Cast number to uint64_t.
1476 * alloc-pool.h (pool_usage::dump): Don't use %zu but PRsa.
1477 (pool_usage::dump_footer): Likewise and also use PRIu64.
1478 * bitmap.h (bitmap_usage::dump): Likewise.
1479 * ggc-common.c (ggc_usage::dump): Likewise.
1480 * ggc-page.c (ggc_print_statistics): Likewise.
1481 * input.c (dump_line_table_statistics): Likewise.
1482 * mem-stats.h (mem_usage::dump): Likewise.
1483 (mem_usage::dump_footer): Likewise.
1484 * rtl.c (dump_rtx_statistics): Likewise.
1485 * tree-cfg.c (dump_cfg_stats): Likewise.
1486 * tree-dfa.c (dump_dfa_stats): Likewise.
1487 * tree-phinodes.c (phinodes_print_statistics): Likewise.
1488 * tree-ssanames (ssanames_print_statistics): Likewise.
1489 * vec.c (vec_usage::dump): Likewise.
1490 (vec_usage::dump_footer): Likewise.
1492 2018-11-16 Richard Biener <rguenther@suse.de>
1494 PR tree-optimization/88011
1495 * tree-vrp.c (extract_range_from_binary_expr): Fix error in
1496 replacing set_value_range_to_undefined and
1497 set_value_range_to_varying with method calls.
1499 2018-11-16 Ilya Leoshkevich <iii@linux.ibm.com>
1501 * config/s390/s390.md
1502 (*r<noxa>sbg_<mode>_srl_bitmask): Do not delegate arithmetic to
1504 (*r<noxa>sbg_<mode>_sll): Likewise.
1505 (*r<noxa>sbg_<mode>_srl): Likewise.
1506 (*r<noxa>sbg_sidi_srl): New pattern.
1508 2018-11-16 Jerome Lambourg <lambourg@adacore.com>
1510 * config/vxworks.h (TARGET_VXWORKS_HAVE_CTORS_DTORS): New macro.
1511 Default to TARGET_VXWORKS_RTP.
1512 (SUPPORTS_INIT_PRIORITY): Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
1513 of TARGET_VXWORKS_RTP.
1514 * config/vxworksae.h: Also define TARGET_VXWORKS_HAVE_CTORS_DTORS.
1515 * config/vxworks.c: Use TARGET_VXWORKS_HAVE_CTORS_DTORS instead
1516 of TARGET_VXWORKS_RTP to set targetm.have_ctors_dtors.
1518 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
1521 * doc/extend.texi (Common Type Attributes): Make it explicit
1522 that attribute "packed" can apply to C++ classes.
1524 2018-11-15 Martin Sebor <msebor@redhat.com>
1528 * tree.c (type_argument_type): New function.
1529 * tree.h (type_argument_type): Declare it.
1530 * gcc/doc/extend.texi (alloc_align): Update and clarify.
1531 (alloc_size, nonnull, sentinel): Same.
1533 2018-11-15 Andrew Stubbs <ams@codesourcery.com>
1534 Kwok Cheung Yeung <kcy@codesourcery.com>
1536 * tree-vect-stmts.c (vectorizable_store): Don't ICE when
1537 int_mode_for_size fails.
1538 (vectorizable_load): Likewise.
1540 2018-11-15 David Malcolm <dmalcolm@redhat.com>
1542 * doc/ux.texi (Group logically-related diagnostics): Move
1543 discussion of auto_diagnostic_group into this new subsection.
1544 Give an example of where this grouping is used.
1546 2018-11-15 David Malcolm <dmalcolm@redhat.com>
1549 * Makefile.in (OBJS): Move json.o to...
1550 (OBJS-libcommon): ...here and add diagnostic-format-json.o.
1551 * common.opt (fdiagnostics-format=): New option.
1552 (diagnostics_output_format): New enum.
1553 * diagnostic-format-json.cc: New file.
1554 * diagnostic.c (default_diagnostic_final_cb): New function, taken
1555 from start of diagnostic_finish.
1556 (diagnostic_initialize): Initialize final_cb to
1557 default_diagnostic_final_cb.
1558 (diagnostic_finish): Move "being treated as errors" messages to
1559 default_diagnostic_final_cb. Call any final_cb.
1560 (default_diagnostic_finalizer): Add diagnostic_t param.
1561 (diagnostic_report_diagnostic): Pass "orig_diag_kind" to
1562 diagnostic_finalizer callback.
1563 * diagnostic.h (enum diagnostics_output_format): New enum.
1564 (diagnostic_finalizer_fn): Reimplement, adding diagnostic_t param.
1565 (struct diagnostic_context): Add "final_cb".
1566 (default_diagnostic_finalizer): Add diagnostic_t param.
1567 (diagnostic_output_format_init): New decl.
1568 * doc/invoke.texi (-fdiagnostics-format): New option.
1569 * dwarf2out.c (gen_producer_string): Ignore
1570 OPT_fdiagnostics_format_.
1571 * gcc.c (driver_handle_option): Handle OPT_fdiagnostics_format_.
1572 * lto-wrapper.c (append_diag_options): Ignore it.
1573 * opts.c (common_handle_option): Handle it.
1575 2018-11-15 David Malcolm <dmalcolm@redhat.com>
1577 PR tree-optimization/88015
1578 * graphite-isl-ast-to-gimple.c
1579 (translate_isl_ast_to_gimple::scop_to_isl_ast): Add missing check
1581 * graphite-sese-to-poly.c (build_poly_scop): Likewise.
1583 2018-11-15 Richard Biener <rguenther@suse.de>
1586 * gimple.c (gimple_call_flags): Union flags from decl, type
1588 * trans-mem.c (is_tm_pure_call): Simplify.
1590 2018-11-15 Richard Biener <rguenther@suse.de>
1592 PR tree-optimization/88031
1593 * tree-vect-loop.c (vectorizable_reduction): Move check
1594 for multiple types earlier so we get the expected dump.
1595 Simplify calls to vectorizable_condition.
1596 * tree-vect-stmts.h (vectorizable_condition): Update prototype.
1597 * tree-vect-stmts.c (vectorizable_condition): Instead of
1598 reduc_def and reduc_index take just a flag. Simplify
1599 code-generation now that we can rely on the defs being set up.
1600 (vectorizable_comparison): Remove unused argument.
1602 2018-11-15 Richard Biener <rguenther@suse.de>
1604 PR tree-optimization/88030
1605 * tree-complex.c (need_eh_cleanup): New global.
1606 (update_complex_assignment): Mark blocks that need EH update.
1607 (expand_complex_comparison): Likewise.
1608 (tree_lower_complex): Allocate and deallocate need_eh_cleanup,
1609 perform EH cleanup and schedule CFG cleanup if that did anything.
1611 2018-11-15 Jakub Jelinek <jakub@redhat.com>
1613 PR rtl-optimization/88018
1614 * cfgrtl.c (fixup_abnormal_edges): Guard moving insns to fallthru edge
1615 on the presence of fallthru edge, rather than if it is a USE or not.
1617 2018-11-15 Richard Biener <rguenther@suse.de>
1620 * tree-data-ref.c (analyze_miv_subscript): Guard calls to
1621 analyze_subscript_affine_affine properly.
1623 2018-11-15 Xianmiao Qu <xianmiao_qu@c-sky.com>
1625 * config/csky/csky-linux-elf.h (LINUX_DYNAMIC_LINKER): Remove.
1626 (GLIBC_DYNAMIC_LINKER): Define.
1627 (LINUX_TARGET_LINK_SPEC): Update the dynamic linker's name.
1629 2018-11-15 Bin Cheng <bin.cheng@linux.alibaba.com>
1631 PR tree-optimization/84648
1632 * tree-ssa-loop-niter.c (adjust_cond_for_loop_until_wrap): New.
1633 (number_of_iterations_cond): Adjust exit cond for loop-until-wrap case
1634 by calling adjust_cond_for_loop_until_wrap.
1636 2018-11-15 Sandra Loosemore <sandra@codesourcery.com>
1639 * doc/extend.texi (Common Function Attributes): Clarify linker
1640 restrictions on "aligned" attribute.
1641 (Common Variable Attributes): Likewise. Mention that linker
1642 restrictions don't apply to stack-allocated variables.
1644 2018-11-15 Joern Rennecke <joern.rennecke@riscy-ip.com>
1646 * gensupport.c (add_predicate_code): Properly handle ZERO_EXTRACT
1649 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
1653 * doc/invoke.texi (Optimize Options): Remove bad example about
1654 interaction between -flto and -O. Replace it with a note that
1655 you need to compile with -O and not just link. Copy-edit -flto
1656 discussion to reduce verbiage and improve flow.
1658 2018-11-14 Sandra Loosemore <sandra@codesourcery.com>
1661 * doc/invoke.texi (Optimize Options): Clarify that -O0 and -Og
1662 also suppress many optimizations. Alphabetize option lists for
1663 -O1, -O2, and -Os. Add list of options disabled with -Og, and
1664 correct documentation for those options to say that.
1665 * opts.c (default_options_table): Sort table by level and option
1666 name, to make it easier to correlate to the manual.
1668 2018-11-14 Uros Bizjak <ubizjak@gmail.com>
1670 * config/i386/i386.c (ix86_print_operand_address_as): Simplify
1671 printing of the names of segment registers.
1673 2018-11-14 Nathan Sidwell <nathan@acm.org>
1677 * dwarf2out.c (dwarf2out_finish): Apply resolve_addr to comdat
1680 2018-11-14 David Malcolm <dmalcolm@redhat.com>
1682 * Makefile.in (CFLAGS-optinfo-emit-json.o): Add $(ZLIBINC).
1684 2018-11-14 Jakub Jelinek <jakub@redhat.com>
1686 P1236R1 - Signed integers are two's complement
1687 * doc/invoke.texi (Wshift-overflow): Adjust documentation for c++2a.
1690 * hash-map.h (hash_map::iterator::reference_pair): New class.
1691 (hash_map::iterator::operator*): Return it rather than std::pair.
1693 2018-11-14 Jeff Law <law@redhat.com>
1695 * optabs.c (expand_binop): Pass INT_MODE to operand_subword_force
1696 iff the operand is a constant.
1698 2018-11-14 Aldy Hernandez <aldyh@redhat.com>
1700 * gimple-ssa-evrp-analyze.c
1701 (evrp_range_analyzer::record_ranges_from_incoming_edge): Rename
1702 ignore_equivs_equal_p to equal_p.
1703 * ipa-cp.c (meet_with_1): Use equal_p instead of
1704 ignore_equivs_equal_p.
1705 * ipa-prop.c (ipa_vr_ggc_hash_traits::equal): Same.
1706 * tree-vrp.c (value_range::ignore_equivs_equal_p): Remove.
1707 (value_range::operator==): Remove.
1708 (value_range::operator!=): Remove.
1709 (vrp_prop::visit_stmt): Use equal_p.
1710 * tree-vrp.h (value_range): Remove operator==, operator!=,
1711 ignore_equivs_equal_p.
1712 * vr-values.c (update_value_range): Use equal_p.
1714 2018-11-14 Michael Matz <matz@suse.de>
1717 * gimplify.c (collect_fallthrough_labels): Add new argument,
1718 return location via that, don't modify statements.
1719 (warn_implicit_fallthrough_r): Adjust call, don't use
1720 statement location directly.
1722 2018-11-14 Richard Biener <rguenther@suse.de>
1725 * tree-data-ref.c (split_constant_offset): Add wrapper
1726 allocating a cache hash-map.
1727 (split_constant_offset_1): Cache results of expanding
1728 expressions from SSA def stmts.
1730 2018-11-14 Richard Biener <rguenther@suse.de>
1733 * tree-data-ref.c (lambda_matrix_row_add): Change const1 argument
1735 (lambda_vector_mult_const): Likewise.
1736 (lambda_matrix_right_hermite): Use lambda_int temporaries.
1738 2018-11-14 Wilco Dijkstra <wdijkstr@arm.com>
1739 Jackson Woodruff <jackson.woodruff@arm.com>
1741 PR 71026/tree-optimization
1742 * match.pd: Simplify floating point comparisons.
1744 2018-11-14 Jakub Jelinek <jakub@redhat.com>
1746 PR rtl-optimization/87817
1747 * config/i386/i386.c (ix86_fold_builtin): For _bzhi_u{32,64} if
1748 last argument has low 8 bits clear, fold to 0.
1750 2018-11-14 Iain Sandoe <iain@sandoe.co.uk>
1752 * tree-vect-data-refs.c (vect_can_force_dr_alignment_p): Cast
1753 MAX_OFILE_ALIGNMENT as needed.
1755 2018-11-14 Richard Biener <rguenther@suse.de>
1757 PR tree-optimization/88019
1758 * graphite-isl-ast-to-gimple.c (ternary_op_to_tree): Rewrite
1759 COND_EXPR arguments to non-trapping overflow.
1761 2018-11-14 Claudiu Zissulescu <claziss@synopsys.com>
1763 * config/arc/arc.h (reg_class): Reorder registers classes, remove
1764 unused register classes.
1765 (REG_CLASS_NAMES): Likewise.
1766 (REG_CLASS_CONTENTS): Likewise.
1767 (FIXED_REGISTERS): Make lp_count fixed.
1768 (BASE_REG_CLASS): Remove ACC16_BASE_REGS reference.
1769 (PROGRAM_COUNTER_REGNO): Remove.
1770 * config/arc/arc.c (arc_conditional_register_usage): Remove unused
1771 register classes, use constants for register numbers, remove
1772 reg_class_contents references.
1773 (arc_process_double_reg_moves): Add asserts.
1774 (arc_secondary_reload): Remove LPCOUNT_REG reference, use
1775 lra_in_progress predicate.
1776 (arc_init_reg_tables): Remove unused register classes.
1777 (arc_register_move_cost): Likewise.
1778 (arc_preferred_reload_class): Likewise.
1779 (hwloop_optimize): Update rtx patterns involving lp_count register.
1780 (arc_return_address_register): Rename ILINK1, INLINK2 regnums macros.
1781 * config/arc/constraints.md ("c"): Choose between GENERAL_REGS and
1782 CHEAP_CORE_REGS. Former one will be used for LRA.
1783 ("Rac"): Choose between GENERAL_REGS and ALL_CORE_REGS. Former
1784 one will be used for LRA.
1785 ("w"): Choose between GENERAL_REGS and WRITABLE_CORE_REGS. Former
1786 one will be used for LRA.
1787 ("W"): Choose between GENERAL_REGS and MPY_WRITABLE_CORE_REGS.
1788 Former one will be used for LRA.
1789 ("f"): Delete constraint.
1792 ("l"): Change it from register constraint to constraint.
1793 * config/arc/arc.md (movqi_insn): Remove unsed lp_count constraints.
1794 (movhi_insn): Likewise.
1795 (movsi_insn): Update pattern.
1798 (stack_tie): Remove 'b' constraint letter.
1800 (R9_REG, R15_REG, R16_REG, R25_REG): Likewise.
1801 (R32_REG, R40_REG, R41_REG, R42_REG, R43_REG, R44_REG): Likewise.
1802 (R57_REG, R59_REG, PCL_REG): Likewise.
1803 (ILINK1_REGNUM): Renamed to ILINK1_REG.
1804 (ILINK2_REGNUM): Renamed to ILINK2_REG.
1806 (SP_REGS): Likewise.
1807 (Rcw): Remove unused reg classes.
1808 * config/arc/predicates.md (dest_reg_operand): Just default on
1809 register_operand predicate.
1810 (mpy_dest_reg_operand): Likewise.
1811 (move_dest_operand): Use macros instead of constants.
1812 * config/arc/arc.opt (mlra): Switch to lra as default.
1814 2018-11-14 Richard Biener <rguenther@suse.de>
1816 PR tree-optimization/87974
1817 * tree-vect-loop.c (vectorizable_reduction): When computing
1818 the vectorized reduction PHI vector type ignore constant
1821 2018-11-14 Jakub Jelinek <jakub@redhat.com>
1823 PR tree-optimization/87977
1824 * tree-ssa-math-opts.c (optimize_recip_sqrt): Don't reuse division
1825 stmt, build a new one and replace the old one with it. Formatting fix.
1826 Call release_ssa_name (x) if !has_other_use and !delete_div.
1827 (pass_cse_reciprocals::execute): Before calling optimize_recip_sqrt
1828 verify lhs of stmt is still def.
1830 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
1832 PR rtl-optimization/87507
1833 * lower-subreg.c (operand_for_swap_move_operator): New function.
1834 (simple_move): Strip simple operators.
1835 (find_pseudo_copy): Likewise.
1836 (resolve_operand_for_swap_move_operator): New function.
1837 (resolve_simple_move): Strip simple operators and swap operands.
1839 2018-11-13 Jakub Jelinek <jakub@redhat.com>
1841 PR tree-optimization/87898
1842 * omp-simd-clone.c (ipa_simd_modify_stmt_ops): Formatting fix.
1843 (ipa_simd_modify_function_body): Remove debug stmts where the first
1844 argument was changed into a non-decl.
1846 2018-11-13 Peter Bergner <bergner@linux.ibm.com>
1848 PR rtl-optimization/87899
1849 * lra-lives.c (start_living): Update white space in comment.
1850 (enum point_type): New.
1851 (sparseset_contains_pseudos_p): New function.
1852 (update_pseudo_point): Likewise.
1853 (make_hard_regno_live): Use HARD_REGISTER_NUM_P macro.
1854 (make_hard_regno_dead): Likewise. Remove ignore_reg_for_conflicts
1855 handling. Move early exit after adding conflicts.
1856 (mark_pseudo_live): Use HARD_REGISTER_NUM_P macro. Add early exit
1857 if regno is already live. Remove all handling of program points.
1858 (mark_pseudo_dead): Use HARD_REGISTER_NUM_P macro. Add early exit
1859 after adding conflicts. Remove all handling of program points and
1860 ignore_reg_for_conflicts.
1861 (mark_regno_live): Use HARD_REGISTER_NUM_P macro. Remove return value
1862 and do not guard call to mark_pseudo_live.
1863 (mark_regno_dead): Use HARD_REGISTER_NUM_P macro. Remove return value
1864 and do not guard call to mark_pseudo_dead.
1865 (check_pseudos_live_through_calls): Use HARD_REGISTER_NUM_P macro.
1866 (process_bb_lives): Use HARD_REGISTER_NUM_P and HARD_REGISTER_P macros.
1867 Use new function update_pseudo_point. Handle register copies by
1868 removing the source register from the live set. Handle INOUT operands.
1869 Update to the next program point using the unused_set, dead_set and
1871 (lra_create_live_ranges_1): Use HARD_REGISTER_NUM_P macro.
1873 2018-11-13 David Malcolm <dmalcolm@redhat.com>
1875 * builtins.c: Replace "source_location" with "location_t".
1876 * diagnostic-show-locus.c: Likewise.
1877 * diagnostic.c: Likewise.
1878 * dumpfile.c: Likewise.
1879 * gcc-rich-location.h: Likewise.
1880 * genmatch.c: Likewise.
1881 * gimple.h: Likewise.
1882 * gimplify.c: Likewise.
1883 * input.c: Likewise.
1884 * input.h: Likewise. Eliminate the typedef.
1885 * omp-expand.c: Likewise.
1886 * selftest.h: Likewise.
1887 * substring-locations.h (get_source_location_for_substring):
1889 (get_location_within_string): ...this.
1890 * tree-cfg.c: Replace "source_location" with "location_t".
1891 * tree-cfgcleanup.c: Likewise.
1892 * tree-diagnostic.c: Likewise.
1893 * tree-into-ssa.c: Likewise.
1894 * tree-outof-ssa.c: Likewise.
1895 * tree-parloops.c: Likewise.
1896 * tree-phinodes.c: Likewise.
1897 * tree-phinodes.h: Likewise.
1898 * tree-ssa-loop-ivopts.c: Likewise.
1899 * tree-ssa-loop-manip.c: Likewise.
1900 * tree-ssa-phiopt.c: Likewise.
1901 * tree-ssa-phiprop.c: Likewise.
1902 * tree-ssa-threadupdate.c: Likewise.
1903 * tree-ssa.c: Likewise.
1904 * tree-ssa.h: Likewise.
1905 * tree-vect-loop-manip.c: Likewise.
1907 2018-11-13 Michael Ploujnikov <michael.ploujnikov@oracle.com>
1909 * doc/extend.texi: Fix typo in the weakref description.
1911 2018-11-13 Richard Biener <rguenther@suse.de>
1913 PR tree-optimization/86991
1914 * tree-vect-loop.c (vect_is_slp_reduction): Delay reduction
1915 group building until we have successfully detected the SLP
1917 (vect_is_simple_reduction): Remove fixup code here.
1919 2018-11-13 David Malcolm <dmalcolm@redhat.com>
1921 * dumpfile.c (VERIFY_DUMP_ENABLED_P): New macro.
1922 (dump_gimple_stmt): Use it.
1923 (dump_gimple_stmt_loc): Likewise.
1924 (dump_gimple_expr): Likewise.
1925 (dump_gimple_expr_loc): Likewise.
1926 (dump_generic_expr): Likewise.
1927 (dump_generic_expr_loc): Likewise.
1928 (dump_printf): Likewise.
1929 (dump_printf_loc): Likewise.
1930 (dump_dec): Likewise.
1931 (dump_dec): Likewise.
1932 (dump_hex): Likewise.
1933 (dump_symtab_node): Likewise.
1934 * gimple-loop-interchange.cc (tree_loop_interchange::interchange):
1935 Guard dump call with dump_enabled_p.
1936 * graphite-isl-ast-to-gimple.c (graphite_regenerate_ast_isl): Likewise.
1937 * graphite-optimize-isl.c (optimize_isl): Likewise.
1938 * graphite.c (graphite_transform_loops): Likewise.
1939 * tree-loop-distribution.c (pass_loop_distribution::execute): Likewise.
1940 * tree-parloops.c (parallelize_loops): Likewise.
1941 * tree-ssa-loop-niter.c (number_of_iterations_exit): Likewise.
1942 * tree-vect-data-refs.c (vect_analyze_group_access_1): Likewise.
1943 (vect_prune_runtime_alias_test_list): Likewise.
1944 * tree-vect-loop.c (vect_update_vf_for_slp): Likewise.
1945 (vect_estimate_min_profitable_iters): Likewise.
1946 * tree-vect-slp.c (vect_record_max_nunits): Likewise.
1947 (vect_build_slp_tree_2): Likewise.
1948 (vect_supported_load_permutation_p): Likewise.
1949 (vect_slp_analyze_operations): Likewise.
1950 (vect_slp_analyze_bb_1): Likewise.
1951 (vect_slp_bb): Likewise.
1952 * tree-vect-stmts.c (vect_analyze_stmt): Likewise.
1953 * tree-vectorizer.c (try_vectorize_loop_1): Likewise.
1954 (pass_slp_vectorize::execute): Likewise.
1955 (increase_alignment): Likewise.
1957 2018-11-13 David Malcolm <dmalcolm@redhat.com>
1960 * ipa-inline.c (report_inline_failed_reason): Guard calls to
1961 cl_target_option_print_diff and cl_optimization_print_diff with
1964 2018-11-13 David Malcolm <dmalcolm@redhat.com>
1966 * doc/invoke.texi (-fsave-optimization-record): Note that the
1967 output is compressed.
1968 * optinfo-emit-json.cc: Include <zlib.h>.
1969 (optrecord_json_writer::write): Compress the output.
1971 2018-11-13 Aldy Hernandez <aldyh@redhat.com>
1973 * tree-vrp.c (value_range_base::dump): Dump type.
1974 Do not use INF nomenclature for 1-bit types.
1975 (dump_value_range): Group all variants to common dumping code.
1976 (debug): New overloaded functions for value_ranges.
1977 (value_range_base::dump): Remove no argument version.
1978 (value_range::dump): Same.
1980 2018-11-13 Richard Biener <rguenther@suse.de>
1982 PR tree-optimization/87931
1983 * tree-vect-loop.c (vect_is_simple_reduction): Restrict
1984 nested cycles we support to latch computations vectorizable_reduction
1987 2018-11-13 Martin Liska <mliska@suse.cz>
1989 PR tree-optimization/87885
1990 * cfghooks.c (account_profile_record): Rename
1992 (profile_record_check_consistency): ... this.
1993 Calculate missing num_mismatched_freq_in.
1994 (profile_record_account_profile): New function
1995 that calculates time and size of a function.
1996 * cfghooks.h (struct profile_record): Remove
1998 (struct cfg_hooks): Remove after_pass flag.
1999 (account_profile_record): Rename to ...
2000 (profile_record_check_consistency): ... this.
2001 (profile_record_account_profile): New.
2002 * cfgrtl.c (rtl_account_profile_record): Remove
2004 * passes.c (check_profile_consistency): Do only
2006 (account_profile): Calculate size and time of
2008 (pass_manager::dump_profile_report): Reformat
2010 (execute_one_ipa_transform_pass): Call
2011 consistency check before clean upand call account_profile
2012 after a clean up is done.
2013 (execute_one_pass): Call check_profile_consistency and
2014 account_profile instead of using after_pass flag..
2015 * tree-cfg.c (gimple_account_profile_record): Likewise.
2017 2018-11-13 Martin Liska <mliska@suse.cz>
2020 * config/i386/i386.c (ix86_option_override_internal): Error
2021 about usage -mabi=ms and -fsanitize={,kernel-}address.
2023 2018-11-13 Andre Vieira <andre.simoesdiasvieira@arm.com>
2025 * config/aarch64/aarch64.c
2026 (aarch64_vectorize_preferred_vector_alignment): Change return type to
2028 (aarch64_simd_vector_alignment_reachable): Adapt to preferred vector
2029 alignment being a poly int.
2030 * doc/tm.texi (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Change
2031 return type to poly_uint64.
2032 * target.def (default_preferred_vector_alignment): Likewise.
2033 * targhooks.c (default_preferred_vector_alignment): Likewise.
2034 * targhooks.h (default_preferred_vector_alignment): Likewise.
2035 * tree-vect-data-refs.c (vect_calculate_target_alignment): Likewise.
2036 (vect_compute_data_ref_alignment): Adapt to vector alignment being a
2038 (vect_update_misalignment_for_peel): Likewise.
2039 (vect_enhance_data_refs_alignment): Likewise.
2040 (vect_find_same_alignment_drs): Likewise.
2041 (vect_duplicate_ssa_name_ptr_info): Likewise.
2042 (vect_setup_realignment): Likewise.
2043 (vect_can_force_dr_alignment_p): Change alignment parameter type to
2045 * tree-vect-loop-manip.c (get_misalign_in_elems): Learn to construct a
2046 mask with a compile time variable vector alignment.
2047 (vect_gen_prolog_loop_niters): Adapt to vector alignment being a poly
2049 (vect_do_peeling): Exit early if vector alignment is not constant.
2050 * tree-vect-stmts.c (ensure_base_align): Adapt to vector alignment being
2052 (vectorizable_store): Likewise.
2053 (vectorizable_load): Likweise.
2054 * tree-vectorizer.h (struct dr_vec_info): Make target_alignment field a
2056 (vect_known_alignment_in_bytes): Adapt to vector alignment being a
2058 (vect_can_force_dr_alignment_p): Change alignment parameter type to
2061 2018-11-13 Richard Biener <rguenther@suse.de>
2063 PR tree-optimization/87962
2064 * tree-vect-loop.c (vect_is_simple_reduction): More reliably
2065 detect outer reduction for disqualifying in-loop uses.
2067 2018-11-13 Richard Biener <rguenther@suse.de>
2069 PR tree-optimization/87967
2070 * tree-vect-loop.c (vect_transform_loop): Also copy PHIs
2071 for constants for the scalar loop.
2073 2018-11-13 Alan Modra <amodra@gmail.com>
2075 * config/rs6000/linux64.h (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Exclude
2076 integer constants when -mcmodel=medium.
2078 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2080 * config/arc/arc.h (FUNCTION_PROFILER): Redefine to empty.
2081 * config/arc/elf.h (PROFILE_HOOK): Define.
2082 * config/arc/linux.h (PROFILE_HOOK): Likewise.
2084 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2086 * config/arc/arc.c (hwloop_optimize): Bailout when detecting a
2087 jump table data in the text section.
2089 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2091 * config/arc/arc.c (arc_eh_return_address_location): Repurpose it
2092 to fit the eh_return pattern.
2093 * config/arc/arc.md (eh_return): Define.
2094 (VUNSPEC_ARC_EH_RETURN): Likewise.
2095 * config/arc/arc-protos.h (arc_eh_return_address_location): Match
2097 * config/arc/arc.h (EH_RETURN_HANDLER_RTX): Remove it.
2100 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2102 * common/config/arc/arc-common.c (arc_option_optimization_table):
2103 Millicode optimization is default on for size optimizations.
2104 * config/arc/arc-protos.h (arc_check_multi): New function.
2105 * config/arc/arc.c (RTX_OK_FOR_OFFSET_P): Rearange.
2106 (ENTER_LEAVE_START_REG): Define.
2107 (ENTER_LEAVE_END_REG): Likewise.
2108 (arc_override_options): Disable millicode when long calls option
2110 (arc_frame_info): Change it from int to bool.
2111 (arc_compute_frame_size): Clean up.
2112 (arc_save_restore): Remove.
2113 (frame_save_reg): New function.
2114 (frame_restore_reg): Likewise.
2115 (arc_enter_leave_p): Likewise.
2116 (arc_save_callee_saves): Likewise.
2117 (arc_restore_callee_saves): Likewise.
2118 (arc_save_callee_enter): Likewise.
2119 (arc_restore_callee_leave): Likewise.
2120 (arc_save_callee_milli): Likewise.
2121 (arc_restore_callee_milli): Likewise.
2122 (arc_expand_prologue): Reimplement to emit enter/leave
2124 (arc_expand_epilogue): Likewise.
2125 (arc_check_multi): New function.
2126 * config/arc/arc.md (push_multi_fp): New pattern.
2127 (push_multi_fp_blink): Likewise.
2128 (pop_multi_fp): Likewise.
2129 (pop_multi_fp_blink): Likewise.
2130 (pop_multi_fp_ret): Likewise.
2131 (pop_multi_fp_blink_ret): Likewise.
2132 * config/arc/arc.opt (mmillicode): Update option.
2133 (mcode-density-frame): New option.
2134 * config/arc/predicates.md (push_multi_operand): New predicate.
2135 (pop_multi_operand): Likewise.
2136 * doc/invoke.texi (ARC): Update ARC options information.
2139 2018-11-13 Claudiu Zissulescu <claziss@synopsys.com>
2141 * config/arc/arc-protos.h (gen_operands_ldd_std): Add.
2142 * config/arc/arc.c (operands_ok_ldd_std): New function.
2143 (mem_ok_for_ldd_std): Likewise.
2144 (gen_operands_ldd_std): Likewise.
2145 * config/arc/arc.md: Add peephole2 rules for std/ldd.
2147 2018-11-13 Eric Botcazou <ebotcazou@adacore.com>
2149 * toplev.c (output_stack_usage): Turn test on flag_stack_usage into
2150 test on stack_usage_file.
2151 (lang_dependent_init): Do not open the .su file if generating LTO.
2153 2018-11-13 Jakub Jelinek <jakub@redhat.com>
2155 PR rtl-optimization/87918
2156 * simplify-rtx.c (simplify_merge_mask): For COMPARISON_P, use
2157 simplify_gen_relational rather than simplify_gen_binary.
2159 2018-11-13 Richard Biener <rguenther@suse.de>
2161 * tree-ssanames.h (set_range_info): Use value_range_base.
2162 (get_range_info): Likewise.
2163 * tree-ssanames.c (set_range_info): Likewise.
2164 (get_range_info): Likewise.
2165 * tree-vrp.c (value_range_base::union_helper): Split
2166 out common parts of value_range[_base]::union_.
2167 (value_range_base::union_): Update.
2168 (value_range::union_): Likewise.
2169 (determine_value_range_1): Use value_range_base.
2170 (determine_value_range): Likewise.
2171 * tree-vrp.h (value_range_base::union_helper): Move ...
2172 (value_range::union_helper): ... from here.
2174 2018-11-13 Alan Modra <amodra@gmail.com>
2176 * config/rs6000/rs6000.c (rs6000_secondary_reload_inner): Negate
2178 (rs6000_secondary_reload_gpr): Don't call find_replacement.
2180 2018-11-13 Sandra Loosemore <sandra@codesourcery.com>
2183 * doc/invoke.texi (Optimize Options): Clarify that the
2184 l1-cache-line-size, l1-cache-size, and l2-cache-size parameters
2185 apply to data cache size.
2187 2018-11-13 Alan Modra <amodra@gmail.com>
2189 * config/rs6000/predicates.md (logical_const_operand),
2190 (logical_operand): Correct comment.
2191 * config/rs6000/rs6000.c (rs6000_emit_prologue): Comment fix.
2193 2018-11-13 Alan Modra <amodra@gmail.com>
2195 * config/rs6000/rs6000.md (addsi3_high): Prefix with '*'.
2197 2018-11-13 Alan Modra <amodra@gmail.com>
2199 * gcc/doc/md.texi (Machine Constraints): Correct rs6000 h constraint
2201 * config/rs6000/rs6000.md (movsi_internal1): Delete MT%0 case
2202 covered by alternative.
2203 (movcc_internal1): Ignore h for register preference.
2204 (mov<mode>_hardfloat64): Likewise.
2205 (mov<mode>_softfloat): Ignore c, l, h for register preference.
2207 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
2209 PR preprocessor/47823
2210 * doc/cpp.texi (Alternatives to Wrapper #ifndef): Move #pragma once
2213 * doc/extend.texi (Pragmas): Note additional pragmas documented
2216 2018-11-13 Kugan Vivekanandarajah <kuganv@linaro.org>
2220 * tree-scalar-evolution.c (expression_expensive_p): Make
2221 BUILTIN POPCOUNT as expensive when backend does not define it.
2223 2018-11-12 Fredrik Noring <noring@nocrew.org>
2225 * config.gcc: Update with-llsc defaults for MIPS r5900.
2227 2018-11-12 Martin Liska <mliska@suse.cz>
2229 PR gcov-profile/87442
2230 * common.opt: Add -fprofile-filter-files and -fprofile-exclude-files
2232 * doc/invoke.texi: Document them.
2233 * tree-profile.c (parse_profile_filter): New.
2234 (parse_profile_file_filtering): Likewise.
2235 (release_profile_file_filtering): Likewise.
2236 (include_source_file_for_profile): Likewise.
2237 (tree_profiling): Filter source files based on the
2238 newly added options.
2240 2018-11-12 Martin Liska <mliska@suse.cz>
2243 * doc/extend.texi: Add missing values for __builtin_cpu_is and
2244 __builtin_cpu_supports for x86 target.
2246 2018-11-12 Sandra Loosemore <sandra@codesourcery.com>
2249 * doc/rtl.texi (Constants): Clarify that mode of "high" doesn't
2251 (Arithmetic): Likewise for "lo_sum".
2253 2018-11-12 Renlin Li <renlin.li@arm.com>
2256 * dse.c (get_stored_val): Add check for compile-time constantness
2259 2018-11-12 Sudakshina Das <sudi.das@arm.com>
2261 * config/arm/arm-cpus.in (armv8_5, sb, predres): New features.
2262 (ARMv8_5a): New fgroup.
2263 (armv8.5-a): New arch.
2264 (armv8-a, armv8.1-a, armv8.2-a, armv8.3-a, armv8.4-a): New options
2266 * config/arm/arm-tables.opt: Regenerate.
2267 * config/arm/t-aprofile: Add matching rules for -march=armv8.5-a
2268 * config/arm/t-arm-elf (all_v8_archs): Add armv8.5-a.
2269 * config/arm/t-multilib (v8_5_a_simd_variants): New variable.
2270 Add matching rules for -march=armv8.5-a and extensions.
2271 * doc/invoke.texi (ARM options): Document -march=armv8.5-a.
2272 Add sb and predres to all armv8-a except armv8.5-a.
2274 2018-11-12 Richard Biener <rguenther@suse.de>
2276 * tree-vrp.h (value_range[_base]::set): Make public. Provide
2277 overload for single value.
2278 (value_range[_base]::set_nonnull): New.
2279 (value_range[_base]::set_null): Likewise.
2280 (value_range): Document bitmap copying behavior, mark
2281 copy constructor and assignment operator deleted.
2282 (value_range::move): New.
2283 (value_range::set_and_canonicalize): Default bitmap to zero.
2284 (set_value_range_to_nonnull): Remove.
2285 (set_value_range_to_null): Likewise.
2286 (set_value_range): Likewise.
2287 (set_value_range_to_value): Likewise.
2288 (extract_range_from_unary_expr): Work on value_range_base.
2289 (extract_range_from_binary_expr_1): Likewise. Rename to...
2290 (extract_range_from_binary_expr): ... this.
2291 * tree-vrp.c (value_range::update): Clear equiv bitmap
2293 (value_range::move): New, move equiv bitmap.
2294 (value_range_base::set_undefined): Avoid assignment.
2295 (value_range::set_undefined): Likewise.
2296 (value_range_base::set_varying): Likewise.
2297 (value_range::set_varying): Likewise.
2298 (set_value_range): Remove.
2299 (value_range_base::set): New overload for value.
2300 (value_range::set): Likewise.
2301 (set_value_range_to_nonnull): Remove.
2302 (value_range_base::set_nonnull): New.
2303 (value_range::set_nonnull): Likewise.
2304 (set_value_range_to_null): Remove.
2305 (value_range_base::set_null): New.
2306 (value_range::set_null): Likewise.
2307 (range_is_null): Work on value_range_base.
2308 (range_is_nonnull): Likewise.
2309 (ranges_from_anti_range): Likewise.
2310 (extract_range_into_wide_ints): Likewise.
2311 (extract_range_from_multiplicative_op): Likewise.
2312 (extract_range_from_binary_expr): Likewise. Update for API changes.
2313 (extract_range_from_unary_expr): Likewise. Remove OBJ_TYPE_REF
2315 (value_range::intersect_helper): Avoid copy and assignment.
2316 (value_range::union_helper): Likewise.
2317 (determine_value_range_1): Adjust.
2318 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::try_find_new_range):
2319 Avoid assignment by using move.
2320 (evrp_range_analyzer::record_ranges_from_stmt): Avoid assignment.
2321 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
2323 * tree-ssanames.c (get_range_info): Likewise.
2324 * vr-values.h (vr_values::get_vr_for_comparison): Adjust API.
2325 * vr-values.c (vr_values::get_value_range): Adjust.
2326 (vr_values::update_value_range): Likewise.
2327 (symbolic_range_based_on_p): Work on value_range_base.
2328 (vr_values::extract_range_from_binary_expr): Use value_range_base.
2329 (vr_values::extract_range_from_unary_expr): Likewise.
2330 (vr_values::extract_range_from_cond_expr): Avoid assignment.
2331 (vr_values::extract_range_from_comparison): Adjust.
2332 (vr_values::check_for_binary_op_overflow): Use value_range_base.
2333 (vr_values::extract_range_basic): Adjust.
2334 (vr_values::adjust_range_with_scev): Likewise.
2335 (vr_values::vrp_visit_assignment_or_call): Likewise.
2336 (vr_values::get_vr_for_comparison): Change API to avoid
2337 assignment and copy construction.
2338 (vr_values::compare_name_with_value): Adjust accordingly.
2339 (vr_values::compare_names): Likewise.
2340 (vr_values::extract_range_from_phi_node): Avoid assignment and
2341 bogus in-place modify of equiv bitmap.
2342 (vr_values::simplify_bit_ops_using_ranges): Use value_range_base.
2343 * ipa-prop.c (ipa_compute_jump_functions_for_edge): Adjust
2344 for extract_range_from_unary_expr API change.
2345 * ipa-cp.c (ipa_vr_operation_and_type_effects): Likewise.
2347 2018-11-12 Eric Botcazou <ebotcazou@adacore.com>
2349 * config/mcore/mcore.h (WORD_REGISTER_OPERATIONS): Remove duplicate.
2351 2018-11-12 Richard Biener <rguenther@suse.de>
2353 * tree-vrp.h (value_range_base::symbolic_p,
2354 value_range_base::constant_p, value_range_base::zero_p,
2355 value_range_base::singleton_p): Move from value_range.
2356 (value_range::dump): Add.
2357 * gimple-ssa-evrp-analyze.c
2358 (evrp_range_analyzer::record_ranges_from_phis): Use set_varying.
2359 * ipa-cp.c (ipcp_vr_lattice::print): Use dump_value_range.
2360 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
2362 * tree-vrp.c (value_range::symbolic_p): Move to value_range_base.
2363 (value_range::constant_p): Likewise.
2364 (value_range::singleton_p): Likewise.
2365 (value_range_base::dump): Add.
2366 (set_value_range_to_undefined): Remove.
2367 (set_value_range_to_varying): Likewise.
2368 (range_int_cst_p): Take value_range_base argument.
2369 (range_int_cst_singleton_p): Likewise.
2370 (value_range_constant_singleton): Likewise.
2371 (vrp_set_zero_nonzero_bits): Likewise.
2372 (extract_range_from_multiplicative_op): Use set_varying.
2373 (extract_range_from_binary_expr_1): Likewise. Use set_undefined.
2374 (extract_range_from_unary_expr): Likewise.
2375 (dump_value_range_base): Change to overload of dump_value_range.
2376 (vrp_prop::vrp_initialize): Use set_varying and set_undefined.
2377 (vrp_prop::visit_stmt): Likewise.
2378 (value_range::intersect_helper): Likewise.
2379 (value_range::union_helper): Likewise.
2380 (determine_value_range_1): Likewise.
2382 2018-11-12 Richard Biener <rguenther@suse.de>
2384 * tree-vrp.c (set_value_range_to_nonnull): Clear equiv.
2385 (set_value_range_to_null): Likewise.
2386 * vr-values.c (vr_values::extract_range_from_comparison):
2387 Clear equiv for constant singleton ranges.
2389 2018-11-12 Wei Xiao <wei3.xiao@intel.com>
2391 * config/i386/sse.md: Combine VFIXUPIMM* patterns
2392 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2393 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2394 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Remove.
2395 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
2396 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
2397 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Remove.
2399 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
2402 * doc/extend.texi (Common Type Attributes): For the align type
2403 attribute, copy language about decreasing alignment from the
2404 corresponding variable attribute.
2406 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
2408 * config/csky/csky.md (*fpuv2_nmulsf3_1, *fpuv2_nmuldf3_1): Handle
2411 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
2414 * doc/invoke.texi (C++ Dialect Options): Add warning about mixing
2415 -frtti and -fno-rtti code.
2417 2018-11-11 Sandra Loosemore <sandra@codesourcery.com>
2420 * doc/extend.texi (Other Builtins): Document probability associated
2421 with __builtin_expect.
2423 2018-11-11 Uros Bizjak <ubizjak@gmail.com>
2426 * config/i386/i386.h (STACK_BOUNDARY): Use TARGET_64BIT_MS_ABI
2427 instead of (TARGET_64BIT && ix86_abi == MS_ABI).
2428 * config/i386/darwin.h (STACK_BOUNDARY): Ditto.
2429 * config/i386/cygming.h (STACK_BOUNDARY): Remove.
2431 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
2433 * config/csky/csky-linux-elf.h (CC1_SPEC): Support -profile.
2435 2018-11-11 Xianmiao Qu <xianmiao_qu@c-sky.com>
2437 * config/csky/csky.h (ASM_PREFERRED_EH_DATA_FORMAT): Define.
2439 2018-11-11 Richard Biener <rguenther@suse.de>
2441 * tree-vrp.h (class value_range_base): New base class for
2442 value_range containing all but the m_equiv member.
2443 (dump_value_range_base): Add.
2444 (range_includes_zero_p): Work on value_range_base.
2445 * tree-vrp.c (value_range_base::set): Split out base handling
2447 (value_range::set): this.
2448 (value_range::set_equiv): New.
2449 (value_range_base::value_range_base): New constructors.
2450 (value_range_base::check): Split out base handling from...
2451 (value_range::check): this.
2452 (value_range::equal_p): Refactor in terms of
2453 ignore_equivs_equal_p which is now member of the base.
2454 (value_range_base::set_undefined): New.
2455 (value_range_base::set_varying): Likewise.
2456 (value_range_base::dump):Split out base handling from...
2457 (value_range::dump): this.
2458 (value_range_base::set_and_canonicalize): Split out base handling
2460 (value_range::set_and_canonicalize): this.
2461 (value_range_base::union_): New.
2462 * ipa-prop.h (struct ipa_jump_func): Use value_range_base *
2464 * ipa-cp.c (class ipcp_vr_lattice): Use value_range_base
2465 instead of value_range everywhere.
2466 (ipcp_vr_lattice::print): Use dump_value_range_base.
2467 (ipcp_vr_lattice::meet_with): Adjust.
2468 (ipcp_vr_lattice::meet_with_1): Likewise.
2469 (ipa_vr_operation_and_type_effects): Likewise.
2470 (propagate_vr_across_jump_function): Likewise.
2471 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Likewise.
2472 (ipa_get_value_range): Likewise.
2473 (ipa_set_jfunc_vr): Likewise.
2474 (ipa_compute_jump_functions_for_edge): Likewise.
2476 2018-11-10 Sandra Loosemore <sandra@codesourcery.com>
2479 * doc/invoke.texi (Optimize Options): Add @opindex entries
2480 for the positive forms of -fno-xxx and -mno-xxx options
2481 that were lacking them.
2483 2018-11-10 Segher Boessenkool <segher@kernel.crashing.org>
2485 * combine.c (make_more_copies): Only make an intermediate copy if the
2486 dest of a move is a pseudo.
2488 2018-11-09 Maya Rashish <coypu@sdf.org>
2491 * config/netbsd-elf.h (NETBSD_STARTFILE_SPEC): Use crtbeginS.o for PIE.
2492 (NETBSD_ENDFILE_SPEC): Use crtendS.o for PIE.
2494 2018-11-09 Sandra Loosemore <sandra@codesourcery.com>
2498 * doc/invoke.texi (Optimize Options): Clarify default behavior
2499 for -fno-toplevel-reorder, -fno-defer-pop, and -fno-branch-count-reg.
2501 2018-11-09 Bernd Edlinger <bernd.edlinger@hotmail.de>
2503 PR tree-optimization/87940
2504 * expr.c (string_constant): Don't strip NOPS in subexpressions.
2505 Fold PLUS_EXPR correctly.
2507 2018-11-09 Ilya Leoshkevich <iii@linux.ibm.com>
2510 * config/s390/s390.c (s390_safe_relative_long_p): New function.
2511 (annotate_constant_pool_refs): Skip insns which support
2512 relative addressing.
2513 (annotate_constant_pool_refs_1): New helper function.
2514 (find_constant_pool_ref): Skip insns which support relative
2516 (find_constant_pool_ref_1): New helper function.
2517 (replace_constant_pool_ref): Skip insns which support
2518 relative addressing.
2519 (replace_constant_pool_ref_1): New helper function.
2520 (s390_mainpool_start): Adapt to the new signature.
2521 (s390_mainpool_finish): Likewise.
2522 (s390_chunkify_start): Likewise.
2523 (s390_chunkify_finish): Likewise.
2524 (pass_s390_early_mach::execute): Likewise.
2525 (s390_prologue_plus_offset): Likewise.
2526 (s390_emit_prologue): Likewise.
2527 (s390_emit_epilogue): Likewise.
2529 2018-11-09 Jakub Jelinek <jakub@redhat.com>
2531 * gimplify.c (gimplify_scan_omp_clauses): Call sorry_at for valid
2532 but unsupported lastprivate with conditional modifier.
2534 2018-11-09 Jeff Law <law@redhat.com>
2536 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): Handle
2537 unused argument better. Add gcc_unreachable to silence warning.
2539 2018-11-09 Martin Sebor <msebor@redhat.com>
2542 * attribs.c (has_attribute): New helper function.
2543 (decls_mismatched_attributes, maybe_diag_alias_attributes): Same.
2544 * attribs.h (decls_mismatched_attributes): Declare.
2545 * cgraphunit.c (handle_alias_pairs): Call maybe_diag_alias_attributes.
2546 (maybe_diag_incompatible_alias): Use OPT_Wattribute_alias_.
2547 * common.opt (-Wattribute-alias): Take an argument.
2548 (-Wno-attribute-alias): New option.
2549 * doc/extend.texi (Common Function Attributes): Document copy.
2550 (Common Variable Attributes): Same.
2551 * doc/invoke.texi (-Wmissing-attributes): Document enhancement.
2552 (-Wattribute-alias): Document new option argument.
2554 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
2556 * config/arm/parsecpu.awk (/alias/): Tighten invisible alias
2557 matching criteria. Remove unused array initializer.
2559 2018-11-09 Bill Schmidt <wschmidt@linux.ibm.com>
2560 Jinsong Ji <jji@us.ibm.com>
2562 * config/rs6000/xmmintrin.h (_mm_cvtss_si32): Fix incorrect
2563 constraints by introducing a new temporary.
2564 (_mm_cvtss_si64): Likewise.
2566 2018-11-09 Martin Liska <mliska@suse.cz>
2568 * common.opt: Add -fipa-stack-alignment flag.
2569 * doc/invoke.texi: Document it.
2570 * final.c (rest_of_clean_state): Guard stack
2571 shrinking with flag.
2573 2018-11-09 Martin Liska <mliska@suse.cz>
2575 * cgraph.h (ipa_discover_readonly_nonaddressable_vars): Rename
2577 (ipa_discover_variable_flags): ... this.
2578 * common.opt: Come up with new flag -fipa-reference-addressable.
2579 * doc/invoke.texi: Document it.
2580 * ipa-reference.c (propagate): Call the renamed fn.
2581 * ipa-visibility.c (whole_program_function_and_variable_visibility):
2583 * ipa.c (ipa_discover_readonly_nonaddressable_vars): Renamed to
2585 (ipa_discover_variable_flags): ... this. Discover
2586 non-addressable variables only with the newly added flag.
2587 * opts.c: Enable the newly added flag with -O1 and higher
2590 2018-11-09 David Malcolm <dmalcolm@redhat.com>
2592 * json.cc (selftest::test_writing_literals): Fix comment.
2594 2018-11-09 Martin Liska <mliska@suse.cz>
2596 * config/aarch64/aarch64.c (aarch64_parse_arch): Do not copy
2597 string to a stack buffer.
2598 (aarch64_parse_cpu): Likewise.
2599 (aarch64_parse_tune): Likewise.
2601 2018-11-09 Richard Biener <rguenther@suse.de>
2603 PR tree-optimization/87953
2604 * tree-vect-loop.c (vectorizable_reduction): For analysis
2605 always pass ops[0] to vectorizable_condition.
2607 2018-11-09 Stafford Horne <shorne@gmail.com>
2608 Richard Henderson <rth@twiddle.net>
2609 Joel Sherrill <joel@rtems.org>
2611 * common/config/or1k/or1k-common.c: New file.
2612 * config/or1k/*: New.
2613 * config.gcc (or1k*-*-*): New.
2614 * configure.ac (or1k*-*-*): New test for openrisc tls.
2615 * configure: Regenerated.
2616 * doc/install.texi: Document OpenRISC triplets.
2617 * doc/invoke.texi: Document OpenRISC arguments.
2618 * doc/md.texi: Document OpenRISC.
2620 2018-11-09 Richard Earnshaw <rearnsha@arm.com>
2622 * config/arm/arm-cpus.in (arm7tdmi): Add an alias for arm7tdmi-s.
2623 (arm7tdmi-s): Delete CPU.
2624 (arm710t): Add aliases for arm720t and arm740t.
2625 (arm720t, arm740t): Delete CPUs.
2626 (arm920t): Add aliases for arm920, arm922t and arm940t.
2627 (arm920, arm922t, arm940t): Delete CPUs.
2628 (arm10tdmi): Add alias for arm1020t.
2629 (arm1020t): Delete CPU.
2630 (arm9e): Add aliases for arm946e-s, arm966e-s and arm968e-s.
2631 (arm946e-s, arm966e-s, arm968e-s): Delete CPUs.
2632 (arm10e): Add aliases for arm1020e and arm1022e.
2633 (arm1020e, arm1022e): Delete CPU.
2634 * config/arm/arm.md (generic_sched): Remove entries that are now
2636 (generic_vfp): Likewise.
2637 * config/arm/arm1020e.md: Simplify tuning selection based on alias
2639 * config/arm/arm-tune.md: Regenerated.
2640 * config/arm/arm-tables.opt: Regenerated.
2642 2018-11-09 Richard Biener <rguenther@suse.de>
2644 PR tree-optimization/87621
2645 * tree-vect-loop.c (vectorizable_reduction): Handle reduction
2646 op with only phi inputs.
2647 * tree-ssa-loop-ch.c: Include tree-ssa-sccvn.h.
2648 (ch_base::copy_headers): Run CSE on copied loop headers.
2649 (pass_ch_vect::process_loop_p): Simplify.
2651 2018-11-09 Alexandre Oliva <oliva@adacore.com>
2653 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Adjust
2654 for non-w64 x86_64 biarch.
2656 2018-11-09 Alexandre Oliva <aoliva@redhat.com>
2658 PR rtl-optimization/86438
2659 * compare-elim.c (try_eliminate_compare): Use SET_SRC instead
2660 of in_b for the compare if in_b is SET_DEST.
2663 * config/i386/i386.c (ix86_const_not_ok_for_debug_p): Reject
2664 non-toplevel UNSPEC.
2666 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
2668 * tree-vrp.c (value_range::check): Do not access internals
2670 (value_range::singleton_p): Same.
2671 (value_range::type): Same.
2672 (vrp_finalize): Use value_range API.
2674 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
2676 * tree-vrp.c (may_contain_p): Do not access m_min/m_max directly.
2678 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
2680 * vr-values.c (vr_values::get_value_range): Use value_range API
2681 instead of piecing together ranges.
2682 (vr_values::update_value_range): Same.
2684 2018-11-08 Aldy Hernandez <aldyh@redhat.com>
2686 * gimple-fold.c (size_must_be_zero_p): Use value_range API instead
2687 of performing ad-hoc calculations.
2688 * tree-ssanames.c (set_range_info): New overloaded function
2689 accepting value_range &.
2690 (get_range_info): Same.
2691 * tree-ssanames.h (set_range_info_raw): Remove.
2692 (set_range_info): New prototype.
2693 (get_range_info): Same.
2694 * tree-vrp.h (value_range::null_p): Rename to zero_p.
2695 * tree-vrp.c (value_range::null_p): Same.
2697 2018-11-09 Jan Hubicka <jh@suse.cz>
2699 * tree.c (fld_type_variant_equal_p): Test user align flag.
2700 (flt_type_variant): Copy user align flag.
2701 (fld_incomplete_type_of): Clear it.
2703 2018-11-09 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2705 * config/arm/neon.md (div<mode>3): New pattern.
2707 2018-11-08 Andi Kleen <ak@linux.intel.com>
2709 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PTWRITE_SET): New.
2710 (OPTION_MASK_ISA_PTWRITE_UNSET): New.
2711 (ix86_handle_option): Handle OPT_mptwrite.
2712 * config/i386/cpuid.h (bit_PTWRITE): Add.
2713 * config/i386/driver-i386.c (host_detect_local_cpu): Detect ptwrite.
2714 * config/i386/i386-builtin.def (BDESC): Add ptwrite32/64.
2715 * config/i386/i386-c.c (ix86_target_macros_internal): Define __PTWRITE__.
2716 * config/i386/i386.c (ix86_target_string): Handle ptwrite.
2717 (ix86_option_override_internal): Handle PTA_PTWRITE.
2718 (ix86_valid_target_attribute_inner_p): Define ptwrite.
2719 (def_builtin2): Force UINT64 to be 64bit only.
2720 * config/i386/i386.h (TARGET_PTWRITE): Add.
2721 (TARGET_PTWRITE_P): Add.
2723 * config/i386/i386.md: Define ptwrite.
2724 * config/i386/i386.opt: Add -mptwrite.
2725 * config/i386/immintrin.h (_ptwrite64): Add.
2727 * doc/extend.texi: Document __builtin_ia32_ptwrite*.
2728 * doc/invoke.texi: Document -mptwrite.
2730 2018-11-08 Peter Bergner <bergner@linux.ibm.com>
2732 PR rtl-optimization/87600
2733 * cfgexpand.c (expand_asm_stmt): Catch illegal asm constraint usage.
2734 * lra-constraints.c (process_alt_operands): Skip illegal hard
2735 register usage. Prefer reloading non hard register operands.
2737 2018-11-08 Sandra Loosemore <sandra@codesourcery.com>
2740 * doc/invoke.texi (Optimize Options): Clarify default behavior
2741 for -fno-sched-interblock and -fno-sched-spec.
2743 2018-11-08 Roman Geissler <roman.geissler@amadeus.com>
2745 * collect2.c (linker_select): Add USE_LLD_LD.
2746 (ld_suffixes): Add ld.lld.
2747 (main): Handle -fuse-ld=lld.
2748 * common.opt (-fuse-ld=lld): New option.
2749 * doc/invoke.texi (-fuse-ld=lld): Document.
2750 * opts.c (common_handle_option): Handle OPT_fuse_ld_lld.
2752 2018-11-08 Paul Koning <ni1d@arrl.net>
2754 * config/pdp11/constraints.md: Add "Z" series constraints for use
2755 with pre-dec and post-inc addressing.
2756 * config/pdp11/pdp11-protos.m (expand_block_move): Delete.
2757 (pdp11_expand_operands): Add int argument (word count).
2758 (pdp11_sp_frame_offset): Delete.
2759 (pdp11_cmp_length): New function.
2760 (pushpop_regeq): New function.
2761 * config/pdp11/pdp11.c (TARGET_STACK_PROTECT_RUNTIME_ENABLED_P):
2763 (pdp11_expand_prologue, pdp11_expand_epilogue): Rewrite for new
2765 (pdp11_initial_elimination_offset): Ditto.
2766 (pdp11_expand_operands): Add word count argument. Bugfixes.
2767 (output_move_multiple): Change how pointer adjustment is done.
2768 (pdp11_gen_int_label): Correct format.
2769 (output_ascii): Ditto.
2770 (pdp11_asm_output_var): Add code for DEC assembler case.
2771 (pdp11_asm_print_operand): Bugfix for CONST_DOUBLE holding integer
2773 (legitimate_const_double_p): Ditto.
2774 (pdp11_register_move_cost): Adjust for new register classes.
2775 (pdp11_regno_reg_class): Ditto.
2776 (expand_block_move): Delete.
2777 (pushpop_regeq): New function.
2778 (pdp11_legitimate_address_p): Bugfix in check for constant
2780 (pdp11_sp_frame_offset): Delete.
2781 (pdp11_reg_save_size): New helper function for new frame layout.
2782 (output_addr_const_pdp11): Remove CONST_DOUBLE case.
2783 (pdp11_expand_shift): Bugfix in check for constant shift count.
2784 (pdp11_shift_length): Ditto.
2785 (pdp11_assemble_shift): Copy input to pdp11_expand_operands.
2786 (pdp11_cmp_length): New function.
2787 * config/pdp11/pdp11.h (TARGET_CPU_CPP_BUILTINS): Add macros for
2788 some compile options.
2789 (FIXED_REGISTERS): Remove HARD_FRAME_POINTER_REGNUM.
2790 (CALL_USED_REGISTERS): Ditto.
2791 (ELIMINABLE_REGS): Ditto.
2792 (REGISTER_NAMES): Ditto.
2793 (reg_class): Add classes NOTR0_REG through NOTSP_REG for use by Z
2795 (REG_CLASS_NAMES): Ditto.
2796 (REG_CLASS_CONTENTS): Ditto. Also remove
2797 HARD_FRAME_POINTER_REGNUM.
2798 (CPU_REG_CLASS): New macro.
2799 (CLASS_MAX_NREGS): Adjust for new register classes.
2800 (FUNCTION_PROFILER): Make no-op.
2801 (may_call_alloca): Remove unused declaration.
2802 (ASM_OUTPUT_ALIGN): Add workaround for PR87795.
2803 (ASM_OUTPUT_SKIP): Fix format.
2804 * config/pdp11/pdp11.md (unspecv): Add UNSPECV_MOVMEM.
2805 (HARD_FRAME_POINTER_REGNUM): Remove.
2807 (*rts): Rename. Remove epilogue related checks.
2808 (cmpsi, cmpdi): New insn.
2809 (cbranch<mode>4): Change to apply to SI and DI modes as well.
2810 (mov<mode>): Change constraints to enforce that push/pop
2811 destination cannot use the same register as source.
2812 (*mov<mode><cc_cc>): Ditto.
2813 (movmemhi, movmemhi1, movmemhi_nocc): Change to expand block move
2814 at assembly output rather than as RTL expander.
2815 (zero_extendqihi2): Bugfix in check for same registers.
2816 (adddi3_nocc): Bugfix in check for constant operand.
2817 (addsi3_nocc): Ditto.
2818 (subdi3_nocc): Ditto.
2819 (subsi3_nocc): Ditto.
2820 (negdi2_nocc): Copy input to pdp11_expand_operands.
2821 (negsi2_nocc): Ditto.
2822 (bswap2_nocc): Ditto.
2823 * config/pdp11/pdp11.opt (mlra): Fix documentation.
2824 * config/pdp11/t-pdp11: Use -Os.
2826 2018-11-08 Richard Earnshaw <rearnsha@arm.com>
2828 * config/arm/parsecpu.awk (/alias/): New parsing rule.
2829 (/begin cpu/): Check that the cpu name hasn't been previously defined.
2830 (gen_comm_data): Print out CPU alias tables.
2831 (check_cpu): Match aliases when checking the CPU name.
2832 * config/arm/arm-protos.h (cpu_alias): New structure.
2833 (cpu_option): Add entry for aliases.
2834 * config/arm/arm-cpus.in (strongarm): Add aliases for strongarm110
2835 strongarm1100 and strongarm1110.
2836 (strongarm110, strongarm1100, strongarm1110): Delete CPU entries.
2837 (config/arm/arm-generic.md): Remove redundant references to
2838 strongarm110, strongarm1100 and strongarm1110.
2839 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
2840 Scan aliases for additional hints.
2841 (arm_parse_cpu_option_name): Also match a cpu name against the list
2843 * config/arm/arm-tables.opt: Regenerated.
2844 * config/arm/arm-tune.md: Regenerated.
2846 2018-11-08 Jakub Jelinek <jakub@redhat.com>
2848 * builtin-types.def (BT_FN_VOID_BOOL, BT_FN_VOID_SIZE_SIZE_PTR,
2849 BT_FN_UINT_UINT_PTR_PTR, BT_FN_UINT_OMPFN_PTR_UINT_UINT,
2850 BT_FN_BOOL_UINT_LONGPTR_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
2851 BT_FN_BOOL_UINT_ULLPTR_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR,
2852 BT_FN_BOOL_LONG_LONG_LONG_LONG_LONG_LONGPTR_LONGPTR_PTR_PTR,
2853 BT_FN_BOOL_BOOL_ULL_ULL_ULL_LONG_ULL_ULLPTR_ULLPTR_PTR_PTR): New.
2854 * gengtype.c (open_base_files): Add omp-general.h.
2855 * gimple.c (gimple_build_omp_critical):
2856 (gimple_build_omp_taskgroup): Add CLAUSES argument. Call
2857 gimple_omp_taskgroup_set_clauses.
2858 (gimple_build_omp_atomic_load): Add mo argument, call
2859 gimple_omp_atomic_set_memory_order.
2860 (gimple_build_omp_atomic_store): Likewise.
2861 (gimple_copy): Adjust handling of GIMPLE_OMP_TASKGROUP.
2862 * gimple.def (GIMPLE_OMP_TASKGROUP): Use GSS_OMP_SINGLE_LAYOUT
2864 (GIMPLE_OMP_TEAMS): Use GSS_OMP_PARALLEL_LAYOUT instead
2865 of GSS_OMP_SINGLE_LAYOUT, adjust comments.
2866 * gimple.h (enum gf_mask): Add GF_OMP_TEAMS_HOST, GF_OMP_TASK_TASKWAIT
2867 and GF_OMP_ATOMIC_MEMORY_ORDER. Remove GF_OMP_ATOMIC_SEQ_CST, use
2868 different value for GF_OMP_ATOMIC_NEED_VALUE.
2869 (struct gimple_statement_omp_taskreg): Add GIMPLE_OMP_TEAMS to
2871 (struct gimple_statement_omp_single_layout): And remove here.
2872 (struct gomp_teams): Inherit from gimple_statement_omp_taskreg rather
2873 than gimple_statement_omp_single_layout.
2874 (is_a_helper <gimple_statement_omp_taskreg *>::test): Allow
2876 (is_a_helper <const gimple_statement_omp_taskreg *>::test): Likewise.
2877 (gimple_omp_subcode): Formatting fix.
2878 (gimple_omp_teams_child_fn, gimple_omp_teams_child_fn_ptr,
2879 gimple_omp_teams_set_child_fn, gimple_omp_teams_data_arg,
2880 gimple_omp_teams_data_arg_ptr, gimple_omp_teams_set_data_arg,
2881 gimple_omp_teams_host, gimple_omp_teams_set_host,
2882 gimple_omp_task_taskwait_p, gimple_omp_task_set_taskwait_p,
2883 gimple_omp_taskgroup_clauses, gimple_omp_taskgroup_clauses_ptr,
2884 gimple_omp_taskgroup_set_clauses): New inline functions.
2885 (gimple_build_omp_atomic_load): Add enum omp_memory_order argument.
2886 (gimple_build_omp_atomic_store): Likewise.
2887 (gimple_omp_atomic_seq_cst_p): Remove.
2888 (gimple_omp_atomic_memory_order): New function.
2889 (gimple_omp_atomic_set_seq_cst): Remove.
2890 (gimple_omp_atomic_set_memory_order): New function.
2891 (gimple_build_omp_taskgroup): Add clauses argument.
2892 * gimple-pretty-print.c (dump_gimple_omp_taskgroup): New function.
2893 (dump_gimple_omp_task): Print taskwait with depend clauses.
2894 (dump_gimple_omp_atomic_load, dump_gimple_omp_atomic_store): Use
2895 dump_omp_atomic_memory_order.
2896 (pp_gimple_stmt_1): Handle GIMPLE_OMP_TASKGROUP.
2897 * gimplify.c (enum gimplify_omp_var_data): Add GOVD_MAP_ALLOC_ONLY,
2898 GOVD_MAP_FROM_ONLY and GOVD_NONTEMPORAL.
2899 (enum omp_region_type): Reserve bits 1 and 2 for auxiliary flags,
2900 renumber values of most of ORT_* enumerators, add ORT_HOST_TEAMS,
2901 ORT_COMBINED_HOST_TEAMS, ORT_TASKGROUP, ORT_TASKLOOP and
2902 ORT_UNTIED_TASKLOOP enumerators.
2903 (enum gimplify_defaultmap_kind): New.
2904 (struct gimplify_omp_ctx): Remove target_map_scalars_firstprivate and
2905 target_map_pointers_as_0len_arrays members, add defaultmap.
2906 (new_omp_context): Initialize defaultmap member.
2907 (gimple_add_tmp_var): Handle ORT_TASKGROUP like ORT_WORKSHARE.
2908 (maybe_fold_stmt): Don't fold even in host teams regions.
2909 (omp_firstprivatize_variable): Handle ORT_TASKGROUP like
2910 ORT_WORKSHARE. Test ctx->defaultmap[GDMK_SCALAR] instead of
2911 ctx->omp_firstprivatize_variable.
2912 (omp_add_variable): Don't add private/firstprivate for VLAs in
2914 (omp_default_clause): Print "taskloop" rather than "task" if
2916 (omp_notice_variable): Handle ORT_TASKGROUP like ORT_WORKSHARE.
2917 Handle new defaultmap clause kinds.
2918 (omp_is_private): Handle ORT_TASKGROUP like ORT_WORKSHARE. Allow simd
2919 iterator to be lastprivate or private. Fix up diagnostics if linear
2920 is used on collapse>1 simd iterator.
2921 (omp_check_private): Handle ORT_TASKGROUP like ORT_WORKSHARE.
2922 (gimplify_omp_depend): New function.
2923 (gimplify_scan_omp_clauses): Add shared clause on parallel for
2924 combined parallel master taskloop{, simd} if taskloop has
2925 firstprivate, lastprivate or reduction clause. Handle
2926 OMP_CLAUSE_REDUCTION_TASK diagnostics. Adjust tests for
2927 ORT_COMBINED_TEAMS. Gimplify depend clauses with iterators. Handle
2928 cancel and simd OMP_CLAUSE_IF_MODIFIERs. Handle
2929 OMP_CLAUSE_NONTEMPORAL. Handle new defaultmap clause kinds. Handle
2930 OMP_CLAUSE_{TASK,IN}_REDUCTION. Diagnose invalid conditional
2932 (gimplify_adjust_omp_clauses_1): Ignore GOVD_NONTEMPORAL. Handle
2933 GOVD_MAP_ALLOC_ONLY and GOVD_MAP_FROM_ONLY.
2934 (gimplify_adjust_omp_clauses): Handle OMP_CLAUSE_NONTEMPORAL. Handle
2935 OMP_CLAUSE_{TASK,IN}_REDUCTION.
2936 (gimplify_omp_task): Handle taskwait with depend clauses.
2937 (gimplify_omp_for): Add shared clause on parallel for combined
2938 parallel master taskloop{, simd} if taskloop has firstprivate,
2939 lastprivate or reduction clause. Use ORT_TASKLOOP or
2940 ORT_UNTIED_TASKLOOP instead of ORT_TASK or ORT_UNTIED_TASK. Adjust
2941 tests for ORT_COMBINED_TEAMS. Handle C++ range for loops with
2942 NULL TREE_PURPOSE in OMP_FOR_ORIG_DECLS. Firstprivatize
2943 __for_end and __for_range temporaries on OMP_PARALLEL for
2944 distribute parallel for{, simd}. Move OMP_CLAUSE_REDUCTION
2945 and OMP_CLAUSE_IN_REDUCTION from taskloop to the task construct
2946 sandwiched in between two taskloops.
2947 (computable_teams_clause): Test ctx->defaultmap[GDMK_SCALAR]
2948 instead of ctx->omp_firstprivatize_variable.
2949 (gimplify_omp_workshare): Set ort to ORT_HOST_TEAMS or
2950 ORT_COMBINED_HOST_TEAMS if not inside of target construct. If
2951 host teams, use gimplify_and_return_first etc. for body like
2952 for target or target data constructs, and at the end call
2953 gimple_omp_teams_set_host on the GIMPLE_OMP_TEAMS object.
2954 (gimplify_omp_atomic): Use OMP_ATOMIC_MEMORY_ORDER instead
2955 of OMP_ATOMIC_SEQ_CST, pass it as new argument to
2956 gimple_build_omp_atomic_load and gimple_build_omp_atomic_store, remove
2957 gimple_omp_atomic_set_seq_cst calls.
2958 (gimplify_expr) <case OMP_TASKGROUP>: Move handling into a separate
2959 case, handle taskgroup clauses.
2960 * lto-streamer-out.c (hash_tree): Handle
2961 OMP_CLAUSE_{TASK,IN}_REDUCTION.
2962 * Makefile.in (GTFILES): Add omp-general.h.
2963 * omp-builtins.def (BUILT_IN_GOMP_TASKWAIT_DEPEND,
2964 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_START,
2965 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_START,
2966 BUILT_IN_GOMP_LOOP_START, BUILT_IN_GOMP_LOOP_ORDERED_START,
2967 BUILT_IN_GOMP_LOOP_DOACROSS_START,
2968 BUILT_IN_GOMP_LOOP_NONMONOTONIC_RUNTIME_NEXT,
2969 BUILT_IN_GOMP_LOOP_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
2970 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_START,
2971 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_START,
2972 BUILT_IN_GOMP_LOOP_ULL_START, BUILT_IN_GOMP_LOOP_ULL_ORDERED_START,
2973 BUILT_IN_GOMP_LOOP_ULL_DOACROSS_START,
2974 BUILT_IN_GOMP_LOOP_ULL_NONMONOTONIC_RUNTIME_NEXT,
2975 BUILT_IN_GOMP_LOOP_ULL_MAYBE_NONMONOTONIC_RUNTIME_NEXT,
2976 BUILT_IN_GOMP_PARALLEL_LOOP_NONMONOTONIC_RUNTIME,
2977 BUILT_IN_GOMP_PARALLEL_LOOP_MAYBE_NONMONOTONIC_RUNTIME,
2978 BUILT_IN_GOMP_PARALLEL_REDUCTIONS, BUILT_IN_GOMP_SECTIONS2_START,
2979 BUILT_IN_GOMP_TEAMS_REG, BUILT_IN_GOMP_TASKGROUP_REDUCTION_REGISTER,
2980 BUILT_IN_GOMP_TASKGROUP_REDUCTION_UNREGISTER,
2981 BUILT_IN_GOMP_TASK_REDUCTION_REMAP,
2982 BUILT_IN_GOMP_WORKSHARE_TASK_REDUCTION_UNREGISTER): New builtins.
2983 * omp-expand.c (workshare_safe_to_combine_p): Return false for
2984 non-worksharing loops.
2985 (omp_adjust_chunk_size): Don't adjust anything if chunk_size is zero.
2986 (determine_parallel_type): Don't combine parallel with worksharing
2987 which has _reductemp_ clause.
2988 (expand_parallel_call): Emit the GOMP_*nonmonotonic_runtime* or
2989 GOMP_*maybe_nonmonotonic_runtime* builtins instead of GOMP_*runtime*
2990 if there is nonmonotonic modifier or if there is no modifier and no
2991 ordered clause. For dynamic and guided schedule without monotonic
2992 and nonmonotonic modifier, default to nonmonotonic.
2993 (expand_omp_for): Likewise. Adjust expand_omp_for_generic caller, use
2994 GOMP_loop{,_ull}{,_ordered,_doacross}_start builtins if there are
2996 (expand_task_call): Add GOMP_TASK_FLAG_REDUCTION flag to flags if
2997 there are any reduction clauses.
2998 (expand_taskwait_call): New function.
2999 (expand_teams_call): New function.
3000 (expand_omp_taskreg): Allow GIMPLE_OMP_TEAMS and call
3001 expand_teams_call for it. Formatting fix. Handle taskwait with
3003 (expand_omp_for_generic): Add SCHED_ARG argument. Handle expansion
3004 of worksharing loops with task reductions.
3005 (expand_omp_for_static_nochunk, expand_omp_for_static_chunk): Handle
3006 expansion of worksharing loops with task reductions.
3007 (expand_omp_sections): Handle expansion of sections with task
3009 (expand_omp_synch): For host teams call expand_omp_taskreg.
3010 (omp_memory_order_to_memmodel): New function.
3011 (expand_omp_atomic_load, expand_omp_atomic_store,
3012 expand_omp_atomic_fetch_op): Use it and gimple_omp_atomic_memory_order
3013 instead of gimple_omp_atomic_seq_cst_p.
3014 (build_omp_regions_1, omp_make_gimple_edges): Treat taskwait with
3015 depend clauses as a standalone directive.
3016 * omp-general.c (enum omp_requires): New variable.
3017 (omp_extract_for_data): Initialize have_reductemp member. Allow
3018 NE_EXPR even in OpenMP loops, transform them into LT_EXPR or
3019 GT_EXPR loops depending on incr sign. Formatting fixes.
3020 * omp-general.h (struct omp_for_data): Add have_reductemp member.
3021 (enum omp_requires): New enum.
3022 (omp_requires_mask): Declare.
3023 * omp-grid.c (grid_eliminate_combined_simd_part): Formatting fix.
3025 * omp-low.c (struct omp_context): Add task_reductions and
3026 task_reduction_map fields.
3027 (is_host_teams_ctx): New function.
3028 (is_taskreg_ctx): Return true also if is_host_teams_ctx.
3029 (use_pointer_for_field): Use is_global_var instead of
3030 TREE_STATIC || DECL_EXTERNAL, and apply only if not privatized
3032 (build_outer_var_ref): Ignore taskgroup outer contexts.
3033 (delete_omp_context): Release task_reductions and task_reduction_map.
3034 (scan_sharing_clauses): Don't add any fields for reduction clause on
3035 taskloop. Handle OMP_CLAUSE__REDUCTEMP_. Handle
3036 OMP_CLAUSE_{IN,TASK}_REDUCTION and OMP_CLAUSE_REDUCTION with task
3037 modifier. Don't ignore shared clauses in is_host_teams_ctx contexts.
3038 Handle OMP_CLAUSE_NONTEMPORAL.
3039 (add_taskreg_looptemp_clauses): Add OMP_CLAUSE__REDUCTEMP_ clause if
3041 (scan_omp_parallel): Add _reductemp_ clause if there are any reduction
3042 clauses with task modifier.
3043 (scan_omp_task): Handle taskwait with depend clauses.
3044 (finish_taskreg_scan): Move field corresponding to _reductemp_ clause
3045 first. Move also OMP_CLAUSE__REDUCTEMP_ clause in front if present.
3046 Handle GIMPLE_OMP_TEAMS like GIMPLE_OMP_PARALLEL.
3047 (scan_omp_for): Fix comment formatting.
3048 (scan_omp_teams): Handle host teams constructs.
3049 (check_omp_nesting_restrictions): Allow teams with no outer
3050 OpenMP context. Adjust diagnostics for teams strictly nested into
3051 some explicit OpenMP construct other than target. Allow OpenMP atomics
3052 inside of simd regions.
3053 (scan_omp_1_stmt): Call scan_sharing_clauses for taskgroups.
3054 (scan_omp_1_stmt) <case GIMPLE_OMP_TEAMS>: Temporarily bump
3055 taskreg_nesting_level while scanning host teams construct.
3056 (task_reduction_read): New function.
3057 (lower_rec_input_clauses): Handle OMP_CLAUSE_REDUCTION on taskloop
3058 construct. Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE__REDUCTEMP_
3059 clauses. Handle OMP_CLAUSE_REDUCTION with task modifier. Remove
3060 second argument create_tmp_var if it is NULL. Don't ignore shared
3061 clauses in is_host_teams_ctx contexts. Handle
3062 OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE on OMP_CLAUSE_FIRSTPRIVATE
3064 (lower_reduction_clauses): Ignore reduction clauses with task
3065 modifier. Remove second argument create_tmp_var if it is NULL.
3066 Initialize OMP_ATOMIC_MEMORY_ORDER to relaxed.
3067 (lower_send_clauses): Ignore reduction clauses with task modifier.
3068 Handle OMP_CLAUSE__REDUCTEMP_. Don't send anything for
3069 OMP_CLAUSE_REDUCTION on taskloop. Handle OMP_CLAUSE_IN_REDUCTION.
3070 (maybe_add_implicit_barrier_cancel): Add OMP_RETURN argument, don't
3071 rely that it is the last stmt in body so far. Ignore outer taskgroup
3073 (omp_task_reductions_find_first, omp_task_reduction_iterate,
3074 lower_omp_task_reductions): New functions.
3075 (lower_omp_sections): Handle reduction clauses with taskgroup
3076 modifiers. Adjust maybe_add_implicit_barrier_cancel caller.
3077 (lower_omp_single): Adjust maybe_add_implicit_barrier_cancel caller.
3078 (lower_omp_for): Likewise. Handle reduction clauses with taskgroup
3080 (lower_omp_taskgroup): Handle taskgroup reductions.
3081 (create_task_copyfn): Copy over OMP_CLAUSE__REDUCTEMP_ pointer.
3082 Handle OMP_CLAUSE_IN_REDUCTION and OMP_CLAUSE_REDUCTION clauses.
3083 (lower_depend_clauses): If there are any
3084 OMP_CLAUSE_DEPEND_DEPOBJ or OMP_CLAUSE_DEPEND_MUTEXINOUTSET
3085 depend clauses, use a new array format. If OMP_CLAUSE_DEPEND_LAST is
3086 seen, assume lowering is done already and return early. Set kind
3087 on artificial depend clause to OMP_CLAUSE_DEPEND_LAST.
3088 (lower_omp_taskreg): Handle reduction clauses with task modifier on
3089 parallel construct. Handle reduction clause on taskloop construct.
3090 Handle taskwait with depend clauses.
3091 (lower_omp_1): Use lower_omp_taskreg instead of lower_omp_teams
3092 for host teams constructs.
3093 * tree.c (omp_clause_num_ops): Add in_reduction, task_reduction,
3094 nontemporal and _reductemp_ clause entries.
3095 (omp_clause_code_name): Likewise.
3096 (walk_tree_1): Handle OMP_CLAUSE_{IN,TASK}_REDUCTION,
3097 OMP_CLAUSE_NONTEMPORAL and OMP_CLAUSE__REDUCTEMP_.
3098 * tree-core.h (enum omp_clause_code): Add
3099 OMP_CLAUSE_{{IN,TASK}_REDUCTION,NONTEMPORAL,_REDUCTEMP_}.
3100 (enum omp_clause_defaultmap_kind, enum omp_memory_order): New.
3101 (struct tree_base): Add omp_atomic_memory_order field into union.
3102 Remove OMP_ATOMIC_SEQ_CST comment.
3103 (enum omp_clause_depend_kind): Add OMP_CLAUSE_DEPEND_MUTEXINOUTSET
3104 and OMP_CLAUSE_DEPEND_DEPOBJ.
3105 (struct tree_omp_clause): Add subcode.defaultmap_kind.
3106 * tree.def (OMP_TASKGROUP): Add another operand, move next to other
3107 OpenMP constructs with body and clauses operands.
3108 * tree.h (OMP_BODY): Use OMP_MASTER instead of OMP_TASKGROUP.
3109 (OMP_CLAUSES): Use OMP_TASKGROUP instead of OMP_SINGLE.
3110 (OMP_TASKGROUP_CLAUSES): Define.
3111 (OMP_CLAUSE_DECL): Use OMP_CLAUSE__REDUCTEMP_ instead of
3112 OMP_CLAUSE__LOOPTEMP_.
3113 (OMP_ATOMIC_SEQ_CST): Remove.
3114 (OMP_ATOMIC_MEMORY_ORDER, OMP_CLAUSE_FIRSTPRIVATE_NO_REFERENCE,
3115 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL): Define.
3116 (OMP_CLAUSE_REDUCTION_CODE, OMP_CLAUSE_REDUCTION_INIT,
3117 OMP_CLAUSE_REDUCTION_MERGE, OMP_CLAUSE_REDUCTION_PLACEHOLDER,
3118 OMP_CLAUSE_REDUCTION_DECL_PLACEHOLDER,
3119 OMP_CLAUSE_REDUCTION_OMP_ORIG_REF): Handle
3120 OMP_CLAUSE_{,IN_,TASK_}REDUCTION.
3121 (OMP_CLAUSE_REDUCTION_TASK, OMP_CLAUSE_REDUCTION_INSCAN,
3122 OMP_CLAUSE_DEFAULTMAP_KIND, OMP_CLAUSE_DEFAULTMAP_CATEGORY,
3123 OMP_CLAUSE_DEFAULTMAP_BEHAVIOR, OMP_CLAUSE_DEFAULTMAP_SET_KIND):
3125 * tree-inline.c (remap_gimple_stmt): Remap taskgroup clauses.
3126 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
3127 OMP_CLAUSE__REDUCTEMP_, OMP_CLAUSE_NONTEMPORAL.
3128 (convert_local_omp_clauses): Likewise. Remove useless test.
3129 * tree-parloops.c (create_call_for_reduction_1): Pass
3130 OMP_MEMORY_ORDER_RELAXED as new argument to
3131 dump_gimple_omp_atomic_load and dump_gimple_omp_atomic_store.
3132 * tree-pretty-print.c (dump_omp_iterators): New function.
3133 (dump_omp_clause): Handle OMP_CLAUSE__REDUCTEMP_,
3134 OMP_CLAUSE_NONTEMPORAL, OMP_CLAUSE_{TASK,IN}_REDUCTION. Print
3135 reduction modifiers. Handle OMP_CLAUSE_DEPEND_DEPOBJ and
3136 OMP_CLAUSE_DEPEND_MUTEXINOUTSET. Print iterators in depend clauses.
3137 Print __internal__ for OMP_CLAUSE_DEPEND_LAST. Handle cancel and
3138 simd OMP_CLAUSE_IF_MODIFIERs. Handle new kinds of
3139 OMP_CLAUSE_DEFAULTMAP. Print conditional: for
3140 OMP_CLAUSE_LASTPRIVATE_CONDITIONAL.
3141 (dump_omp_atomic_memory_order): New function.
3142 (dump_generic_node): Use it. Print taskgroup clauses. Print
3143 taskwait with depend clauses.
3144 * tree-pretty-print.h (dump_omp_atomic_memory_order): Declare.
3145 * tree-streamer-in.c (unpack_ts_omp_clause_value_fields):
3146 Handle OMP_CLAUSE_{TASK,IN}_REDUCTION.
3147 * tree-streamer-out.c (pack_ts_omp_clause_value_fields,
3148 write_ts_omp_clause_tree_pointers): Likewise.
3150 2018-11-08 David Malcolm <dmalcolm@redhat.com>
3153 * doc/invoke.texi (-fdump-ipa-): Document the "-optimized",
3154 "-missed", "-note", and "-all" sub-options.
3155 * ipa-inline.c (caller_growth_limits): Port from fprintf to dump
3157 (can_early_inline_edge_p): Likewise.
3158 (want_early_inline_function_p): Likewise.
3159 (want_inline_self_recursive_call_p): Likewise.
3160 (recursive_inlining): Likewise.
3161 (inline_small_functions): Likewise.
3162 (flatten_function): Likewise.
3163 (ipa_inline): Likewise.
3164 (inline_always_inline_functions): Likewise.
3165 (early_inline_small_functions): Likewise.
3166 (early_inliner): Likewise.
3167 * tree-inline.c (expand_call_inline): Likewise.
3169 2018-11-08 David Malcolm <dmalcolm@redhat.com>
3171 * pretty-print.c (pp_format): Handle %f.
3172 (selftest::test_pp_format): Add test of %f.
3173 * pretty-print.h (pp_double): New macro.
3175 2018-11-08 David Malcolm <dmalcolm@redhat.com>
3177 * dump-context.h (ASSERT_IS_CGRAPH_NODE): New macro.
3178 * dumpfile.c (make_item_for_dump_cgraph_node): Move to before...
3179 (dump_pretty_printer::decode_format): Implement "%C" for
3181 (selftest::test_capture_of_dump_calls): Rename "where" to
3182 "stmt_loc". Convert test_decl to a function decl and set its
3183 location. Add a symbol_table_test RAII instance and a
3184 cgraph_node, using it to test "%C" and dump_symtab_node.
3186 2018-11-08 Eric Botcazou <ebotcazou@adacore.com>
3189 * cgraphclones.c (duplicate_thunk_for_node): Also set DECL_IGNORED_P.
3191 2018-11-08 David Malcolm <dmalcolm@redhat.com>
3193 * cgraph.c: Include "selftest.h".
3194 (saved_symtab): New variable.
3195 (selftest::symbol_table_test::symbol_table_test): New ctor.
3196 (selftest::symbol_table_test::~symbol_table_test): New dtor.
3197 (selftest::test_symbol_table_test): New test.
3198 (selftest::cgraph_c_tests): New.
3199 * cgraph.h (saved_symtab): New decl.
3200 (selftest::symbol_table_test): New class.
3201 * selftest-run-tests.c (selftest::run_tests): Call
3202 selftest::cgraph_c_tests.
3203 * selftest.h (selftest::cgraph_c_tests): New decl.
3205 2018-11-08 Richard Biener <rguenther@suse.de>
3207 * tree-data-ref.h (lambda_int): New typedef.
3208 (lambda_vector_gcd): Adjust.
3209 (lambda_vector_new): Likewise.
3210 (lambda_matrix_new): Likewise.
3211 * tree-data-ref.c (print_lambda_vector): Adjust.
3213 2018-11-08 Richard Biener <rguenther@suse.de>
3215 PR tree-optimization/87929
3216 * tree-complex.c (expand_complex_comparison): Clean EH.
3218 2018-11-08 Martin Liska <mliska@suse.cz>
3220 * doc/extend.texi: Reword.
3221 * predict.c (expr_expected_value_1): Likewise.
3223 2018-11-08 Richard Biener <rguenther@suse.de>
3225 PR tree-optimization/87913
3226 * tree-ssa-phiopt.c (minmax_replacement): Turn EQ/NE compares
3227 of extreme values to ordered comparisons.
3229 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
3232 * doc/invoke.texi (Code Gen Options): Clarify -fno-common behavior.
3234 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
3236 * doc/invoke.texi: Remove leading dash from @opindex entries
3237 throughout the file.
3239 2018-11-07 Sandra Loosemore <sandra@codesourcery.com>
3242 * doc/invoke.texi (Option Summary): Add -e and --entry.
3243 (Link Options): Likewise.
3245 2018-11-07 Nathan Sidwell <nathan@acm.org>
3248 * Makefile.in (bitmap.o-warn): Add -Wno-error to unbreak
3249 --disable-checking bootstrap.
3251 2018-11-07 Hafiz Abid Qadeer <abidh@codesourcery.com>
3253 * configure: Regenerated.
3255 2018-11-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3258 * stor-layout.c (compute_record_mode): Set TYPE_MODE of UNION_TYPE
3259 to the mode of the widest field iff the widest field has mode class
3260 MODE_INT, or MODE_PARTIAL_INT and the union would be passed by
3263 2018-11-07 Nikolai Merinov <n.merinov@inango-systems.com>
3265 * common.opt: Add -Wattribute-warning.
3266 * doc/invoke.texi: Add documentation for -Wno-attribute-warning.
3267 * expr.c (expand_expr_real_1): Add new attribute to warning_at
3268 call to allow user configure behavior of "warning" attribute.
3270 2018-11-07 Segher Boessenkool <segher@kernel.crashing.org>
3272 * target.def: Put @: after every vs., e.g., and i.e. where it is
3273 followed by whitespace.
3274 * doc/extend.texi: Ditto.
3275 * doc/fragments.texi: Ditto.
3276 * doc/gimple.texi: Ditto.
3277 * doc/implement-c.texi: Ditto.
3278 * doc/install.texi: Ditto.
3279 * doc/invoke.texi: Ditto.
3280 * doc/md.texi: Ditto.
3281 * doc/plugins.texi: Ditto.
3282 * doc/rtl.texi: Ditto.
3283 * doc/sourcebuild.texi: Ditto.
3284 * doc/tm.texi.in: Ditto.
3285 * doc/ux.texi: Ditto.
3286 * doc/tm.texi: Regenerate.
3288 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3290 * config/arm/arm-cpus.in (ares): New entry.
3291 * config/arm/arm-tables.opt: Regenerate.
3292 * config/arm/arm-tune.md: Likewise.
3293 * doc/invoke.texi (ARM Options): Document ares.
3295 2018-11-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3297 * config/aarch64/aarch64-cores.def (ares): Define.
3298 * config/aarch64/aarch64-tune.md: Regenerate.
3299 * doc/invoke.texi (AArch64 Options): Document ares value for mtune.
3301 2018-11-07 Jan Hubicka <jh@suse.cz>
3303 * tree.c (fld_incomplete_type_of): Clear TREE_ADDRESSABLE.
3304 (free_lang_data_in_decl): Set TREE_ADDRESSABLE for public vars and
3305 functions; clear TYPE_DECL_SUPPRESS_DEBUG and DECL_MODE for
3308 2018-11-07 Richard Biener <rguenther@suse.de>
3310 PR tree-optimization/87914
3311 * tree-vect-loop.c (vect_is_simple_reduction): Improve detection
3313 (vectorizable_reduction): Handle shifts and rotates by dispatching
3314 to vectorizable_shift.
3315 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Handle
3316 in-loop uses of vect_nested_cycle defs. Merge cycle and internal
3318 (vectorizable_shift): Export and handle being called as
3320 (vect_analyze_stmt): Call vectorizable_shift after
3321 vectorizable_reduction.
3322 * tree-vectorizer.h (vectorizable_shift): Declare.
3324 2018-11-07 Jan Hubicka <jh@suse.cz>
3326 * ipa-devirt.c (odr_types_equivalent_p): Expect constants
3327 than const decls in TREE_VALUE of enum.
3328 (dump_type_inheritance_graph): Improve duplicate dumping.
3329 (free_enum_values): New.
3330 (build_type_inheritance_graph): Use it.
3331 * tree.c (free_lang_data_in_type): Free TYPE_VALUES of enums
3332 which are not main variants or not ODR types.
3333 (verify_type_variant): Expect variants to have no TYPE_VALUES.
3335 2018-11-07 Richard Biener <rguenther@suse.de>
3337 * ipa-inline.c (want_inline_small_function_p): Compute
3338 big_speedup_p lazily and last.
3340 2018-11-07 Jan Hubicka <jh@suse.cz>
3342 * tree.c (fld_type_variant_equal_p): Skip TYPE_ALIGN check when
3343 building incomplete variant of complete type.
3344 (fld_type_variant): Do not copy TYPE_ALIGN when building incomplete
3345 variant of complete type.
3347 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3349 * config/mips/mips.c: Fix typo in documentation of
3350 mips_loongson_ext2_prefetch_cookie.
3351 (mips_option_override): fix brain twister logical.
3352 * config/mips/mips.h: Fix typo in documentation of
3353 ISA_HAS_CTZ_CTO and define pattern.
3354 * config/mips/mips.md (prefetch): Hoist EXT2 above
3356 (prefetch_indexed): Hoist EXT2 above the EXT block.
3358 2018-11-07 Jan Hubicka <jh@suse.cz>
3360 * tree.c (free_lang_data_in_type): Add fld parameter; simplify
3361 return and parameter types of function and method types.
3362 (free_lang_data_in_cgraph): Update.
3364 2018-11-07 Martin Liska <mliska@suse.cz>
3366 PR rtl-optimization/87868
3367 * postreload-gcse.c (eliminate_partially_redundant_load): Set
3368 threshold to max_count if we would overflow.
3369 * profile-count.h: Make max_count a public constant.
3371 2018-11-07 Martin Liska <mliska@suse.cz>
3373 * mem-stats.h: Fix GNU coding style.
3375 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3377 * config/mips/gs264e.md: New.
3378 * config/mips/mips-cpus.def: Define gs264e.
3379 * config/mips/mips-tables.opt: Regenerate.
3380 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
3382 (mips_issue_rate): Add support for gs264e.
3383 (mips_multipass_dfa_lookahead): Likewise.
3384 * config/mips/mips.h: Define TARGET_GS264E and TUNE_GS264E.
3385 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs264e.
3386 (MIPS_ASE_MSA_SPEC): New.
3387 (BASE_DRIVER_SELF_SPECS): march=gs264e implies -mmsa.
3388 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS264E.
3389 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS264E.
3390 * config/mips/mips.md: Include gs264e.md.
3391 (processor): Add gs264e.
3392 * config/mips/mips.opt (MSA): Use Mask instead of Var.
3393 * doc/invoke.texi: Add gs264e to supported architectures.
3395 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3397 * config/mips/gs464e.md: New.
3398 * config/mips/mips-cpus.def: Define gs464e.
3399 * config/mips/mips-tables.opt: Regenerate.
3400 * config/mips/mips.c (mips_rtx_cost_data): Add DEFAULT_COSTS for
3402 (mips_issue_rate): Add support for gs464e.
3403 (mips_multipass_dfa_lookahead): Likewise.
3404 (mips_option_override): Enable MMI, EXT and EXT2 for gs464e.
3405 * config/mips/mips.h: Define TARGET_GS464E and TUNE_GS464E.
3406 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464e.
3407 (ISA_HAS_FUSED_MADD4): Enable for TARGET_GS464E.
3408 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_GS464E.
3409 * config/mips/mips.md: Include gs464e.md.
3410 (processor): Add gs464e.
3411 * doc/invoke.texi: Add gs464e to supported architectures.
3413 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3415 * config/mips/loongson3a.md: Rename to ...
3416 * config/mips/gs464.md: ... here.
3417 * config/mips/mips-cpus.def: Define gs464; Add loongson3a
3418 as an alias of gs464 processor.
3419 * config/mips/mips-tables.opt: Regenerate.
3420 * config/mips/mips.c (mips_issue_rate): Use PROCESSOR_GS464
3421 instead of PROCESSOR_LOONGSON_3A.
3422 (mips_multipass_dfa_lookahead): Use TUNE_GS464 instead of
3424 (mips_option_override): Enable MMI and EXT for gs464.
3425 * config/mips/mips.h: Rename TARGET_LOONGSON_3A to TARGET_GS464;
3426 Rename TUNE_LOONGSON_3A to TUNE_GS464.
3427 (MIPS_ISA_LEVEL_SPEC): Infer mips64r2 from gs464.
3428 (ISA_HAS_ODD_SPREG, ISA_AVOID_DIV_HILO, ISA_HAS_FUSED_MADD4,
3429 ISA_HAS_UNFUSED_MADD4): Use TARGET_GS464 instead of
3431 * config/mips/mips.md: Include gs464.md instead of loongson3a.md.
3432 (processor): Add gs464;
3433 * doc/invoke.texi: Add gs464 to supported architectures.
3435 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3437 * config/mips/mips-protos.h
3438 (mips_loongson_ext2_prefetch_cookie): New prototype.
3439 * config/mips/mips.c (mips_loongson_ext2_prefetch_cookie): New.
3440 (mips_option_override): Enable TARGET_LOONGSON_EXT when
3441 TARGET_LOONGSON_EXT2 is true.
3442 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Define
3443 __mips_loongson_ext2, __mips_loongson_ext_rev=2.
3444 (ISA_HAS_CTZ_CTO): New, true if TARGET_LOONGSON_EXT2.
3445 (ISA_HAS_PREFETCH): Include TARGET_LOONGSON_EXT and
3446 TARGET_LOONGSON_EXT2.
3447 (ASM_SPEC): Add mloongson-ext2 and mno-loongson-ext2.
3448 (define_insn "ctz<mode>2"): New insn pattern.
3449 (define_insn "prefetch"): Include TARGET_LOONGSON_EXT2.
3450 (define_insn "prefetch_indexed_<mode>"): Include
3451 TARGET_LOONGSON_EXT and TARGET_LOONGSON_EXT2.
3452 * config/mips/mips.opt (-mloongson-ext2): Add option.
3453 * gcc/doc/invoke.texi (-mloongson-ext2): Document.
3455 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3457 * config/mips/mips.h (TARGET_CPU_CPP_BUILTINS): Add
3458 __mips_loongson_ext.
3459 (MIPS_ASE_LOONGSON_EXT_SPEC): New.
3460 (BASE_DRIVER_SELF_SPECS): march=loongson3a implies
3462 (ASM_SPEC): Add mloongson-ext and mno-loongson-ext.
3463 * config/mips/mips.md (mul<mode>3, mul<mode>3_mul3_nohilo,
3464 <u>div<mode>3, <u>mod<mode>3, prefetch): Use TARGET_LOONGSON_EXT
3465 instead of TARGET_LOONGSON_3A.
3466 * config/mips/mips.opt (-mloongson-ext): Add option.
3467 * gcc/doc/invoke.texi (-mloongson-ext): Document.
3469 2018-11-07 Chenghua Xu <paul.hua.gm@gmail.com>
3471 * config.gcc (extra_headers): Add loongson-mmiintrin.h.
3472 * config/mips/loongson.md: Move to ...
3473 * config/mips/loongson-mmi.md: here; Adjustment.
3474 * config/mips/loongson.h: Move to ...
3475 State as deprecated. Include loongson-mmiintrin.h for back
3476 compatibility and warning.
3477 * config/mips/loongson-mmiintrin.h: ... here.
3478 * config/mips/mips.c (mips_hard_regno_mode_ok_uncached,
3479 mips_vector_mode_supported_p, AVAIL_NON_MIPS16): Use
3480 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
3481 (mips_option_override): Make sure MMI use hard float;
3482 (mips_shift_truncation_mask, mips_expand_vpc_loongson_even_odd,
3483 mips_expand_vpc_loongson_pshufh, mips_expand_vpc_loongson_bcast,
3484 mips_expand_vector_init): Use TARGET_LOONGSON_MMI instead of
3485 TARGET_LOONGSON_VECTORS.
3486 * gcc/config/mips/mips.h (TARGET_LOONGSON_VECTORS): Delete.
3487 (TARGET_CPU_CPP_BUILTINS): Add __mips_loongson_mmi.
3488 (MIPS_ASE_DSP_SPEC, MIPS_ASE_LOONGSON_MMI_SPEC): New.
3489 (BASE_DRIVER_SELF_SPECS): march=loongson2e/2f/3a implies
3491 (SHIFT_COUNT_TRUNCATED): Use TARGET_LOONGSON_MMI instead of
3492 TARGET_LOONGSON_VECTORS.
3493 * gcc/config/mips/mips.md (MOVE64, MOVE128): Use
3494 TARGET_LOONGSON_MMI instead of TARGET_LOONGSON_VECTORS.
3495 (Loongson MMI patterns): Include loongson-mmi.md instead of
3497 * gcc/config/mips/mips.opt (-mloongson-mmi): New option.
3498 * gcc/doc/invoke.texi (-mloongson-mmi): Document.
3500 2018-11-07 Richard Biener <rguenther@suse.de>
3503 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Fixup
3504 BLOCK_ABSTRACT_ORIGIN to be the ultimate origin.
3506 2018-11-07 Alexandre Oliva <aoliva@redhat.com>
3508 PR rtl-optimization/87874
3509 * lra.c (lra_substitute_pseudo): Do not create a subreg for
3512 2018-11-06 Aaron Sawdey <acsawdey@linux.ibm.com>
3514 * config/rs6000/rs6000.md (bswap<mode>2): Force address into register
3515 if not in indexed or indirect form.
3516 (bswap<mode>2_load): Change predicate to indexed_or_indirect_operand.
3517 (bswap<mode>2_store): Ditto.
3519 2018-11-06 Richard Earnshaw <rearnsha@arm.com>
3521 * config/aarch64/aarch64.md (speculation_tracker): Set the mode for
3524 2018-11-06 Richard Biener <rguenther@suse.de>
3526 PR tree-optimization/86850
3527 * vec.h (vec<T, va_heap, vl_ptr>::splice): Check src.length ()
3528 instead of src.m_vec.
3530 2018-11-06 Jan Hubicka <jh@suse.cz>
3532 * tree.c (fld_simplified_type_name): Break out form ...
3533 (free_lang_data_in_type): ... here.
3534 (fld_type_variant_equal_p): Use it.
3536 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3538 * config/default-d.c: Include memmodel.h.
3540 * config/sol2-d.c: New file.
3541 * config/t-sol2 (sol2-d.o): New rule.
3542 * config.gcc <*-*-solaris2*>: Set d_target_objs,
3543 target_has_targetdm.
3545 2018-11-06 Jan Hubicka <jh@suse.cz>
3547 * tree.c (fld_type_variant): Also copy alignment; be sure that
3548 new variant is equal.
3550 2018-11-06 Ilya Leoshkevich <iii@linux.ibm.com>
3553 * config/s390/s390.md: Add relative_long attribute.
3555 2018-11-06 Jan Hubicka <jh@suse.cz>
3557 * ipa-pure-const.c (check_decl): Do not test TYPE_NEEDS_CONSTRUCTING.
3558 * lto-streamer-out.c (hash_tree): Do not hash TYPE_NEEDS_CONSTRUCTING.
3559 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Do not
3560 stream TYPE_NEEDS_CONSTRUCTING.
3561 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
3562 * tree.c (free_lang_data_in_type): Clear TYPE_NEEDS_CONSTRUCTING.
3564 2018-11-06 Richard Biener <rguenther@suse.de>
3566 * tree-vect-slp.c (vect_slp_bb): Move opening of vect_slp_analyze_bb
3568 (vect_slp_analyze_bb_1): ... here to avoid hiding optimized locations.
3570 2018-11-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
3572 * gcc/config/msp430/msp430.h (REG_CLASS_CONTENTS): Add R0 to
3573 REG_CLASS_CONTENTS[GEN_REGS].
3574 (REGNO_REG_CLASS): Return NO_REGS for R2 and R3.
3576 2018-11-06 Jan Hubicka <jh@suse.cz>
3578 * tree.c (fld_simplified_type_of): Clear TYPELESS_STORAGE flag.
3580 2018-11-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3583 * config/sol2.h (ASAN_CC1_SPEC): Define.
3584 (LD_WHOLE_ARCHIVE_OPTION): Define.
3585 (LD_NO_WHOLE_ARCHIVE_OPTION): Define.
3586 (ASAN_REJECT_SPEC): Provide default.
3587 (LIBASAN_EARLY_SPEC): Define.
3588 (LIBTSAN_EARLY_SPEC): Define.
3589 (LIBLSAN_EARLY_SPEC): Define.
3590 * config/i386/sol2.h (CC1_SPEC): Redefine.
3591 (ASAN_REJECT_SPEC): Define.
3593 * config/sparc/sparc.c (sparc_asan_shadow_offset): Declare.
3594 (TARGET_ASAN_SHADOW_OFFSET): Define.
3595 (sparc_asan_shadow_offset): New function.
3596 * config/sparc/sol2.h (CC1_SPEC): Append ASAN_CC1_SPEC.
3597 (ASAN_REJECT_SPEC): Define.
3599 2018-11-06 Jan Hubicka <jh@suse.cz>
3601 * tree.c (fld_type_variant): Copy canonical type.
3602 (fld_incomplete_type_of): Check that canonical types looks sane;
3603 copy canonical type.
3604 (verify_type): Accept when incomplete type has complete canonical type.
3606 2018-11-06 Jan Hubicka <jh@suse.cz>
3608 * tree.c (free_lang_data): Reset overwite_assembler_name,
3609 print_xnode, print_decl, print_type and print_identifier of
3612 2018-11-06 Richard Biener <rguenther@suse.de>
3614 PR tree-optimization/87889
3615 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
3616 Do nothing if old and new arg are the same
3618 2018-11-06 Andreas Krebbel <krebbel@linux.ibm.com>
3621 * config/s390/s390.md ("*r<noxa>sbg_di_rotl"): Remove mode
3622 attributes for operands 3 and 4.
3624 2018-11-06 Richard Biener <rguenther@suse.de>
3627 * simplify-rtx.c (simplify_binary_operation_1): Add pattern
3628 matching bitfield insertion.
3630 2018-11-06 Alexandre Oliva <aoliva@redhat.com>
3632 * auto-inc-dec.c: Include valtrack.h. Improve comments.
3633 (reg_next_debug_use): New.
3634 (attempt_change): Propagate adjusted expression into affected
3636 (merge_in_block): Track uses in debug insns.
3637 (pass_inc_dec::execute): Allocate and release
3640 2018-11-06 Wei Xiao <wei3.xiao@intel.com>
3642 * config/i386/avx512fintrin.h: Update VFIXUPIMM* intrinsics.
3643 (_mm512_fixupimm_round_pd): Update parameters and builtin.
3644 (_mm512_maskz_fixupimm_round_pd): Ditto.
3645 (_mm512_fixupimm_round_ps): Ditto.
3646 (_mm512_maskz_fixupimm_round_ps): Ditto.
3647 (_mm_fixupimm_round_sd): Ditto.
3648 (_mm_maskz_fixupimm_round_sd): Ditto.
3649 (_mm_fixupimm_round_ss): Ditto.
3650 (_mm_maskz_fixupimm_round_ss): Ditto.
3651 (_mm512_fixupimm_pd): Ditto.
3652 (_mm512_maskz_fixupimm_pd): Ditto.
3653 (_mm512_fixupimm_ps): Ditto.
3654 (_mm512_maskz_fixupimm_ps): Ditto.
3655 (_mm_fixupimm_sd): Ditto.
3656 (_mm_maskz_fixupimm_sd): Ditto.
3657 (_mm_fixupimm_ss): Ditto.
3658 (_mm_maskz_fixupimm_ss): Ditto.
3659 (_mm512_mask_fixupimm_round_pd): Update builtin.
3660 (_mm512_mask_fixupimm_round_ps): Ditto.
3661 (_mm_mask_fixupimm_round_sd): Ditto.
3662 (_mm_mask_fixupimm_round_ss): Ditto.
3663 (_mm512_mask_fixupimm_pd): Ditto.
3664 (_mm512_mask_fixupimm_ps): Ditto.
3665 (_mm_mask_fixupimm_sd): Ditto.
3666 (_mm_mask_fixupimm_ss): Ditto.
3667 * config/i386/avx512vlintrin.h:
3668 (_mm256_fixupimm_pd): Update parameters and builtin.
3669 (_mm256_maskz_fixupimm_pd): Ditto.
3670 (_mm256_fixupimm_ps): Ditto.
3671 (_mm256_maskz_fixupimm_ps): Ditto.
3672 (_mm_fixupimm_pd): Ditto.
3673 (_mm_maskz_fixupimm_pd): Ditto.
3674 (_mm_fixupimm_ps): Ditto.
3675 (_mm_maskz_fixupimm_ps): Ditto.
3676 (_mm256_mask_fixupimm_pd): Update builtin.
3677 (_mm256_mask_fixupimm_ps): Ditto.
3678 (_mm_mask_fixupimm_pd): Ditto.
3679 (_mm_mask_fixupimm_ps): Ditto.
3680 * config/i386/i386-builtin-types.def: Add new types and remove useless ones.
3681 * config/i386/i386-builtin.def: Update builtin definitions.
3682 * config/i386/i386.c: Handle new builtin types and remove useless ones.
3683 * config/i386/sse.md: Update VFIXUPIMM* patterns.
3684 (<avx512>_fixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
3685 (<avx512>_fixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
3686 (<avx512>_fixupimm<mode>_mask<round_saeonly_name>): Update.
3687 (avx512f_sfixupimm<mode>_maskz<round_saeonly_expand_name>): Update.
3688 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>): Update.
3689 (avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Update.
3690 * config/i386/subst.md:
3691 (round_saeonly_sd_mask_operand4): Add new subst_attr.
3692 (round_saeonly_sd_mask_op4): Ditto.
3693 (round_saeonly_expand_operand5): Ditto.
3694 (round_saeonly_expand): Update.
3696 2018-11-05 Max Filippov <jcmvbkbc@gmail.com>
3698 * config/xtensa/uclinux.h (XTENSA_ALWAYS_PIC): Change to 0.
3700 2018-11-05 Segher Boessenkool <segher@kernel.crashing.org>
3702 PR rtl-optimization/87871
3703 * combine.c (make_more_copies): Skip if dest is frame_pointer_rtx.
3705 2018-11-05 Paul Koning <ni1d@arrl.net>
3707 * doc/sourcebuild.texi (target attributes): Document new "inf"
3708 effective target keyword.
3710 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
3712 * config/s390/s390.c (s390_register_move_cost): Increase costs for
3713 moves involving the CC reg.
3715 2018-11-05 Richard Biener <rguenther@suse.de>
3717 PR tree-optimization/87873
3718 * tree-ssa-loop-manip.h (split_loop_exit_edge): Add copy_constants_p
3720 * tree-ssa-loop-manip.c (split_loop_exit_edge): Likewise.
3721 * tree-vect-loop.c (vect_transform_loop): When splitting the
3722 loop exit also create forwarder PHIs for constants.
3723 * tree-vect-loop-manip.c (slpeel_duplicate_current_defs_from_edges):
3724 Handle constant to_arg, add extra checking we match up the correct
3727 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
3729 * config/s390/s390.md: QImode and HImode for load on condition.
3731 2018-11-05 Robin Dapp <rdapp@linux.ibm.com>
3733 * config/s390/predicates.md: Fix typo.
3734 * config/s390/s390.md: Allow immediates for load on condition.
3736 2018-11-05 Martin Liska <mliska@suse.cz>
3738 * alloc-pool.h (struct pool_usage): Use SIZE_AMOUNT.
3739 * bitmap.h (struct bitmap_usage): Likewise.
3740 * ggc-common.c (SCALE): Remove.
3742 (struct ggc_usage): Use SIZE_AMOUNT. And update
3744 * ggc-page.c (SCALE): Remove.
3745 (STAT_LABEL): Remove.
3746 (ggc_print_statistics): Use SIZE_AMOUNT.
3747 * gimple.h (SCALE): Remove.
3749 * input.c (ONE_K): Remove.
3752 (STAT_LABEL): Likewise.
3753 (FORMAT_AMOUNT): Likewise.
3754 (dump_line_table_statistics): Use SIZE_AMOUNT.
3755 * mem-stats.h (struct mem_usage): Likewise.
3756 * rtl.c (dump_rtx_statistics): Likewise.
3757 (rtx_alloc_counts): Change type to size_t.
3758 (rtx_alloc_sizes): Likewise.
3759 (rtx_count_cmp): New.
3760 (dump_rtx_statistics): Sort first based on counts.
3761 * tree.c (tree_nodes_cmp): New.
3762 (tree_codes_cmp): New.
3763 (dump_tree_statistics): Sort first based on counts.
3764 * system.h (ONE_K): New.
3766 (SIZE_SCALE): Likewise.
3767 (SIZE_LABEL): Likewise.
3768 (SIZE_AMOUNT): Likewise.
3769 * tree-cfg.c (dump_cfg_stats): Use SIZE_AMOUNT.
3770 * tree-dfa.c (dump_dfa_stats): Likewise.
3771 * tree-phinodes.c (phinodes_print_statistics): Likewise.
3772 * tree-ssanames.c (ssanames_print_statistics): Likewise.
3773 * tree.c (dump_tree_statistics): Likewise.
3774 * vec.c (struct vec_usage): Likewise.
3775 * trans-mem.c (tm_mangle): Enlarge buffer in order to not
3776 trigger a -Werror=format-overflow with
3777 --enable-gather-detailed-stats.
3779 2018-11-05 Martin Liska <mliska@suse.cz>
3781 * mem-stats.h (mem_alloc_description::release_instance_overhead):
3783 * vec.c (struct vec_usage): Register m_element_size.
3784 (vec_prefix::register_overhead): New arguments: elements and
3786 (vec_prefix::release_overhead): Subtract elements.
3787 * vec.h (struct vec_prefix): Change signature.
3788 (va_heap::reserve): Pass proper arguments.
3789 (va_heap::release): Likewise.
3791 2018-11-05 Martin Liska <mliska@suse.cz>
3793 * mem-stats.h (mem_alloc_description::get_list): Fix GNU coding
3797 2018-11-05 Richard Biener <rguenther@suse.de>
3799 * tree-scalar-evolution.h (final_value_replacement_loop): Update
3801 * tree-scalar-evolution.c (final_value_replacement_loop): Return
3802 whether anything was done.
3803 (scev_const_prop): Remove constant propagation part, fold
3805 * tree-ssa-loop.c (pass_scev_cprop::execute): ... here.
3806 (pass_data_scev_cprop): TODO_cleanup_cfg is now done
3809 2018-11-05 Jakub Jelinek <jakub@redhat.com>
3811 PR tree-optimization/87859
3812 * gimple-ssa-store-merging.c (struct merged_store_group): Add
3813 only_constants and first_nonmergeable_order members.
3814 (merged_store_group::merged_store_group): Initialize them.
3815 (merged_store_group::do_merge): Clear only_constants member if
3816 adding something other than INTEGER_CST store.
3817 (imm_store_chain_info::coalesce_immediate_stores): Don't merge
3818 stores with order >= first_nonmergeable_order. Use
3819 merged_store->only_constants instead of always recomputing it.
3820 Set merged_store->first_nonmergeable_order if we've skipped any
3821 stores. Attempt to merge overlapping INTEGER_CST stores that
3822 we would otherwise skip.
3825 * match.pd (X + Y < X): Don't optimize if TYPE_OVERFLOW_SANITIZED.
3827 2018-11-05 Xuepeng Guo <xuepeng.guo@intel.com>
3830 * config/i386/emmintrin.h (__v16qs): New to cope with option
3832 (_mm_cmpeq_epi8): Replace __v16qi with __v16qs.
3833 (_mm_cmplt_epi8): Likewise.
3834 (_mm_cmpgt_epi8): Likewise.
3836 2018-11-05 Richard Biener <rguenther@suse.de>
3838 PR rtl-optimization/87852
3839 * fwprop.c (use_killed_between): Only consider single-defs of the
3840 use whose definition statement dominates the use.
3842 2018-11-05 Martin Liska <mliska@suse.cz>
3845 * doc/invoke.texi: Remove options that are
3846 not disabled with -Os.
3848 2018-11-05 Martin Liska <mliska@suse.cz>
3851 * doc/extend.texi: Update constrain about the last argument
3852 of __builtin_expect_with_probability.
3854 2018-11-05 Martin Liska <mliska@suse.cz>
3857 * predict.c (expr_expected_value_1): Verify
3858 that last argument is a real constants and emit
3861 2018-11-05 Martin Liska <mliska@suse.cz>
3863 PR gcov-profile/77698
3864 * ipa-profile.c (ipa_profile): Adjust hotness threshold
3867 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3869 PR tree-optimization/86572
3870 * builtins.c (c_strlen): Handle negative offsets in a safe way.
3872 2018-11-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
3874 PR tree-optimization/87672
3875 * gimple-fold.c (gimple_fold_builtin_stxcpy_chk): Gimplify.
3876 * tree-ssa-strlen.c (handle_builtin_strcat): Adjust object size.
3878 2018-11-04 Uros Bizjak <ubizjak@gmail.com>
3881 * cfgexpand.c (pass_expand::execute): Move the call to
3882 finish_eh_generation in front of the call to expand_stack_alignment.
3884 2018-11-04 Venkataramanan Kumar <venkataramanan.kumar@amd.com>
3886 * common/config/i386/i386-common.c (processor_alias_table): Add
3888 * config.gcc (i[34567]86-*-linux* | ...): Add znver2.
3889 (case ${target}): Add znver2.
3890 * config/i386/driver-i386.c: (host_detect_local_cpu): Let
3891 -march=native recognize znver2 processors.
3892 * config/i386/i386-c.c (ix86_target_macros_internal): Add znver2.
3893 * config/i386/i386.c (m_znver2): New definition.
3894 (m_ZNVER): New definition.
3895 (m_AMD_MULTIPLE): Includes m_znver2.
3896 (processor_cost_table): Add znver2 entry.
3897 (processor_target_table): Add znver2 entry.
3898 (get_builtin_code_for_version): Set priority for
3900 (processor_model): Add M_AMDFAM17H_ZNVER2.
3901 (arch_names_table): Ditto.
3902 (ix86_reassociation_width): Include znver2.
3903 * config/i386/i386.h (TARGET_znver2): New definition.
3904 (struct ix86_size_cost): Add TARGET_ZNVER2.
3905 (enum processor_type): Add PROCESSOR_ZNVER2.
3906 * config/i386/i386.md (define_attr "cpu"): Add znver2.
3907 * config/i386/x86-tune-costs.h: (processor_costs) Add znver2 costs.
3908 * config/i386/x86-tune-sched.c: (ix86_issue_rate): Add znver2.
3909 (ix86_adjust_cost): Add znver2.
3910 * config/i386/x86-tune.def: Replace m_ZNVER1 by m_ZNVER.
3911 * gcc/doc/extend.texi: Add details about znver2.
3912 * gcc/doc/invoke.texi: Add details about znver2.
3914 2018-11-03 Sandra Loosemore <sandra@codesourcery.com>
3918 * config/nios2/nios2.c (nios2_rtx_costs): Recognize <mul>sidi3
3921 2018-11-02 Aaron Sawdey <acsawdey@linux.ibm.com>
3923 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Pay
3924 attention to TARGET_AVOID_XFORM and BYTES_BIG_ENDIAN.
3926 2018-11-02 Richard Earnshaw <rearnsha@arm.com>
3928 * config/aarch64/aarch64.c ((aarch64_override_options): Disable
3929 shrink-wrapping when -mtrack-speculation.
3931 2018-11-02 Richard Biener <rguenther@suse.de>
3933 * tree-ssa-coalesce.c (struct coalesce_list): Add obstack member.
3934 (pop_cost_one_pair): Do not free pair.
3935 (pop_best_coalesce): Likewise.
3936 (create_coalesce_list): Initialize obstack.
3937 (delete_coalesce_list): Free obstack.
3938 (find_coalesce_pair): Obstack-allocate coalesce pairs.
3939 (add_cost_one_coalesce): Likewise.
3940 (struct live_track): Remove bitmap pointer indirections.
3941 (new_live_track): Adjust.
3942 (delete_live_track): Likewise.
3943 (live_track_remove_partition): Likewise.
3944 (live_track_add_partition): Likewise.
3945 (live_track_live_p): Likewise.
3946 (live_track_process_def): Likewise.
3947 (live_track_clear_base_vars): Likewise.
3949 2018-11-02 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
3951 * configure.ac (gcc_cv_as_sparc_register_op): Remove.
3952 * configure: Regenerate.
3953 * config.in: Regenerate.
3954 * config/sparc/sparc.c (sparc_output_scratch_registers): Remove
3955 HAVE_AS_REGISTER_PSEUDO_OP guard.
3956 * config/sparc/sparc.h (ASM_ARCH64_SPEC): Remove
3957 !HAVE_AS_REGISTER_PSEUDO_OP support.
3958 (ASM_DECLARE_REGISTER_GLOBAL): Remove HAVE_AS_REGISTER_PSEUDO_OP
3961 2018-11-02 Richard Biener <rguenther@suse.de>
3963 * dwarf2out.c (build_abbrev_table): Guard lookup_external_ref call
3966 2018-11-02 Richard Biener <rguenther@suse.de>
3968 PR tree-optimization/87776
3969 * tree-ssa-sccvn.c (do_rpo_vn): Do not mark backedges
3970 executable when iterating but running into rpo-vn-max-loop-depth
3971 and not eliding the iteration.
3973 2018-11-30 Jan Hubicka <jh@suse.cz>
3975 * tree.c (free_lang_data_in_decl): Free TREE_TYPE and alignment of
3978 2018-11-01 Aaron Sawdey <acsawdey@linux.ibm.com>
3980 * config/rs6000/rs6000-protos.h (rs6000_address_for_fpconvert): Remove
3982 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem):
3983 Combine with rs6000_address_for_fpconvert.
3984 (rs6000_address_for_fpconvert): Combine with
3985 rs6000_force_indexed_or_indirect_mem.
3986 (rs6000_expand_vector_init): Change function call from
3987 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
3988 * config/rs6000/rs6000.md (floatsi<mode>2_lfiwax): Change call from
3989 rs6000_address_for_fpconvert to rs6000_force_indexed_or_indirect_mem.
3990 (floatsi<mode>2_lfiwax_mem): Ditto.
3991 (floatunssi<mode>2_lfiwzx): Ditto.
3992 (floatunssi<mode>2_lfiwzx_mem): Ditto.
3993 (float<QHI:mode><FP_ISA3:mode>2): Ditto.
3994 (floatuns<QHI:mode><FP_ISA3:mode>2): Ditto.
3995 (fix_trunc<mode>si2_stfiwx): Ditto.
3996 (fixuns_trunc<mode>si2_stfiwx): Ditto.
3997 (float_<mode>si2_hw): Ditto.
3998 (floatuns_<mode>si2_hw): Ditto.
3999 * config/rs6000/vsx.md (*vsx_extract_si): Ditto.
4000 (vsx_splat_<mode>): Ditto.
4002 2018-11-01 Joseph Myers <joseph@codesourcery.com>
4004 * configure.ac (gcc_cv_as_ia64_ltoffx_ldxmov_relocs): Add newline
4005 at end of assembler input text.
4006 * configure: Regenerate.
4008 2018-11-01 Jakub Jelinek <jakub@redhat.com>
4010 PR tree-optimization/87826
4011 * gimple-ssa-store-merging.c (do_shift_rotate): Punt if count is
4012 negative or larger or equal to type's precision.
4014 2018-10-31 Alexandre Oliva <aoliva@redhat.com>
4016 * opts.c (default_options_table): Do not enable
4017 OPT_fdelayed_branch at -Og.
4018 * doc/invoke.texi (-fdelayed-branch): Document it.
4020 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
4022 * optabs-libfuncs.c (build_libfunc_function_visibility):
4023 New, split out from...
4024 (build_libfunc_function): ... here.
4025 (init_one_libfunc_visibility): New, split out from ...
4026 (init_one_libfunc): ... here.
4028 * config/aarch64/atomics.md (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse):
4029 scratch register need not be early-clobber. Document the reason
4030 why we cannot use ST<OP>.
4032 2018-10-31 Joseph Myers <joseph@codesourcery.com>
4035 * configure.ac: Remove AC_PREREQ. Use AC_LANG_SOURCE. Use single
4036 line for second argument of AC_DEFINE_UNQUOTED.
4037 * doc/install.texi (Tools/packages necessary for modifying GCC):
4038 Update to autoconf 2.69 and automake 1.15.1.
4039 * aclocal.m4, config.in, configure: Regenerate.
4041 2018-10-31 Pat Haugen <pthaugen@us.ibm.com>
4043 * tree-ssa-loop-ivcanon.c (try_peel_loop): Fix typo and remove dead
4046 2018-10-31 Martin Liska <mliska@suse.cz>
4049 * common/config/aarch64/aarch64-common.c (aarch64_parse_extension):
4050 Add new argument invalid_extension.
4051 (aarch64_get_all_extension_candidates): New function.
4052 (aarch64_rewrite_selected_cpu): Add NULL to function call.
4053 * config/aarch64/aarch64-protos.h (aarch64_parse_extension): Add
4055 (aarch64_get_all_extension_candidates): New function.
4056 * config/aarch64/aarch64.c (aarch64_parse_arch): Add new
4057 argument invalid_extension.
4058 (aarch64_parse_cpu): Likewise.
4059 (aarch64_print_hint_for_extensions): New function.
4060 (aarch64_validate_mcpu): Provide hint about invalid extension.
4061 (aarch64_validate_march): Likewise.
4062 (aarch64_handle_attr_arch): Pass new argument.
4063 (aarch64_handle_attr_cpu): Provide hint about invalid extension.
4064 (aarch64_handle_attr_isa_flags): Likewise.
4066 2018-10-31 Richard Biener <rguenther@suse.de>
4070 * tree-outof-ssa.c (insert_backedge_copies): Restrict
4071 copy generation to useful cases. Place the copy before
4072 the definition of the backedge value when possible.
4074 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4076 * config/arc/arc-arch.h (ARC_TUNE_ARC7XX): New tune value.
4077 * config/arc/arc.c (arc_active_insn): New function.
4078 (check_store_cacheline_hazard): Likewise.
4079 (workaround_arc_anomaly): Use check_store_cacheline_hazard.
4080 (arc_override_options): Disable delay slot scheduler for older
4082 (arc_store_addr_hazard_p): New implementation, old one renamed to
4084 (arc_store_addr_hazard_internal_p): Renamed.
4085 (arc_reorg): Don't combine into brcc instructions which are part
4086 of hardware hazard solution.
4087 * config/arc/arc.md (attr tune): Consider new arc7xx tune value.
4088 (tune_arc700): Likewise.
4089 * config/arc/arc.opt (arc7xx): New tune value.
4090 * config/arc/arc700.md: Improve A7 scheduler.
4092 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4094 * config/arc/arc.c (arc_override_options): Remove
4095 TARGET_COMPACT_CASESI.
4096 * config/arc/arc.h (ASM_OUTPUT_ADDR_DIFF_ELT): Update.
4097 (CASE_VECTOR_MODE): Likewise.
4098 (CASE_VECTOR_PC_RELATIVE): Likewise.
4099 (CASE_VECTOR_SHORTEN_MODE): Likewise.
4100 (CASE_VECTOR_SHORTEN_MODE1): Delete.
4101 (ADDR_VEC_ALIGN): Update.
4102 (ASM_OUTPUT_CASE_LABEL): Undefine.
4103 (ASM_OUTPUT_BEFORE_CASE_LABEL): Undefine.
4104 (TARGET_BI_BIH): Define.
4105 (DEFAULT_BRANCH_INDEX): Likewise.
4106 * config/arc/arc.md (casesi): Rework to accept BI/BIH
4107 instructions, remove compact_casesi use case.
4108 (casesi_compact_jump): Remove.
4109 (casesi_dispatch): New pattern.
4110 * config/arc/arc.opt: Add mbranch-index option. Deprecate
4111 compact_casesi option.
4112 * doc/invoke.texi: Document mbranch-index option.
4114 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4116 * config/arc/arc.c (arc_get_tp): Remove function.
4117 (arc_emit_call_tls_get_addr): Likewise.
4118 (arc_call_tls_get_addr): New function.
4119 (arc_legitimize_tls_address): Make use of arc_call_tls_get_addr.
4120 * config/arc/arc.md (tls_load_tp_soft): Remove.
4121 (tls_gd_get_addr): Likewise.
4123 2018-10-31 Claudiu Zissulescu <claziss@synopsys.com>
4125 * config/arc/arc.md (mulsi3): Remove call to mulsi_600_lib.
4126 (mulsi3_600_lib): Remove pattern.
4127 (umulsi3_highpart_600_lib_le): Likewise.
4128 (umulsi3_highpart): Remove call to umulsi3_highpart_600_lib_le.
4129 (umulsidi3): Remove call to umulsidi3_600_lib.
4130 (umulsidi3_600_lib): Remove pattern.
4131 (peephole2): Remove peephole using the above deprecated patterns.
4133 2018-10-31 Thomas Preud'homme <thomas.preudhomme@linaro.org>
4136 * config/arm/arm.c (arm_option_check_internal): Disable the combined
4137 use of -mslow-flash-data and -mword-relocations.
4138 (arm_option_override): Enable -mword-relocations if -fpic or -fPIC.
4139 * config/arm/arm.md (SYMBOL_REF MOVT splitter): Stop checking for
4141 * doc/invoke.texi (-mword-relocations): Mention conflict with
4143 (-mslow-flash-data): Reciprocally.
4145 2018-10-31 Richard Henderson <richard.henderson@linaro.org>
4147 * config/aarch64/aarch64.c (aarch64_hard_regno_mode_ok): Force
4148 16-byte modes held in GP registers to use an even regno.
4150 * config/aarch64/aarch64.c (aarch64_emit_bic): Remove.
4151 (aarch64_atomic_ldop_supported_p): Remove.
4152 (aarch64_gen_atomic_ldop): Remove.
4153 * config/aarch64/atomic.md (atomic_<atomic_optab><ALLI>):
4154 Fully expand LSE operations here.
4155 (atomic_fetch_<atomic_optab><ALLI>): Likewise.
4156 (atomic_<atomic_optab>_fetch<ALLI>): Likewise.
4157 (aarch64_atomic_<ATOMIC_LDOP><ALLI>_lse): Drop atomic_op iterator
4158 and use ATOMIC_LDOP instead; use register_operand for the input;
4159 drop the split and emit insns directly.
4160 (aarch64_atomic_fetch_<ATOMIC_LDOP><ALLI>_lse): Likewise.
4161 (aarch64_atomic_<atomic_op>_fetch<ALLI>_lse): Remove.
4162 (@aarch64_atomic_load<ATOMIC_LDOP><ALLI>): Remove.
4164 * config/aarch64/aarch64.c (aarch64_emit_atomic_swap): Remove.
4165 (aarch64_gen_atomic_ldop): Don't call it.
4166 * config/aarch64/atomics.md (atomic_exchange<ALLI>):
4167 Use aarch64_reg_or_zero.
4168 (aarch64_atomic_exchange<ALLI>): Likewise.
4169 (aarch64_atomic_exchange<ALLI>_lse): Remove split; remove & from
4170 operand 0; use aarch64_reg_or_zero for input; merge ...
4171 (@aarch64_atomic_swp<ALLI>): ... this and remove.
4173 * config/aarch64/aarch64.c (aarch64_gen_compare_reg_maybe_ze): New.
4174 (aarch64_split_compare_and_swap): Use it.
4175 (aarch64_expand_compare_and_swap): Likewise. Remove convert_modes;
4176 test oldval against the proper predicate.
4177 * config/aarch64/atomics.md (@atomic_compare_and_swap<ALLI>):
4178 Use nonmemory_operand for expected.
4179 (cas_short_expected_pred): New.
4180 (@aarch64_compare_and_swap<SHORT>): Use it; use "rn" not "rI" to match.
4181 (@aarch64_compare_and_swap<GPI>): Use "rn" not "rI" for expected.
4182 * config/aarch64/predicates.md (aarch64_plushi_immediate): New.
4183 (aarch64_plushi_operand): New.
4185 * config/aarch64/aarch64.c (aarch64_expand_compare_and_swap):
4186 Force oldval into the rval register for TARGET_LSE; emit the compare
4187 during initial expansion so that it may be deleted if unused.
4188 (aarch64_gen_atomic_cas): Remove.
4189 * config/aarch64/atomics.md (@aarch64_compare_and_swap<SHORT>_lse):
4190 Change =&r to +r for operand 0; use match_dup for operand 2;
4191 remove is_weak and mod_f operands as unused. Drop the split
4193 (@aarch64_atomic_cas<SHORT>): ... this pattern's output; remove.
4194 (@aarch64_compare_and_swap<GPI>_lse): Similarly.
4195 (@aarch64_atomic_cas<GPI>): Similarly.
4197 2018-10-31 Richard Biener <rguenther@suse.de>
4199 * tree-eh.c (replace_trapping_overflow): Simplify ABS_EXPR case
4202 2018-10-31 Sameera Deshpande <sameera.deshpande@linaro.org>
4204 * config/aarch64/aarch64-cores.def (saphira): Use saphira pipeline.
4205 * config/aarch64/aarch64.md: Include saphira.md
4206 * config/aarch64/saphira.md: New file for pipeline description.
4208 2018-10-30 Martin Sebor <msebor@redhat.com>
4211 * gimple-ssa-sprintf.c (format_directive): Use %G to include
4213 (sprintf_dom_walker::compute_format_length):
4214 Avoid setting POSUNDER4K here.
4215 (get_destination_size): Handle null argument values.
4216 (get_user_idx_format): New function.
4217 (sprintf_dom_walker::handle_gimple_call): Handle all printf-like
4218 functions, including user-defined with attribute format printf.
4219 Use %G to include inlining context.
4220 Set POSUNDER4K here.
4222 2018-10-30 Jan Hubicka <jh@suse.cz>
4224 * params.def (lto-partitions): Bump from 32 to 128.
4226 2018-10-30 Jan Hubicka <jh@suse.cz>
4229 (free_lang_data_d, add_tree_to_fld_list, fld_worklist_push): Move
4231 (free_lang_data_in_type): Forward declare.
4232 (fld_type_variant_equal_p): New function.
4233 (fld_type_variant): New function
4234 (fld_incomplete_types): New hash.
4235 (fld_incomplete_type_of): New function
4236 (fld_simplfied-type): New function.
4237 (free_lang_data_in_decl): Add fld parameter; simplify type of FIELD_DECL
4238 (free_lang_data): Allocate and free fld_incomplete_type; update call
4239 of free_lang_data_in_decl.
4241 2018-10-30 Eric Botcazou <ebotcazou@adacore.com>
4243 * gcov.c (output_lines): Remove duplicate line.
4245 2018-10-30 Aaron Sawdey <acsawdey@linux.ibm.com>
4247 * config/rs6000/rs6000.md (bswapdi2): Force address into register
4248 if not in indexed or indirect form.
4249 (bswapdi2_load): Change predicate to indexed_or_indirect_operand.
4250 (bswapdi2_store): Ditto.
4251 * config/rs6000/rs6000.c (rs6000_force_indexed_or_indirect_mem): New
4253 * config/rs6000/rs6000-protos.h (rs6000_force_indexed_or_indirect_mem):
4254 Prototype for helper function.
4256 2018-10-30 Martin Sebor <msebor@redhat.com>
4258 * doc/extend.texi (optimize): Clarify/expand attribute documentation.
4259 (target, pragma GCC optimize, pragma GCC target): Ditto.
4261 2018-10-30 Jonathan Wakely <jwakely@redhat.com>
4263 * doc/extend.texi: Fix prototype and description of
4264 __builtin_expect_with_probability.
4266 2018-10-30 Michael Ploujnikov <michael.ploujnikov@oracle.com>
4268 * cgraph.h (clone_function_name_1): Replaced by new
4269 clone_function_name_numbered that takes name as string; for
4270 privatize_symbol_name_1 use only.
4271 (clone_function_name): Renamed to
4272 clone_function_name_numbered to be explicit about numbering.
4273 (clone_function_name): New two-argument function that does
4274 not number its output.
4275 (clone_function_name): New three-argument function that
4276 takes a number to append to its output.
4277 * cgraphclones.c (duplicate_thunk_for_node):
4278 (clone_function_name_1): Renamed.
4279 (clone_function_name_numbered): Two new functions.
4280 (clone_function_name): Improved documentation.
4281 (cgraph_node::create_virtual_clone): Use clone_function_name_numbered.
4282 * config/rs6000/rs6000.c (make_resolver_func): Ditto.
4283 * final.c (final_scan_insn_1): Use the new clone_function_name
4285 * multiple_target.c (create_dispatcher_calls): Ditto.
4286 (create_target_clone): Ditto.
4287 * omp-expand.c (grid_expand_target_grid_body): Ditto.
4288 * omp-low.c (create_omp_child_function_name): Ditto.
4289 * omp-simd-clone.c (simd_clone_create): Ditto.
4290 * symtab.c (simd_symtab_node::noninterposable_alias): Use the
4291 new clone_function_name without numbering.
4293 2018-10-30 Richard Earnshaw <rearnsha@arm.com>
4295 * alloc-pool.h (base_pool_allocator <TBlockAllocator>::initialize):
4296 Assert that the allocation size is not zero.
4298 2018-10-30 Richard Biener <rguenther@suse.de>
4300 PR tree-optimization/87800
4301 * tree-vect-slp.c (vect_build_slp_tree_2): Reject any
4302 non-induction or reduction PHIs.
4304 2018-10-30 Sameera Deshpande <sameera.deshpande@linaro.org
4306 * config/aarch64/falkor-tag-collision-avoidance.c
4307 (execute_tag_collision_avoidance): Call df_note_add_problem.
4309 2018-10-30 Martin Liska <mliska@suse.cz>
4311 * doc/extend.texi: Fix typo in documentation
4312 of __builtin_expect_with_probability.
4314 2018-10-29 David Malcolm <dmalcolm@redhat.com>
4317 * input.c (get_substring_ranges_for_loc): Detect if
4318 linemap_resolve_location gives us a NULL map, and reject
4321 2018-10-29 Iain Buclaw <ibuclaw@gdcproject.org>
4323 * config.gcc (xstormy16-*-elf): Set tm_d_file.
4325 2018-10-29 David Malcolm <dmalcolm@redhat.com>
4326 Martin Sebor <msebor@redhat.com>
4327 Manuel Lopez-Ibanez <manu@gcc.gnu.org>
4329 * doc/ux.texi (Quoting): New subsection, adapted from material at
4330 https://gcc.gnu.org/wiki/DiagnosticsGuidelines written by
4331 MartinSebor and ManuelLopezIbanez.
4332 (Fix-it hints): Note that fix-it hints shouldn't be marked for
4335 2018-10-29 Kugan Vivekanandarajah <kuganv@linaro.org>
4338 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Fix niter
4341 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
4343 * gcc.target/powerpc/ssse3-check.h: Remove inaccurate comment.
4345 2018-10-29 Paul Koning <ni1d@arrl.net>
4347 * doc/rtl.texi (CONST_WIDE_INT_ELT): Give correct macro name.
4349 2018-10-29 Paul A. Clarke <pc@us.ibm.com>
4351 * gcc/config/rs6000/mmintrin.h (_mm_packs_pi16, _mm_packs_pi32,
4352 _mm_packs_pu16, _mm_unpackhi_pi8, _mm_unpacklo_pi8, _mm_add_pi8,
4353 _mm_add_pi16, _mm_add_pi32, _mm_sub_pi8, _mm_sub_pi16, _mm_sub_pi32,
4354 _mm_cmpgt_pi8, _mm_cmpeq_pi16, _mm_cmpgt_pi16, _mm_cmpeq_pi32,
4355 _mm_cmpgt_pi32, _mm_adds_pi8, _mm_adds_pi16, _mm_adds_pu8,
4356 _mm_adds_pu16, _mm_subs_pi8, _mm_subs_pi16, _mm_subs_pu8,
4357 _mm_subs_pu16, _mm_madd_pi16, _mm_mulhi_pi16, _mm_mullo_pi16,
4358 _mm_sll_pi16, _mm_sra_pi16, _mm_srl_pi16, _mm_set1_pi16, _mm_set1_pi8):
4359 Change 'vector' to '__vector'.
4360 * gcc/config/rs6000/xmmintrin.h (_mm_cvtps_pi32, _mm_cvttps_pi32,
4361 _mm_cvtps_pi16, _mm_cvtps_pi8, _mm_max_pi16, _mm_max_pu8, _mm_min_pi16,
4362 _mm_min_pu8, _mm_mulhi_pu16, _mm_shuffle_pi16, _mm_avg_pu8,
4363 _mm_avg_pu16): Likewise. And, whitespace corrections.
4365 2018-10-29 Richard Biener <rguenther@suse.de>
4367 PR tree-optimization/87785
4368 * tree-vect-slp.c (vect_gather_slp_loads): Only gather
4371 2018-10-29 Olivier Hainque <hainque@adacore.com>
4373 * VXWORKS_NET_LIBS_RTP: New macro, network part of VXWORKS_LIBS_RTP.
4375 2018-10-29 Olivier Hainque <hainque@adacore.com>
4377 * config/vxworks.h (VXWORKS_PERSONALITY): New VxWorks
4378 ports configuration macro, defaults to "gnu".
4379 (VXWORKS_OS_CPP_BUILTINS): Use it. Feed TOOL and
4380 TOOL_FAMILY instead of _VX_TOOL/_VXTOOL_FAMILY.
4382 2018-10-29 Olivier Hainque <hainque@adacore.com>
4384 * config/vxworks.h (CLEAR_INSN_CACHE): #undef before #define.
4386 2018-10-29 Richard Biener <rguenther@suse.de>
4388 PR tree-optimization/87790
4389 * tree-vect-slp.c (vect_mark_slp_stmts): Simplify.
4390 (vect_make_slp_decision): Adjust.
4391 (vect_slp_analyze_bb_1): Likewise.
4392 (vect_detect_hybrid_slp_stmts): Properly union SLP type over
4395 2018-10-29 Richard Biener <rguenther@suse.de>
4397 PR tree-optimization/87785
4398 * tree-vect-slp.c (vect_build_slp_tree_2): Remove loads argument
4400 (vect_build_slp_tree): Likewise.
4401 (vect_gather_slp_loads): New function.
4402 (vect_analyze_slp_instance): Gather loads separately from the
4405 2018-10-29 Martin Liska <mliska@suse.cz>
4407 * Makefile.in: Make dependency to json.o.
4408 * doc/gcov.texi: Document new JSON format, remove
4409 old intermediate format documentation.
4410 * gcov.c (struct function_info): Come up with m_name and
4412 (function_info::function_info): Initialize it.
4413 (function_info::~function_info): Release it.
4414 (main): Rename flag_intermediate_format to flag_json_format.
4415 (print_usage): Describe --json-format.
4416 (process_args): Set flag_json_format.
4417 (output_intermediate_line): Remove.
4418 (output_intermediate_json_line): Likewise.
4419 (get_gcov_intermediate_filename): Return new extension
4421 (output_intermediate_file): Implement JSON emission.
4422 (output_json_intermediate_file): Implement JSON emission.
4423 (generate_results): Use ::get_name for function name.
4424 Handle JSON output file.
4425 (read_graph_file): Use ::get_name instead of cplus_demangle.
4426 (read_count_file): Likewise.
4427 (solve_flow_graph): Likewise.
4428 (add_line_counts): Likewise.
4429 (accumulate_line_counts): Use new flag_json_format.
4430 (output_function_details): Use ::get_name instead of cplus_demangle.
4431 (output_lines): Likewise.
4432 * json.cc (test_writing_literals): Add new tests.
4433 * json.h (class literal): Add new boolean constructor.
4435 2018-10-29 Segher Boessenkool <segher@kernel.crashing.org>
4437 PR rtl-optimization/87701
4438 PR rtl-optimization/87780
4439 * combine.c (make_more_copies): Rewrite.
4441 2018-10-28 Kugan Vivekanandarajah <kuganv@linaro.org>
4443 * doc/generic.texi (ABSU_EXPR): Document.
4444 * match.pd (absu(x)*absu(x) -> x*x): Handle.
4445 (absu(absu(X)) -> absu(X)): Likewise.
4446 (absu(-X) -> absu(X)): Likewise.
4447 (absu(X) where X is nonnegative -> X): Likewise.
4449 2018-10-28 Iain Buclaw <ibuclaw@gdcproject.org>
4451 * Makefile.in (tm_d_file_list, tm_d_include_list): New variables.
4452 (TM_D_H, D_TARGET_DEF, D_TARGET_H, D_TARGET_OBJS): New variables.
4453 (tm_d.h, cs-tm_d.h, default-d.o): New rules.
4454 (d/d-target-hooks-def.h, s-d-target-hooks-def-h): New rules.
4455 (s-tm-texi): Also check timestamp on d-target.def.
4456 (generated_files): Add TM_D_H and d-target-hooks-def.h.
4457 (build/genhooks.o): Also depend on D_TARGET_DEF.
4458 * config.gcc (tm_d_file, d_target_objs, target_has_targetdm): New
4460 * config/aarch64/aarch64-d.c: New file.
4461 * config/aarch64/aarch64-linux.h (GNU_USER_TARGET_D_CRITSEC_SIZE):
4463 * config/aarch64/aarch64-protos.h (aarch64_d_target_versions): New
4465 * config/aarch64/aarch64.h (TARGET_D_CPU_VERSIONS): Define.
4466 * config/aarch64/t-aarch64 (aarch64-d.o): New rule.
4467 * config/arm/arm-d.c: New file.
4468 * config/arm/arm-protos.h (arm_d_target_versions): New prototype.
4469 * config/arm/arm.h (TARGET_D_CPU_VERSIONS): Define.
4470 * config/arm/linux-eabi.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
4471 * config/arm/t-arm (arm-d.o): New rule.
4472 * config/default-d.c: New file.
4473 * config/glibc-d.c: New file.
4474 * config/gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4475 * config/i386/i386-d.c: New file.
4476 * config/i386/i386-protos.h (ix86_d_target_versions): New prototype.
4477 * config/i386/i386.h (TARGET_D_CPU_VERSIONS): Define.
4478 * config/i386/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
4479 (GNU_USER_TARGET_D_CRITSEC_SIZE): Define.
4480 * config/i386/t-i386 (i386-d.o): New rule.
4481 * config/kfreebsd-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4482 * config/kopensolaris-gnu.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4483 * config/linux-android.h (ANDROID_TARGET_D_OS_VERSIONS): Define.
4484 * config/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4485 * config/mips/linux-common.h (EXTRA_TARGET_D_OS_VERSIONS): Define.
4486 * config/mips/mips-d.c: New file.
4487 * config/mips/mips-protos.h (mips_d_target_versions): New prototype.
4488 * config/mips/mips.h (TARGET_D_CPU_VERSIONS): Define.
4489 * config/mips/t-mips (mips-d.o): New rule.
4490 * config/powerpcspe/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4491 * config/powerpcspe/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4492 * config/powerpcspe/powerpcspe-d.c: New file.
4493 * config/powerpcspe/powerpcspe-protos.h (rs6000_d_target_versions):
4495 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
4496 Support GNU D by using 0 as the language type.
4497 * config/powerpcspe/powerpcspe.h (TARGET_D_CPU_VERSIONS): Define.
4498 * config/powerpcspe/t-powerpcspe (powerpcspe-d.o): New rule.
4499 * config/riscv/riscv-d.c: New file.
4500 * config/riscv/riscv-protos.h (riscv_d_target_versions): New
4502 * config/riscv/riscv.h (TARGET_D_CPU_VERSIONS): Define.
4503 * config/riscv/t-riscv (riscv-d.o): New rule.
4504 * config/rs6000/linux.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4505 * config/rs6000/linux64.h (GNU_USER_TARGET_D_OS_VERSIONS): Define.
4506 * config/rs6000/rs6000-d.c: New file.
4507 * config/rs6000/rs6000-protos.h (rs6000_d_target_versions): New
4509 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
4510 Support GNU D by using 0 as the language type.
4511 * config/rs6000/rs6000.h (TARGET_D_CPU_VERSIONS): Define.
4512 * config/rs6000/t-rs6000 (rs6000-d.o): New rule.
4513 * config/s390/s390-d.c: New file.
4514 * config/s390/s390-protos.h (s390_d_target_versions): New prototype.
4515 * config/s390/s390.h (TARGET_D_CPU_VERSIONS): Define.
4516 * config/s390/t-s390 (s390-d.o): New rule.
4517 * config/sparc/sparc-d.c: New file.
4518 * config/sparc/sparc-protos.h (sparc_d_target_versions): New
4520 * config/sparc/sparc.h (TARGET_D_CPU_VERSIONS): Define.
4521 * config/sparc/t-sparc (sparc-d.o): New rule.
4522 * config/t-glibc (glibc-d.o): New rule.
4523 * configure: Regenerated.
4524 * configure.ac (tm_d_file): New variable.
4525 (tm_d_file_list, tm_d_include_list, d_target_objs): Add substitutes.
4526 * doc/contrib.texi (Contributors): Add self for the D frontend.
4527 * doc/frontends.texi (G++ and GCC): Mention D as a supported language.
4528 * doc/install.texi (Configuration): Mention libphobos as an option for
4529 --enable-shared. Mention d as an option for --enable-languages.
4530 (Testing): Mention check-d as a target.
4531 * doc/invoke.texi (Overall Options): Mention .d, .dd, and .di as file
4532 name suffixes. Mention d as a -x option.
4533 * doc/sourcebuild.texi (Top Level): Mention libphobos.
4534 * doc/standards.texi (Standards): Add section on D language.
4535 * doc/tm.texi: Regenerated.
4536 * doc/tm.texi.in: Add @node for D language and ABI, and @hook for
4537 TARGET_CPU_VERSIONS, TARGET_D_OS_VERSIONS, and TARGET_D_CRITSEC_SIZE.
4538 * dwarf2out.c (is_dlang): New function.
4539 (gen_compile_unit_die): Use DW_LANG_D for D.
4540 (declare_in_namespace): Return module die for D, instead of adding
4541 extra declarations into the namespace.
4542 (gen_namespace_die): Generate DW_TAG_module for D.
4543 (gen_decl_die): Handle CONST_DECLSs for D.
4544 (dwarf2out_decl): Likewise.
4545 (prune_unused_types_walk_local_classes): Handle DW_tag_interface_type.
4546 (prune_unused_types_walk): Handle DW_tag_interface_type same as other
4547 kinds of aggregates.
4548 * gcc.c (default_compilers): Add entries for .d, .dd and .di.
4549 * genhooks.c: Include d/d-target.def.
4551 2018-10-28 Iain Sandoe <iain@sandoe.co.uk>
4554 * config/rs6000/darwin.h (STACK_BOUNDARY): New.
4555 (RS6000_STARTING_FRAME_OFFSET): Adjust to preserve 16byte alignment.
4556 (STACK_DYNAMIC_OFFSET): Likewise.
4558 2018-10-27 Sandra Loosemore <sandra@codesourcery.com>
4561 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Fix
4564 2018-10-26 Tulio Magno Quites Machado Filho <tuliom@linux.ibm.com>
4566 * doc/extend.texi (PowerPC builtins): Fix __builtin_unpack_ibm128
4567 return type and other typos.
4569 2018-10-26 Aaron Sawdey <acsawdey@linux.ibm.com>
4571 * config/rs6000/rs6000-string.c (expand_strncmp_gpr_sequence): Change to
4572 a shorter sequence with fewer branches.
4573 (emit_final_str_compare_gpr): Ditto.
4575 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
4577 * config/rs6000/tmmintrin.h: New file.
4578 * config.gcc (powerpc*-*-*): Add tmmintrin.h to extra_headers.
4580 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
4582 * config/rs6000/mmintrin.h: Enable 32bit compilation.
4583 * config/rs6000/xmmintrin.h: Likewise.
4585 2018-10-26 Paul A. Clarke <pc@us.ibm.com>
4587 * config/rs6000/xmmintrin.h (_mm_extract_pi16): Fix for big-endian.
4589 2018-10-26 Richard Biener <rguenther@suse.de>
4591 * tree-vect-slp.c (vect_mark_slp_stmts): Add visited hash_set
4593 (vect_mark_slp_stmts_relevant): Likewise.
4594 (vect_detect_hybrid_slp_stmts): Likewise.
4595 (vect_bb_slp_scalar_cost): Likewise.
4596 (vect_remove_slp_scalar_calls): Likewise.
4598 2018-10-26 Jan Hubicka <jh@suse.cz>
4600 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix recursion.
4601 (warn_types_mismatch): Fix walk of DECL_NAME.
4602 (odr_types_equivalent_p): Fix overactive assert.
4604 2018-10-26 Richard Biener <rguenther@suse.de>
4606 PR tree-optimization/87105
4607 * tree-vectorizer.h (_slp_tree::refcnt): New member.
4608 * tree-vect-slp.c (vect_free_slp_tree): Decrement and honor
4610 (vect_create_new_slp_node): Initialize refcnt to one.
4612 (scalar_stmts_set_t, bst_fail): Remove.
4613 (vect_build_slp_tree_2): Add bst_map argument and adjust calls.
4614 (vect_build_slp_tree): Add bst_map argument and lookup
4615 already created SLP nodes.
4616 (vect_print_slp_tree): Handle a SLP graph, print SLP node
4618 (vect_slp_rearrange_stmts): Handle a SLP graph.
4619 (vect_analyze_slp_instance): Adjust and free SLP nodes from
4620 the CSE map. Fix indenting.
4621 (vect_schedule_slp_instance): Add short-cut.
4623 2018-10-26 Martin Liska <mliska@suse.cz>
4626 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Hash with
4627 addr_expr and not with pointers.
4629 2018-10-26 Jan Hubicka <jh@suse.cz>
4631 * tree.c (free_lang_data_in_type): Only check main variants.
4632 * ipa-devirt.c (warn_odr): Make static.
4633 (types_same_for_odr): Drop strict variant.
4634 (types_odr_comparable): Likewise.
4635 (odr_or_derived_type_p): Look for main variants.
4636 (odr_name_hasher::equal): Cleanup comment.
4637 (odr_subtypes_equivalent): Add warn and warned arguments; check main
4639 (type_variants_equivalent_p): break out from ...
4640 (odr_types_equivalent): ... here; go for main variants where needed.
4641 (warn_odr): ... here; turn static.
4642 (warn_types_mismatch): Compare mangled names of main variants.
4643 * ipa-utils.h (types_odr_comparable): Drop strict parameter.
4644 (type_with_linkage_p): Sanity check that we look at main variant.
4645 * lto.c (lto_read_decls): Only consider main variant to be ODR type.
4646 * tree.h (types_same_for_odr): Drop strict argument.
4648 2018-10-26 Richard Biener <rguenther@suse.de>
4650 PR tree-optimization/87746
4651 * tree-vect-data-refs.c (vect_update_misalignment_for_peel):
4652 Simplify and fix WRT strided store groups with size not
4653 equal to step in element count.
4654 (vect_analyze_group_access_1): Dump the whole group.
4656 2018-10-25 Carl Love <cel@us.ibm.com>
4658 * config/rs6000/rs6000-c.c (P9V_BUILTIN_VEC_VSCEDPGT,
4659 P9V_BUILTIN_VEC_VSCEDPLT, P9V_BUILTIN_VEC_VSCEDPEQ,
4660 P9V_BUILTIN_VEC_VSCEDPUO): Rename base overloaded name. Add quad
4661 precicion entry for each overloaded builtin.
4662 * config/rs6000/rs6000-builtin.def (VSCEDPGT, VSCEDPLT, VSCEDPEQ,
4663 VSCEDPUO): Rename overloaded name.
4664 (VSCEDPGT, VSCEQPGT, VSCEDPLT, VSCEQPLT, VSCEDPEQ, VSCEQPEQ,
4665 VSCEDPUO, VSCEQPUO): Add defitions for overloaded builtins.
4666 * config/rs6000/vsx.md (xscmpexpqp_<code>_<mode>): Add
4667 define_expand for xscmpexqp instruction.
4668 (*xscmpexpqp): Add define_insn for the xscmpexqp instruction.
4670 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
4671 Jinsong Ji <jji@us.ibm.com>
4673 * config/rs6000/emmintrin.h (_mm_slli_epi16): Replace deprecated
4674 function with vec_sl.
4675 (_mm_slli_epi32): Likewise.
4676 (_mm_slli_epi64): Likewise.
4677 (_mm_srai_epi16): Replace deprecated function with vec_sra.
4678 (_mm_srai_epi32): Likewise.
4679 (_mm_srli_epi16): Replace deprecated function with vec_sr.
4680 (_mm_srli_epi32): Likewise.
4681 (_mm_srli_epi64): Likewise.
4682 (_mm_sll_epi16): Replace deprecated function with vec_sl.
4683 (_mm_sll_epi32): Likewise.
4684 (_mm_sll_epi64): Likewise.
4685 (_mm_sra_epi16): Replace deprecated function with vec_sra.
4686 (_mm_sra_epi32): Likewise.
4687 (_mm_srl_epi16): Replace deprecated function with vec_sr.
4688 (_mm_srl_epi32): Likewise.
4689 (_mm_srl_epi64): Likewise.
4691 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
4692 Jinsong Ji <jji@us.ibm.com>
4694 * gcc/config/rs6000/emmintrin.h (_mm_sll_epi16): Replace
4695 comparison operators with vec_cmp* for compatibility due to
4696 unfortunate history; clean up formatting and use types more
4698 (_mm_sll_epi32): Likewise.
4699 (_mm_sll_epi64): Likewise.
4700 (_mm_srl_epi16): Likewise.
4701 (_mm_srl_epi32): Likewise.
4702 (_mm_srl_epi64): Likewise.
4704 2018-10-25 Bill Schmidt <wschmidt@linux.ibm.com>
4705 Jinsong Ji <jji@us.ibm.com>
4707 * config/rs6000/emmintrin.h (_mm_sll_epi64): Remove wrong cast.
4708 * config/rs6000/xmmintrin.h (_mm_min_ps): Change m's type to
4709 __vector __bool int. Use vec_cmpgt in preference to deprecated
4710 function vec_vcmpgtfp.
4711 (_mm_max_ps): Likewise.
4713 2018-10-25 Jeff Law <law@redhat.com>
4715 * config/rl78/rl78.c (insn_ok_now): Always re-recognize the insn
4718 2018-10-25 Martin Sebor <msebor@redhat.com>
4720 * doc/extend.texi (aligned): Expand attribute description.
4721 (Alignment): Rename section. Discuss function arguments.
4723 2018-10-25 Jan Hubicka <jh@suse.cz>
4725 * ipa-devirt.c (main_odr_variant): Remove.
4726 (hash_odr_name, types_same_for_odr, types_odr_comparable,
4727 odr_name_hasher::equal, odr_subtypes_equivalent_p):
4728 Drop use of main_odr_variant.
4729 (add_type_duplicate): Silence confused warnings on integer types.
4730 (get_odr_type): Always look for main variant.
4731 (register_odr_type): Simplify.
4733 2018-10-25 Richard Biener <rguenther@suse.de>
4735 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses):
4736 Initialize ng to silence error with release checking bootstrap.
4738 2018-10-25 Richard Biener <rguenther@suse.de>
4740 * tree-if-conv.c: Include tree-ssa-sccvn.h.
4741 (tree_if_conversion): Run CSE on the if-converted loop body.
4743 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
4745 * config/s390/constraints.md (ZL): New constraint.
4746 * config/s390/s390.c (legitimate_pic_operand_p): Accept LARL
4748 * config/s390/s390.md (movdi_larl): Remove.
4749 (movdi_64): Add the LARL alternative.
4751 2018-10-25 Ilya Leoshkevich <iii@linux.ibm.com>
4754 * rtl.c (RTX_CODE_HWINT_P_1): New helper macro.
4755 (RTX_CODE_HWINT_P): New macro.
4756 (rtx_code_size): Use RTX_CODE_HWINT_P ().
4758 2018-10-25 Jan Hubicka <jh@suse.cz>
4760 * ipa-devirt.c (odr_types_equivalent_p): Do not ICE if one of types
4763 2018-10-25 Richard Biener <rguenther@suse.de>
4765 PR tree-optimization/87665
4766 PR tree-optimization/87745
4767 * tree-vectorizer.h (get_earlier_stmt): Remove.
4768 (get_later_stmt): Pick up UID from the original non-pattern stmt.
4770 2018-10-25 Sam Tebbs <sam.tebbs@arm.com>
4772 * options.texi (Deprecated): Move list to Var section.
4774 2018-10-24 Bill Schmidt <wschmidt@linux.ibm.com>
4775 Jinsong Ji <jji@us.ibm.com>
4777 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Change deprecated
4778 __vector long to __vector long long.
4779 (_mm_cvtpd_ps): Likewise.
4780 (_mm_cvttpd_epi32): Likewise.
4781 (_mm_cvtpi32_pd): Likewise.
4782 (_mm_unpackhi_epi64): Likewise.
4783 (_mm_unpacklo_epi64): Likewise.
4785 2018-10-24 Segher Boessenkool <segher@kernel.crashing.org>
4787 PR rtl-optimization/87720
4788 * combine.c (make_more_copies): Skip if the dest is pc_rtx.
4790 2018-10-24 Alexandre Oliva <aoliva@redhat.com>
4792 * gimple-ssa-isolate-paths.c
4793 (find_implicit_erroneous_behavior): Do not change code if the
4794 pass is running for warnings only.
4795 (find_explicit_erroneous_behavior): Likewise.
4797 2018-10-24 Michael Meissner <meissner@linux.ibm.com>
4799 * config/rs6000/rs6000.c (TARGET_MANGLE_DECL_ASSEMBLER_NAME):
4800 Define as rs6000_mangle_decl_assembler_name.
4801 (rs6000_mangle_decl_assembler_name): If the user switched from IBM
4802 long double to IEEE long double, switch the names of the long
4803 double built-in functions to be <func>f128 instead of <func>l.
4805 2018-10-24 Martin Sebor <msebor@redhat.com>
4807 * doc/extend.texi (nonnull): List no-argument form. Reference
4808 -fno-delete-null-pointer-checks and -fisolate-erroneous-paths-attribute.
4810 2018-10-24 Richard Biener <rguenther@suse.de>
4812 * tree-ssa-sccvn.c (do_rpo_vn): Free rpo_state.
4814 2018-10-24 Martin Liska <mliska@suse.cz>
4816 PR tree-optimization/84436
4817 * tree-switch-conversion.c (switch_conversion::contains_same_values_p):
4819 (switch_conversion::contains_linear_function_p): New.
4820 (switch_conversion::build_one_array): Support linear
4821 transformation on input.
4822 * tree-switch-conversion.h (struct switch_conversion): Add
4823 contains_linear_function_p declaration.
4825 2018-10-24 Richard Biener <rguenther@suse.de>
4827 * varasm.c (const_hash_1): Return hash of ADDR_EXPR
4828 if its argument is CONSTANT_CLASS_P.
4830 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
4832 * ipa-utils.h (type_with_linkage_p): No longer check for TYPE_STUB_DECL;
4833 it is wrong for forward declarations.
4835 2018-10-24 Ilya Leoshkevich <iii@linux.ibm.com>
4837 * config/s390/s390.c (s390_check_qrst_address): Add the missing
4838 SYMBOL_REF_P () check.
4840 2018-10-24 Richard Biener <rguenther@suse.de>
4842 PR tree-optimization/87105
4843 * tree-vect-data-refs.c (vect_analyze_group_access_1): Adjust
4844 dump classification.
4845 (vect_analyze_data_ref_accesses): Handle duplicate loads and
4846 stores by splitting the affected group after the fact.
4847 * tree-vect-slp.c (vect_build_slp_tree_2): Dump when we
4848 fail the SLP build because of size constraints.
4850 2018-10-24 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
4852 * configure.ac (gcc_cv_ld_aligned_shf_merge): New test.
4853 * configure: Regenerate.
4854 * config.in: Regenerate.
4855 * varasm.c (mergeable_string_section): Use readonly_data_section
4856 if linker doesn't support SHF_MERGE with alignment > 8.
4857 (mergeable_constant_section): Likewise.
4859 2018-10-24 Richard Biener <rguenther@suse.de>
4861 PR tree-optimization/84013
4862 * tree-ssa-structalias.c (struct msdi_data): New struct for
4863 marshalling data to walk_stmt_load_store_ops.
4864 (maybe_set_dependence_info): Refactor as callback for
4865 walk_stmt_load_store_ops.
4866 (compute_dependence_clique): Set restrict info on all stmt kinds.
4868 2018-10-24 Martin Liska <mliska@suse.cz>
4870 * cgraph.c (cgraph_node::dump):
4871 Remove reduntant dumps and make tp_first_run dump more compact.
4873 2018-10-24 Richard Biener <rguenther@suse.de>
4875 PR tree-optimization/87665
4876 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Adjust
4879 2018-10-12 Jeff Law <law@redhat.com>
4881 * config/h8300/h8300.c (h8300_expand_prologue): Fix stm generation
4884 2018-10-23 Richard Biener <rguenther@suse.de>
4886 * tree-vrp.c (add_assert_info): Guard dump_printf with
4888 * gimple-ssa-evrp-analyze.c
4889 (evrp_range_analyzer::record_ranges_from_incoming_edge):
4890 Use value_range::ignore_equivs_equal_p.
4892 2018-10-23 Richard Biener <rguenther@suse.de>
4894 PR tree-optimization/87105
4895 PR tree-optimization/87608
4896 * passes.def (pass_all_early_optimizations): Add early phi-opt
4898 * tree-ssa-phiopt.c (value_replacement): Ignore NOPs and predicts in
4899 addition to debug stmts.
4900 (tree_ssa_phiopt_worker): Add early_p argument, do only min/max
4901 and abs replacement early.
4902 * tree-cfg.c (gimple_empty_block_p): Likewise.
4904 2018-10-23 Richard Earnshaw <rearnsha@arm.com>
4907 * config.gcc (arm*-*-netbsdelf*): Default to StrongARM if no CPU
4908 specified to configure.
4909 (arm*-*-*): Use ARM7TDMI as the target CPU if no default provided.
4911 2018-10-23 Richard Biener <rguenther@suse.de>
4913 PR tree-optimization/87700
4914 * tree-ssa-copy.c (set_copy_of_val): Fix change detection logic.
4916 2018-10-23 Jakub Jelinek <jakub@redhat.com>
4919 * config/i386/avx512vlintrin.h (_mm_mask_mullo_epi32): Change type of
4920 second argument from __mmask16 to __mmask8.
4921 * config/i386/avx512vlbwintrin.h (_mm_mask_packus_epi32,
4922 _mm_mask_packs_epi32): Likewise.
4923 * config/i386/avx512pfintrin.h (_mm512_mask_prefetch_i64scatter_ps):
4925 (_mm512_mask_prefetch_i64scatter_pd): Likewise. Formatting fix.
4927 2018-10-23 Richard Biener <rguenther@suse.de>
4929 * tree-vect-stmts.c (vect_analyze_stmt): Fix typo in comment.
4931 2018-10-23 Richard Biener <rguenther@suse.de>
4933 PR tree-optimization/86144
4934 * tree-vect-stmts.c (vect_analyze_stmt): Prefer -mveclibabi
4935 over simd attribute.
4937 2018-10-23 Richard Biener <rguenther@suse.de>
4939 PR tree-optimization/87693
4940 * tree-ssa-threadedge.c (thread_around_empty_blocks): Handle
4941 the case we do not find the taken edge.
4943 2018-10-22 Bill Schmidt <wschmidt@linux.ibm.com>
4944 Jinsong Ji <jji@us.ibm.com>
4946 * config/rs6000/emmintrin.h (_MM_SHUFFLE2): Comment cleanup.
4947 (_mm_store_pd): Use unaligned vector type for pointer cast.
4948 (_mm_maskmoveu_si128): Likewise.
4949 * config/rs6000/xmmintrin.h (__m128_u): New typedef.
4950 (_mm_store_ps): Use unaligned vector type for pointer cast.
4952 2018-10-22 Paul Koning <ni1d@arrl.net>
4954 * symtab.c (symtab_node::increase_alignment): Correct max
4957 2018-10-22 Yury Gribov <tetra2005@gmail.com>
4959 PR tree-optimization/87633
4960 * match.pd: Do not generate unordered integer comparisons.
4962 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
4964 PR rtl-optimization/87600
4965 * combine.c: Add include of expr.h.
4966 (cant_combine_insn_p): Do not combine moves from any hard non-fixed
4967 register to a pseudo.
4968 (make_more_copies): New function, add a copy to a new pseudo after
4969 the moves from hard registers into pseudos.
4970 (rest_of_handle_combine): Declare rebuild_jump_labels_after_combine
4971 later. Call make_more_copies.
4973 2018-10-22 Andrew Stubbs <ams@codesourcery.com>
4975 * lra-constraints.c (process_alt_operands): New local array,
4976 matching_early_clobber. Check matching_early_clobber before
4977 decrementing reject, and set matching_early_clobber after.
4979 2018-10-22 Segher Boessenkool <segher@kernel.crashing.org>
4982 * config/rs6000/rs6000.c (print_operand_address): For unexpected RTL
4983 call output_addr_const and hope for the best.
4985 2018-10-22 Richard Biener <rguenther@suse.de>
4987 * gimple-ssa-evrp-analyze.c
4988 (evrp_range_analyzer::record_ranges_from_incoming_edge): Be
4989 smarter about what ranges to use.
4990 * tree-vrp.c (add_assert_info): Dump here.
4991 (register_edge_assert_for_2): Instead of here at multiple but
4994 * gcc.dg/tree-ssa/evrp12.c: New testcase.
4995 * gcc.dg/predict-6.c: Adjust.
4996 * gcc.dg/tree-ssa/vrp33.c: Disable EVRP.
4997 * gcc.dg/tree-ssa/vrp02.c: Likewise.
4998 * gcc.dg/tree-ssa/cunroll-9.c: Likewise.
5000 2018-10-22 Steven Bosscher <steven@gcc.gnu.org>
5001 Richard Biener <rguenther@suse.de>
5003 * bitmap.h: Update data structure documentation, including a
5004 description of bitmap views as either linked-lists or splay trees.
5005 (struct bitmap_element_def): Update comments for splay tree bitmaps.
5006 (struct bitmap_head_def): Likewise.
5007 (bitmap_list_view, bitmap_tree_view): New prototypes.
5008 (bitmap_initialize_stat): Initialize a bitmap_head's indx and
5010 (bmp_iter_set_init): Assert the iterated bitmaps are in list form.
5011 (bmp_iter_and_init, bmp_iter_and_compl_init): Likewise.
5012 * bitmap.c (bitmap_elem_to_freelist): Unregister overhead of a
5013 released bitmap element here.
5014 (bitmap_element_free): Remove.
5015 (bitmap_elt_clear_from): Work on splay tree bitmaps.
5016 (bitmap_list_link_element): Renamed from bitmap_element_link. Move
5017 this function similar ones such that linked-list bitmap implementation
5018 functions are grouped.
5019 (bitmap_list_unlink_element): Renamed from bitmap_element_unlink,
5020 and moved for grouping.
5021 (bitmap_list_insert_element_after): Renamed from
5022 bitmap_elt_insert_after, and moved for grouping.
5023 (bitmap_list_find_element): New function spliced from bitmap_find_bit.
5024 (bitmap_tree_link_left, bitmap_tree_link_right,
5025 bitmap_tree_rotate_left, bitmap_tree_rotate_right, bitmap_tree_splay,
5026 bitmap_tree_link_element, bitmap_tree_unlink_element,
5027 bitmap_tree_find_element): New functions for splay-tree bitmap
5029 (bitmap_element_link, bitmap_element_unlink, bitmap_elt_insert_after):
5030 Renamed and moved, see above entries.
5031 (bitmap_tree_listify_from): New function to convert part of a splay
5032 tree bitmap to a linked-list bitmap.
5033 (bitmap_list_view): Convert a splay tree bitmap to linked-list form.
5034 (bitmap_tree_view): Convert a linked-list bitmap to splay tree form.
5035 (bitmap_find_bit): Remove.
5036 (bitmap_clear, bitmap_clear_bit, bitmap_set_bit,
5037 bitmap_single_bit_set_p, bitmap_first_set_bit, bitmap_last_set_bit):
5038 Handle splay tree bitmaps.
5039 (bitmap_copy, bitmap_count_bits, bitmap_and, bitmap_and_into,
5040 bitmap_elt_copy, bitmap_and_compl, bitmap_and_compl_into,
5041 bitmap_compl_and_into, bitmap_elt_ior, bitmap_ior, bitmap_ior_into,
5042 bitmap_xor, bitmap_xor_into, bitmap_equal_p, bitmap_intersect_p,
5043 bitmap_intersect_compl_p, bitmap_ior_and_compl,
5044 bitmap_ior_and_compl_into, bitmap_set_range, bitmap_clear_range,
5045 bitmap_hash): Reject trying to act on splay tree bitmaps. Make
5046 corresponding changes to use linked-list specific bitmap_element
5047 manipulation functions as applicable for efficiency.
5048 (bitmap_tree_to_vec): New function.
5049 (debug_bitmap_elt_file): New function split out from ...
5050 (debug_bitmap_file): ... here. Handle splay tree bitmaps.
5051 (bitmap_print): Likewise.
5053 PR tree-optimization/63155
5054 * tree-ssa-propagate.c (ssa_prop_init): Use tree-view for the
5056 * tree-ssa-coalesce.c (coalesce_ssa_name): Populate used_in_copies
5059 2018-10-22 Martin Liska <mliska@suse.cz>
5061 PR tree-optimization/87686
5063 2018-08-29 Martin Liska <mliska@suse.cz>
5065 * tree-switch-conversion.c (switch_conversion::expand):
5066 Strenghten assumption about gswitch statements.
5068 2018-10-22 Martin Liska <mliska@suse.cz>
5070 * ipa-icf.c (sem_item::compare_attributes): Remove.
5071 (sem_item::compare_referenced_symbol_properties): Use
5072 attribute_list_equal instead.
5073 (sem_function::equals_wpa): Likewise.
5074 * ipa-icf.h: Remove compare_attributes.
5076 2018-10-22 Richard Biener <rguenther@suse.de>
5079 * mem-stats.h (mem_usage::operator==): Fix pasto.
5081 2018-10-22 Richard Biener <rguenther@suse.de>
5083 PR tree-optimization/87640
5084 * tree-vrp.c (set_value_range_with_overflow): Decompose
5086 (extract_range_from_binary_expr_1): Adjust.
5088 2018-10-22 Martin Jambor <mjambor@suse.cz>
5090 * tree-eh.h (stmt_could_throw_p): Add function parameter.
5091 (stmt_can_throw_external): Likewise.
5092 (stmt_can_throw_internal): Likewise.
5093 * tree-eh.c (lower_eh_constructs_2): Pass cfun to stmt_could_throw_p.
5094 (lower_eh_constructs_2): Likewise.
5095 (stmt_could_throw_p): Add fun parameter, use it instead of cfun.
5096 (stmt_can_throw_external): Likewise.
5097 (stmt_can_throw_internal): Likewise.
5098 (maybe_clean_eh_stmt_fn): Pass cfun to stmt_could_throw_p.
5099 (maybe_clean_or_replace_eh_stmt): Pass cfun to stmt_could_throw_p.
5100 (maybe_duplicate_eh_stmt_fn): Pass new_fun to stmt_could_throw_p.
5101 (maybe_duplicate_eh_stmt): Pass cfun to stmt_could_throw_p.
5102 (pass_lower_eh_dispatch::execute): Pass cfun to
5103 stmt_can_throw_external.
5104 (cleanup_empty_eh): Likewise.
5105 (verify_eh_edges): Pass cfun to stmt_could_throw_p.
5106 * cgraph.c (cgraph_edge::set_call_stmt): Pass a function to
5107 stmt_can_throw_external instead of pushing it to cfun.
5108 (symbol_table::create_edge): Likewise.
5109 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Pass cfun to
5110 stmt_can_throw_internal.
5111 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Pass cfun
5112 to stmt_could_throw_p.
5113 * gimple-ssa-store-merging.c (handled_load): Pass cfun to
5114 stmt_can_throw_internal.
5115 (pass_store_merging::execute): Likewise.
5116 * gimple-ssa-strength-reduction.c
5117 (find_candidates_dom_walker::before_dom_children): Pass cfun to
5119 * gimplify-me.c (gimple_regimplify_operands): Pass cfun to
5120 stmt_can_throw_internal.
5121 * ipa-pure-const.c (check_call): Pass cfun to stmt_could_throw_p and
5122 to stmt_can_throw_external.
5123 (check_stmt): Pass cfun to stmt_could_throw_p.
5124 (check_stmt): Pass cfun to stmt_can_throw_external.
5125 (pass_nothrow::execute): Likewise.
5126 * trans-mem.c (expand_call_tm): Pass cfun to stmt_can_throw_internal.
5127 * tree-cfg.c (is_ctrl_altering_stmt): Pass cfun to
5128 stmt_can_throw_internal.
5129 (verify_gimple_in_cfg): Pass cfun to stmt_could_throw_p.
5130 (stmt_can_terminate_bb_p): Pass cfun to stmt_can_throw_external.
5131 (gimple_purge_dead_eh_edges): Pass cfun to stmt_can_throw_internal.
5132 * tree-complex.c (expand_complex_libcall): Pass cfun to
5133 stmt_could_throw_p and to stmt_can_throw_internal.
5134 (expand_complex_multiplication): Pass cfun to stmt_can_throw_internal.
5135 * tree-inline.c (copy_edges_for_bb): Likewise.
5136 (maybe_move_debug_stmts_to_successors): Likewise.
5137 * tree-outof-ssa.c (ssa_is_replaceable_p): Pass cfun to
5139 * tree-parloops.c (oacc_entry_exit_ok_1): Likewise.
5140 * tree-sra.c (scan_function): Pass cfun to stmt_can_throw_external.
5141 * tree-ssa-alias.c (stmt_kills_ref_p): Pass cfun to
5142 stmt_can_throw_internal.
5143 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Likewise.
5144 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Pass cfun to
5146 (mark_aliased_reaching_defs_necessary_1): Pass cfun to
5147 stmt_can_throw_internal.
5148 * tree-ssa-forwprop.c (pass_forwprop::execute): Likewise.
5149 * tree-ssa-loop-im.c (movement_possibility): Pass cfun to
5151 * tree-ssa-loop-ivopts.c (find_givs_in_stmt_scev): Likewise.
5152 (add_autoinc_candidates): Pass cfun to stmt_can_throw_internal.
5153 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
5154 (convert_mult_to_fma_1): Likewise.
5155 (convert_to_divmod): Likewise.
5156 * tree-ssa-phiprop.c (propagate_with_phi): Likewise.
5157 * tree-ssa-pre.c (compute_avail): Pass cfun to stmt_could_throw_p.
5158 * tree-ssa-propagate.c
5159 (substitute_and_fold_dom_walker::before_dom_children): Likewise.
5160 * tree-ssa-reassoc.c (suitable_cond_bb): Likewise.
5161 (maybe_optimize_range_tests): Likewise.
5162 (linearize_expr_tree): Likewise.
5163 (reassociate_bb): Likewise.
5164 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
5165 * tree-ssa-scopedtables.c (hashable_expr_equal_p): Likewise.
5166 * tree-ssa-strlen.c (adjust_last_stmt): Likewise.
5167 (handle_char_store): Likewise.
5168 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Pass cfun to
5169 stmt_can_throw_internal.
5170 * tree-vect-patterns.c (check_bool_pattern): Pass cfun to
5172 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
5173 (vectorizable_call): Pass cfun to stmt_can_throw_internal.
5174 (vectorizable_simd_clone_call): Likewise.
5175 * value-prof.c (gimple_ic): Pass cfun to stmt_could_throw_p.
5176 (gimple_stringop_fixed_value): Likewise.
5178 2018-10-22 Ilya Leoshkevich <iii@linux.ibm.com>
5180 * config/s390/s390.c (s390_loadrelative_operand_p): Accept
5181 literal pool references.
5182 (s390_check_qrst_address): Adapt to the new behavior of
5183 s390_loadrelative_operand_p ().
5185 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
5188 * config/i386/sse.md (*andnot<mode>3_bcst): New.
5190 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
5193 * config/i386/sse.md (*<code><mode>3_bcst): New.
5195 2018-10-22 H.J. Lu <hongjiu.lu@intel.com>
5198 * config/i386/sse.md (avx512bcst): Updated for V4SI, V2DI, V8SI,
5199 V4DI, V16SI and V8DI.
5200 (*sub<mode>3<mask_name>_bcst): New.
5201 (*add<mode>3<mask_name>_bcst): Likewise.
5203 2018-10-21 Bill Schmidt <wschmidt@linux.ibm.com>
5204 Jinsong Ji <jji@us.ibm.com>
5206 * config/rs6000/emmintrin.h (_mm_movemask_pd): Replace __vector
5207 __m64 with __vector unsigned long long for compatibility.
5208 (_mm_movemask_epi8): Likewise.
5209 * config/rs6000/xmmintrin.h (_mm_cvtps_pi32): Likewise.
5210 (_mm_cvttps_pi32): Likewise.
5211 (_mm_cvtpi32_ps): Likewise.
5212 (_mm_cvtps_pi16): Likewise.
5213 (_mm_loadh_pi): Likewise.
5214 (_mm_storeh_pi): Likewise.
5215 (_mm_movehl_ps): Likewise.
5216 (_mm_movelh_ps): Likewise.
5217 (_mm_loadl_pi): Likewise.
5218 (_mm_storel_pi): Likewise.
5219 (_mm_movemask_ps): Likewise.
5220 (_mm_shuffle_pi16): Likewise.
5222 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
5225 * config/i386/avx512fintrin.h (_mm512_fnmsub_round_pd): Use
5226 __builtin_ia32_vfnmsubpd512_mask.
5227 (_mm512_mask_fnmsub_round_pd): Likewise.
5228 (_mm512_fnmsub_pd): Likewise.
5229 (_mm512_mask_fnmsub_pd): Likewise.
5230 (_mm512_maskz_fnmsub_round_pd): Use
5231 __builtin_ia32_vfnmsubpd512_maskz.
5232 (_mm512_maskz_fnmsub_pd): Likewise.
5233 (_mm512_fnmsub_round_ps): Use __builtin_ia32_vfnmsubps512_mask.
5234 (_mm512_mask_fnmsub_round_ps): Likewise.
5235 (_mm512_fnmsub_ps): Likewise.
5236 (_mm512_mask_fnmsub_ps): Likewise.
5237 (_mm512_maskz_fnmsub_round_ps): Use
5238 __builtin_ia32_vfnmsubps512_maskz.
5239 (_mm512_maskz_fnmsub_ps): Likewise.
5240 * config/i386/avx512vlintrin.h (_mm256_mask_fnmsub_pd): Use
5241 __builtin_ia32_vfnmsubpd256_mask.
5242 (_mm256_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256_maskz.
5243 (_mm_mask_fnmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
5244 (_mm_maskz_fnmsub_pd): Use __builtin_ia32_vfnmsubpd128_maskz.
5245 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
5246 (_mm256_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_mask.
5247 (_mm256_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps256_maskz.
5248 (_mm_mask_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_mask.
5249 (_mm_maskz_fnmsub_ps): Use __builtin_ia32_vfnmsubps128_maskz.
5250 * config/i386/fmaintrin.h (_mm_fnmsub_pd): Use
5251 __builtin_ia32_vfnmsubpd.
5252 (_mm256_fnmsub_pd): Use __builtin_ia32_vfnmsubpd256.
5253 (_mm_fnmsub_ps): Use __builtin_ia32_vfnmsubps.
5254 (_mm256_fnmsub_ps): Use __builtin_ia32_vfnmsubps256.
5255 (_mm_fnmsub_sd): Use __builtin_ia32_vfnmsubsd3.
5256 (_mm_fnmsub_ss): Use __builtin_ia32_vfnmsubss3.
5257 * config/i386/i386-builtin.def: Add
5258 __builtin_ia32_vfnmsubpd256_mask,
5259 __builtin_ia32_vfnmsubpd256_maskz,
5260 __builtin_ia32_vfnmsubpd128_mask,
5261 __builtin_ia32_vfnmsubpd128_maskz,
5262 __builtin_ia32_vfnmsubps256_mask,
5263 __builtin_ia32_vfnmsubps256_maskz,
5264 __builtin_ia32_vfnmsubps128_mask,
5265 __builtin_ia32_vfnmsubps128_maskz,
5266 __builtin_ia32_vfnmsubpd512_mask,
5267 __builtin_ia32_vfnmsubpd512_maskz,
5268 __builtin_ia32_vfnmsubps512_mask,
5269 __builtin_ia32_vfnmsubps512_maskz, __builtin_ia32_vfnmsubss3,
5270 __builtin_ia32_vfnmsubsd3, __builtin_ia32_vfnmsubps,
5271 __builtin_ia32_vfnmsubpd, __builtin_ia32_vfnmsubps256 and.
5272 __builtin_ia32_vfnmsubpd256.
5273 * config/i386/sse.md (fma4i_fnmsub_<mode>): New.
5274 (<avx512>_fnmsub_<mode>_maskz<round_expand_name>): Likewise.
5275 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_1):
5277 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_2):
5279 (*<sd_mask_codefor>fma_fnmsub_<mode><sd_maskz_name>_bcst_3):
5281 (fmai_vmfnmsub_<mode><round_name>): Likewise.
5283 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
5286 * config/i386/avx512fintrin.h (_mm512_fnmadd_round_pd): Use
5287 __builtin_ia32_vfnmaddpd512_mask.
5288 (_mm512_mask_fnmadd_round_pd): Likewise.
5289 (_mm512_fnmadd_pd): Likewise.
5290 (_mm512_mask_fnmadd_pd): Likewise.
5291 (_mm512_maskz_fnmadd_round_pd): Use
5292 __builtin_ia32_vfnmaddpd512_maskz.
5293 (_mm512_maskz_fnmadd_pd): Likewise.
5294 (_mm512_fnmadd_round_ps): Use __builtin_ia32_vfnmaddps512_mask.
5295 (_mm512_mask_fnmadd_round_ps): Likewise.
5296 (_mm512_fnmadd_ps): Likewise.
5297 (_mm512_mask_fnmadd_ps): Likewise.
5298 (_mm512_maskz_fnmadd_round_ps): Use
5299 __builtin_ia32_vfnmaddps512_maskz.
5300 (_mm512_maskz_fnmadd_ps): Likewise.
5301 * config/i386/avx512vlintrin.h (_mm256_mask_fnmadd_pd): Use
5302 __builtin_ia32_vfnmaddpd256_mask.
5303 (_mm256_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256_maskz.
5304 (_mm_mask_fnmadd_pd): Use __builtin_ia32_vfmaddpd128_mask
5305 (_mm_maskz_fnmadd_pd): Use __builtin_ia32_vfnmaddpd128_maskz.
5306 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
5307 (_mm256_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_mask.
5308 (_mm256_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps256_maskz.
5309 (_mm_mask_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_mask.
5310 (_mm_maskz_fnmadd_ps): Use __builtin_ia32_vfnmaddps128_maskz.
5311 * config/i386/fmaintrin.h (_mm_fnmadd_pd): Use
5312 __builtin_ia32_vfnmaddpd.
5313 (_mm256_fnmadd_pd): Use __builtin_ia32_vfnmaddpd256.
5314 (_mm_fnmadd_ps): Use __builtin_ia32_vfnmaddps.
5315 (_mm256_fnmadd_ps): Use __builtin_ia32_vfnmaddps256.
5316 (_mm_fnmadd_sd): Use __builtin_ia32_vfnmaddsd3.
5317 (_mm_fnmadd_ss): Use __builtin_ia32_vfnmaddss3.
5318 * config/i386/i386-builtin.def: Add
5319 __builtin_ia32_vfnmaddpd256_mask,
5320 __builtin_ia32_vfnmaddpd256_maskz,
5321 __builtin_ia32_vfnmaddpd128_mask,
5322 __builtin_ia32_vfnmaddpd128_maskz,
5323 __builtin_ia32_vfnmaddps256_mask,
5324 __builtin_ia32_vfnmaddps256_maskz,
5325 __builtin_ia32_vfnmaddps128_mask,
5326 __builtin_ia32_vfnmaddps128_maskz,
5327 __builtin_ia32_vfnmaddpd512_mask,
5328 __builtin_ia32_vfnmaddpd512_maskz,
5329 __builtin_ia32_vfnmaddps512_mask,
5330 __builtin_ia32_vfnmaddps512_maskz, __builtin_ia32_vfnmaddss3,
5331 __builtin_ia32_vfnmaddsd3, __builtin_ia32_vfnmaddps,
5332 __builtin_ia32_vfnmaddpd, __builtin_ia32_vfnmaddps256 and.
5333 __builtin_ia32_vfnmaddpd256.
5334 * config/i386/sse.md (fma4i_fnmadd_<mode>): New.
5335 (<avx512>_fnmadd_<mode>_maskz<round_expand_name>): Likewise.
5336 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_1):
5338 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_2):
5340 (*<sd_mask_codefor>fma_fnmadd_<mode><sd_maskz_name>_bcst_3):
5342 (fmai_vmfnmadd_<mode><round_name>): Likewise.
5344 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
5347 * config/i386/avx512fintrin.h (_mm512_fmsub_round_pd): Use
5348 __builtin_ia32_vfmsubpd512_mask.
5349 (_mm512_mask_fmsub_round_pd): Likewise.
5350 (_mm512_fmsub_pd): Likewise.
5351 (_mm512_mask_fmsub_pd): Likewise.
5352 (_mm512_maskz_fmsub_round_pd): Use
5353 __builtin_ia32_vfmsubpd512_maskz.
5354 (_mm512_maskz_fmsub_pd): Likewise.
5355 (_mm512_fmsub_round_ps): Use __builtin_ia32_vfmsubps512_mask.
5356 (_mm512_mask_fmsub_round_ps): Likewise.
5357 (_mm512_fmsub_ps): Likewise.
5358 (_mm512_mask_fmsub_ps): Likewise.
5359 (_mm512_maskz_fmsub_round_ps): Use
5360 __builtin_ia32_vfmsubps512_maskz.
5361 (_mm512_maskz_fmsub_ps): Likewise.
5362 * config/i386/avx512vlintrin.h (_mm256_mask_fmsub_pd): Use
5363 __builtin_ia32_vfmsubpd256_mask.
5364 (_mm256_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd256_maskz.
5365 (_mm_mask_fmsub_pd): Use __builtin_ia32_vfmaddpd128_mask
5366 (_mm_maskz_fmsub_pd): Use __builtin_ia32_vfmsubpd128_maskz.
5367 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
5368 (_mm256_mask_fmsub_ps): Use __builtin_ia32_vfmsubps256_mask.
5369 (_mm256_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps256_maskz.
5370 (_mm_mask_fmsub_ps): Use __builtin_ia32_vfmsubps128_mask.
5371 (_mm_maskz_fmsub_ps): Use __builtin_ia32_vfmsubps128_maskz.
5372 * config/i386/fmaintrin.h (_mm_fmsub_pd): Use
5373 __builtin_ia32_vfmsubpd.
5374 (_mm256_fmsub_pd): Use __builtin_ia32_vfmsubpd256.
5375 (_mm_fmsub_ps): Use __builtin_ia32_vfmsubps.
5376 (_mm256_fmsub_ps): Use __builtin_ia32_vfmsubps256.
5377 (_mm_fmsub_sd): Use __builtin_ia32_vfmsubsd3.
5378 (_mm_fmsub_ss): Use __builtin_ia32_vfmsubss3.
5379 * config/i386/i386-builtin.def: Add
5380 __builtin_ia32_vfmsubpd256_mask,
5381 __builtin_ia32_vfmsubpd256_maskz,
5382 __builtin_ia32_vfmsubpd128_mask,
5383 __builtin_ia32_vfmsubpd128_maskz,
5384 __builtin_ia32_vfmsubps256_mask,
5385 __builtin_ia32_vfmsubps256_maskz,
5386 __builtin_ia32_vfmsubps128_mask,
5387 __builtin_ia32_vfmsubps128_maskz,
5388 __builtin_ia32_vfmsubpd512_mask,
5389 __builtin_ia32_vfmsubpd512_maskz,
5390 __builtin_ia32_vfmsubps512_mask,
5391 __builtin_ia32_vfmsubps512_maskz, __builtin_ia32_vfmsubss3,
5392 __builtin_ia32_vfmsubsd3, __builtin_ia32_vfmsubps,
5393 __builtin_ia32_vfmsubpd, __builtin_ia32_vfmsubps256 and.
5394 __builtin_ia32_vfmsubpd256.
5395 * config/i386/sse.md (fma4i_fmsub_<mode>): New.
5396 (<avx512>_fmsub_<mode>_maskz<round_expand_name>): Likewise.
5397 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_1):
5399 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_2):
5401 (*<sd_mask_codefor>fma_fmsub_<mode><sd_maskz_name>_bcst_3):
5403 (fmai_vmfmsub_<mode><round_name>): Likewise.
5405 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
5407 * config/i386/sse.md (*<plusminus_insn><mode>3<mask_name>_bcst_1):
5408 Remove plus. Renamed to ...
5409 (*sub<mode>3<mask_name>_bcst): This.
5410 (*add<mode>3<mask_name>_bcst_2): Renamede to ...
5411 (*add<mode>3<mask_name>_bcst): This.
5413 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
5416 * config/i386/sse.md (*mul<mode>3<mask_name>_bcst): New.
5418 2018-10-21 H.J. Lu <hongjiu.lu@intel.com>
5421 * i386/avx512vlintrin.h (_mm256_or_epi32): New.
5422 (_mm_or_epi32): Likewise.
5423 (_mm256_xor_epi32): Likewise.
5424 (_mm_xor_epi32): Likewise.
5425 (_mm256_or_epi64): Likewise.
5426 (_mm_or_epi64): Likewise.
5427 (_mm256_xor_epi64): Likewise.
5428 (_mm_xor_epi64): Likewise.
5430 2018-10-20 H.J. Lu <hongjiu.lu@intel.com>
5433 * config/i386/sse.md (*<avx512>_div<mode>3<mask_name>_bcst): New.
5435 2018-10-20 Jakub Jelinek <jakub@redhat.com>
5438 * varasm.c (decode_addr_const): Handle COMPOUND_LITERAL_EXPR.
5440 2018-10-20 Andreas Schwab <schwab@linux-m68k.org>
5442 * doc/ux.texi: Move @section directly after @node.
5444 2018-10-19 Jakub Jelinek <jakub@redhat.com>
5448 * omp-low.c (check_omp_nesting_restrictions): Diagnose ordered without
5449 depend closely nested inside of loop with ordered clause with
5452 2018-10-19 David Malcolm <dmalcolm@redhat.com>
5454 * Makefile.in (TEXI_GCCINT_FILES): Add ux.texi.
5455 * doc/gccint.texi: Include ux.texi and use it in top-level menu.
5456 * doc/ux.texi: New file.
5458 2018-10-19 Segher Boessenkool <segher@kernel.crashing.org>
5460 * config/rs6000/rs6000.h (REG_ALLOC_ORDER): Move 68 (that is, CR0) to
5461 be the first CR field allocated.
5463 2018-10-19 Richard Biener <rguenther@suse.de>
5466 * config/i386/i386.c (ix86_builtin_vectorization_cost): Use
5467 TYPE_VECTOR_SUBPARTS and avoid relying on vector mode.
5469 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
5472 * config/i386/sse.md
5473 (*<plusminus_insn><mode>3<mask_name>_bcst_1): New.
5474 (*add<mode>3<mask_name>_bcst_2): Likewise.
5476 2018-10-19 H.J. Lu <hongjiu.lu@intel.com>
5478 * config/i386/sse.md
5479 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
5480 Replace nonimmediate_operand with register_operand.
5481 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
5483 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
5486 2018-10-19 Ilya Leoshkevich <iii@linux.ibm.com>
5488 PR rtl-optimization/87596
5489 * lra-constraints.c (spill_hard_reg_in_range): Use INSN_P () +
5490 lra_get_insn_recog_data () instead of lra_insn_recog_data[]
5491 for instructions in FROM..TO range.
5493 2018-10-19 Eric Botcazou <ebotcazou@adacore.com>
5495 * cfgexpand.c (expand_one_var): Use specific wording in error message
5496 for non-local frame variables.
5497 * stor-layout.c (layout_decl): Do not issue a warning for them.
5499 2018-10-19 Robin Dapp <rdapp@linux.ibm.com>
5501 * haifa-sched.c (priority): Add force_recompute parameter.
5502 (apply_replacement): Call priority () with force_recompute = true.
5503 (restore_pattern): Likewise.
5505 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
5507 * simplify-rtx.c (simplify_subreg): Limit mask of vec_merge to
5508 HOST_BITS_PER_WIDE_INT.
5509 (test_vector_ops_duplicate): Likewise.
5511 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
5514 * config/i386/sse.md (VF_AVX512): New.
5515 (avx512bcst): Likewise.
5516 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_1):
5518 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_2):
5520 (*<sd_mask_codefor>fma_fmadd_<mode><sd_maskz_name>_bcst_3):
5523 2018-10-18 Jonathan Wakely <jwakely@redhat.com>
5525 * doc/invoke.texi (-dumpversion): Improve grammar.
5526 (-dumpfullversion): Make more consistent with -dumpversion.
5528 2018-10-18 Uros Bizjak <ubizjak@gmail.com>
5530 * config/i386/i386.c (ix86_emit_fp_unordered_jump):
5531 Set JUMP_LABEL to the jump insn.
5532 (ix86_emit_i387_log1p): Use ix86_expand_branch to expand branch.
5533 Predict emitted jump and add label to jump insn.
5535 2018-10-18 David Malcolm <dmalcolm@redhat.com>
5537 PR tree-optimization/87562
5538 * input.c (get_substring_ranges_for_loc): Use
5539 LRK_SPELLING_LOCATION rather than LRK_MACRO_EXPANSION_POINT when
5540 getting the linemap for the endpoint. Verify that it's either
5541 in the same linemap as the start point's spelling location, or
5542 at least in the same file.
5544 2018-10-18 Richard Biener <rguenther@suse.de>
5546 * config/i386/i386.c (ix86_builtin_vectorization_cost): Do not
5547 feed width-specific load/store costs through ix86_vec_cost.
5548 * config/i386/x86-tune-costs.h (athlon_cost): Adjust.
5549 (k8_cost): Likewise.
5550 (bdver_cost): Likewise.
5551 (znver1_cost): Likewise.
5552 (btver1_cost): Likewise.
5553 (btver2_cost): Likewise.
5555 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
5557 * simplify-rtx.c (simplify_subreg): Call simplify_gen_subreg
5558 to simplify subreg of vec_merge.
5560 2018-10-18 Richard Biener <rguenther@suse.de>
5562 * config/i386/i386.c: Fix costing of vector FMA.
5564 2018-10-18 Richard Biener <rguenther@suse.de>
5566 * config/i386/i386.c (ix86_vec_cost): Remove !parallel path
5568 (ix86_builtin_vectorization_cost): For vec_construct properly
5569 cost insertion into SSE regs.
5570 (...): Adjust calls to ix86_vec_cost.
5572 2018-10-18 Richard Biener <rguenther@suse.de>
5576 2018-02-07 Richard Biener <rguenther@suse.de>
5578 PR tree-optimization/84204
5579 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
5582 2018-10-18 H.J. Lu <hongjiu.lu@intel.com>
5585 * simplify-rtx.c (simplify_subreg): Simplify subreg of vec_merge
5587 (test_vector_ops_duplicate): Add test for a scalar subreg of a
5588 VEC_MERGE of a VEC_DUPLICATE.
5590 2018-10-17 Joseph Myers <joseph@codesourcery.com>
5592 * doc/cpp.texi (__STDC_VERSION__): Document C2X handling.
5593 * doc/invoke.texi (-std=c2x, -std=gnu2x): Document new options.
5594 * doc/standards.texi (C Language): Document C2X.
5595 * dwarf2out.c (highest_c_language), config/rl78/rl78.c
5596 (rl78_option_override): Handle "GNU C2X" language name.
5598 2018-10-17 Joseph Myers <joseph@codesourcery.com>
5600 * doc/invoke.texi (-std=c17), doc/standards.texi (C Language):
5601 Document C17 as published in 2018.
5603 2018-10-17 Eric Botcazou <ebotcazou@adacore.com>
5606 * fold-const.c (fold_truth_andor_1): If the right side is not constant,
5607 bail out if both sides do not have the same storage order.
5609 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
5611 * bitmap.c (bitmap_head::dump): New.
5612 * bitmap.h (bitmap_head): Add dump().
5613 * gimple-ssa-evrp-analyze.c
5614 (evrp_range_analyzer::try_find_new_range): Adjust for value_range API.
5615 (evrp_range_analyzer::set_ssa_range_info): Same.
5616 (evrp_range_analyzer::record_ranges_from_phis): Same.
5617 (evrp_range_analyzer::record_ranges_from_stmt): Same.
5618 * gimple-ssa-evrp.c (evrp_dom_walker::before_dom_children): Same.
5619 * gimple-ssa-sprintf.c (get_int_range): Same.
5620 (format_integer): Same.
5621 (sprintf_dom_walker::handle_gimple_call): Same.
5622 * ipa-cp.c (ipcp_vr_lattice::meet_with_1): Same.
5623 (ipcp_vr_lattice::top_p): Same.
5624 (ipcp_vr_lattice::bottom_p): Same.
5625 (ipcp_vr_lattice::set_to_bottom): Same.
5626 (ipa_vr_operation_and_type_effects): Same.
5627 (propagate_vr_across_jump_function): Same.
5628 (ipcp_store_vr_results): Same.
5629 * ipa-prop.c (struct ipa_vr_ggc_hash_traits): Same.
5630 (ipa_print_node_jump_functions_for_edge): Same.
5631 (ipa_get_value_range): Same.
5632 (ipa_compute_jump_functions_for_edge): Same.
5633 (ipa_write_jump_function): Same.
5634 * tree-ssa-dom.c (simplify_stmt_for_jump_threading): Same.
5635 * tree-ssa-threadedge.c (record_temporary_equivalences_from_phis):
5637 * vr-values.c (set_value_range_to_nonnegative): Same.
5638 (set_value_range_to_truthvalue): Same.
5639 (vr_values::get_value_range): Same.
5640 (vr_values::set_defs_to_varying): Same.
5641 (vr_values::update_value_range): Same.
5642 (symbolic_range_based_on_p): Same.
5643 (vr_values::op_with_boolean_value_range_p): Same.
5644 (vr_values::extract_range_for_var_from_comparison_expr): Same.
5645 (vr_values::extract_range_from_ssa_name): Same.
5646 (vr_values::extract_range_from_binary_expr): Same.
5647 (vr_values::extract_range_from_unary_expr): Same.
5648 (vr_values::extract_range_from_cond_expr): Same.
5649 (vr_values::extract_range_from_comparison): Same.
5650 (vr_values::check_for_binary_op_overflow): Same.
5651 (vr_values::extract_range_basic): Same.
5652 (vr_values::extract_range_from_assignment): Same.
5653 (compare_ranges): Same.
5654 (compare_range_with_value): Same.
5655 (vr_values::adjust_range_with_scev): Same.
5656 (vrp_valueize): Same.
5657 (vrp_valueize_1): Same.
5658 (vr_values::get_vr_for_comparison): Same.
5659 (vr_values::compare_name_with_value): Same.
5660 (vr_values::compare_names): Same.
5661 (vr_values::vrp_evaluate_conditional): Same.
5662 (find_case_label_ranges): Same.
5663 (vr_values::vrp_visit_switch_stmt): Same.
5664 (vr_values::extract_range_from_phi_node): Same.
5665 (vr_values::simplify_div_or_mod_using_ranges): Same.
5666 (vr_values::simplify_bit_ops_using_ranges): Same.
5667 (test_for_singularity): Same.
5668 (range_fits_type_p): Same.
5669 (vr_values::simplify_cond_using_ranges_1): Same.
5670 (vr_values::simplify_switch_using_ranges): Same.
5671 (vr_values::simplify_float_conversion_using_ranges): Same.
5672 (vr_values::two_valued_val_range_p): Same.
5673 (vr_values::add_equivalence): Move to value_range::equiv_add.
5674 * vr-values.h (vr_values::add_equivalence): Remove.
5675 (VR_INITIALIZER): Remove.
5676 * tree-vrp.c (value_range::set): New.
5677 (value_range::equiv_add): New.
5678 (value_range::value_range): New.
5679 (value_range::deep_copy): New.
5680 (value_range::check): New.
5681 (value_range::equal_p): New.
5682 (value_range::ignore_equivs_equal_p): New.
5683 (value_range::operator==): New.
5684 (value_range::operator!=): New.
5685 (value_range::symbolic_p): New.
5686 (value_range::numeric_p): New.
5687 (value_range::set_undefined): New.
5688 (value_range::set_varying): New.
5689 (value_range::may_contain_p): New.
5690 (value_range::equiv_clear): New.
5691 (value_range::singleton_p): New.
5692 (value_range::intersect): New.
5693 (value_range::dump): New.
5694 (value_range::set_and_canonicalize): New.
5695 (set_value_range): Adjust for value_range API.
5696 (set_value_range_to_undefined): Same.
5697 (set_value_range_to_varying): Same.
5698 (set_and_canonicalize_value_range): Same.
5699 (set_value_range_to_nonnull): Same.
5700 (set_value_range_to_null): Same.
5701 (range_is_null): Same.
5702 (range_is_nonnull): Same.
5703 (range_int_cst_p): Same.
5704 (range_int_cst_singleton_p): Same.
5705 (symbolic_range_p): Same.
5706 (range_includes_zero_p): Same.
5707 (value_range_constant_singleton): Same.
5708 (vrp_set_zero_nonzero_bits): Same.
5709 (ranges_from_anti_range): Same.
5710 (extract_range_into_wide_ints): Same.
5711 (extract_range_from_multiplicative_op): Same.
5712 (set_value_range_with_overflow): Same.
5713 (extract_range_from_binary_expr_1): Same.
5714 (extract_range_from_unary_expr): Same.
5715 (dump_value_range): Same.
5716 (debug_value_range): Same.
5717 (vrp_prop::check_array_ref): Same.
5718 (vrp_prop::check_mem_ref): Same.
5719 (vrp_prop::vrp_initialize): Same.
5720 (vrp_prop::visit_stmt): Same.
5721 (intersect_ranges): Same.
5722 (vrp_prop::visit_phi): Same.
5723 (vrp_prop::vrp_finalize): Same.
5724 (determine_value_range_1): Same.
5725 (determine_value_range): Same.
5726 (vrp_intersect_ranges_1): Rename to...
5727 (vrp_intersect_1): this.
5728 (vrp_intersect_ranges): Rename to...
5729 (value_range::intersect_helper): ...this.
5730 (vrp_meet_1): Rename to...
5731 (value_range::union_helper): ...this.
5732 (vrp_meet): Rename to...
5733 (value_range::union_): ...this.
5734 (copy_value_range): Remove.
5735 * tree-vrp.h (struct value_range): Rewrite into a proper class.
5736 (value_range::vrtype): New.
5737 (value_range::type): New.
5738 (value_range::equiv): New.
5739 (value_range::min): New.
5740 (value_range::max): New.
5741 (value_range::varying_p): New.
5742 (value_range::undefined_p): New.
5743 (value_range::null_p): New.
5744 (value_range::equiv_add): New.
5745 (copy_value_range): Remove.
5747 2018-10-17 David Malcolm <dmalcolm@redhat.com>
5749 * Makefile.in (SELFTEST_TARGETS): New.
5750 (selftest) Change from s-selftest-c to $(SELFTEST_TARGETS).
5751 (C_SELFTEST_FLAGS, C_SELFTEST_DEPS, s-selftest-c, selftest-c-gdb)
5752 (selftest-gdb, selftest-c-valgrind, selftest-valgrind): Move to
5754 (CPP_SELFTEST_FLAGS, CPP_SELFTEST_DEPS, s-selftest-c++)
5755 (selftest-c++-gdb, selftest-c++-valgrind): Move to
5757 * configure: Regenerate.
5758 * configure.ac (selftest_languages): New.
5760 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
5762 * tree-vrp.c (extract_range_from_multiplicative_op): Remove
5763 overflow wraps argument.
5764 (extract_range_from_binary_expr_1): Do not pass overflow wraps to
5765 wide_int_range_multiplicative_op.
5766 * wide-int-range.cc (wide_int_range_mult_wrapping): Remove
5767 overflow wraps argument.
5768 (wide_int_range_multiplicative_op): Same.
5769 (wide_int_range_lshift): Same.
5770 (wide_int_range_div): Same.
5771 * wide-int-range.h (wide_int_range_multiplicative_op): Same.
5772 (wide_int_range_lshift): Same.
5773 (wide_int_range_div): Same.
5775 2018-10-17 Aldy Hernandez <aldyh@redhat.com>
5777 * wide-int-range.h (wide_int_range_shift_undefined_p): Adjust to
5778 use sign as argument.
5779 * tree-vrp.c (extract_range_from_binary_expr_1): Pass sign to
5780 wide_int_range_shift_undefined_p.
5782 2018-10-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5784 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>):
5786 (@despeculate_copy<ALLI_TI:mode>): ... This.
5787 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): Remove
5790 2018-10-16 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
5792 * config.gcc: Obsolete *-*-solaris2.10*.
5793 * doc/install.texi (Specific, *-*-solaris2*): Document it.
5795 2018-10-12 Jeff Law <law@redhat.com>
5797 * config/ft32/ft32.md (ft32_general_movsrc_operand): Disable
5798 reg + sym +- const_int addressing modes.
5800 2018-10-15 David Malcolm <dmalcolm@redhat.com>
5802 * common.opt (fdiagnostics-minimum-margin-width=): New option.
5803 * diagnostic-show-locus.c (layout::layout): Apply the minimum
5805 (layout::start_annotation_line): Only print up to 3 of the
5806 margin character, to avoid touching the left-hand side.
5807 (selftest::test_diagnostic_show_locus_fixit_lines): Update for
5808 minimum margin width, as set by test_diagnostic_context's ctor.
5809 (selftest::test_fixit_insert_containing_newline): Likewise.
5810 (selftest::test_fixit_insert_containing_newline_2): Likewise.
5811 (selftest::test_line_numbers_multiline_range): Clear
5812 dc.min_margin_width.
5813 * diagnostic.c (diagnostic_initialize): Initialize
5815 * diagnostic.h (struct diagnostic_context): Add field
5817 * doc/invoke.texi: Add -fdiagnostics-minimum-margin-width=.
5818 * opts.c (common_handle_option): Handle
5819 OPT_fdiagnostics_minimum_margin_width_.
5820 * selftest-diagnostic.c
5821 (selftest::test_diagnostic_context::test_diagnostic_context):
5822 Initialize min_margin_width to 6.
5823 * toplev.c (general_init): Initialize global_dc->min_margin_width.
5825 2018-10-15 David Malcolm <dmalcolm@redhat.com>
5827 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
5828 Fix usage of "error_at_rich_loc" in the comment.
5830 2018-10-15 Renlin Li <renlin.li@arm.com>
5833 * tree-vectorizer.c (try_vectorize_loop_1): Don't use
5834 if-conversioned loop when it contains ifn with types not
5835 supported by backend.
5836 * internal-fn.c (expand_direct_optab_fn): Add an assert.
5837 (direct_internal_fn_supported_p): New helper function.
5838 * internal-fn.h (direct_internal_fn_supported_p): Declare.
5840 2018-10-15 Jakub Jelinek <jakub@redhat.com>
5843 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
5844 Remove OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
5845 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
5847 2018-10-15 Bin Cheng <bin.cheng@linux.alibaba.com>
5849 PR tree-optimization/87022
5850 * tree-loop-distribution.c (pg_add_dependence_edges): Check all
5851 bits in dist vector rather than the first one.
5853 2018-10-15 Richard Biener <rguenther@suse.de>
5856 * tree-ssa-structalias.c (struct vls_data): Add escaped_p member.
5857 (visit_loadstore): When a used restrict tag escaped verify that
5858 the points-to solution of "other" pointers do not include
5860 (compute_dependence_clique): If a used restrict tag escaped
5861 communicated that down to visit_loadstore.
5863 2018-10-15 Andreas Krebbel <krebbel@linux.ibm.com>
5865 * config/s390/s390.c (s390_expand_vec_init): Force vector element
5866 into reg if it isn't a general operand.
5868 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
5871 * config/i386/sse.md (*vec_dupv2di): Add register source to
5874 2018-10-14 H.J. Lu <hongjiu.lu@intel.com>
5877 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512F_UNSET):
5878 Add OPTION_MASK_ISA_AVX512IFMA_UNSET,
5879 OPTION_MASK_ISA_AVX5124FMAPS_UNSET and
5880 OPTION_MASK_ISA_AVX5124VNNIW_UNSET.
5882 2018-10-13 Eric Botcazou <ebotcazou@adacore.com>
5884 * dwarf2cfi.c (struct dw_trace_info): Add args_size_defined_for_eh.
5885 (notice_args_size): Set it in the current trace if no insn that can
5886 throw internally has been seen yet.
5887 (connect_traces): When connecting args_size between traces, allow the
5888 incoming values not to match if there is an insn setting it before the
5889 first insn that can throw internally; in that case, force the creation
5890 of a CFI note on this latter insn.
5892 2018-10-13 Jonathan Wakely <jwakely@redhat.com>
5894 * opt-problem.h (opt_wrapper): Use template-argument-list when naming
5895 the base class, because using the injected-class-name was not clearly
5896 specified until DR 176.
5898 2018-10-12 Paul Koning <ni1d@arrl.net>
5900 * config/pdp11/pdp11.md (doloop_end): New expander.
5901 (doloop_end_insn): renamed from "doloop_end".
5902 (addqi3): New pattern.
5903 (subqi3): New pattern.
5904 * config/pdp11/predicates.md (incdec_operand): New predicate.
5906 2018-10-12 Yury Gribov <tetra2005@gmail.com>
5909 * real.c (format_helper::can_represent_integral_type_p): New function
5910 * real.h (format_helper::can_represent_integral_type_p): Ditto.
5911 * match.pd: New pattern.
5913 2018-10-12 Alexandre Oliva <oliva@adacore.com>
5915 * configure.ac: Introduce --enable-large-address-aware
5916 to define MINGW_DEFAULT_LARGE_ADDR_AWARE.
5917 * doc/install.texi: Document it.
5918 * configure, config.in: Rebuilt.
5919 * config/i386/mingw32.h (LINK_SPEC_LARGE_ADDR_AWARE): Define,
5920 based on MINGW_DEFAULT_LARGE_ADDR_AWARE.
5921 (LINK_SPEC): Insert it.
5922 * config/i386/mingw-w64.h: Likewise.
5924 * cygming.h (HAVE_GAS_ALIGNED_COMM): Fallback-define.
5926 2018-10-12 Peter Bergner <bergner@linux.ibm.com>
5928 PR rtl-optimization/87600
5929 * ira-lives (non_conflicting_reg_copy_p): Disable for non LRA targets.
5931 2018-10-12 Paul Koning <ni1d@arrl.net>
5933 * doc/md.texi (doloop_end): Document that the pattern code may
5934 need to check operand mode.
5936 2018-10-12 Wilco Dijkstra <wdijkstr@arm.com>
5938 * config/aarch64/aarch64.md (zero_extendsidi2_aarch64): Add alternatives
5939 to zero-extend between int and floating-point registers.
5940 (load_pair_zero_extendsidi2_aarch64): Add alternative for zero-extended
5941 ldp into floating-point registers. Add type and arch attributes.
5942 (zero_extend<SHORT:mode><GPI:mode>2_aarch64): Add arch attribute.
5943 Use f_loads for type attribute.
5945 2018-10-11 Martin Sebor <msebor@redhat.com>
5947 * doc/extend.texi (attribute packed): Correct typos.
5949 2018-10-11 Martin Sebor <msebor@redhat.com>
5951 * doc/extend.texi (attribute flatten): Mention interaction with
5954 2018-10-11 Jan Hubicka <hubicka@ucw.cz>
5957 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
5958 Set new_decl virtual flag to zero.
5960 2018-10-11 Martin Sebor <msebor@redhat.com>
5963 * doc/extend.texi (attribute format_arg): Discuss using multiple
5964 attributes on a single function.
5966 2018-10-11 Giuliano Belinassi <giuliano.belinassi@usp.br>
5968 PR tree-optimization/86829
5969 * match.pd (sin (atan (x))): New simplification rules.
5970 (cos (atan (x))): Likewise.
5971 * real.c (build_sinatan_real): New function.
5972 * real.h (build_sinatan_real): Prototype.
5974 2018-10-11 Will Schmidt <will_schmidt@vnet.ibm.com>
5976 * config/rs6000/rs6000.c (map_to_integral_tree_type): New helper
5978 (fold_mergeeo_helper): New helper function.
5979 (rs6000_gimple_fold_builtin): Add hooks for vec_mergee and vec_mergeo
5980 intrinsics. Correct some whitespace indentation issues.
5982 2018-10-11 Wilco Dijkstra <wdijkstr@arm.com>
5985 * config/aarch64/aarch64.c (aarch64_mask_and_shift_for_ubfiz_p):
5986 Use HOST_WIDE_INT_1U for shift.
5988 2018-10-11 Doug Rupp <rupp@adacore.com>
5989 Olivier Hainque <hainque@adacore.com>
5991 * config/rs6000/vxworks.h (VXWORKS_RELAX_LINK_SPEC): New macro.
5992 Pass --relax to the linker for RTPs.
5993 (LINK_SPEC): Append VXWORKS_RELAX_LINK_SPEC.
5995 2018-10-11 Andrew Stubbs <ams@codesourcery.com>
5996 Jan Hubicka <jh@suse.cz>
5997 Martin Jambor <mjambor@suse.cz>
5999 * print-rtl.c (print_rtx_operand_codes_E_and_V): Print how many times
6000 the same elements are repeated rather than printing all of them.
6001 * read-rtl.c (rtx_reader::read_rtx_operand): Recognize and expand
6002 "repeated" elements.
6003 * read-rtl-function.c (test_loading_repeat): New function.
6004 (read_rtl_function_c_tests): Call test_loading_repeat.
6005 * rtl-tests.c (test_dumping_repeat): New function.
6006 (rtl_tests_c_tests): Call test_dumping_repeat.
6008 2018-10-11 Richard Biener <rguenther@suse.de>
6010 * config/i386/x86-tune-costs.h (bdver?_memcpy, bdver?_memset,
6011 bdver?_cost): Unify to ...
6012 (bdver_memcpy, bdver_memset, bdver_cost): ... this.
6013 * config/i386/i386.c (processor_cost_table): Adjust.
6015 2018-10-10 Eric Botcazou <ebotcazou@adacore.com>
6018 * cgraphunit.c (cgraph_node::expand_thunk): Force DECL_IGNORED_P on
6019 the thunk when expanding to GIMPLE.
6021 2018-10-10 Bernd Edlinger <bernd.edlinger@hotmail.de>
6023 * varasm.c (mergeable_string_section): Don't try to move zero-length
6024 strings to the merge section.
6026 2018-10-10 Uros Bizjak <ubizjak@gmail.com>
6029 * config/i386/mmx.md (const_vector 0 -> mem splitter): New splitter.
6031 2018-10-10 Jakub Jelinek <jakub@redhat.com>
6034 * config/i386/i386-builtin.def (IX86_BUILTIN_RDPMC): Move from args set
6035 to special_args set.
6037 2018-10-10 Richard Biener <rguenther@suse.de>
6039 * config/i386/sse.md (reduc_plus_scal_v8df, reduc_plus_scal_v4df,
6040 reduc_plus_scal_v2df, reduc_plus_scal_v16sf, reduc_plus_scal_v8sf,
6041 reduc_plus_scal_v4sf): Merge into pattern reducing to half width
6042 and recursing and pattern terminating the recursion on SSE
6043 vector width using ix86_expand_reduc.
6044 (reduc_sminmax_scal_<mode>): Split into part reducing to half
6045 width and recursing and SSE2 vector variant doing the final
6046 reduction with ix86_expand_reduc.
6047 (reduc_uminmax_scal_<mode>): Likewise for the AVX512 variants
6048 with terminating the recursion at AVX level, splitting that
6051 2018-10-09 David Malcolm <dmalcolm@redhat.com>
6053 * genmatch.c (error_cb): Rename to...
6054 (diagnostic_cb): ...this, converting int params to enums.
6055 (fatal_at): Update for renaming.
6056 (warning_at): Likewise.
6058 * input.c (selftest::ebcdic_execution_charset::apply):
6059 Update for renaming of...
6060 (selftest::ebcdic_execution_charset::on_error): ...this, renaming
6062 (selftest::ebcdic_execution_charset::on_diagnostic): ...this,
6063 converting level and reason to enums.
6064 (class selftest::lexer_error_sink): Rename to...
6065 (class selftest::lexer_test_options): ...this, renaming field
6066 "m_errors" to "m_diagnostics".
6067 (selftest::lexer_test_options::apply): Update for renaming of...
6068 (selftest::lexer_test_options::on_error): ...this, renaming to...
6069 (selftest::lexer_test_options::on_diagnostic): ...this
6070 converting level and reason to enums.
6071 (selftest::test_lexer_string_locations_raw_string_unterminated):
6072 Update for renamings.
6073 * opth-gen.awk (struct cpp_reason_option_codes_t): Use enum for
6076 2018-10-09 Paul A. Clarke <pc@us.ibm.com>
6078 * config.gcc (powerpc*-*-*): Add pmmintrin.h to extra_headers.
6079 * config/rs6000/pmmintrin.h: New file.
6081 2018-10-09 Eric Botcazou <ebotcazou@adacore.com>
6083 PR tree-optimization/86659
6084 * gimple-match.h (gimple_match_op constructors): Initialize reverse.
6086 2018-10-09 Richard Biener <rguenther@suse.de>
6088 PR tree-optimization/63155
6089 * tree-ssa-structalias.c: Include tree-ssa.h.
6090 (get_constraint_for_ssa_var): For undefs return nothing_id.
6091 (find_func_aliases): Cleanup PHI handling.
6093 2018-10-09 Richard Biener <rguenther@suse.de>
6095 * tree-vectorizer.c (dump_stmt_cost): Fix cut&paste missing
6098 2018-10-09 Martin Liska <mliska@suse.cz>
6100 * asan.c (asan_emit_stack_protection): If a stack variable
6101 is located in a same file as current function, then emit
6102 line info into variable definition string.
6104 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
6106 * print-rtl.c (rtx_writer::print_rtx_operand_code_i): Print column
6109 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
6111 * cgraphunit.c (cgraph_node::expand_thunk): Do not force DECL_IGNORED_P
6114 2018-10-08 Eric Botcazou <ebotcazou@adacore.com>
6116 PR tree-optimization/86659
6117 * gimple-match.h (struct gimple_match_op): Add reverse field.
6118 (gimple_match_op::set_op): New overloaded method.
6119 * gimple-match-head.c (maybe_build_generic_op) <BIT_FIELD_REF>: Set
6120 the REF_REVERSE_STORAGE_ORDER flag on the value.
6121 (gimple_simplify) <GIMPLE_ASSIGN>: For BIT_FIELD_REF, propagate the
6122 REF_REVERSE_STORAGE_ORDER flag and avoid simplifying if it is set.
6124 2018-10-08 Richard Sandiford <richard.sandiford@arm.com>
6127 * gimple-ssa-backprop.c (backprop::intersect_uses): Use
6128 FOR_EACH_IMM_USE_FAST instead of FOR_EACH_IMM_USE_STMT.
6130 2018-10-08 H.J. Lu <hongjiu.lu@intel.com>
6133 * config/i386/avx512fintrin.h (_mm512_mask_fmaddsub_round_pd):
6134 Defined with __builtin_ia32_vfmaddsubpd512_mask.
6136 2018-10-08 Richard Biener <rguenther@suse.de>
6138 * config/i386/x86-tune-costs.h (znver1_cost): Make AVX256 vector loads
6139 cost the same as AVX128 ones.
6141 2018-10-08 Paul Koning <ni1d@arrl.net>
6143 * config/pdp11/pdp11-protos.h (output_block_move): Remove.
6144 (expand_block_move): New function.
6145 * config/pdp11/pdp11.c (output_block_move): Remove.
6146 (expand_block_move): New function.
6147 * config/pdp11/pdp11.h (MOVE_RATIO): New definition.
6148 * config/pdp11/pdp11.md (movmemhi): Use expand_block_move.
6149 (*movmemhi1): Remove.
6151 2018-10-08 Robin Dapp <rdapp@linux.ibm.com>
6153 * config/s390/2827.md: Increase latencies for some FP instructions.
6155 2018-10-08 Richard Biener <rguenther@suse.de>
6157 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
6159 * tree-vectorizer.c (dump_stmt_cost): Add cost param and dump it.
6160 * tree-vectorizer.h (dump_stmt_cost): Adjust.
6161 (add_stmt_cost): Dump return value of the hook.
6163 2018-10-08 Richard Biener <rguenther@suse.de>
6165 PR tree-optimization/63155
6166 * tree-ssa-propagate.c (add_ssa_edge): Do cheap check first.
6167 (ssa_propagation_engine::ssa_propagate): Remove redundant
6168 bitmap bit clearing.
6170 2018-10-05 Peter Bergner <bergner@linux.ibm.com>
6172 PR rtl-optimization/86939
6173 PR rtl-optimization/87479
6174 * ira.h (non_conflicting_reg_copy_p): New prototype.
6175 * ira-lives.c (ignore_reg_for_conflicts): New static variable.
6176 (make_hard_regno_dead): Don't add conflicts for register
6177 ignore_reg_for_conflicts.
6178 (make_object_dead): Likewise.
6179 (non_conflicting_reg_copy_p): New function.
6180 (process_bb_node_lives): Set ignore_reg_for_conflicts for copies.
6181 Remove special conflict handling of REAL_PIC_OFFSET_TABLE_REGNUM.
6182 * lra-lives.c (ignore_reg_for_conflicts): New static variable.
6183 (make_hard_regno_dead): Don't add conflicts for register
6184 ignore_reg_for_conflicts. Remove special conflict handling of
6185 REAL_PIC_OFFSET_TABLE_REGNUM. Remove now unused argument
6186 check_pic_pseudo_p and update callers.
6187 (mark_pseudo_dead): Don't add conflicts for register
6188 ignore_reg_for_conflicts.
6189 (process_bb_lives): Set ignore_reg_for_conflicts for copies.
6191 2018-10-05 Andrew Waterman <andrew@sifive.com>
6192 Jim Wilson <jimw@sifive.com>
6194 * config/riscv/riscv.md (f<quiet_pattern>_quiet<ANYF:mode><X:mode>4):
6195 Add define_expand. Add ! HONOR_SNANS check to current pattern. Add
6196 new pattern using HONOR_SNANS that emits one extra instruction.
6198 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6200 * config/rs6000/rs6000.md (unnamed mfcr scc_comparison_operator
6201 patterns): Merge SI and DI patterns to a GPR pattern.
6202 (unnamed define_insn and define_split for record form of that): Merge
6203 to a single define_insn_and_split pattern.
6205 2018-10-05 David Malcolm <dmalcolm@redhat.com>
6208 * input.c (expand_location_to_spelling_point): Add param "aspect"
6209 and use rather than hardcoding LOCATION_ASPECT_CARET.
6210 (get_substring_ranges_for_loc): Handle the case of a single token
6211 within a macro expansion.
6212 * input.h (expand_location_to_spelling_point): Add "aspect" param,
6213 defaulting to LOCATION_ASPECT_CARET.
6215 2018-10-05 Paul Koning <ni1d@arrl.net>
6217 * config/pdp11/pdp11.c (TARGET_CXX_GUARD_TYPE): Define.
6218 (TARGET_CXX_CLASS_DATA_ALWAYS_COMDAT): Define.
6219 (TARGET_CXX_LIBRARY_RTTI_COMDAT): Define.
6220 (pdp11_guard_type): New function.
6222 2018-10-05 Paul Koning <ni1d@arrl.net>
6224 * config/pdp11/pdp11.h (FLOAT_TYPE_SIZE): Always 32.
6225 * config/pdp11/pdp11.opt (mfloat32): Remove.
6227 * doc/invoke.texi (pdp11 -mfloat32): Remove:
6228 (pdp11 -mfloat64): Remove.
6230 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
6232 * config/i386/i386.md (*cmpxf_cc_i387): Remove pattern.
6233 (*cmp<mode>_cc_i387): Ditto.
6234 (*cmpu<mode>_cc_i387): Ditto.
6235 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6236 * config/i386/i386.c (ix86_expand_fp_compare): Remove
6238 <case IX86_FPCMP_SAHF>: Do not generate pattern with HImode clobber.
6239 Emit x86_sahf_1 pattern.
6240 (ix86_expand_compare): Update call to ix86_expand_fp_compare.
6241 (ix86_expand_carry_flag_compare): Ditto.
6243 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
6245 * config/i386/i386.md (*cmpxf_i387): Change operand 2 predicate
6246 to reg_or_0_operand. Add "C" constraint.
6247 (*cmpxf_cc_i387): Ditto.
6248 (*cmp<mode>_i387): Change operand 2 predicate
6249 to nonimm_or_0_operand. Add "C" constraint.
6250 (*cmp<mode>_cc_i387): Ditto.
6251 (*cmp<mode>_0_i387): Remove insn pattern.
6252 (*cmp<mode>_0_cc_i387): Ditto.
6254 2018-10-05 Uros Bizjak <ubizjak@gmail.com>
6256 * config/i386/constraints.md ("C"): Do not depend on TARGET_SSE.
6257 * config/i386/predicates.md (nonimm_or_0_operand): Rename
6258 from vector_move_operand. Update all uses.
6260 2018-10-05 Martin Sebor <msebor@redhat.com>
6262 PR tree-optimization/87490
6263 * builtins.c (expand_builtin_strnlen): Handle a null data.decl
6266 2018-10-05 Richard Biener <rguenther@suse.de>
6268 PR tree-optimization/63155
6269 * tree-ssa-ccp.c (ccp_propagate::visit_phi): Avoid excess
6270 vertical space in dumpfiles.
6271 * tree-ssa-propagate.h
6272 (ssa_propagation_engine::process_ssa_edge_worklist): Remove.
6273 * tree-ssa-propagate.c (cfg_blocks_back): New global.
6274 (ssa_edge_worklist_back): Likewise.
6275 (curr_order): Likewise.
6276 (cfg_blocks_get): Remove abstraction.
6277 (cfg_blocks_add): Likewise.
6278 (cfg_blocks_empty_p): Likewise.
6279 (add_ssa_edge): Add to current or next worklist based on
6281 (add_control_edge): Likewise.
6282 (ssa_propagation_engine::process_ssa_edge_worklist): Fold
6284 (ssa_propagation_engine::ssa_propagate): ... here. Unify
6285 iteration from CFG and SSA edge worklist so we process
6286 everything in RPO order, prioritizing forward progress
6288 (ssa_prop_init): Allocate new worklists, do not dump
6290 (ssa_prop_fini): Free new worklists.
6292 2018-10-05 Richard Biener <rguenther@suse.de>
6294 * tree-core.h (tree_block::abstract_flag): Remove.
6295 (tree_block::block_num): Make full 32bits.
6296 * tree.def (BLOCK): Remove docs about BLOCK_ABSTRACT.
6297 * tree.h (BLOCK_ABSTRACT): Remove.
6298 * dwarf2out.c (gen_lexical_block_die): Remove dead code
6299 resulting from BLOCK_ABSTRACT being always false.
6300 (gen_inlined_subroutine_die): Likewise.
6301 (gen_block_die): Likewise.
6302 * tree.c (block_ultimate_origin): Likewise.
6303 * tree-pretty-print.c (dump_block_node): Remove code dealing
6304 with BLOCK_ABSTRACT.
6305 * tree-ssa-live.c (dump_scope_block): Likewise.
6306 * tree-streamer-in.c (unpack_ts_block_value_fields): Likewise.
6307 * tree-streamer-out.c (pack_ts_block_value_fields): Likewise.
6309 2018-10-05 Richard Biener <rguenther@suse.de>
6311 * config/i386/i386.c (ix86_add_stmt_cost): When scalar cost
6312 is asked for initialize mode to the component mode of the
6315 2018-10-05 H.J. Lu <hongjiu.lu@intel.com>
6318 * config/i386/gnu-user.h (ASM_SPEC): Don't pass -msse2avx to
6319 assembler for -mavx.
6320 * config/i386/gnu-user64.h (ASM_SPEC): Likewise.
6322 2018-10-05 Segher Boessenkool <segher@kernel.crashing.org>
6325 * config/rs6000/rs6000-builtin.def (RS6000_BUILTIN_SET_FPSCR_DRN): Use
6327 * config/rs6000/rs6000.md (rs6000_set_fpscr_rn): Require the operand
6328 to be DImode. When using mffscrn, force the operand to a register.
6330 2018-10-04 Uros Bizjak <ubizjak@gmail.com>
6332 * config/i386/i386.md (*fop_<X87MODEF:mode>_2_i387): Macroize insn
6333 from *fop_<MODEF:mode>_2_i387 and *fop_xf_2_i387 using
6334 X87MODEF mode iterator.
6335 (*fop_<X87MODEF:mode>_3_i387): Macroize insn from
6336 *fop_<MODEF:mode>_3_i387 and *fop_xf_3_i387 using
6337 X87MODEF mode iterator.
6339 2018-10-04 Vinay Kumar <vinay.kumar@blackfigtech.com>
6341 * doc/invoke.texi (-Wno-prio-ctor-dtor): Document new warning
6342 -Wno-prio-ctor-dtor.
6344 2018-10-04 David Malcolm <dmalcolm@redhat.com>
6346 * Makefile.in (OBJS): Add opt-problem.o.
6347 * dump-context.h: Include "selftest.h.
6348 (selftest::temp_dump_context): New forward decl.
6349 (class dump_context): Make friend of class
6350 selftest::temp_dump_context.
6351 (dump_context::dump_loc_immediate): New decl.
6352 (class dump_pretty_printer): Move here from dumpfile.c.
6353 (class temp_dump_context): Move to namespace selftest.
6354 (temp_dump_context::temp_dump_context): Add param
6355 "forcibly_enable_dumping".
6356 (selftest::verify_dumped_text):
6357 (ASSERT_DUMPED_TEXT_EQ): Move here from dumpfile.c.
6358 (selftest::verify_item):
6359 (ASSERT_IS_TEXT): Move here from dumpfile.c.
6360 (ASSERT_IS_TREE): Likewise.
6361 (ASSERT_IS_GIMPLE): Likewise.
6362 * dumpfile.c (dump_context::dump_loc): Move immediate dumping
6364 (dump_context::dump_loc_immediate): ...this new function.
6365 (class dump_pretty_printer): Move to dump-context.h.
6366 (dump_switch_p_1): Don't enable MSG_PRIORITY_REEMITTED.
6367 (opt_info_switch_p_1): Enable MSG_PRIORITY_REEMITTED.
6368 (temp_dump_context::temp_dump_context): Move to "selftest"
6369 namespace. Add param "forcibly_enable_dumping", and use it to
6370 conditionalize the use of m_pp;
6371 (selftest::verify_dumped_text): Make non-static.
6372 (ASSERT_DUMPED_TEXT_EQ): Move to dump-context.h.
6373 (selftest::verify_item): Make non-static.
6374 (ASSERT_IS_TEXT): Move to dump-context.h.
6375 (ASSERT_IS_TREE): Likewise.
6376 (ASSERT_IS_GIMPLE): Likewise.
6377 (selftest::test_capture_of_dump_calls): Pass "true" for new
6378 param of temp_dump_context.
6379 * dumpfile.h (enum dump_flag): Add MSG_PRIORITY_REEMITTED, adding
6380 it to MSG_ALL_PRIORITIES. Update values of TDF_COMPARE_DEBUG and
6382 * opt-problem.cc: New file.
6383 * opt-problem.h: New file.
6384 * optinfo-emit-json.cc
6385 (selftest::test_building_json_from_dump_calls): Pass "true" for
6386 new param of temp_dump_context.
6387 * optinfo.cc (optinfo_kind_to_dump_flag): New function.
6388 (optinfo::emit_for_opt_problem): New function.
6389 (optinfo::emit): Clarity which emit_item is used.
6390 * optinfo.h (optinfo::get_dump_location): New accessor.
6391 (optinfo::emit_for_opt_problem): New decl.
6392 (optinfo::emit): Make const.
6393 * selftest-run-tests.c (selftest::run_tests): Call
6394 selftest::opt_problem_cc_tests.
6395 * selftest.h (selftest::opt_problem_cc_tests): New decl.
6396 * tree-data-ref.c (dr_analyze_innermost): Convert return type from
6397 bool to opt_result, converting fprintf messages to
6398 opt_result::failure_at calls. Add "stmt" param for use by the
6400 (create_data_ref): Pass "stmt" to the dr_analyze_innermost call.
6401 (runtime_alias_check_p): Convert return type from bool to
6402 opt_result, converting dump_printf calls to
6403 opt_result::failure_at, using the statement DDR_A for their
6405 (find_data_references_in_stmt): Convert return type from bool to
6406 opt_result, converting "return false" to opt_result::failure_at
6408 * tree-data-ref.h: Include "opt-problem.h".
6409 (dr_analyze_innermost): Convert return type from bool to opt_result,
6410 and add a const gimple * param.
6411 (find_data_references_in_stmt): Convert return type from bool to
6413 (runtime_alias_check_p): Likewise.
6414 * tree-predcom.c (find_looparound_phi): Pass "init_stmt" to
6415 dr_analyze_innermost.
6416 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test):
6417 Convert return type from bool to opt_result, adding a message for
6418 the PARAM_VECT_MAX_VERSION_FOR_ALIAS_CHECKS zero case.
6419 (vect_analyze_data_ref_dependence): Convert return type from bool
6420 to opt_result. Change sense of return type from "false"
6421 effectively meaning "no problems" to "false" meaning a problem,
6422 so that "return false" becomes "return opt_result::success".
6423 Convert "return true" calls to opt_result::failure_at, using
6424 the location of statement A rather than vect_location.
6425 (vect_analyze_data_ref_dependences): Convert return type from bool
6427 (verify_data_ref_alignment): Likewise, converting dump_printf_loc
6428 calls to opt_result::failure_at, using the stmt location rather
6430 (vect_verify_datarefs_alignment): Convert return type from bool
6432 (vect_enhance_data_refs_alignment): Likewise. Split local "stat"
6433 into multiple more-tightly-scoped copies.
6434 (vect_analyze_data_refs_alignment): Convert return type from bool
6436 (vect_analyze_data_ref_accesses): Likewise, converting a
6437 "return false" to a "return opt_result::failure_at", adding a
6439 (vect_prune_runtime_alias_test_list): Convert return type from
6440 bool to opt_result, converting dump_printf_loc to
6441 opt_result::failure_at. Add a %G to show the pertinent statement,
6442 and use the stmt's location rather than vect_location.
6443 (vect_find_stmt_data_reference): Convert return type from
6444 bool to opt_result, converting dump_printf_loc to
6445 opt_result::failure_at, using stmt's location.
6446 (vect_analyze_data_refs): Convert return type from bool to
6447 opt_result. Convert "return false" to "return
6448 opt_result::failure_at", adding messages as needed.
6449 * tree-vect-loop.c (vect_determine_vf_for_stmt_1): Convert return
6450 type from bool to opt_result.
6451 (vect_determine_vf_for_stmt): Likewise.
6452 (vect_determine_vectorization_factor): Likewise, converting
6453 dump_printf_loc to opt_result::failure_at, using location of phi
6454 rather than vect_location.
6455 (vect_analyze_loop_form_1): Convert return type from bool to
6456 opt_result, converting dump_printf_loc calls, retaining the use of
6458 (vect_analyze_loop_form): Convert return type from loop_vec_info
6459 to opt_loop_vec_info.
6460 (vect_analyze_loop_operations): Convert return type from bool to
6461 opt_result, converting dump_printf_loc calls, using the location
6462 of phi/stmt rather than vect_location where available. Convert
6463 various "return false" to "return opt_result::failure_at" with
6464 "unsupported phi" messages.
6465 (vect_get_datarefs_in_loop): Convert return type from bool to
6466 opt_result. Add a message for the
6467 PARAM_LOOP_MAX_DATAREFS_FOR_DATADEPS failure.
6468 (vect_analyze_loop_2): Convert return type from bool to
6469 opt_result. Ensure "ok" is set to a opt_result::failure_at before
6470 each "goto again;", adding new messages where needed.
6471 Add "unsupported grouped {store|load}" messages.
6472 (vect_analyze_loop): Convert return type from loop_vec_info to
6474 * tree-vect-slp.c (vect_analyze_slp): Convert return type from
6476 * tree-vect-stmts.c (process_use): Likewise, converting
6477 dump_printf_loc call and using stmt location, rather than
6479 (vect_mark_stmts_to_be_vectorized): Likeise.
6480 (vect_analyze_stmt): Likewise, adding a %G.
6481 (vect_get_vector_types_for_stmt): Convert return type from bool to
6482 opt_result, converting dump_printf_loc calls and using stmt
6483 location, rather than vect_location.
6484 (vect_get_mask_type_for_stmt): Convert return type from tree to
6485 opt_tree, converting dump_printf_loc calls and using stmt location.
6486 * tree-vectorizer.c: Include "opt-problem.h.
6487 (try_vectorize_loop_1): Flag "Analyzing loop at" dump message as
6488 MSG_PRIORITY_INTERNALS. Convert local "loop_vinfo" from
6489 loop_vec_info to opt_loop_vec_info. If if fails, and dumping is
6490 enabled, use it to report at the top level "couldn't vectorize
6491 loop" followed by the problem.
6492 * tree-vectorizer.h (opt_loop_vec_info): New typedef.
6493 (vect_mark_stmts_to_be_vectorized): Convert return type from bool
6495 (vect_analyze_stmt): Likewise.
6496 (vect_get_vector_types_for_stmt): Likewise.
6497 (tree vect_get_mask_type_for_stmt): Likewise.
6498 (vect_analyze_data_ref_dependences): Likewise.
6499 (vect_enhance_data_refs_alignment): Likewise.
6500 (vect_analyze_data_refs_alignment): Likewise.
6501 (vect_verify_datarefs_alignment): Likewise.
6502 (vect_analyze_data_ref_accesses): Likewise.
6503 (vect_prune_runtime_alias_test_list): Likewise.
6504 (vect_find_stmt_data_reference): Likewise.
6505 (vect_analyze_data_refs): Likewise.
6506 (vect_analyze_loop): Convert return type from loop_vec_info to
6508 (vect_analyze_loop_form): Likewise.
6509 (vect_analyze_slp): Convert return type from bool to opt_result.
6511 2018-10-04 David Malcolm <dmalcolm@redhat.com>
6513 * doc/invoke.texi (-fopt-info): Document new "internals"
6515 * dump-context.h (dump_context::apply_dump_filter_p): New decl.
6516 * dumpfile.c (dump_options): Update for renaming of MSG_ALL to
6518 (optinfo_verbosity_options): Add "internals".
6519 (kind_as_string): Update for renaming of MSG_ALL to MSG_ALL_KINDS.
6520 (dump_context::apply_dump_filter_p): New member function.
6521 (dump_context::dump_loc): Use apply_dump_filter_p rather than
6522 explicitly masking the dump_kind.
6523 (dump_context::begin_scope): Increment the scope depth first. Use
6524 apply_dump_filter_p rather than explicitly masking the dump_kind.
6525 (dump_context::emit_item): Use apply_dump_filter_p rather than
6526 explicitly masking the dump_kind.
6527 (dump_dec): Likewise.
6528 (dump_hex): Likewise.
6529 (dump_switch_p_1): Default to MSG_ALL_PRIORITIES.
6530 (opt_info_switch_p_1): Default to MSG_PRIORITY_USER_FACING.
6531 (opt_info_switch_p): Update handling of default
6532 MSG_OPTIMIZED_LOCATIONS to cope with default of
6533 MSG_PRIORITY_USER_FACING.
6534 (dump_basic_block): Use apply_dump_filter_p rather than explicitly
6535 masking the dump_kind.
6536 (selftest::test_capture_of_dump_calls): Update test_dump_context
6537 instances to use MSG_ALL_KINDS | MSG_PRIORITY_USER_FACING rather
6538 than MSG_ALL. Generalize scope test to be run at all four
6539 combinations of with/without MSG_PRIORITY_USER_FACING and
6540 MSG_PRIORITY_INTERNALS, adding examples of explicit priority
6541 for each of the two values.
6542 * dumpfile.h (enum dump_flag): Add comment about the MSG_* flags.
6543 Rename MSG_ALL to MSG_ALL_KINDS. Add MSG_PRIORITY_USER_FACING,
6544 MSG_PRIORITY_INTERNALS, and MSG_ALL_PRIORITIES, updating the
6545 values for TDF_COMPARE_DEBUG and TDF_ALL_VALUES.
6546 (AUTO_DUMP_SCOPE): Add a note to the comment about the interaction
6547 with MSG_PRIORITY_*.
6548 * tree-vect-loop-manip.c (vect_loop_versioning): Mark versioning
6549 dump messages as MSG_PRIORITY_USER_FACING.
6550 * tree-vectorizer.h (DUMP_VECT_SCOPE): Add a note to the comment
6551 about the interaction with MSG_PRIORITY_*.
6553 2018-10-04 Bernd Edlinger <bernd.edlinger@hotmail.de>
6555 * varasm.c (output_constant): Add new parameter merge_strings.
6556 Make strings properly zero terminated in merge string sections.
6557 (mergeable_string_section): Don't fail if the last char is non-zero.
6558 (assemble_variable_contents): Handle merge string sections.
6559 (assemble_variable): Likewise.
6560 (assemble_constant_contents): Likewise.
6561 (output_constant_def_contents): Likewise.
6562 (output_constructor_array_range,
6563 output_constructor_regular_field): Adjust call to output_constant.
6564 (output_object_block): Adjust call to assemble_constant_contents
6565 and assemble_variable_contents.
6567 2018-10-04 Martin Liska <mliska@suse.cz>
6570 * cgraphunit.c (process_function_and_variable_attributes):
6571 Warn about a function with alias attribute and a body.
6573 2018-10-04 Martin Liska <mliska@suse.cz>
6576 * multiple_target.c (redirect_to_specific_clone): New function.
6577 (ipa_target_clone): Use it.
6578 * tree-inline.c: Fix comment.
6580 2018-10-04 David Malcolm <dmalcolm@redhat.com>
6582 * dumpfile.c (gcc::dump_manager::dump_manager): Initialize new
6584 (gcc::dump_manager::~dump_manager): Free m_optinfo_filename.
6585 (gcc::dump_manager::register_pass): New member function, adapted
6586 from loop body in gcc::pass_manager::register_pass, adding a
6587 call to update_dfi_for_opt_info.
6588 (gcc::dump_manager::opt_info_enable_passes): Store the
6589 -fopt-info options into the new fields. Move the loop
6591 (gcc::dump_manager::update_dfi_for_opt_info): ...this new member
6593 * dumpfile.h (struct opt_pass): New forward decl.
6594 (gcc::dump_manager::register_pass): New decl.
6595 (gcc::dump_manager::update_dfi_for_opt_info): New decl.
6596 (class gcc::dump_manager): Add fields "m_optgroup_flags",
6597 "m_optinfo_flags", and "m_optinfo_filename".
6598 * passes.c (gcc::pass_manager::register_pass): Move all of the
6599 dump-handling code to gcc::dump_manager::register_pass.
6601 2018-10-04 Peter Bergner <bergner@linux.ibm.com>
6603 PR rtl-optimization/87466
6604 * target.def (setjmp_preserves_nonvolatile_regs_p): New target hook.
6605 * doc/tm.texi.in (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P): New hook.
6606 * doc/tm.texi: Regenerate.
6607 * ira-lives.c (process_bb_node_lives): Use the new target hook.
6608 * lra-lives.c (process_bb_lives): Likewise.
6609 * config/rs6000/rs6000.c (TARGET_SETJMP_PRESERVES_NONVOLATILE_REGS_P):
6612 2018-10-04 Tamar Christina <tamar.christina@arm.com>
6614 * params.c (add_params): Fix initialization.
6616 2018-10-04 Martin Liska <mliska@suse.cz>
6618 PR gcov-profile/84107
6619 * tree-profile.c (init_ic_make_global_vars):
6620 Remove ic_void_ptr_var and ic_gcov_type_ptr_var.
6621 Come up with new ic_tuple* variables. Emit
6622 __gcov_indirect_call{,_topn} variables.
6623 (gimple_gen_ic_profiler): Access the variable
6625 (gimple_gen_ic_func_profiler): Access
6626 __gcov_indirect_call.callee field.
6627 (gimple_init_gcov_profiler): Use ptr_type_node.
6628 * value-prof.c (gimple_ic): Use ptr_type_node.
6630 2018-10-04 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6632 PR tree-optimization/85787
6633 * ipa-pure-const.c (malloc_candidate_p_1): Move most of
6634 malloc_candidate_p into this function and add support for
6635 detecting multiple phis.
6636 (DUMP_AND_RETURN): Move from malloc_candidate_p into top-level macro.
6638 2018-10-04 Martin Liska <mliska@suse.cz>
6641 * ipa-inline.c (inline_to_all_callers_1):
6642 Call ultimate_alias_target for node being inlined.
6644 2018-10-03 Jeff Law <law@redhat.com>
6646 * gimple-ssa-sprintf.c (format_string): Do not hardcode size of
6648 * tree.c (get_typenode_from_name): Moved from fortran/trans-types.c.
6649 * tree.h (get_typenode_from_name): Prototype.
6651 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
6653 * config/i386/i386.md (*cmp<X87MODEF:mode>_<SWI24:mode>_i387):
6654 Change operand 2 predicate to nonimmediate_operand.
6655 (*cmp<X87MODEF:mode>_<SWI24:mode>_cc_i387): Ditto.
6657 2018-10-03 Martin Sebor <msebor@redhat.com>
6658 Jeff Law <law@redhat.com>
6660 * gimple-ssa-sprintf.c (struct fmtresult): Add new member and
6662 (get_string_length): Detect unterminated arrays.
6663 (format_string): Same.
6664 (format_directive): Warn about unterminated arrays.
6665 (handle_gimple_call): Mark statements with no_warning as needed.
6667 2018-10-03 Jim Wilson <jimw@sifive.com>
6669 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): For ABI_ILP32E,
6670 also define __riscv_abi_rve. Delete trailing white space.
6672 2018-10-03 Paul Koning <ni1d@arrl.net>
6674 Enable LRA register allocator for PDP11.
6675 * config/pdp11/constraints.md (Q): Use define_memory_constraint.
6678 * config/pdp11/pdp11.c (pdp11_lra_p): New function.
6679 * config/pdp11/pdp11.opt (-mlra): New option.
6680 * doc/invoke.texi (PDP-11 Options): Document -mlra.
6682 2018-10-03 Uros Bizjak <ubizjak@gmail.com>
6684 * config/i386/i386.md (*<absneg:code>extendsfdf2): Remove.
6685 (*<absneg:code>extend<mode>xf2): Ditto.
6687 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
6689 PR tree-optimization/87415
6690 * tree-vrp.c (set_value_range_with_overflow): Special case one bit
6693 2018-10-02 Jeff Law <law@redhat.com>
6695 * gimple-fold.c (get_range_strlen): Only set *nonstr when
6696 an unterminated string is discovered. Bubble up range
6697 even for unterminated strings.
6698 (gimple_fold_builtin_strlen): Do not fold if get_range_strlen
6699 indicates the string was not terminated via NONSTR.
6701 2018-10-03 Aldy Hernandez <aldyh@redhat.com>
6703 * tree-vrp.c (extract_range_from_unary_expr): Special case all
6704 pointer conversions.
6705 Do not do anything special for anti-ranges.
6707 2018-10-03 Jérôme Lambourg <lambourg@adacore.com>
6709 * config/arm/vxworks.h (ARM_TARGET2_DWARF_FORMAT): Adjust to
6710 DW_EH_PE_pcrel | DW_EH_PE_indirect for RTPs.
6712 2018-10-03 Martin Liska <mliska@suse.cz>
6714 PR gcov-profile/86109
6715 * coverage.c (coverage_begin_function): Do not
6716 mark lambdas as artificial.
6717 * tree-core.h (struct GTY): Remove tm_clone_flag
6718 and introduce new lambda_function.
6719 * tree.h (DECL_LAMBDA_FUNCTION): New macro.
6721 2018-10-02 Aaron Sawdey <acsawdey@linux.ibm.com>
6724 * config/rs6000/rs6000-string.c (expand_strn_compare): Check that both
6725 P8_VECTOR and VSX are enabled.
6727 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
6729 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
6730 0x3907 as CPU model number.
6732 2018-10-02 Andreas Krebbel <krebbel@linux.ibm.com>
6734 * common/config/s390/s390-common.c: Rename PF_ARCH12 to PF_Z14.
6735 * config/s390/s390.h (enum processor_flags): Rename PF_ARCH12 to
6736 PF_Z14. Rename TARGET_CPU_ARCH12 to TARGET_CPU_Z14,
6737 TARGET_CPU_ARCH12_P to TARGET_CPU_Z14_P, TARGET_ARCH12 to
6738 TARGET_Z14, and TARGET_ARCH12_P to TARGET_Z14_P.
6739 * config/s390/s390.md: Likewise. Rename also the cpu attribute
6740 value from arch12 to z14.
6742 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
6744 * config/i386/i386.md (fxam<mode>2_i387_with_temp): Remove.
6746 (isinf<mode>2): Ditto.
6748 2018-10-02 Uros Bizjak <ubizjak@gmail.com>
6750 * config/i386/i386.c (ix86_emit_i387_round): Extend op1 to XFmode
6751 before emitting fxam. Perform calculations in XFmode.
6753 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
6755 * match.pd (((X /[ex] A) +- B) * A): New transformation.
6757 2018-10-02 Marc Glisse <marc.glisse@inria.fr>
6760 * fold-const.c (fold_plusminus_mult_expr): Handle complex and vectors.
6761 * tree.c (signed_or_unsigned_type_for): Handle complex.
6763 2018-10-02 Jeff Law <law@redhat.com>
6765 * gimple-fold.c (get_range_strlen): Remove dead code.
6767 2018-10-02 Martin Sebor <msebor@redhat.com>
6768 Jeff Law <law@redhat.com>
6770 * builtins.c (unterminated_array): Add new arguments.
6771 If argument is not terminated, bubble up size and exact
6773 (expand_builtin_strnlen): Detect, avoid expanding
6774 and diagnose unterminated arrays.
6775 (c_strlen): Fill in offset of start of unterminated strings.
6776 * builtins.h (unterminated_array): Update prototype.
6778 2018-10-02 Richard Biener <rguenther@suse.de>
6780 * config/i386/sse.md (reduc_plus_scal_v4df): Avoid the use
6781 of haddv4df, first reduce to SSE width and exploit the fact
6782 that we only need element zero with the reduction result.
6783 (reduc_plus_scal_v2df): Likewise.
6785 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
6787 * dojump.h (do_jump): Delete.
6788 (do_jump_1): Likewise.
6789 (split_comparison): Move around.
6790 * dojump.c (do_jump): Make static.
6791 (do_jump_1): Likewise.
6792 (jumpifnot): Move around.
6793 (jumpifnot_1): Likewise.
6795 (jumpif_1): Likewise.
6796 * expr.c (expand_expr_real_1): Call jumpif[not] instead of do_jump.
6798 2018-10-02 Eric Botcazou <ebotcazou@adacore.com>
6800 * reorg.c (make_return_insns): Use emit_copy_of_insn_after for the
6801 insns in the delay slot and add_insn_after for the jump insn.
6803 2018-10-02 Richard Biener <rguenther@suse.de>
6805 * tree-inline.c (expand_call_inline): Use the location of
6806 the callee declaration for the inline-entry marker.
6807 * final.c (notice_source_line): Remove special-casing of
6808 NOTE_INSN_INLINE_ENTRY.
6810 2018-10-01 Carl Love <cel@us.ibm.com>
6813 * config/rs6000/rs6000-builtin.def (__builtin_mffsl): New.
6814 (__builtin_mtfsb0): New.
6815 (__builtin_mtfsb1): New.
6816 ( __builtin_set_fpscr_rn): New.
6817 (__builtin_set_fpscr_drn): New.
6818 * config/rs6000/rs6000.c (rs6000_expand_mtfsb_builtin): Add.
6819 (rs6000_expand_set_fpscr_rn_builtin): Add.
6820 (rs6000_expand_set_fpscr_drn_builtin): Add.
6821 (rs6000_expand_builtin): Add case statement entries for
6822 RS6000_BUILTIN_MTFSB0, RS6000_BUILTIN_MTFSB1,
6823 RS6000_BUILTIN_SET_FPSCR_RN, RS6000_BUILTIN_SET_FPSCR_DRN,
6824 RS6000_BUILTIN_MFFSL.
6825 (rs6000_init_builtins): Add ftype initialization and def_builtin
6826 calls for __builtin_mffsl, __builtin_mtfsb0, __builtin_mtfsb1,
6827 __builtin_set_fpscr_rn, __builtin_set_fpscr_drn.
6828 * config/rs6000.md (rs6000_mtfsb0, rs6000_mtfsb1, rs6000_mffscrn,
6829 rs6000_mffscdrn): Add define_insn.
6830 (rs6000_set_fpscr_rn, rs6000_set_fpscr_drn): Add define_expand.
6831 * doc/extend.texi: Add documentation for the builtins.
6833 2018-10-01 Richard Biener <rguenther@suse.de>
6835 PR tree-optimization/87465
6836 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Fix typo
6837 causing branch miscounts.
6839 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6841 * common/config/aarch64/aarch64-common.c (TARGET_OPTION_DEFAULT_PARAM,
6842 aarch64_option_default_param): New.
6843 (params.h): Include.
6844 (TARGET_OPTION_VALIDATE_PARAM, aarch64_option_validate_param): New.
6845 * config/aarch64/aarch64.c (aarch64_override_options_internal): Simplify
6846 stack-clash protection validation code.
6848 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6850 * params.c (validate_param): New.
6851 (add_params): Use it.
6852 (set_param_value): Refactor param validation into validate_param.
6853 (diagnostic.h): Include.
6854 * diagnostic.h (diagnostic_ready_p): New.
6856 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6858 * params.c (set_param_value):
6859 Add index of parameter being validated.
6860 * common/common-target.def (option_validate_param): New.
6861 * common/common-targhooks.h (default_option_validate_param): New.
6862 * common/common-targhooks.c (default_option_validate_param): New.
6863 * doc/tm.texi.in (TARGET_OPTION_VALIDATE_PARAM): New.
6864 * doc/tm.texi: Regenerate.
6866 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6869 * config/aarch64/aarch64.c (aarch64_override_options_internal):
6870 Add validation for stack-clash parameters and set defaults.
6872 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6875 * configure.ac: Add stack-clash-protection-guard-size.
6876 * doc/install.texi: Document it.
6877 * config.in (DEFAULT_STK_CLASH_GUARD_SIZE): New.
6878 * params.def: Update comment for guard-size.
6879 (PARAM_STACK_CLASH_PROTECTION_GUARD_SIZE,
6880 PARAM_STACK_CLASH_PROTECTION_PROBE_INTERVAL): Update description.
6881 * configure: Regenerate.
6883 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6886 * config/aarch64/aarch64.h (STACK_CLASH_MIN_BYTES_OUTGOING_ARGS,
6887 STACK_DYNAMIC_OFFSET): New.
6888 * config/aarch64/aarch64.c (aarch64_layout_frame):
6889 Update outgoing args size.
6890 (aarch64_stack_clash_protection_alloca_probe_range,
6891 TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
6893 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6896 * explow.c (anti_adjust_stack_and_probe_stack_clash): Support custom
6898 * target.def (stack_clash_protection_alloca_probe_range): New.
6899 (stack_clash_protection_final_dynamic_probe): Remove.
6900 * targhooks.h (default_stack_clash_protection_alloca_probe_range) New.
6901 (default_stack_clash_protection_final_dynamic_probe): Remove.
6902 * targhooks.c: Likewise.
6903 * doc/tm.texi.in (TARGET_STACK_CLASH_PROTECTION_ALLOCA_PROBE_RANGE): New.
6904 (TARGET_STACK_CLASH_PROTECTION_FINAL_DYNAMIC_PROBE): Remove.
6905 * doc/tm.texi: Regenerate.
6907 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6910 * config/aarch64/aarch64-protos.h (aarch64_output_probe_sve_stack_clash): New.
6911 * config/aarch64/aarch64.c (aarch64_output_probe_sve_stack_clash,
6912 aarch64_clamp_to_uimm12_shift): New.
6913 (aarch64_allocate_and_probe_stack_space): Add SVE specific section.
6914 * config/aarch64/aarch64.md (probe_sve_stack_clash): New.
6916 2018-10-01 Tamar Christina <tamar.christina@arm.com>
6919 * config/aarch64/aarch64.c (aarch64_layout_frame): Add assert.
6921 2018-10-01 Jeff Law <law@redhat.com>
6922 Richard Sandiford <richard.sandiford@linaro.org>
6923 Tamar Christina <tamar.christina@arm.com>
6926 * config/aarch64/aarch64.md
6927 (probe_stack_range): Add k (SP) constraint.
6928 * config/aarch64/aarch64.h (STACK_CLASH_CALLER_GUARD,
6929 STACK_CLASH_MAX_UNROLL_PAGES): New.
6930 * config/aarch64/aarch64.c (aarch64_output_probe_stack_range): Emit
6931 stack probes for stack clash.
6932 (aarch64_allocate_and_probe_stack_space): New.
6933 (aarch64_expand_prologue): Use it.
6934 (aarch64_expand_epilogue): Likewise and update IP regs re-use criteria.
6935 (aarch64_sub_sp): Add emit_move_imm optional param.
6937 2018-10-01 MCC CS <deswurstes@users.noreply.github.com>
6939 PR tree-optimization/87261
6940 * match.pd: Remove trailing whitespace.
6941 Add (x & y) | ~(x | y) -> ~(x ^ y),
6942 (~x | y) ^ (x ^ y) -> x | ~y and (x ^ y) | ~(x | y) -> ~(x & y)
6944 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
6946 * config/arc/arc.md (*add_n): Clean up pattern, update instruction
6948 (ashlsi3_insn): Update instruction constraints.
6949 (ashrsi3_insn): Likewise.
6950 (rotrsi3): Likewise.
6951 (add_shift): Likewise.
6952 * config/arc/constraints.md (Csz): New 32 bit constraint. It
6953 avoids placing in the limm field small constants which, otherwise,
6954 could end into a small instruction.
6956 2018-10-01 Claudiu Zissulescu <claziss@synopsys.com>
6958 * config/arc/arc.md (maddsidi4_split): Don't use dmac if the
6959 destination register is not odd-even.
6960 (umaddsidi4_split): Likewise.
6962 2018-10-01 Richard Biener <rguenther@suse.de>
6964 * tree-inline.c (expand_call_inline): Store origin of fn
6965 in BLOCK_ABSTRACT_ORIGIN for the inline BLOCK.
6966 * tree.c (block_ultimate_origin): Simplify and do some
6969 2018-09-30 Uros Bizjak <ubizjak@gmail.com>
6971 * config/i386/mmx.md (EMMS): New int iterator.
6972 (emms): New int attribute.
6973 (mmx_<emms>): Macroize insn from *mmx_emms and *mmx_femms using
6974 EMMS int iterator. Explicitly declare clobbers.
6975 (mmx_emms): Remove expander.
6977 * config/i386/predicates.md (emms_operation): Remove predicate.
6978 (vzeroall_pattern): New predicate.
6979 (vzeroupper_pattern): Rename from vzeroupper_operation.
6980 * config/i386/i386.c (ix86_avx_u128_mode_after): Use
6981 vzeroupper_pattern and vzeroall_pattern predicates.
6983 2018-09-30 Peter Bergner <bergner@linux.ibm.com>
6985 PR rtl-optimization/86939
6986 * ira-lives.c (make_hard_regno_born): Rename from this...
6987 (make_hard_regno_live): ... to this. Remove update to conflict
6988 information. Update function comment.
6989 (make_hard_regno_dead): Add conflict information update. Update
6991 (make_object_born): Rename from this...
6992 (make_object_live): ... to this. Remove update to conflict information.
6993 Update function comment.
6994 (make_object_dead): Add conflict information update. Update function
6996 (mark_pseudo_regno_live): Call make_object_live.
6997 (mark_pseudo_regno_subword_live): Likewise.
6998 (mark_hard_reg_dead): Update function comment.
6999 (mark_hard_reg_live): Call make_hard_regno_live.
7000 (process_bb_node_lives): Likewise.
7001 * lra-lives.c (make_hard_regno_born): Rename from this...
7002 (make_hard_regno_live): ... to this. Remove update to conflict
7003 information. Remove now uneeded check_pic_pseudo_p argument.
7004 Update function comment.
7005 (make_hard_regno_dead): Add check_pic_pseudo_p argument and add update
7006 to conflict information. Update function comment.
7007 (mark_pseudo_live): Remove update to conflict information. Update
7009 (mark_pseudo_dead): Add conflict information update.
7010 (mark_regno_live): Call make_hard_regno_live.
7011 (mark_regno_dead): Call make_hard_regno_dead with new arguement.
7012 (process_bb_lives): Call make_hard_regno_live and make_hard_regno_dead.
7014 2018-09-29 H.J. Lu <hongjiu.lu@intel.com>
7017 * config/i386/i386.c (construct_container): Use TImode for
7018 BLKmode values in 2 integer registers.
7020 2018-09-29 Jeff Law <law@redhat.com>
7022 * builtins.c (unterminated_array): Pass in c_strlen_data * to
7023 c_strlen rather than just a tree *.
7024 (c_strlen): Change NONSTR argument to a c_strlen_data pointer.
7025 Update recursive calls appropriately. If caller did not provide a
7026 suitable data pointer, create a local one. When a non-terminated
7027 string is discovered, bubble up information about the string via the
7028 c_strlen_data object.
7029 * builtins.h (c_strlen): Update prototype.
7030 (c_strlen_data): New structure.
7031 * gimple-fold.c (get_range_strlen): Update calls to c_strlen.
7032 For a type 2 call, if c_strlen indicates a non-terminated string
7033 use the length of the non-terminated string.
7034 (gimple_fold_builtin_stpcpy): Update calls to c_strlen.
7036 2018-09-29 Jakub Jelinek <jakub@redhat.com>
7039 * config/i386/avx512fintrin.h (_mm512_abs_pd, _mm512_mask_abs_pd): Use
7040 __m512d type for __A argument rather than __m512.
7042 2018-09-28 John David Anglin <danglin@gcc.gnu.org>
7044 * match.pd (simple_comparison): Don't optimize if either operand is
7045 a function pointer when target needs function pointer canonicalization.
7047 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
7049 * config/rs6000/driver-rs6000.c (asm_names): Adjust the entries for
7050 power5 .. power9 to remove indirection.
7051 * config/rs6000/rs6000.h (ASM_CPU_POWER5_SPEC, ASM_CPU_POWER6_SPEC,
7052 ASM_CPU_POWER7_SPEC, ASM_CPU_POWER8_SPEC, ASM_CPU_POWER9_SPEC,
7053 ASM_CPU_476_SPEC): Delete.
7054 (ASM_CPU_SPEC): Adjust.
7055 (EXTRA_SPECS): Delete asm_cpu_power5, asm_cpu_power6, asm_cpu_power7,
7056 asm_cpu_power8, asm_cpu_power9, asm_cpu_476.
7058 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
7060 * config.in: Delete HAVE_AS_DCI.
7061 * config/powerpcspe/powerpcspe.h: Treat HAVE_AS_DCI as always true.
7062 * config/rs6000/rs6000.h: Ditto.
7063 * configure.ac: Delete HAVE_AS_DCI.
7064 * configure: Regenerate.
7066 2018-09-28 Segher Boessenkool <segher@kernel.crashing.org>
7068 * config.in (HAVE_AS_LWSYNC): Delete.
7069 * config/powerpcspe/powerpcspe.h (TARGET_LWSYNC_INSTRUCTION): Delete.
7070 * config/powerpcspe/sync.md (*lwsync): Always generate lwsync, never
7072 * config/rs6000/rs6000.h (TARGET_LWSYNC_INSTRUCTION): Delete.
7073 * config/rs6000/sync.md (*lwsync): Always generate lwsync, never do it
7075 * configure.ac: Delete HAVE_AS_LWSYNC.
7076 * configure: Regenerate.
7078 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
7079 Pierre-Marie de Rodat <derodat@adacore.com>
7081 * calls.c (expand_call): Try to do a tail call for thunks at -O0 too.
7082 * cgraph.h (struct cgraph_thunk_info): Add indirect_offset.
7083 (cgraph_node::create_thunk): Add indirect_offset parameter.
7084 (thunk_adjust): Likewise.
7085 * cgraph.c (cgraph_node::create_thunk): Add indirect_offset parameter
7086 and initialize the corresponding field with it.
7087 (cgraph_node::dump): Dump indirect_offset field.
7088 * cgraphclones.c (duplicate_thunk_for_node): Deal with indirect_offset.
7089 * cgraphunit.c (cgraph_node::analyze): Be prepared for external thunks.
7090 (thunk_adjust): Add indirect_offset parameter and deal with it.
7091 (cgraph_node::expand_thunk): Deal with the indirect_offset field and
7092 pass it to thunk_adjust. Do not call the target hook if it's non-zero
7093 or if the thunk is external or local. Fix formatting. Do not chain
7094 the RESULT_DECL to BLOCK_VARS. Pass the static chain to the target,
7095 if any, in the GIMPLE representation.
7096 * ipa-icf.c (sem_function::equals_wpa): Deal with indirect_offset.
7097 * lto-cgraph.c (lto_output_node): Write indirect_offset field.
7098 (input_node): Read indirect_offset field.
7099 * tree-inline.c (expand_call_inline): Pass indirect_offset field in the
7100 call to thunk_adjust.
7101 * tree-nested.c (struct nesting_info): Add thunk_p field.
7102 (create_nesting_tree): Set it.
7103 (convert_all_function_calls): Copy static chain from targets to thunks.
7104 (finalize_nesting_tree_1): Return early for thunks.
7105 (unnest_nesting_tree_1): Do not finalize thunks.
7106 (gimplify_all_functions): Do not gimplify thunks.
7108 2018-09-28 David Malcolm <dmalcolm@redhat.com>
7110 * opt-suggestions.c (option_proposer::build_option_suggestions):
7111 Release "option_values".
7113 2018-09-28 David Malcolm <dmalcolm@redhat.com>
7115 * coverage.c (get_coverage_counts): Convert problem-reporting dump
7116 messages from MSG_OPTIMIZED_LOCATIONS to MSG_MISSED_OPTIMIZATION.
7117 * dumpfile.c (kind_as_string): New function.
7118 (dump_loc): Rather than a hardcoded prefix of "note: ", use
7119 kind_as_string to vary the prefix based on dump_kind.
7120 (selftest::test_capture_of_dump_calls): Update for above.
7122 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
7124 * config/i386/i386.h (SSE_REGNO): Fix check for FIRST_REX_SSE_REG.
7125 (GET_SSE_REGNO): Rename from SSE_REGNO. Update all uses for rename.
7127 2018-09-28 Uros Bizjak <ubizjak@gmail.com>
7129 * config/i386/i386.h (CC_REGNO): Remove FPSR_REG.
7130 * config/i386/i386.c (ix86_fixed_condition_code_regs): Use
7131 INVALID_REGNUM instead of FPSR_REG.
7132 (ix86_md_asm_adjust): Do not clobber FPSR_REG.
7133 * config/i386/i386.md: Update comment of FP compares.
7134 (fldenv): Do not clobber FPSR_REG.
7136 2018-09-28 Richard Biener <rguenther@suse.de>
7138 * tree.h (BLOCK_ORIGIN): New.
7139 * omp-expand.c (grid_expand_target_grid_body): Assign
7140 BLOCK_ORIGIN to BLOCK_ABSTRACT_ORIGIN.
7141 * tree-inline.c (remap_block): Likewise.
7142 * auto-profile.c (get_function_decl_from_block): Simplify
7143 by eliding the BLOCK_ABSTRACT_ORIGIN chasing.
7144 * langhooks.c (lhd_print_error_function): Likewise.
7145 * optinfo-emit-json.cc (optrecord_json_writer::inlining_chain_to):
7147 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
7148 * tree.c (block_nonartificial_location): Likewise.
7149 (block_ultimate_origin): Likewise.
7150 * tree-pretty-print.c (percent_K_format): Likewise. Remove
7151 no longer needed LTO case.
7153 2018-09-28 Andrew Stubbs <ams@codesourcery.com>
7154 Jan Hubicka <jh@suse.cz>
7155 Martin Jambor <mjambor@suse.cz>
7157 * simplify-rtx.c (simplify_merge_mask): New function.
7158 (simplify_ternary_operation): Use it, also see if VEC_MERGEs with the
7159 same masks are used in op1 or op2.
7160 (test_vec_merge): New function.
7161 (test_vector_ops): Call test_vec_merge.
7163 2018-09-28 Eric Botcazou <ebotcazou@adacore.com>
7165 * config/sparc/sparc-protos.h (sparc_branch_cost): Declare.
7166 * config/sparc/sparc.h (BRANCH_COST): Call sparc_branch_cost.
7167 * config/sparc/sparc.c (struct processor_costs): Add branch_cost field.
7168 (cypress_costs): Set it.
7169 (supersparc_costs): Likewise.
7170 (hypersparc_costs): Likewise.
7171 (leon_cost): Likewise.
7172 (leon3_costs): Likewise.
7173 (sparclet_costs): Likewise.
7174 (ultrasparc_costs): Likewise.
7175 (ultrasparc_costs): Likewise.
7176 (niagara_costs): Likewise.
7177 (niagara2_costs): Likewise.
7178 (niagara3_costs): Likewise.
7179 (niagara4_costs): Likewise.
7180 (niagara7_costs): Likewise.
7181 (m8_costs): Likewise.
7182 (TARGET_CAN_FOLLOW_JUMP): Define.
7183 (pass_work_around_errata::gate): Minor tweak.
7184 (sparc_option_override): Remove MASK_FSMULD mask for V7 processors.
7185 Do not set both MASK_VIS4 and MASK_VIS4B for M8 processor.
7186 Automaitcally clear MASK_FSMULD mask for V7 processors.
7187 (sparc_can_follow_jump): New static function.
7188 (output_ubranch): Deal with CROSSING_JUMP_P.
7189 (sparc_use_sched_lookahead): Rewrite using switch statement.
7190 (sparc_issue_rate): Reorder.
7191 (sparc_branch_cost): New function.
7193 2018-09-27 Martin Sebor <msebor@redhat.com>
7195 * tree.h (tree_to_shwi): Add attribute nonnull and pure.
7196 (tree_to_poly_int64, tree_to_uhwi, tree_to_poly_uint64): Same.
7197 (int_fits_type_p): Same.
7199 2018-09-27 Uros Bizjak <ubizjak@gmail.com>
7201 * config/i386/i386.md (FPCR_REG): Remove.
7202 (UNSPEC_FLDCW): Remove.
7203 (x86_fnstcw_1): Use (const_int 0) instead of FPCR_REG.
7204 (x86_fldcw_1): Remove insn pattern.
7205 (fnstenv): Do not clobber FPCR_REG.
7207 * config/i386/i386.h (FIXED_REGISTERS) Remove fpsr register.
7208 (CALL_USED_REGISTERS): Ditto.
7209 (REG_ALLOC_ORDER): Ditto.
7210 (REG_CLASS_CONTENTS): Ditto.
7211 (HI_REGISTER_NAMES): Ditto.
7212 (ADDITIONAL_REGISTER_NAMES): Use defines instead
7213 of numerical constants.
7214 * config/i386/i386.c (regclass_map): Remove fpsr register.
7215 (dbx_register_map): Ditto.
7216 (dbx64_register_map): Ditto.
7217 (svr4_dbx_register_map): Ditto.
7218 (print_reg): Do not handle FPCR_REG.
7220 2018-09-27 Segher Boessenkool <segher@kernel.crashing.org>
7223 * config.in (HAVE_AS_CMPB, HAVE_AS_DFP, HAVE_AS_FPRND, HAVE_AS_MFPGPR,
7224 HAVE_AS_POPCNTB, HAVE_AS_POPCNTD, HAVE_AS_POWER8, HAVE_AS_POWER9):
7225 Delete, always treat as true.
7226 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
7227 Ditto. Simplify remaining code.
7228 * config/powerpcspe/powerpcspe.h: Ditto.
7229 * config/rs6000/rs6000.c (rs6000_option_override_internal): Ditto.
7230 Simplify remaining code.
7231 (rs6000_expand_builtin): Ditto.
7232 * config/rs6000/rs6000.h: Ditto.
7233 * configure.ac: Ditto.
7234 * configure: Regenerate.
7236 2018-09-27 Martin Liska <mliska@suse.cz>
7238 * coverage.c (get_coverage_counts): Revert the formatting
7239 of missing profile opt info.
7241 2018-09-27 Richard Biener <rguenther@suse.de>
7245 * dwarf2out.c (set_block_origin_self): Do not mark outermost
7246 block as we do not output that.
7247 (gen_inlined_subroutine_die): Elide the originally outermost
7248 block, matching what we do for concrete instances.
7249 (decls_for_scope): Add parameter specifying whether to recurse
7252 2018-09-27 Andrew Stubbs <ams@codesourcery.com>
7253 Tom de Vries <tom@codesourcery.com>
7257 * expmed.c (emit_cstore): Fix handling of result_mode == BImode and
7258 STORE_FLAG_VALUE == 1.
7260 2018-09-27 Andreas Krebbel <krebbel@linux.ibm.com>
7262 * config/s390/s390.md (PPA_TX_ABORT, PPA_OOO_BARRIER): New
7263 constant definitions.
7264 ("tx_assist"): Replace magic number with PPA_TX_ABORT.
7265 ("*ppa"): Enable pattern also for -march=zEC12 -mno-htm.
7266 ("speculation_barrier"): New expander definition.
7268 2018-09-26 Indu Bhagat <indu.bhagat@oracle.com>
7270 PR gcov-profile/86957
7271 * common.opt: New warning option -Wmissing-profile.
7272 * coverage.c (get_coverage_counts): Add warning for missing .gcda file.
7273 * doc/invoke.texi: Document -Wmissing-profile.
7275 2018-09-26 Jim Wilson <jimw@sifive.com>
7277 * config/riscv/riscv.md (subsi3_extended2): Add J constraint.
7278 (negdi2, negsi2, negsi2_extended, negsi2_extended2): New.
7280 2018-09-26 Martin Sebor <msebor@redhat.com>
7282 * tree.c (zerop): Change return type to bool.
7283 (integer_zerop, integer_onep, integer_each_onep): Same.
7284 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
7285 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
7286 (real_onep, real_minus_onep, chain_index): Same.
7287 (print_type_hash_statistics, type_list_equal): Same.
7288 * tree.h (zerop): Same.
7289 (zerop, integer_zerop, integer_onep, integer_each_onep): Same.
7290 (integer_all_onesp, integer_minus_onep, integer_pow2p): Same.
7291 (integer_nonzerop, integer_truep, tree_ctz, real_zerop): Same.
7292 (real_onep, real_minus_onep, chain_index): Same.
7293 (print_type_hash_statistics, type_list_equal): Same.
7295 2018-09-26 Jim Wilson <jimw@sifive.com>
7297 * config/riscv/riscv.h (FUNCTION_ARG_REGNO_P): Fix comment.
7299 2018-09-26 Jakub Jelinek <jakub@redhat.com>
7302 * config/i386/i386.c: Include debug.h and dwarf2out.h.
7303 (output_indirect_thunk): Emit DW_CFA_def_cfa_offset after the
7306 2018-09-25 Andrew Stubbs <ams@codesourcery.com>
7308 * builtins.c (get_builtin_sync_mem): Force address mode conversion.
7310 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
7312 * config/i386/i386.h (enum reg_class): Remove FP_TOP_SSE_REGS
7313 and FP_SECOND_SSE_REGS.
7314 (REG_CLASS_NAMES): Ditto.
7315 (REG_CLASS_CONTENTS): Ditto.
7316 * config/i386/i386.c (ix86_preferred_reload_class) Do not handle
7317 FP_TOP_SSE_REGS and FP_SECOND_SSE_REGS classes.
7318 (ix86_preferred_output_reload_class): Ditto.
7319 * config/i386/i386.md (fix_trunc<mode>_i387_fisttp): Change "=&1f"
7320 clobber constraint to "=&f".
7321 (fix_truncdi_i387): Ditto.
7322 (lrintxfdi2): Ditto.
7323 (fistdi2_<rounding>): Ditto.
7324 (fpremxf4_i387): Change "=u" constraint to "=f".
7325 (fprem1xf4_i387): Ditto.
7327 (fptanxf4_i387): Ditto.
7328 (fxtractxf3_i387): Ditto.
7329 (fscalexf4_i387): Ditto.
7330 (atan2xf3): Change "u" constraint to "f".
7331 (fyl2xxf3_i387): Ditto.
7332 (fyl2xp1xf3_i387): Ditto.
7334 2018-09-26 Uros Bizjak <ubizjak@gmail.com>
7337 * config/i386/i386.h (NUM_MODES_FOR_MODE_SWITCHING): Update
7338 for removed I387_MASK_PM entity.
7341 2018-09-26 Jeff Law <law@redhat.com>
7343 2018-09-26 Alexey Neyman <stilor@att.net>
7345 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
7346 headers are no longer pulled in by <isl/val.h>.
7348 2018-09-26 Richard Biener <rguenther@suse.de>
7351 * dwarf2out.c (gen_lexical_block_die): Do not equate inline
7352 or concrete instance DIE to the tree. Create abstract origin
7353 attributes also for concrete instances.
7355 2018-09-26 Alexey Neyman <stilor@att.net>
7357 * graphite.h: Include <isl/id.h> and <isl/space.h>; these
7358 headers are no longer pulled in by <isl/val.h>.
7360 2018-09-26 Matthew Malcomson <matthew.malcomson@arm.com>
7362 * config/arm/arm.c (arm_split_compare_and_swap, arm_split_atomic_op):
7363 Use new helper functions.
7364 * config/arm/sync.md (atomic_load<mode>, atomic_store<mode>):
7365 Use new helper functions.
7366 * config/arm/aarch-common-protos.h (aarch_mm_needs_acquire,
7367 aarch_mm_needs_release): New declarations.
7368 * config/arm/aarch-common.c (aarch_mm_needs_acquire,
7369 aarch_mm_needs_release): New.
7371 2018-09-26 Eric Botcazou <ebotcazou@adacore.com>
7373 * config/arm/arm.c (arm_reorg): Skip Thumb reorg pass for thunks.
7374 (arm32_output_mi_thunk): Deal with long calls.
7376 2018-09-26 Richard Biener <rguenther@suse.de>
7380 * tree-inline.c (expand_call_inline): When the location
7381 of the call is UNKNOWN_LOCATION use DECL_SOURCE_LOCATION
7382 or BUILTINS_LOCATION for the BLOCK_SOURCE_LOCATION of
7383 the inserted BLOCK to make inlined_function_outer_scope_p
7385 * dwarf2out.c (add_call_src_coords_attributes): Do not add
7386 coords for reserved locations.
7388 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
7390 * config/rs6000/rs6000.md (*movcc_internal1): Use set_attr_alternative.
7391 (*call_indirect_nonlocal_sysv<mode>): Ditto.
7392 (*call_value_indirect_nonlocal_sysv<mode>): Ditto.
7393 (*sibcall_nonlocal_sysv<mode>): Ditto.
7394 (*sibcall_value_nonlocal_sysv<mode>): Ditto.
7395 (<bd>_<mode>): Ditto.
7396 (<bd>tf_<mode>): Ditto.
7398 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
7400 * config/rs6000/altivec.md (*altivec_mov<mode>): Write the output
7401 control string as a list of templates instead of as C code.
7402 (*altivec_movti): Ditto.
7403 * config/rs6000/darwin.md (movdf_low_di): Ditto.
7405 2018-09-25 Jim Wilson <jimw@sifive.com>
7407 * config/riscv/riscv.c (riscv_split_symbol): Mark auipc label as weak
7408 when target symbol is weak.
7410 2018-09-25 Bernd Edlinger <bernd.edlinger@hotmail.de>
7413 * builtins.c (unterminated_array): Simplify.
7414 * expr.c (string_constant): Handle SSA_NAME. Add more exceptions
7415 where pointer arithmetic is safe.
7417 2018-09-25 Segher Boessenkool <segher@kernel.crashing.org>
7420 * config/rs6000/altivec.md (altivec_vspltb): Use
7421 const_0_to_15_operand instead of u5bit_cint_operand.
7422 (*altivec_vspltb_internal): Ditto.
7423 (altivec_vspltb_direct): Ditto.
7424 (altivec_vsplth): Use const_0_to_7_operand instead of
7426 (*altivec_vsplth_internal): Ditto.
7427 (altivec_vsplth_direct): Ditto.
7428 (altivec_vspltw): Use const_0_to_3_operand instead of
7430 (*altivec_vspltw_internal): Ditto.
7431 (altivec_vspltw_direct): Ditto.
7432 (altivec_vspltsf): Ditto.
7433 (*altivec_vspltsf_internal): Ditto.
7434 * config/rs6000/rs6000.c (rs6000_expand_binop_builtin): Handle the
7435 various splats with the proper size immediate. Reorder the various
7436 cases by ascending size of immediate, and put all such together.
7438 2018-09-25 Richard Biener <rguenther@suse.de>
7441 * dwarf2out.c (add_AT_external_die_ref): Remove now redundant
7443 (maybe_create_die_with_external_ref): Do not create
7444 DW_TAG_imported_unit here.
7445 (add_abstract_origin_attribute): Handle external BLOCK refs.
7446 (dwarf2out_abstract_function): Simplify LTO case.
7447 (dwarf2out_early_finish): Create DW_TAG_imported_unit explicitely
7448 rather than using maybe_create_die_with_external_ref.
7450 2018-09-25 Uros Bizjak <ubizjak@gmail.com>
7453 * config/i386/i386.md (frndintxf2_mask_pm): Remove.
7454 (frndintxf2_mask_pm_i387): Ditto.
7455 (nearbyintxf2): Rewrite expander pattern to match rintxf2.
7456 Enable for !flag_trapping_math.
7457 (nearbyint<mode>2): Enable x87 modes for !flag_trapping_math.
7458 Enable SSE modes for TARGET_SSE4_1 and expand them with round insn.
7459 Change operand 1 predicate to nonimmediate_operand.
7460 (attr "i387_cw"): Remove mask_pm.
7461 * config/i386/i386.h (enum ix86_stack_slot): Remove SLOT_CW_MASK_PM.
7462 (enum ix86_entity): Remove I387_MASK_PM.
7463 * config/i386/i386.c (ix86_i387_mode_needed): Do not
7464 handle I387_MASK_PM.
7465 (ix86_mode_needed): Ditto.
7466 (ix86_mode_after): Ditto.
7467 (ix86_mode_entry): Ditto.
7468 (ix86_mode_exit): Ditto.
7469 (emit_i387_cw_initialization): Do not handle I387_CW_MASK_PM.
7471 2018-09-25 Jakub Jelinek <jakub@redhat.com>
7473 * vr-values.c (vr_values::vr_values): Initialize to_remove_edges and
7474 to_update_switch_stmts to vNULL instead of calling create on them
7477 2018-09-25 Richard Biener <rguenther@suse.de>
7479 PR tree-optimization/87402
7480 * tree-ssa-sccvn.c (SSA_VISITED): Remove unused function.
7481 (visit_phi): Re-instantiate handling of supposed to be VARYING
7482 but non-VARYING backedge value.
7484 2018-09-25 Richard Biener <rguenther@suse.de>
7487 * dwarf2out.c (struct sym_off_pair): New.
7488 (external_die_map): New global.
7489 (lookup_decl_die): When in LTO create DIEs lazily from the
7491 (lookup_block_die): New function, create DIEs lazily in LTO.
7492 (equate_block_to_die): New function.
7493 (dwarf2out_die_ref_for_decl): During WPA get the association
7494 from the external DIE map.
7495 (dwarf2out_register_external_die): Record mapping into the
7497 (maybe_create_die_with_external_ref): New function split out from
7498 DIE generation part of old dwarf2out_register_external_die.
7499 (add_abstract_origin_attribute): Do not return the DIE. When
7500 in LTO reference externals directly.
7501 (dwarf2out_abstract_function): When in LTO ignore calls for
7502 decls with external DIEs (already present abstract instances).
7503 (gen_call_site_die): Adjust.
7504 (add_high_low_attributes): Likewise.
7505 (gen_lexical_block_die): Likewise.
7506 (gen_inlined_subroutine_die): Likewie.
7507 (gen_block_die): Likewise.
7508 (dwarf2out_inline_entry): Likewise.
7509 (dwarf2out_early_finish): In LTRANS phase create DW_TAG_imported_unit
7512 2018-09-25 Martin Liska <mliska@suse.cz>
7514 * ipa-fnsummary.c (estimate_node_size_and_time): Scale by two
7515 integers and not by a float value.
7517 2018-09-25 Martin Liska <mliska@suse.cz>
7520 * dbgcnt.c (dbg_cnt_process_single_pair): Return false
7522 * dumpfile.c (dump_enable_all): Remove extra parenthesis.
7523 * gcov-tool.c: Declare the function with ATTRIBUTE_NORETURN.
7524 * godump.c (go_format_type): Remove extra parenthesis.
7526 2018-09-25 Martin Liska <mliska@suse.cz>
7528 * alias.c (set_dest_equal_p): Remove unused function.
7529 * config/i386/i386.c (def_builtin_pure2): Likewise.
7530 * diagnostic-show-locus.c (class layout): Remove
7532 (layout::layout): Likewise here.
7533 * dump-context.h (class temp_dump_context): Likewise.
7534 * dwarf2out.c (add_AT_fde_ref): Remove unused function.
7535 (add_AT_loclistsptr): Likewise.
7536 (add_AT_offset): Likewise.
7537 (get_AT_hi_pc): Likewise.
7538 (is_comdat_die): Likewise.
7539 (type_is_enum): Likewise.
7540 (ceiling): Likewise.
7541 (add_AT_vms_delta): Likewise.
7542 (is_class_die): Likewise.
7543 * edit-context.c (class line_event): Remove unused field.
7544 * graphite-sese-to-poly.c (tree_int_to_gmp): Remove
7546 * ipa-cp.c (ipa_get_vr_lat): Likewise.
7547 * lra-constraints.c (ok_for_index_p_nonstrict): Likewise.
7548 (ok_for_base_p_nonstrict): Likewise.
7549 * tree-chrec.c (is_not_constant_evolution): Likewise.
7550 (chrec_fold_poly_cst): Likewise.
7551 * tree-if-conv.c (has_pred_critical_p): Likewise.
7552 * tree-ssa-coalesce.c (print_exprs): Likewise.
7553 * tree-ssa-pre.c (bitmap_set_contains_expr): Likewise.
7554 * tree-ssa-uninit.c (is_and_or_or_p): Likewise.
7555 * tree-vrp.c (value_ranges_intersect_p): Likewise.
7556 (value_range_nonnegative_p): Likewise.
7558 2018-09-25 Martin Liska <mliska@suse.cz>
7560 * config/powerpcspe/powerpcspe.c (rs6000_output_function_epilogue):
7561 Do not handle "GNU Pascal".
7562 * config/rs6000/rs6000.c (rs6000_output_function_epilogue):
7564 * config/sparc/sparc.c (sparc_pass_by_reference): Remove Pascal
7565 from documentation. Likewise.
7566 * dbxout.c (dbxout_range_type): Likewise.
7567 * doc/cpp.texi: Likewise.
7568 * doc/extend.texi: Likewise.
7569 * doc/frontends.texi: Likewise.
7570 * doc/invoke.texi: Remove Pascal entry.
7571 * tree.def (CLEANUP_POINT_EXPR): Likewise.
7572 * doc/rtl.texi (MODE_FUNCTION): Remove not used entry.
7574 2018-09-25 Martin Liska <mliska@suse.cz>
7577 * doc/invoke.texi: Document all parameters and remove default
7580 2018-09-25 Ilya Leoshkevich <iii@linux.ibm.com>
7583 * rtl.c (rtx_code_size): Take into account that EQ_ATTR_ALT
7584 contains HOST_WIDE_INTs when computing its size.
7586 2018-09-24 Jim Wilson <jimw@sifive.com>
7589 * config/riscv/riscv.h (STACK_BOUNDARY): Test riscv_abi == ABI_ILP32E
7591 (ABI_STACK_BOUNDARY, MAX_ARGS_IN_REGISTERS): Likewise.
7593 2018-09-24 Andrew Pinski <apinski@marvell.com>
7595 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Don't
7596 access prev before checking it for NULLness in the
7597 AARCH64_FUSE_CMP_BRANCH case.
7599 2018-09-24 H.J. Lu <hongjiu.lu@intel.com>
7602 * config/i386/i386.c (rest_of_insert_endbranch): Set
7603 endbr_queued_at_entrance to true and don't insert ENDBR if
7604 x86_function_profiler will be called.
7605 (x86_function_profiler): Insert ENDBR if endbr_queued_at_entrance
7607 * config/i386/i386.h (machine_function): Add
7608 endbr_queued_at_entrance.
7610 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
7612 * genattrtab.c (mk_attr_alt): Use alternative_mask.
7613 (attr_rtx_1): Adjust caching to match the new EQ_ATTR_ALT field
7615 (check_attr_test): Use alternative_mask.
7616 (get_attr_value): Likewise.
7617 (compute_alternative_mask): Use alternative_mask and XWINT.
7618 (make_alternative_compare): Use alternative_mask.
7619 (attr_alt_subset_p): Use XWINT.
7620 (attr_alt_subset_of_compl_p): Likewise.
7621 (attr_alt_intersection): Use alternative_mask and XWINT.
7622 (attr_alt_union): Likewise.
7623 (attr_alt_complement): Use HOST_WIDE_INT and XWINT.
7624 (mk_attr_alt): Use alternative_mask and HOST_WIDE_INT.
7625 (simplify_test_exp): Use alternative_mask and XWINT.
7626 (write_test_expr): Use alternative_mask and XWINT, adjust bit
7627 number calculation to support 64 bits. Generate code that
7628 checks 64-bit masks.
7629 (main): Use alternative_mask.
7630 * rtl.def (EQ_ATTR_ALT): Change field types from ii to ww.
7632 2018-09-24 Ilya Leoshkevich <iii@linux.ibm.com>
7635 * config/s390/s390.c (s390_emit_epilogue): Do not use PARALLEL
7636 RETURN+USE when returning via %r14.
7638 2018-09-24 Martin Liska <mliska@suse.cz>
7640 * gcov.c (output_lines): Print colorization legend
7641 for both flag_use_colors and flag_use_hotness_colors.
7644 2018-09-24 Martin Liska <mliska@suse.cz>
7646 * coverage.c (get_coverage_counts): Use warning_at
7647 with current_function_decl location. Use %qD in warning
7650 2018-09-24 Martin Liska <mliska@suse.cz>
7652 * memory-block.h (memory_block_pool::release): Annotate with
7653 valgrind that the memory is not accessible.
7655 2018-09-24 Martin Liska <mliska@suse.cz>
7658 * asan.c: Make asan_handled_variables extern.
7660 * cfgexpand.c (expand_stack_vars): Make sure
7661 a representative is unpoison if another
7662 variable in the partition is handled by
7663 use-after-scope sanitization.
7665 2018-09-24 Richard Biener <rguenther@suse.de>
7667 PR tree-optimization/63155
7668 * tree-ssa-propagate.c (add_ssa_edge): Avoid adding PHIs to
7669 the worklist when the edge of the respective argument isn't
7672 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
7674 * config/i386/i386.h (enum reg_class): Rename MASK_REGS to
7675 ALL_MASK_REGS and MASK_EVEX_REGS to MASK_REGS.
7676 (MASK_CLASS_P): Update for rename.
7677 (MAYBE_MASK_CLASS_P): Ditto.
7678 (REG_CLASS_NAMES): Update.
7679 (REG_CLASS_CONTENT): Update.
7680 * config/i386/i386.c (regclass_map): Update for MASK_REG
7681 and ALL_MASK_REGS rename.
7682 * config/i386/constraints.md (Yk): Update for rename.
7685 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
7687 * config/i386/i386.h (enum reg_class): Remove
7688 EVEX_SSE_REGS and MOD4_SSE_REGS.
7689 (REG_CLASS_NAMES): Update.
7690 (REG_CLASS_CONTENT): Update.
7691 * config/i386/i386.c (regclass_map): Declare AVX-512 SSE
7692 registers as ALL_SSE_REGS.
7693 (ix86_additional_allocno_class_p): Remove.
7694 (TARGET_ADDITIONAL_ALLOCNO_CLASS_P): Remove.
7695 (ix86_register_priority): Lower priority of EVEX SSE registers.
7696 Use IN_RANGE macro where appropriate.
7697 (ix86_hard_regno_mode_ok): Merge AVX-5124FMAPS and
7698 AVX-5124VNNIW checks.
7699 (ix86_modes_tieable_p): Tie 512-bit SSE modes.
7700 * config/i386/sse.md (avx5124fmaddps_4fmaddps)
7701 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddps_maskz)
7702 (avx5124fmaddps_4fmaddss, avx5124fmaddps_4fmaddss_mask)
7703 (avx5124fmaddps_4fmaddss_maskz, avx5124fmaddps_4fnmaddps)
7704 (avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddps_maskz)
7705 (avx5124fmaddps_4fnmaddss, avx5124fmaddps_4fnmaddss_mask)
7706 (avx5124fmaddps_4fnmaddss_maskz, avx5124vnniw_vp4dpwssd)
7707 (avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssd_maskz)
7708 (avx5124vnniw_vp4dpwssds, avx5124vnniw_vp4dpwssds_mask)
7709 (avx5124vnniw_vp4dpwssds_maskz): Use "v" instead of "Yh" constraint.
7710 * config/i386/constraints.md (Yh): Remove.
7712 2018-09-23 Uros Bizjak <ubizjak@gmail.com>
7714 * config/i386/i386.c (regclass_map): Declare integer REX registers
7717 2018-09-23 Gerald Pfeifer <gerald@pfeifer.com>
7719 * doc/service.texi (Service): Switch the fsf.org link to https.
7721 2018-09-22 Chung-Ju Wu <jasonwucj@gmail.com>
7724 * config/nds32/nds32.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
7725 Define to speculation_safe_value_not_needed.
7727 2018-09-21 Florian Weimer <fweimer@redhat.com>
7730 * doc/extend.texi (Common Function Attributes): Mention that
7731 noreturn suppresses tail call optimization.
7733 2018-09-21 Jeff Law <law@redhat.com>
7735 * gimple-ssa-evrp.c (evrp_dom_walker::cleanup): Call
7736 vr_values::cleanup_edges_and_switches.
7737 * tree-vrp.c (to_remove_edges, to_update_switch_stmts): Moved into
7739 (identify_jump_threads): Remove EDGE_IGNORE handling.
7740 (execute_vrp): Move handling of to_remove_edges and
7741 to_update_switch_stmts into vr_values class member functions.
7742 * tree-vrp.h (switch_update, to_remove_edges): Remove declarations.
7743 (to_update_switch_stmts): Likewise.
7744 * vr-values.c: Include cfghooks.h.
7745 (vr_values::vr_values): Initialize to_remove_edges and
7746 to_update_switch_stmts.
7747 (vr_values::~vr_values): Verify to_remove_edges and
7748 to_update_switch_stmts are empty.
7749 (vr_values::simplify_switch_using_ranges): Set EDGE_IGNORE as needed.
7750 (vr_values::cleanup_edges_and_switches): New member function.
7751 * vr-values.h (vr_values): Add cleanup_edges_and_switches member
7752 function. Add new data members.
7754 2018-09-21 David Malcolm <dmalcolm@redhat.com>
7756 PR tree-optimization/87309
7757 * dumpfile.c (dump_context::begin_scope): Filter the dump_loc
7758 calls with pflags and alt_flags.
7759 (selftest::test_capture_of_dump_calls): Add test of interaction of
7760 MSG_OPTIMIZED_LOCATIONS with AUTO_DUMP_SCOPE.
7762 2018-09-21 Olivier Hainque <hainque@adacore.com>
7764 * config.gcc: Factorize and comment inclusion of vxworks-dummy.h.
7766 2018-09-21 Olivier Hainque <hainque@adacore.com>
7768 * config/vxworks.h (CLEAR_INSN_CACHE): #define to 1.
7770 2018-09-21 Olivier Hainque <hainque@adacore.com>
7772 * config/vxworks.h (VXWORKS_LIBGCC_SPEC): Remove -lc_internal.
7773 Merge block comment with the one ahead of VXWORKS_LIBS_RTP. Then:
7774 (VXWORKS_LIBS_RTP): Minor reordering.
7776 2018-09-21 Olivier Hainque <hainque@adacore.com>
7778 * config/vxworks.h (STARTFILE_PREFIX_SPEC): Define.
7779 (VXWORKS_LIBS_DIR_RTP): Remove definition and use.
7781 2018-09-21 Olivier Hainque <hainque@adacore.com>
7783 * config/vxworks.h (SIZE_TYPE): Account for TARGET_VXWORKS64.
7784 (PTRDIFF_TYPE): Likewise.
7786 2018-09-21 Olivier Hainque <hainque@adacore.com>
7788 * config.gcc: Enforce def of TARGET_VXWORKS64 to 1 from
7789 triplet, similar to support for VxWorks7.
7790 * config/vxworks-dummy.h: Provide a default definition
7791 of TARGET_VXWORKS64 to 0.
7793 2018-09-21 Olivier Hainque <hainque@adacore.com>
7795 * config/vxworks.h (TARGET_VXWORKS7): Move default definition ...
7796 * config/vxworks-dummy.h: here.
7798 2018-09-21 Olivier Hainque <hainque@adacore.com>
7800 * config.gcc: Prepend vxworks-dummy.h to tm_file for powerpc*
7802 2018-09-21 Shaokun Zhang <zhangshaokun@hisilicon.com>
7803 Bo Zhou <zbo.zhou@hisilicon.com>
7805 * config/aarch64/aarch64-cores.def (tsv110): New CPU.
7806 * config/aarch64/aarch64-tune.md: Regenerated.
7807 * doc/invoke.texi (AArch64 Options/-mtune): Add "tsv110".
7808 * config/aarch64/aarch64.c (tsv110_tunings): New tuning table.
7809 * config/aarch64/aarch64-cost-tables.h: Add "tsv110" extra costs.
7811 2018-09-21 Andrew Stubbs <ams@codesourcery.com>
7812 Julian Brown <julian@codesourcery.com>
7814 * builtins.c (get_builtin_sync_mem): Handle address spaces.
7816 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
7818 * config/rs6000/rs6000.c (rs6000_function_ok_for_sibcall): Return false
7819 if the call takes a static chain.
7821 2018-09-21 Martin Liska <mliska@suse.cz>
7823 * auto-profile.c (autofdo_source_profile::read): Do not
7825 (read_profile): Do not add working sets.
7826 (read_autofdo_file): Remove sum_all.
7827 (afdo_callsite_hot_enough_for_early_inline): Remove const
7829 * coverage.c (struct counts_entry): Remove gcov_summary.
7830 (read_counts_file): Read new GCOV_TAG_OBJECT_SUMMARY,
7831 do not support GCOV_TAG_PROGRAM_SUMMARY.
7832 (get_coverage_counts): Remove summary and expected
7834 * coverage.h (get_coverage_counts): Likewise.
7835 * doc/gcov-dump.texi: Remove -w option.
7836 * gcov-dump.c (dump_working_sets): Remove.
7837 (main): Do not support '-w' option.
7838 (print_usage): Likewise.
7839 (tag_summary): Likewise.
7840 * gcov-io.c (gcov_write_summary): Do not dump
7842 (gcov_read_summary): Likewise.
7843 (gcov_histo_index): Remove.
7844 (gcov_histogram_merge): Likewise.
7845 (compute_working_sets): Likewise.
7846 * gcov-io.h (GCOV_TAG_OBJECT_SUMMARY): Mark
7848 (GCOV_TAG_PROGRAM_SUMMARY): Mark it obsolete.
7849 (GCOV_TAG_SUMMARY_LENGTH): Adjust.
7850 (GCOV_HISTOGRAM_SIZE): Remove.
7851 (GCOV_HISTOGRAM_BITVECTOR_SIZE): Likewise.
7852 (struct gcov_summary): Simplify rapidly just
7853 to runs and sum_max fields.
7854 (gcov_histo_index): Remove.
7855 (NUM_GCOV_WORKING_SETS): Likewise.
7856 (compute_working_sets): Likewise.
7857 * gcov-tool.c (print_overlap_usage_message): Remove
7858 trailing empty line.
7859 * gcov.c (read_count_file): Read GCOV_TAG_OBJECT_SUMMARY.
7860 (output_lines): Remove program related line.
7861 * ipa-profile.c (ipa_profile): Do not consider GCOV histogram.
7862 * lto-cgraph.c (output_profile_summary): Do not stream GCOV
7864 (input_profile_summary): Do not read it.
7865 (merge_profile_summaries): And do not merge it.
7866 (input_symtab): Do not call removed function.
7867 * modulo-sched.c (sms_schedule): Do not print sum_max.
7868 * params.def (HOT_BB_COUNT_FRACTION): Reincarnate param that was
7869 removed when histogram method was invented.
7870 (HOT_BB_COUNT_WS_PERMILLE): Mention that it's used only in LTO
7872 * postreload-gcse.c (eliminate_partially_redundant_load): Fix
7874 * predict.c (get_hot_bb_threshold): Use HOT_BB_COUNT_FRACTION
7875 and dump selected value.
7876 * profile.c (add_working_set): Remove.
7877 (get_working_sets): Likewise.
7878 (find_working_set): Likewise.
7879 (get_exec_counts): Do not work with working sets.
7880 (read_profile_edge_counts): Do not inform as sum_max is removed.
7881 (compute_branch_probabilities): Likewise.
7882 (compute_value_histograms): Remove argument for call of
7883 get_coverage_counts.
7884 * profile.h: Do not make gcov_summary const.
7886 2018-09-21 Monk Chiang <sh.chiang04@gmail.com>
7888 * config.gcc (nds32*-*-*): Set TARGET_DEFAULT_TLSDESC_TRAMPOLINE=0.
7890 2018-09-21 Eric Botcazou <ebotcazou@adacore.com>
7892 PR tree-optimization/86990
7893 * gimple-ssa-store-merging.c (imm_store_chain_info:coalesce_immediate):
7894 Check that the entire merged store group is made of constants only for
7897 2018-09-20 Allan Sandfeld Jensen <allan.jensen@qt.io>
7899 * gcc.c (LINK_COMMAND_SPEC): Handle -r like -nostdlib.
7900 (VTABLE_VERIFICATION_SPEC): Likewise.
7901 (SANITIZER_EARLY_SPEC): Likewise.
7902 (SANITIZER_SPEC): Likewise.
7903 * config/darwin.h (LINK_COMMAND_SPEC): Likewise.
7904 * doc/invoke.texi (Link Options): Document -r.
7906 2018-09-20 Richard Biener <rguenther@suse.de>
7909 * gimplify.c (gimplify_expr): Retain alignment of
7910 addressable lvalue in dereference.
7912 2018-09-20 Alexandre Oliva <aoliva@redhat.com>
7915 * configure.ac: Check for .loc is_stmt support.
7916 * configure, config.in: Rebuilt.
7917 * dwarf2out.c (dwarf2out_source_line): Skip is_stmt
7920 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7922 * config/rs6000/rs6000.opt (misel=no, misel=yes): Delete.
7923 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -misel=yes and
7926 2018-09-20 Segher Boessenkool <segher@kernel.crashing.org>
7928 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
7930 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete
7933 2018-09-20 Marek Polacek <polacek@redhat.com>
7935 * doc/invoke.texi: Add -Wno-init-list-lifetime to C++ Language Options.
7937 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
7939 PR tree-optimization/87288
7940 * tree-vect-loop.c (vect_analyze_loop_2): Take PEELING_FOR_GAPS
7941 into account when determining PEELING_FOR_NITERS.
7943 2018-09-20 Richard Sandiford <richard.sandiford@arm.com>
7945 PR tree-optimization/86877
7946 * tree-vect-loop.c (vect_analyze_loop_2): Call
7947 vect_verify_datarefs_alignment.
7949 2018-09-19 Marek Polacek <polacek@redhat.com>
7951 * doc/invoke.texi: Document -Wclass-conversion.
7953 2018-09-19 John David Anglin <danglin@gcc.gnu.org>
7955 * config/pa/pa.c (pa_adjust_priority): Delete.
7956 (TARGET_SCHED_ADJUST_PRIORITY): Delete define.
7958 * config/pa/pa.md (atomic_storeqi): Restore deleted expander.
7959 (atomic_storehi): Likewise.
7960 (atomic_storesi): Likewise.
7961 (atomic_loaddi): Restore compare and swap exchange loop code.
7963 2018-09-19 Segher Boessenkool <segher@kernel.crashing.org>
7965 PR rtl-optimization/86902
7966 * combine.c (try_combine): When changing the CC mode used, don't change
7967 an unrelated mode in other_insn to that new CC mode.
7969 2018-09-19 David Malcolm <dmalcolm@redhat.com>
7971 * tree-data-ref.c (runtime_alias_check_p): Use formatted printing
7972 with %T in place of calls to dump_generic_expr.
7973 (prune_runtime_alias_test_list): Likewise.
7974 (create_runtime_alias_checks): Likewise.
7975 * tree-vect-data-refs.c (vect_check_nonzero_value): Likewise.
7976 (vect_analyze_data_ref_dependence): Likewise.
7977 (vect_slp_analyze_data_ref_dependence): Likewise.
7978 (vect_record_base_alignment): Likewise. Use %G in place of call
7979 to dump_gimple_stmt.
7980 (vect_compute_data_ref_alignment): Likewise.
7981 (verify_data_ref_alignment): Likewise.
7982 (vect_find_same_alignment_drs): Likewise.
7983 (vect_analyze_group_access_1): Likewise.
7984 (vect_analyze_data_ref_accesses): Likewise.
7985 (dependence_distance_ge_vf): Likewise.
7986 (dump_lower_bound): Likewise.
7987 (vect_prune_runtime_alias_test_list): Likewise.
7988 (vect_find_stmt_data_reference): Likewise.
7989 (vect_analyze_data_refs): Likewise.
7990 (vect_create_addr_base_for_vector_ref): Likewise.
7991 (vect_create_data_ref_ptr): Likewise.
7992 * tree-vect-loop-manip.c (vect_set_loop_condition): Likewise.
7993 (vect_can_advance_ivs_p): Likewise.
7994 (vect_update_ivs_after_vectorizer): Likewise.
7995 (vect_gen_prolog_loop_niters): Likewise.
7996 (vect_prepare_for_masked_peels): Likewise.
7997 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
7998 (vect_determine_vectorization_factor): Likewise.
7999 (vect_is_simple_iv_evolution): Likewise.
8000 (vect_analyze_scalar_cycles_1): Likewise.
8001 (vect_analyze_loop_operations): Likewise.
8002 (report_vect_op): Likewise.
8003 (vect_is_slp_reduction): Likewise.
8004 (check_reduction_path): Likewise.
8005 (vect_is_simple_reduction): Likewise.
8006 (vect_create_epilog_for_reduction): Likewise.
8007 (vect_finalize_reduction:): Likewise.
8008 (vectorizable_induction): Likewise.
8009 (vect_transform_loop_stmt): Likewise.
8010 (vect_transform_loop): Likewise.
8011 (optimize_mask_stores): Likewise.
8012 * tree-vect-patterns.c (vect_pattern_detected): Likewise.
8013 (vect_split_statement): Likewise.
8014 (vect_recog_over_widening_pattern): Likewise.
8015 (vect_recog_average_pattern): Likewise.
8016 (vect_determine_min_output_precision_1): Likewise.
8017 (vect_determine_precisions_from_range): Likewise.
8018 (vect_determine_precisions_from_users): Likewise.
8019 (vect_mark_pattern_stmts): Likewise.
8020 (vect_pattern_recog_1): Likewise.
8021 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
8022 (vect_record_max_nunits): Likewise.
8023 (vect_build_slp_tree_1): Likewise.
8024 (vect_build_slp_tree_2): Likewise.
8025 (vect_print_slp_tree): Likewise.
8026 (vect_analyze_slp_instance): Likewise.
8027 (vect_detect_hybrid_slp_stmts): Likewise.
8028 (vect_detect_hybrid_slp_1): Likewise.
8029 (vect_slp_analyze_operations): Likewise.
8030 (vect_slp_analyze_bb_1): Likewise.
8031 (vect_transform_slp_perm_load): Likewise.
8032 (vect_schedule_slp_instance): Likewise.
8033 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
8034 (vect_mark_stmts_to_be_vectorized): Likewise.
8035 (vect_init_vector_1): Likewise.
8036 (vect_get_vec_def_for_operand): Likewise.
8037 (vect_finish_stmt_generation_1): Likewise.
8038 (vect_check_load_store_mask): Likewise.
8039 (vectorizable_call): Likewise.
8040 (vectorizable_conversion): Likewise.
8041 (vectorizable_operation): Likewise.
8042 (vectorizable_load): Likewise.
8043 (vect_analyze_stmt): Likewise.
8044 (vect_is_simple_use): Likewise.
8045 (vect_get_vector_types_for_stmt): Likewise.
8046 (vect_get_mask_type_for_stmt): Likewise.
8047 * tree-vectorizer.c (increase_alignment): Likewise.
8049 2018-09-19 Andrew Stubbs <ams@codesourcery.com>
8051 * doc/rtl.texi: Adjust vec_select description.
8052 * simplify-rtx.c (simplify_binary_operation_1): Allow VEC_SELECT to use
8053 non-constant selectors.
8055 2018-09-19 Matthew Malcomson <matthew.malcomson@arm.com>
8057 * config/aarch64/aarch64-protos.h
8058 (aarch64_offset_9bit_signed_unscaled_p): New declaration.
8059 * config/aarch64/aarch64.md (arches): New "rcpc8_4" attribute value.
8060 (arch_enabled): Add check for "rcpc8_4" attribute value of "arch".
8061 * config/aarch64/aarch64.h (AARCH64_FL_RCPC8_4): New bitfield.
8062 (AARCH64_FL_FOR_ARCH8_4): Include AARCH64_FL_RCPC8_4.
8063 (AARCH64_FL_PROFILE): Move index so flags are ordered.
8064 (AARCH64_ISA_RCPC8_4): New flag.
8065 * config/aarch64/aarch64.c (offset_9bit_signed_unscaled_p): Renamed
8066 to aarch64_offset_9bit_signed_unscaled_p.
8067 * config/aarch64/atomics.md (atomic_store<mode>): Allow offset
8069 * config/aarch64/constraints.md (Ust): New constraint.
8070 * config/aarch64/predicates.md.
8071 (aarch64_9bit_offset_memory_operand): New predicate.
8072 (aarch64_rcpc_memory_operand): New predicate.
8074 2018-09-19 Eric Botcazou <ebotcazou@adacore.com>
8076 PR rtl-optimization/87361
8077 * rtlanal.c (nonzero_bits1): Revert accidental change.
8079 2018-09-19 Richard Biener <rguenther@suse.de>
8081 PR tree-optimization/87349
8082 PR tree-optimization/87342
8083 * tree-ssa-sccvn.c (do_rpo_vn): Iterate max_rpo computation.
8085 2018-09-18 Marek Polacek <polacek@redhat.com>
8087 P1064R0 - Allowing Virtual Function Calls in Constant Expressions
8088 * gimple-fold.c (gimple_get_virt_method_for_vtable): Adjust assert.
8090 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
8092 * config/rs6000/rs6000.md: Remove old "Cygnus sibcall" comment.
8094 2018-09-18 Segher Boessenkool <segher@kernel.crashing.org>
8096 PR rtl-optimization/86882
8097 * rtlanal.c (reg_overlap_mentioned_p): Handle CLOBBER.
8099 2018-09-18 Uros Bizjak <ubizjak@gmail.com>
8101 * config/i386/i386.md (*<code>extend<mode>xf2): Macroize insn from
8102 *<code>extendsfxf2 and *<code>extenddfxf2 using MODEF mode iterator.
8104 2018-09-18 Jonathan Wakely <jwakely@redhat.com>
8107 * doc/invoke.texi (Link Options): Fix formatting and grammar.
8109 2018-09-18 Richard Biener <rguenther@suse.de>
8112 * tree-ssa-coalesce.c (tree_int_map_hasher): Remove.
8113 (compute_samebase_partition_bases): Likewise.
8114 (coalesce_ssa_name): Always use compute_optimized_partition_bases.
8115 (gimple_can_coalesce_p): Simplify.
8117 2018-09-18 Hans-Peter Nilsson <hp@bitrange.com>
8119 Handle a library implementation of ffs calling __builtin_ffs.
8120 * config/mmix/mmix.c (TARGET_INIT_LIBFUNCS): Override with...
8121 (mmix_init_libfuncs): New function: make __builtin_ffs expand
8124 2018-09-17 David Malcolm <dmalcolm@redhat.com>
8126 * diagnostic-show-locus.c (class layout_range): Add field
8128 (layout_range::layout_range): Add "original_idx" param and use it
8129 to initialize new field.
8130 (make_range): Use 0 for original_idx.
8131 (layout::layout): Pass in index to calls to
8132 maybe_add_location_range.
8133 (layout::maybe_add_location_range): Add param "original_idx" and
8134 pass it on to layout_range.
8135 (layout::print_any_labels): Pass on range->m_original_idx to
8137 (gcc_rich_location::add_location_if_nearby): Use 0 for
8139 * gcc-rich-location.h (text_range_label::get_text): Update for new
8141 (range_label_for_type_mismatch::get_text): Likewise.
8143 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
8145 * config/i386/i386.c (ix86_emit_i387_log1p): Emit fldln2 earlier.
8147 2018-09-17 David Malcolm <dmalcolm@redhat.com>
8149 * gimple-ssa-sprintf.c (fmtwarn): Update for introduction of
8150 format_string_diagnostic_t.
8151 (fmtwarn_n): Likewise.
8152 * substring-locations.c
8153 (format_string_diagnostic_t::format_string_diagnostic_t) New ctor.
8154 (format_warning_n_va): Convert to...
8155 (format_string_diagnostic_t::emit_warning_n_va): ...this.
8156 (format_warning_va): Convert to...
8157 (format_string_diagnostic_t::emit_warning_va): ...this.
8158 (format_warning_at_substring): Convert to...
8159 (format_string_diagnostic_t::emit_warning): ...this.
8160 (format_warning_at_substring_n): Convert to...
8161 (format_string_diagnostic_t::emit_warning_n): ...this.
8162 * substring-locations.h (class format_string_diagnostic_t): New
8164 (format_warning_va): Convert to
8165 format_string_diagnostic_t::emit_warning_va.
8166 (format_warning_n_va): Convert to
8167 format_string_diagnostic_t::emit_warning_n_va.
8168 (format_warning_at_substring): Convert to
8169 format_string_diagnostic_t::emit_warning.
8170 (format_warning_at_substring_n): Convert to
8171 format_string_diagnostic_t::emit_warning_n.
8173 2018-09-17 Cesar Philippidis <cesar@codesourcery.com>
8174 Bernd Schmidt <bernds_cb1@t-online.de>
8176 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with
8179 2018-09-17 Uros Bizjak <ubizjak@gmail.com>
8181 * config/i386/i386.md (truncxf<mode>2_i387_noop_unspec): Change
8182 operand 0 predicate to nonimmediate operand.
8183 (rint<mode>2_frndint): Remove insn pattern.
8184 (rint<mode>2): Change operand 1 predicate to general_operand.
8185 Extend operand 1 to XFmode and generate rintxf2 insn.
8186 (frndintxf2_<rounding>): Rename from frndint<mode>2_<rounding>.
8187 Do not use X87MODEF mode macro.
8188 (frndintxf2_<rounding>_i387): Rename from
8189 frndint<mode>2_<rounding>_i387. Do not use X87MODEF mode macro.
8190 (<rounding_insn><mode>2): For non-SSE modes, extend operand 1
8191 to XFmode and generate significandxf3 insn.
8193 2018-09-17 Richard Biener <rguenther@suse.de>
8195 PR tree-optimization/87328
8196 * tree-ssa-sccvn.c (process_bb): Remove assertion about not
8197 visiting unexecutable backedges when not iterating.
8198 (do_rpo_vn): Mark all edges not executable even when not
8201 2018-09-17 Martin Jambor <mjambor@suse.cz>
8204 * doc/invoke.texi (Warning Options): Likewise.
8206 2018-09-17 Richard Biener <rguenther@suse.de>
8208 PR tree-optimization/87301
8209 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup): Properly
8210 clean EH info from leftover copy assignments.
8212 2018-09-17 Martin Liska <mliska@suse.cz>
8214 PR gcov-profile/85871
8215 * gcov.c (output_intermediate_file): Fix out of bounds
8218 2018-09-17 Vineet Gupta <vgupta@synopsys.com>
8220 * config/arc/arc.c: Object attributes for core4 not reflected
8222 * config/arc/arc.h: Don't restrict DBNZ to core3 (core4 includes
8225 2018-09-17 Alexey Brodkin <abrodkin@synopsys.com>
8227 * config/arc/linux.h (LINK_EH_SPEC): Add missing space.
8229 2018-09-17 Martin Liska <mliska@suse.cz>
8231 * doc/gcov.texi: Document new option --use-hotness-colors.
8232 * gcov.c (struct source_info): Declare new field.
8233 (source_info::source_info): Set default for maximum_count.
8234 (print_usage): Add new -q option.
8235 (process_args): Process it.
8236 (accumulate_line_info): Save src->maximum_count.
8237 (output_line_beginning): Make color line number if
8238 flag_use_hotness_colors is set.
8239 (output_line_details): Pass default argument value.
8240 (output_lines): Pass src->maximum_count.
8242 2018-09-17 Martin Liska <mliska@suse.cz>
8244 * common/config/i386/i386-common.c (ix86_get_valid_option_values):
8245 Use processor_names table.
8246 * config/i386/i386.c (ix86_default_align): Use
8247 processor_cost_table for alignment values.
8248 (ix86_option_override_internal): Use processor_names.
8249 (ix86_function_specific_print): Likewise.
8250 * config/i386/i386.h (struct processor_costs):
8251 Add alignment values.
8252 (struct ptt): Remove and replace with const char *.
8253 * config/i386/x86-tune-costs.h (struct processor_costs):
8254 Declare default alignments for all costs.
8256 2018-09-17 Aldy Hernandez <aldyh@redhat.com>
8258 * tree-vrp.c (extract_range_from_unary_expr): Do not special case
8259 symbolics or VR_VARYING ranges for ABS_EXPR.
8260 * wide-int-range.cc (wide_int_range_abs): Return positive numbers
8261 when range will wrap.
8263 2018-09-15 Eric Botcazou <ebotcazou@adacore.com>
8266 * cfgexpand.c (expand_gimple_basic_block): Be prepared for a BARRIER
8267 before and after a JUMP_TABLE_DATA.
8269 2018-09-14 John David Anglin <danglin@gcc.gnu.org>
8272 * dojump.c (do_compare_and_jump): Canonicalize function pointers
8273 when one operand is a function pointer. Use POINTER_TYPE_P and
8274 FUNC_OR_METHOD_TYPE_P.
8275 * expr.c (do_store_flag): Use POINTER_TYPE_P and FUNC_OR_METHOD_TYPE_P.
8276 * fold-const.c (build_range_check): Likewise.
8277 * match.pd (simple_comparison): Likewise.
8279 2018-09-14 David Malcolm <dmalcolm@redhat.com>
8282 * spellcheck.c (get_edit_distance_cutoff): New function.
8283 (selftest::test_edit_distance_unit_test_oneway): Rename to...
8284 (selftest::test_get_edit_distance_one_way): ...this.
8285 (selftest::test_get_edit_distance_unit): Rename to...
8286 (selftest::test_get_edit_distance_both_ways): ...this.
8287 (selftest::test_edit_distances): Move tests to this new function,
8288 and test some more pairs of strings. Update for above renaming.
8289 (selftest::get_old_cutoff): New function.
8290 (selftest::test_get_edit_distance_cutoff): New function.
8291 (selftest::assert_suggested_for): New function.
8292 (ASSERT_SUGGESTED_FOR): New macro.
8293 (selftest::assert_not_suggested_for): New function.
8294 (ASSERT_NOT_SUGGESTED_FOR): New macro.
8295 (selftest::test_suggestions): New function.
8296 (selftest::spellcheck_c_tests): Move test_get_edit_distance_unit
8297 tests to selftest::test_edit_distances and call it. Add calls to
8298 selftest::test_get_edit_distance_cutoff and
8299 selftest::test_suggestions.
8300 * spellcheck.h (get_edit_distance_cutoff): New function declaration.
8301 (best_match::consider): Replace hard-coded cutoff calculation with
8303 (best_match::get_cutoff): New declaration.
8304 (best_match::get_best_meaningful_candidate): Likewise.
8306 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
8308 * builtins.c (fold_builtin_strlen): Remove TODO comment.
8310 2018-09-14 Bernd Edlinger <bernd.edlinger@hotmail.de>
8313 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
8315 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
8316 terminated string literal.
8318 2018-09-14 Martin Sebor <msebor@redhat.com>
8320 * builtins.c (unterminated_array): Handle ARRAY_REF.
8321 (expand_builtin_stpcpy_1): Detect unterminated char arrays.
8322 * builtins.h (unterminated_array): Declare extern.
8323 * gimple-fold.c (gimple_fold_builtin_stpcpy): Detect unterminated
8325 (gimple_fold_builtin_sprintf): Propagate NO_WARNING to transformed
8328 2018-09-14 Martin Sebor <msebor@redhat.com>
8329 Jeff Law <law@redhat.com>
8331 * builtins.c (unterminated_array): New.
8332 (expand_builtin_strcpy): Adjust.
8333 (expand_builtin_strcpy_args): Detect unterminated arrays.
8334 * gimple-fold.c (get_maxval_strlen): Add argument. Detect
8335 unterminated arrays.
8336 * gimple-fold.h (get_maxval_strlen): Add argument.
8337 (gimple_fold_builtin_strcpy): Detec unterminated arrays.
8339 * gimple-fold.c (get_range_strlen): Add argument.
8340 (get_maxval_strlen): Adjust.
8341 * gimple-fold.h (get_range_strlen): Add argument.
8343 2018-09-14 Wei Xiao <wei3.xiao@intel.com>
8345 * config/i386/movdirintrin.h: Fix copyright year.
8347 2018-09-14 Uros Bizjak <ubizjak@gmail.com>
8349 * reg-stack.c: Include regs.h.
8350 (replace_reg): Assert that mode is MODE_FLOAT or MODE_COMPLEX_FLOAT.
8351 (emit_pop_insn): Default pop insn mode to the reg_raw_mode of
8352 FIRST_STACK_REG, not DFmode.
8353 (emit_swap_insn): Default swap insn mode to the reg_raw_mode of
8354 FIRST_STACK_REG, not XFmode. Explicitly construct swap RTX.
8355 (change stack): Default register mode to the reg_raw_mode of
8356 FIRST_STACK_REG, not DFmode.
8357 * config/i386/i386.md (*swap<mode>): Remove insn pattern.
8358 (*swapxf): Rename from swapxf.
8360 2018-09-14 Carl Love <cel@us.ibm.com>
8362 * config/rs6000/emmintrin.h: Add _MM_SHUFFLE2.
8363 * config/rs6000/xmmintrin.h: Add _MM_SHUFFLE.
8365 2018-09-14 Segher Boessenkool <segher@kernel.crashing.org>
8368 * config/rs6000/rs6000.md (*mov<mode>_hardfloat64): Add Z to the Y
8371 2018-09-14 Sam Tebbs <sam.tebbs@arm.com>
8374 * config/aarch64/aarch64.md (*aarch64_bfxilsi_uxtw): Define.
8376 2018-09-14 Jason Merrill <jason@redhat.com>
8378 Fix --enable-gather-detailed-mem-stats.
8379 * hash-table.c (hash_table_usage): Change from variable to function.
8380 * hash-table.h: Adjust.
8381 * Makefile.in: Add missing dependencies on hash-table.h.
8383 2018-09-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8385 PR tree-optimization/87259
8387 (pass_cse_reciprocals::execute): Run optimize_recip_sqrt after
8388 execute_cse_reciprocals_1 has tried transforming.
8390 2018-09-14 Aldy Hernandez <aldyh@redhat.com>
8392 * tree-vrp.c (extract_range_from_binary_expr_1): Normalize
8393 VR_VARYING for PLUS/MINUS_EXPR.
8395 2018-09-14 Ilya Leoshkevich <iii@linux.ibm.com>
8397 * config/s390/s390-passes.def (INSERT_PASS_BEFORE): Improve
8400 2018-09-14 Richard Biener <rguenther@suse.de>
8403 * tree-ssa-coalesce.c (ssa_conflicts_merge): Remove conflict
8404 bits for the merged partition.
8406 2018-09-13 Martin Sebor <msebor@redhat.com>
8407 Bernd Edlinger <bernd.edlinger@hotmail.de>
8409 * builtins.h (c_srlen): Add argument.
8410 * builtins.c (warn_string_no_nul): New function.
8411 (c_strlen): Add argument and use it. Update recursive calls.
8412 Pass DECL argument to string_constant to get info on non
8413 terminated strings. Update *NONSTR as needed.
8414 (fold_builtin_strlen): Add argument to calls to c_strlen.
8415 Warn for unterminated arrays.
8416 (warn_string_no_null): Add prototype.
8417 * expr.c (string_constant): Update arguments. Update recursive
8418 calls appropriately. Detect missing NUL terminator and outermost
8419 declaration its missing in.
8420 Improve checks for arrays with nonzero lower bound or elements
8421 that are not a single byte. Simplify offset computation.
8422 Simplify checks for non-NUL terminated strings.
8423 * gimple-fold.c (get_range_strlen): Add argument to c_strlen call.
8424 * gimple-ssa-sprintf.c (get_string_length): Remove unnecessary code.
8426 2018-09-13 Bernd Edlinger <bernd.edlinger@hotmail.de>
8428 * builtins.c (c_strlen): Handle not zero terminated STRING_CSTs
8430 * fold-const.c (c_getstr): Fix function comment. Remove unused third
8431 argument. Fix range checks.
8432 * fold-const.h (c_getstr): Adjust protoype.
8433 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid folding when
8434 string is constant but contains no NUL byte.
8436 * expr.c (string_constant): Adjust function comment.
8437 Remove bogus check for zero termination.
8439 * fold-const.c (c_getstr): Clamp STRING_LENGTH to STRING_SIZE.
8441 * varasm.c (compare_constant): Compare type size of STRING_CSTs.
8442 (get_constant_size): Don't make STRING_CSTs larger than they are.
8443 (check_string_literal): New check function for STRING_CSTs.
8444 (output_constant): Use it.
8446 2018-09-13 Eric Botcazou <ebotcazou@adacore.com>
8449 * config/visium/visium.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define.
8451 2018-09-13 Richard Biener <rguenther@suse.de>
8453 PR tree-optimization/87263
8454 * tree-ssa-sccvn.c (visit_phi): Revert some earlier changes.
8455 (struct unwind_state): Add max_rpo field.
8456 (do_rpo_vn): Allow up-to-date loop state to be used when not iterating.
8457 Compute max_rpo, the max RPO number a block can be backwards reached
8458 from. Re-write non-iterating mode to a RPO ordered worklist approach,
8459 separating it from the iterating mode.
8461 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
8463 * haifa-sched.c (rank_for_schedule): Schedule by INSN_COST.
8464 (rfs_decision): New scheduling decision.
8466 2018-09-13 Richard Biener <rguenther@suse.de>
8469 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Fix assert.
8470 (vn_nary_op_insert_pieces_predicated): Do not write useless
8471 valid_dominated_by_p entry outside of the allocated storage.
8473 2018-09-13 Omar Sandoval <osandov@osandov.com>
8474 Tom de Vries <tdevries@suse.de>
8477 * dwarf2out.c (is_c): New function.
8478 (add_subscript_info): Add DW_AT_count of 0 for C zero-length arrays.
8480 2018-09-13 Sam Tebbs <sam.tebbs@arm.com>
8483 * config/aarch64/aarch64.md (*aarch64_bfxil):
8485 * config/aarch64/constraints.md (Ulc): Define.
8486 * config/aarch64/aarch64-protos.h (aarch64_high_bits_all_ones_p):
8488 * config/aarch64/aarch64.c (aarch64_high_bits_all_ones_p):
8491 2018-09-13 Vlad Lazar <vlad.lazar@arm.com>
8493 * config/aarch64/aarch64.h (TARGET_COMPUTE_FRAME_LAYOUT): Define.
8494 * config/aarch64/aarch64.c (aarch64_expand_prologue): Remove
8495 aarch64_layout_frame call.
8496 (aarch64_expand_epilogue): Likewise.
8497 (aarch64_initial_elimination_offset): Likewise.
8498 (aarch64_get_separate_components): Likewise.
8499 (aarch64_use_return_insn_p): Likewise.
8500 (aarch64_layout_frame): Remove unneeded check.
8502 2018-09-13 Jakub Jelinek <jakub@redhat.com>
8504 * configure.ac: Only append
8505 " : (reconfigured) $TOPLEVEL_CONFIGURE_ARGUMENTS" to
8506 gcc_config_arguments if it was never reconfigured or last reconfigure
8507 was with different arguments.
8508 * configure: Regenerated.
8510 2018-09-13 Jakub Jelinek <jakub@redhat.com>
8511 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8514 * expr.c (maybe_optimize_pow2p_mod_cmp): New function.
8515 (maybe_optimize_mod_cmp): Use it if integer_pow2p treeop1.
8517 2018-09-13 Jakub Jelinek <jakub@redhat.com>
8519 PR tree-optimization/87287
8520 * fold-const.c (fold_binary_loc) <case EQ_EXPR>: Move signed modulo
8521 X % C == 0 to X % (unsigned) C == 0 optimization to ...
8522 * match.pd (X % C == 0): ... here. New optimization.
8524 2018-09-12 Jakub Jelinek <jakub@redhat.com>
8527 * expr.h (maybe_optimize_mod_cmp): Declare.
8528 * expr.c (mod_inv): New function.
8529 (maybe_optimize_mod_cmp): New function.
8530 (do_store_flag): Use it.
8531 * cfgexpand.c (expand_gimple_cond): Likewise.
8533 2018-09-09 Cesar Philippidis <cesar@codesourcery.com>
8534 Julian Brown <julian@codesourcery.com>
8537 * gimplify.c (gimplify_scan_omp_clauses): Set
8538 target_firstprivatize_array_bases in OpenACC parallel and kernels
8539 region contexts. Remove GOMP_MAP_FIRSTPRIVATE_REFERENCE clauses from
8540 OpenACC data regions.
8542 2018-09-12 Uros Bizjak <ubizjak@gmail.com>
8544 * config/i386/i386.md (sqrt_extend<mode>xf3_i387): Remove.
8545 (sqrt<mode>2): Extend operand 1 to XFmode and generate sqrtxf3 insn.
8547 2018-09-12 Richard Biener <rguenther@suse.de>
8549 PR tree-optimization/87280
8550 * tree-ssa-sccvn.c (process_bb): Handle the case of executable
8551 edge but unreachable target.
8552 (do_rpo_vn): For conservatively handling a PHI only mark
8553 the backedge executable but not the block reachable.
8555 2018-09-12 Richard Biener <rguenther@suse.de>
8557 PR tree-optimization/87266
8558 * tree-ssa-sccvn.c (do_rpo_vn): Always iterate to not yet
8561 2018-09-12 Andreas Krebbel <krebbel@linux.ibm.com>
8563 * config/s390/s390.md (PFPO_RND_MODE_DFP, PFPO_RND_MODE_BFP): New
8565 ("trunc<BFP:mode><DFP_ALL:mode>2")
8566 ("trunc<DFP_ALL:mode><BFP:mode>2")
8567 ("extend<BFP:mode><DFP_ALL:mode>2")
8568 ("extend<DFP_ALL:mode><BFP:mode>2"): Set proper rounding mode
8569 according to the target operand type.
8571 2018-09-12 Jakub Jelinek <jakub@redhat.com>
8572 Andreas Krebbel <krebbel@linux.ibm.com>
8574 PR tree-optimization/86844
8575 * gimple-ssa-store-merging.c
8576 (imm_store_chain_info::coalesce_immediate): For overlapping stores, if
8577 there are any overlapping stores in between them, make sure they are
8578 also coalesced or we give up completely.
8580 2018-09-12 Jakub Jelinek <jakub@redhat.com>
8583 * fold-const.c (fold_ternary_loc) <case COND_EXPR>: Verify also that
8584 BIT_AND_EXPR's second operand is a power of two. Formatting fix.
8586 2018-09-12 Tom de Vries <tdevries@suse.de>
8588 * common.opt (gdescribe-dies): Add option.
8589 * dwarf2out.c (add_name_and_src_coords_attributes): Add description
8590 attribute for artifical and nameless decls.
8591 (dwarf2out_register_external_die): Add description attribute to
8592 external reference die.
8593 (add_desc_attribute): New functions.
8594 (gen_subprogram_die): Add description attribute to
8595 DW_TAG_call_site_parameter.
8596 * tree-pretty-print.c (print_generic_expr_to_str): New function.
8597 * tree-pretty-print.h (print_generic_expr_to_str): Declare.
8598 * doc/invoke.texi (@item Debugging Options): Add -gdescribe-dies and
8600 (@item -gdescribe-dies): Add.
8602 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
8604 * tree-vrp.c (vrp_shift_undefined_p): Remove.
8605 (extract_range_from_binary_expr_1: Call
8606 wide_int_range_shift_undefined_p instead of vrp_shift_undefined_p.
8607 * wide-int-range.h (wide_int_range_shift_undefined_p): Do not
8610 2018-09-12 Aldy Hernandez <aldyh@redhat.com>
8612 * gimple-ssa-warn-alloca.c
8613 (alloca_type_and_limit::alloca_type_and_limit): Initialize limit
8614 field for ALLOCA_BOUND_*_LARGE.
8616 2018-09-11 Nathan Sidwell <nathan@acm.org>
8618 * gcc.c (load_specs, execute, run_attempt): Use %qs not '%s'.
8620 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
8622 * reg-stack.c (subst_asm_stack_regs): Call replace_reg also
8623 for clobbers. Remove obsolete comment.
8625 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
8627 * config/i386/i386.md (define_attr "type"): Remove mpxmov, mpxmk,
8628 mpxchk, mpxld and mpxst types.
8629 (define_attr length_immediate): Remove all processing of mpx types.
8630 (define_attr prefix_0f): Ditto.
8631 (define_attr memory): Ditto.
8633 2018-09-11 Uros Bizjak <ubizjak@gmail.com>
8635 * config/i386/i386.md (fyl2x_extend<mode>xf3_i387): Remove.
8636 (log<mode>2): Change operand 1 predicate to general_operand.
8637 Extend operand 1 to XFmode and generate logxf3 insn.
8638 (log10<mode>2): Change operand 1 predicate to general_operand.
8639 Extend operand 1 to XFmode and generate log10xf3 insn.
8640 (log2<mode>2): Change operand 1 predicate to general_operand.
8641 Extend operand 1 to XFmode and generate log2xf3 insn.
8642 (fyl2xp1_extend<mode>xf3_i387): Remove.
8643 (log1p<mode>2): Change operand 1 predicate to general_operand.
8644 Extend operand 1 to XFmode and generate log1pxf3 insn.
8645 (fxtract_extend<mode>xf3_i387): Remove.
8646 (logb<mode>2): Change operand 1 predicate to general_operand.
8647 Extend operand 1 to XFmode and generate logbxf3 insn.
8648 (ilogb<mode>2): Change operand 1 predicate to general_operand.
8649 Extend operand 1 to XFmode and generate fxtractxf3_i387 insn.
8650 (significand<mode>2): Change operand 1 predicate to general_operand.
8651 Extend operand 1 to XFmode and generate significandxf3 insn.
8653 2018-09-11 Nathan Sidwell <nathan@acm.org>
8655 * gcc.c (perror_with_name, pfatal_with_name): Delete.
8656 (load_specs): Use fatal_error.
8657 (DELETE_IF_ORDINARY, process_command): Use error.
8658 (execute, run_attempt): Use fatal_error.
8660 2018-09-11 Andrew Stubbs <ams@codesourcery.com>
8662 * diagnostic-core.h (sorry_at): New prototype.
8663 * diagnostic.c (sorry_at): New function.
8665 2018-09-11 Aldy Hernandez <aldyh@redhat.com>
8667 * tree-vrp.c (extract_range_from_binary_expr_1): Treat all divisions
8668 by zero as VR_UNDEFINED.
8670 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
8672 * config/i386/i386.md (<sincos>xf2): Rename from *<sincos>xf2_i387.
8673 (*<sincos>_extend<mode>xf2_i387): Remove insn pattern.
8674 (<sincos>mode2): New expander.
8675 (sincos_extend<mode>xf3_i387): Remove insn pattern.
8676 (sincos -> sin, cos splitters): Remove splitter patterns.
8677 (sincos<mode>3): Change operand 2 predicate to general_operand.
8678 Extend operand 2 to XFmode and generate sincosxf3 insn.
8679 (fptanxf4_i387): Change mode of operands 0 and 3 to SFmode.
8680 Change operand 3 predicate to const1_operand.
8681 (fptan_extend<mode>xf4_i387): Remove insn pattern.
8682 (tanxf2): Update operands in the call to fptanxf4_i387.
8683 (tan<mode>2): Change operand 1 predicate to general_operand.
8684 Extend operand 1 to XFmode and generate tanxf3 insn.
8685 (atan2xf3): Rename from *fpatanxf3_i387.
8686 (fpatan_extend<mode>xf3_i387): Remove insn pattern.
8687 (atan2xf3): Remove expander.
8688 (atan2<mode<3): Change operand 1 and 2 predicates to general_operand.
8689 Extend operands 1 and 2 to XFmode and generate atan2xf3 insn.
8690 (atan<mode>2): Change operand 1 predicate to general_operand.
8691 Extend operand 1 to XFmode and generate atanxf3 insn.
8693 2018-09-10 Uros Bizjak <ubizjak@gmail.com>
8695 * config/i386/i386.md (x87/SSE constant load splitter): Use
8696 memory_operand instead of nonimmediate_operand for input operand
8699 2018-09-09 Uros Bizjak <ubizjak@gmail.com>
8701 * config/i386/i386.md (float partial SSE register stall splitter): Move
8702 splitter near its instruction pattern.
8703 (float_extend partial SSE register stall splitter): Ditto.
8704 (float_truncate partial SSE register stall splitter): Ditto.
8706 2018-09-09 Hans-Peter Nilsson <hp@bitrange.com>
8709 * config/mmix/mmix.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
8710 to speculation_safe_value_not_needed.
8713 * config/mmix/mmix.c (mmix_assemble_integer): Handle byte-size
8714 non-CONST_INT rtx:es using assemble_integer_with_op ".byte".
8715 (MMIX_CFUN_NEEDS_SAVED_EH_RETURN_ADDRESS): Don't call
8716 leaf_function_p, instead use has_hard_reg_initial_val.
8718 2018-09-09 Nathan Sidwell <nathan@acm.org>
8720 * gcc.h (pfatal_with_name): Don't declare here.
8721 * gcc.c (pfatal_with_name): Make static.
8723 2018-09-09 Xianmiao Qu <xianmiao_qu@c-sky.com>
8725 * config/csky/csky.md (*cskyv2_adddi3, *cskyv2_subdi3): Add
8728 2018-09-08 John David Anglin <danglin@gcc.gnu.org>
8730 PR rtl-optimization/85458
8731 * sel-sched.c (sel_target_adjust_priority): Allow backend adjust
8732 priority hook to reduce the priority of EXPR.
8734 2018-09-07 Uros Bizjak <ubizjak@gmail.com>
8736 * config/i386/i386.md (float<SWI48x:mode><MODEF:mode>2) Enable
8737 DImode for x87 on 32bit targets. Conditionally disable x87 modes
8738 with X87_ENABLE_FLOAT. Remove preparation code.
8739 (*float<SWI48:mode><MODEF:mode>2): Rename from
8740 *float<SWI48:mode><MODEF:mode>2_mixed. Handle x87, SSE and mixed
8741 math using "enabled" attribute.
8742 (*floatdi<MODEF:mode>2_i387): Rename from
8743 *float<SWI48x:mode><MODEF:mode>2_i387. Handle only DImode and
8744 enable for 32bit targets only.
8745 (floatdi<X87MODEF:mode>2_i387_with_xmm pre-reload splitter): New
8747 (floatdi<X87MODEF:mode>2_i387_with_xmm): Use register_operand
8748 as operand 1 predicate. Rewrite as define_insn_and_split.
8749 (floatdi<X87MODEF:mode>2_i387_with_xmm memory input splitter): Remove.
8751 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
8753 * reg-stack.c (get_true_reg) <case FLOAT_TRUNCATE>: Reorder
8754 to fallthru to FLOAT case.
8756 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
8759 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Update logic
8760 around folding of vec_sl to handle out of range shift values.
8762 2018-09-06 Uros Bizjak <ubizjak@gmail.com>
8764 * config/i386/i386.md (fix_trunc<mode>_fisttp_i387_1): Remove.
8765 Update callers to gen_fix_trunc<mode>_i387_fisttp
8766 (fix_trunc<mode>_i387_fisttp): Change operand 0 predicate to
8767 nonimmediate_operand.
8768 (fix_trunc<mode>_i387_fisttp_with_temp): Remove insn pattern
8769 and corresponding splitters.
8770 (*fix_trunc<mode>_i387_1): Always expand with fix_trunc<mode>_i387.
8771 (fix_truncdi_i387): Change operand 0 predicate to nonimmediate_operand.
8772 (fix_truncdi_i387_with_temp): Remove insn pattern
8773 and corresponding splitters.
8774 (fix_trunc<mode>_i387): Change operand 0 predicate to
8775 nonimmediate_operand.
8776 (fix_trunc<mode>_i387_with_temp): Remove insn pattern
8777 and corresponding splitters.
8778 (*fistdi2_1): Remove.
8780 (fistdi2_with_temp): Remove insn pattern and corresponding splitters.
8781 (lrintxfdi2): Remove expander. Reimplement as define_insn.
8782 (*fist<mode>2_1): Remove.
8783 (fist<mode>2): Ditto.
8784 (fist<mode>2_with_temp): Remove insn pattern and corresponding
8786 (lrintxf<mode>2): Remove expander. Reimplement as define_insn.
8787 (*fist<mode>2_<rounding>_1): Always expand with fist<mode>2_<rounding>.
8788 (fistdi2_<rounding>): Change operand 0 predicate to
8789 nonimmediate_operand.
8790 (fistdi2_<rounding>_with_temp): Remove insn pattern
8791 and corresponding splitters.
8792 (fist<mode>2_<rounding>): Change operand 0 predicate to
8793 nonimmediate_operand.
8794 (fist<mode>2_<rounding>_with_temp): Remove insn pattern
8795 and corresponding splitters.
8797 (*fixuns_trunc<mode>si2_avx512f_zext): Depend on TARGET_SSE_MATH.
8799 2018-09-06 Bernd Edlinger <bernd.edlinger@hotmail.de>
8801 * varasm.c (output_constructor_regular_field): Check TYPE_SIZE_UNIT of
8804 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
8806 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support for
8807 early gimple folding of vec_splat().
8808 * tree-vect-generic.c: Remove static from tree_vec_extract() definition.
8809 * gimple-fold.h: Add an extern define for tree_vec_extract().
8811 2018-09-06 Will Schmidt <will_schmidt@vnet.ibm.com>
8813 * config/rs6000/rs6000.c (fold_mergehl_helper): Add types_compatible_p
8814 wrappers around TREE_TYPE comparisons.
8816 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
8819 * config/s390/predicates.md: Add nonsym_memory_operand.
8820 * config/s390/s390.c (s390_legitimize_cs_operand): If operand
8821 contains a SYMBOL_REF, load it into an intermediate pseudo.
8822 (s390_emit_compare_and_swap): Legitimize operand.
8823 * config/s390/s390.md: Use the new nonsym_memory_operand
8824 with UNSPECV_CAS patterns.
8826 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
8829 * config/s390/s390-passes.def: New file.
8830 * config/s390/s390-protos.h (class rtl_opt_pass): Add forward
8832 (make_pass_s390_early_mach): Add declaration.
8833 * config/s390/s390.c (make_pass_s390_early_mach):
8834 (s390_option_override): Remove dynamic registration.
8835 * config/s390/t-s390: Add s390-passes.def.
8837 2018-09-06 Ilya Leoshkevich <iii@linux.ibm.com>
8839 * config/s390/s390.c (s390_decompose_constant_pool_ref):
8840 Remove UNSPEC_LTREL_BASE check.
8841 (annotate_constant_pool_refs): Likewise.
8842 (find_constant_pool_ref): Likewise.
8843 (find_ltrel_base): Removed.
8844 (replace_ltrel_base): Removed.
8845 (s390_mainpool_finish): Remove replace_ltrel_base call.
8846 (s390_chunkify_start): Remove pending LTREL_BASE logic.
8847 (s390_chunkify_finish): Remove replace_ltrel_base call.
8848 * config/s390/s390.md: Remove UNSPEC_LTREL_BASE.
8850 2018-09-06 Hans-Peter Nilsson <hp@axis.com>
8853 * config/cris/cris.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
8854 to speculation_safe_value_not_needed.
8856 2018-09-05 Cesar Philippidis <cesar@codesourcery.com>
8857 Bernd Schmidt <bernds_cb1@t-online.de>
8859 * config/nvptx/nvptx-opts.h: New file.
8860 * config/nvptx/nvptx.c (nvptx_file_start): Print the correct .target.
8861 * config/nvptx/nvptx.h: Include "nvptx-opts.h".
8863 (TARGET_SM35): New macro.
8864 * config/nvptx/nvptx.md (atomic_fetch_<logic><mode>): Enable with the
8866 * config/nvptx/nvptx.opt (ptx_isa, sm_30, sm_35): New enum and its
8868 (misa=): New option.
8869 * doc/invoke.texi (Nvidia PTX Options): Document -misa.
8871 2018-09-05 Uros Bizjak <ubizjak@gmail.com>
8873 * config/i386/i386.md (truncdfsf2): Remove expander.
8874 (truncdfsf2_with_temp): Ditto.
8875 (truncxf<mode>2): Ditto.
8876 (*truncdfsf_fast_mixed): Remove insn pattern.
8877 (*truncdfsf_fast_i387): Ditto.
8878 (*truncdfsf_mixed): Ditto.
8879 (*truncdfsf_i387): Ditto.
8880 (*truncdfsf2_i387_1): Ditto.
8881 (*truncxfsf2_mixed): Ditto.
8882 (*truncxfdf2_mixed): Ditto.
8883 (*truncxf<mode>2_i387_noop): Ditto. Update callers
8884 to call gen_truncxf<mode>2 instead.
8885 (*truncxf<mode>2_i387): Remove.
8886 (reg->reg splitters): Remove splitter pattern.
8887 (reg->mem splitters): Ditto.
8889 (truncdfsf2): New insn pattern.
8890 (truncxf<mode>2): Ditto.
8892 2018-09-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8894 * tree-ssa-math-opts.c (is_mult_by): New function.
8895 (is_square_of): Use the above.
8896 (optimize_recip_sqrt): New function.
8897 (pass_cse_reciprocals::execute): Use the above.
8899 2018-09-05 Richard Biener <rguenther@suse.de>
8902 * tree-ssa-sccvn.c (rpo_elim::eliminate_push_avail): Make sure
8903 to zero-init the emplaced vec.
8905 2018-09-05 Martin Liska <mliska@suse.cz>
8907 PR tree-optimization/87205
8908 * tree-switch-conversion.c (pass_lower_switch::execute):
8909 Group cases for switch statements.
8911 2018-09-05 Richard Biener <rguenther@suse.de>
8913 PR tree-optimization/87217
8914 * tree-ssa-sccvn.c (vuse_valueize): New.
8915 (vn_reference_lookup_pieces): Use it.
8916 (vn_reference_lookup): Likewise.
8918 2018-09-05 Nathan Sidwell <nathan@acm.org>
8921 * stor-layout.c (place_field): Scan forwards to check last
8922 bitfield when ms_bitfield_placement is in effect.
8924 2018-09-05 Richard Biener <rguenther@suse.de>
8927 * tree-vect-stmts.c (vectorizable_simd_clone_call): Fix bogus
8930 2018-09-05 Siddhesh Poyarekar <siddhesh@sourceware.org>
8931 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
8933 * config/aarch64/falkor-tag-collision-avoidance.c: New file.
8934 * config.gcc (extra_objs): Build it.
8935 * config/aarch64/t-aarch64 (falkor-tag-collision-avoidance.o):
8937 * config/aarch64/aarch64-passes.def
8938 (pass_tag_collision_avoidance): New pass.
8939 * config/aarch64/aarch64.c (qdf24xx_tunings): Add
8940 AARCH64_EXTRA_TUNE_RENAME_LOAD_REGS to tuning_flags.
8941 (aarch64_classify_address): Remove static qualifier.
8942 (aarch64_address_info, aarch64_address_type): Move to...
8943 * config/aarch64/aarch64-protos.h: ... here.
8944 (make_pass_tag_collision_avoidance): New function.
8945 * config/aarch64/aarch64-tuning-flags.def (rename_load_regs):
8948 2018-09-05 Martin Liska <mliska@suse.cz>
8950 * doc/gcov.texi: Update documentation of humar
8952 * gcov.c (format_count): Print one decimal place, it provides
8953 more fine number of situations like '1G' vs. '1.4G'.
8955 2018-09-05 Martin Liska <mliska@suse.cz>
8958 * config/rs6000/rs6000.opt: Mark the option as Deprecated.
8959 * optc-gen.awk: Allow 'Var' for Deprecated options in order
8960 to generate a MASK value.
8962 2018-09-04 H.J. Lu <hongjiu.lu@intel.com>
8965 * dwarf2out.c (based_loc_descr): Allow hard frame pointer even
8966 if frame pointer isn't used.
8967 (compute_frame_pointer_to_fb_displacement): Likewise.
8969 2018-09-04 Jakub Jelinek <jakub@redhat.com>
8972 * common/config/i386/i386-common.c (OPTION_MASK_ISA_XSAVEOPT_SET,
8973 OPTION_MASK_ISA_XSAVES_SET, OPTION_MASK_ISA_XSAVEC_SET): Use
8974 OPTION_MASK_ISA_XSAVE_SET instead of OPTION_MASK_ISA_XSAVE.
8975 (OPTION_MASK_ISA_XSAVE_UNSET): Add OPTION_MASK_ISA_XSAVES_UNSET
8976 and OPTION_MASK_ISA_XSAVEC_UNSET.
8978 2018-09-04 Max Filippov <jcmvbkbc@gmail.com>
8980 * config/xtensa/xtensa.c (xtensa_expand_atomic): Reorder AND and
8981 XOR operations in NAND case.
8983 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
8985 * wide-int-range.cc (wide_int_range_convert): New.
8986 * wide-int-range.h (wide_int_range_convert): New.
8987 * tree-vrp.c (extract_range_from_unary_expr): Abstract wide int
8988 code into wide_int_range_convert.
8989 (extract_range_into_wide_ints): Do not munge anti range constants
8990 into the entire domain. Just return the range back.
8992 2018-09-04 Martin Liska <mliska@suse.cz>
8994 * genmatch.c (output_line_directive): Add new argument
8996 (dt_simplify::gen_1): Encapsulate dump within __builtin_expect.
8998 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
9000 * doc/invoke.texi (Option Summary): Add whitespace.
9002 * doc/invoke.texi (Option Summary): Add -Waligned-new.
9004 2018-09-04 Richard Biener <rguenther@suse.de>
9006 PR tree-optimization/87211
9007 * tree-ssa-sccvn.c (visit_phi): When value-numbering to a
9008 backedge value we're supposed to treat as VARYING also number
9009 the PHI to VARYING in case it got a different value-number already.
9011 2018-09-04 Aldy Hernandez <aldyh@redhat.com>
9013 * tree-vrp.c (vrp_can_optimize_bit_op): Remove.
9014 (extract_range_from_binary_expr_1): Do not call
9015 vrp_can_optimize_bit_op.
9016 * wide-int-range.cc (wide_int_range_can_optimize_bit_op): Make
9018 (wide_int_range_get_mask_and_bounds): New.
9019 (wide_int_range_optimize_bit_op): New.
9020 (wide_int_range_bit_ior): Call wide_int_range_optimize_bit_op.
9021 (wide_int_range_bit_and): Same.
9022 * wide-int-range.h (wide_int_range_can_optimize_bit_op): Remove.
9023 (wide_int_range_optimize_bit_op): New.
9024 (wide_int_range_get_mask_and_bounds): New.
9026 2018-09-04 Richard Biener <rguenther@suse.de>
9028 PR tree-optimization/87176
9029 * tree-ssa-sccvn.c (visit_phi): Remove redundant allsame
9030 variable. When value-numbering a virtual PHI node make sure
9031 to not value-number to the backedge value.
9033 2018-09-04 Jonathan Wakely <jwakely@redhat.com>
9035 * doc/extend.texi (Long Long, Hex Floats): Document support for
9036 long long and hex floats in more recent versions of ISO C++.
9038 2018-09-03 Richard Biener <rguenther@suse.de>
9040 PR tree-optimization/87177
9041 * tree-ssa-sccvn.c (vuse_ssa_val): Revert previous change, keep
9044 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9046 * bb-reorder.c (edge_order): Convert to C-qsort-style
9047 tri-state comparator.
9048 (reorder_basic_blocks_simple): Change std::stable_sort to
9051 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9053 * tree-loop-distribution.c (offset_cmp): Convert to C-qsort-style
9054 tri-state comparator.
9055 (fuse_memset_builtins): Change std::stable_sort to gcc_stablesort.
9057 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9059 * sort.cc (struct sort_ctx): New field 'nlim'. Use it...
9060 (mergesort): ... here as maximum count for using netsort.
9061 (gcc_qsort): Set nlim to 3 if stable sort is requested.
9062 (gcc_stablesort): New.
9063 * system.h (gcc_stablesort): Declare.
9065 2018-09-03 Alexander Monakov <amonakov@ispras.ru>
9067 * sort.cc (gcc_qsort) [CHECKING_P]: Call qsort_chk.
9068 * system.h (qsort): Always redirect to gcc_qsort. Update comment.
9069 * vec.c (qsort_chk): Do not call gcc_qsort. Update comment.
9071 2018-09-03 Segher Boessenkool <segher@kernel.crashing.org>
9073 * config/rs6000/rs6000.md (*mov<mode>_hardfloat32): Remove %U from the
9074 lxsdx and stxsdx alternatives.
9075 (*mov<mode>_hardfloat64): Ditto.
9076 * config/rs6000/vsx.md (*vsx_extract_<mode>_store): Ditto.
9078 2018-09-03 Richard Biener <rguenther@suse.de>
9080 PR tree-optimization/87200
9081 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Valueize a
9084 2018-09-03 Martin Liska <mliska@suse.cz>
9086 PR tree-optimization/87201
9087 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
9088 Fix parenthesis in an expression.
9090 2018-09-03 Richard Biener <rguenther@suse.de>
9092 PR tree-optimization/87197
9093 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Mark the new def
9094 visited. CSE the VN_INFO hashtable lookup.
9096 PR tree-optimization/87169
9097 * tree-ssa-sccvn.c (do_rpo_vn): When marking loops for not
9098 iterating make sure there's no extra backedges from irreducible
9099 regions feeding the header. Mark the destination block
9102 2018-09-03 Martin Liska <mliska@suse.cz>
9105 * common/common-target.def: Add TARGET_GET_VALID_OPTION_VALUES.
9106 * common/common-targhooks.c (default_get_valid_option_values):
9108 * common/common-targhooks.h (default_get_valid_option_values):
9110 * common/config/i386/i386-common.c: Move processor_target_table
9112 (ix86_get_valid_option_values): New function.
9113 (TARGET_GET_VALID_OPTION_VALUES): New macro.
9114 * config/i386/i386.c (struct ptt): Move to i386-common.c.
9115 (PTA_*): Move all defined masks into i386-common.c.
9116 (ix86_function_specific_restore): Use new processor_cost_table.
9117 * config/i386/i386.h (struct ptt): Moved from i386.c.
9118 (struct pta): Likewise.
9119 * doc/tm.texi: Document new TARGET_GET_VALID_OPTION_VALUES.
9120 * doc/tm.texi.in: Likewise.
9121 * opt-suggestions.c (option_proposer::suggest_option):
9122 Pass prefix to build_option_suggestions.
9123 (option_proposer::get_completions): Likewise.
9124 (option_proposer::build_option_suggestions): Use the new target
9126 * opts.c (struct option_help_tuple): New struct.
9127 (print_filtered_help): Use the new target hook.
9129 2018-09-03 Martin Liska <mliska@suse.cz>
9132 * predict.c (set_even_probabilities): Add likely_edges
9133 argument and handle cases where we have precisely one
9135 (combine_predictions_for_bb): Catch also likely_edges.
9136 (tree_predict_by_opcode): Handle gswitch statements.
9137 * tree-cfg.h (find_case_label_for_value): New declaration.
9138 (find_taken_edge_switch_expr): Likewise.
9139 * tree-switch-conversion.c (switch_decision_tree::balance_case_nodes):
9140 Find pivot in decision tree based on probabily, not by number of
9143 2018-09-02 Gerald Pfeifer <gerald@pfeifer.com>
9145 * doc/standards.texi (Standards): Update Objective-C reference.
9147 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
9149 * doc/install.texi (Prerequisites): Update link for MPC.
9151 2018-09-01 Michael Matz <matz@suse.de>
9153 PR tree-optimization/87074
9154 * gimple-loop-jam.c (unroll_jam_possible_p): Check loop exit
9155 PHIs for outer-loop uses.
9157 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
9159 * doc/generic.texi (OpenMP): Adjust link to openmp.org.
9160 * doc/invoke.texi (C Dialect Options): Ditto.
9162 2018-09-01 Gerald Pfeifer <gerald@pfeifer.com>
9164 * doc/install.texi (Prerequisites): Adjust link mpfr.org.
9166 2018-08-31 Richard Biener <rguenther@suse.de>
9168 PR tree-optimization/87168
9169 * tree-ssa-sccvn.c (SSA_VAL): Add visited output parameter.
9170 (rpo_elim::eliminate_avail): When OP was not visited it must
9173 2018-08-31 David Malcolm <dmalcolm@redhat.com>
9175 * tree-vrp.c (copy_value_range): Convert param "from" from
9176 "value_range *" to "const value_range *".
9177 (range_is_null): Likewise for param "vr".
9178 (range_int_cst_p): Likewise.
9179 (range_int_cst_singleton_p): Likewise.
9180 (symbolic_range_p): Likewise.
9181 (value_ranges_intersect_p): Likewise for both params.
9182 (value_range_nonnegative_p): Likewise for param "vr".
9183 (value_range_constant_singleton): Likewise.
9184 (vrp_set_zero_nonzero_bits): Likewise for param "ar".
9185 (extract_range_into_wide_ints): Likewise for param "vr".
9186 (extract_range_from_multiplicative_op): Likewise for params "vr0"
9188 (vrp_can_optimize_bit_op): Likewise.
9189 (extract_range_from_binary_expr_1): Likewise for params "vr0_" and
9191 (extract_range_from_unary_expr): Likewise.
9192 (debug_value_range): Likewise for param "vr".
9193 (value_range::dump): Add "const" qualifier.
9194 (vrp_prop::check_array_ref): Convert local "vr" from
9195 "value_range *" to "const value_range *".
9196 (vrp_prop::check_mem_ref): Likewise.
9197 (vrp_prop::visit_stmt): Likewise for local "old_vr".
9198 (vrp_intersect_ranges_1): Likewise for param "vr_1".
9199 (vrp_intersect_ranges): Likewise.
9200 (simplify_stmt_for_jump_threading): Likewise for local "vr".
9201 (vrp_prop::vrp_finalize): Likewise.
9202 * tree-vrp.h (value_range::dump): Add "const" qualifier.
9203 (vrp_intersect_ranges): Add "const" qualifier to params as above.
9204 (extract_range_from_unary_expr): Likewise.
9205 (value_range_constant_singleton): Likewise.
9206 (symbolic_range_p): Likewise.
9207 (copy_value_range): Likewise.
9208 (extract_range_from_binary_expr_1): Likewise.
9209 (range_int_cst_p): Likewise.
9210 (vrp_set_zero_nonzero_bits): Likewise.
9211 (range_int_cst_singleton_p): Likewise.
9213 2018-08-31 Vlad Lazar <vlad.lazar@arm.com>
9215 * config/aarch64/arm_neon.h (vabsd_s64): New.
9216 (vnegd_s64): Likewise.
9218 2018-08-31 Martin Jambor <mjambor@suse.cz>
9220 * ipa-cp.c (estimate_local_effects): Replace wrong MAX with MIN.
9222 2018-08-31 Martin Liska <mliska@suse.cz>
9224 * ipa-icf.c (sem_item::add_type): Use
9225 sem_item::m_type_hash_cache.
9226 * ipa-icf.h: Move the cache from sem_item_optimizer
9229 2018-08-31 Nathan Sidwell <nathan@acm.org>
9231 * doc/extend.texi (Backwards Compatibility): Remove implicit
9232 extern C leeway of () being (...).
9234 2018-08-31 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9236 * ipa-inline.c (can_inline_edge_by_limits_p): Fix typos in comment.
9238 2018-08-31 Segher Boessenkool <segher@kernel.crashing.org>
9242 * config/rs6000/rs6000.md (lround<mode>di2): Gate on TARGET_FPRND.
9244 2018-08-31 Jakub Jelinek <jakub@redhat.com>
9247 * expmed.c (expand_mult_const): Use immed_wide_int_const instead of
9248 gen_int_mode. Formatting fixes.
9250 2018-08-30 Sandra Loosemore <sandra@codesourcery.com>
9252 * target.def (custom_function_descriptors): Improve documentation.
9253 * doc/tm.texi.in (Trampolines): Expand discussion of function
9254 descriptors and move TARGET_CUSTOM_FUNCTION_DESCRIPTORS to the
9255 beginning of the section.
9256 * doc/tm.texi: Regenerated.
9258 2018-08-30 Jose E. Marchesi <jose.marchesi@oracle.com>
9260 * cfg.h (class auto_edge_flag): Spell out the template-id of the
9261 base class in the initializer list. This is a workaround for
9262 building with older GCC.
9263 (class auto_bb_flag): Likewise.
9265 2018-08-30 Aaron Sawdey <acsawdey@linux.ibm.com>
9267 * config/rs6000/altivec.md (altivec_eq<mode>): Remove star.
9268 (altivec_vcmpequ<VI_char>_p): Remove star.
9269 * config/rs6000/rs6000-string.c (do_load_for_compare): Support
9271 (expand_strncmp_vec_sequence): New function.
9272 (emit_final_str_compare_vec): New function.
9273 (expand_strn_compare): Add support for vector strncmp.
9274 * config/rs6000/rs6000.opt (-mstring-compare-inline-limit): Change
9275 length specification to bytes.
9276 * config/rs6000/vsx.md (vsx_ld_elemrev_v16qi_internal): Remove star.
9277 (vcmpnezb_p): New pattern.
9278 * doc/invoke.texi (RS/6000 and PowerPC Options): Update documentation
9279 for option -mstring-compare-inline-limit.
9281 2018-08-30 Thiago Macieira <thiago.macieira@intel.com>
9283 * config/i386/i386.c (PTA_WESTMERE): Remove PTA_AES.
9284 (PTA_SKYLAKE): Add PTA_AES.
9285 (PTA_GOLDMONT): Likewise.
9287 2018-08-29 Jan Hubicka <jh@suse.cz>
9290 * lto-opts.c (lto_write_options): Always stream PIC/PIE mode.
9291 * lto-wrapper.c (merge_and_complain): Fix merging of PIC/PIE.
9293 2018-08-29 Jan Hubicka <jh@suse.cz>
9295 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not follow
9297 (hash_tree): Do not visit TYPE_STUB_DECL.
9298 * tree-streamer-out.c (write_ts_type_common_tree_pointers): Do not
9299 stream TYPE_STUB_DECL.
9300 * tree-streamer-in.c (lto_input_ts_type_common_tree_pointers): Likewise.
9301 * ipa-utils.h (type_with_linkage_p): Do not rely on TYPE_STUB_DECL
9302 after free_lang_data.
9303 (type_in_anonymous_namespace_p): Likewise.
9305 2018-08-29 Jan Hubicka <jh@suse.cz>
9307 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
9308 comment that it has to be even number.
9309 (class sreal): Change m_sig type to int32_t.
9310 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
9311 int64_t for temporary calculations.
9312 (sreal_verify_basics): Drop one bit from minimum and maximum.
9314 2018-08-30 Richard Biener <rguenther@suse.de>
9316 PR tree-optimization/87147
9317 * tree-ssa-sccvn.c (SSA_VISITED): New function.
9318 (visit_phi): When the degenerate result is from the backedge and
9319 we didn't visit its definition yet drop to VARYING.
9320 (do_rpo_vn): Properly mark blocks with incoming backedges as executable.
9322 2018-08-29 Jan Hubicka <jh@suse.cz>
9324 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not walk
9326 (hash_tree): Likewise.
9328 2018-08-29 Jan Hubicka <jh@suse.cz>
9330 * tree.c (find_decls_types_r): Walk also TYPE_NEXT_PTR_TO
9331 and TYPE_NEXT_REF_TO.
9333 2018-08-29 Jan Hubicka <jh@suse.cz>
9335 * sreal.h (SREAL_PART_BITS): Change to 31; remove seemingly unnecessary
9336 comment that it has to be even number.
9337 (class sreal): Change m_sig type to int32_t.
9338 * sreal.c (sreal::dump, sreal::to_int, opreator+, operator-): Use
9339 int64_t for temporary calculations.
9340 (sreal_verify_basics): Drop one bit from minimum and maximum.
9342 2018-08-30 Tamar Christina <tamar.christina@arm.com>
9344 * config/aarch64/aarch64.c (aarch64_expand_movmem): Set TImode max.
9346 2018-08-30 Vlad Lazar <vlad.lazar@arm.com>
9349 * expmed.c (canonicalize_comparison): Use wi::sub instead of wi::add
9350 if to_add is negative.
9352 2018-08-29 Bernd Edlinger <bernd.edlinger@hotmail.de>
9355 * builtins.c (c_strlen): Improve range checks.
9357 2018-08-29 Martin Sebor <msebor@redhat.com>
9358 Jeff Law <law@redhat.com>
9360 PR tree-optimization/86714
9361 PR tree-optimization/86711
9362 * builtins.c (c_strlen): Add arguments to call to string_constant.
9363 * expr.c (string_constant): Add argument. Detect missing nul
9364 terminator and outermost declaration it's missing in.
9365 * expr.h (string_constant): Add argument.
9366 * fold-const.c (read_from_constant_string): Add arguments to call to
9368 (c_getstr): Likewise.
9369 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
9371 * tree-ssa-strlen.c (get_stridx): Likewise.
9373 2018-08-29 Jan Hubicka <jh@suse.cz>
9375 * tree-streamer-in.c (lto_input_ts_function_decl_tree_pointers):
9376 Do not stream DECL_VINDEX.
9377 * tree-streamer-out.c (write_ts_function_decl_tree_pointers): Likewise.
9378 * tree.c (free_lang_data_in_decl): Clear DECL_VINDEX.
9379 (decl_function_context): Use DECL_VIRTUAL_P rather than DECL_VINDEX.
9381 2018-08-29 Richard Biener <rguenther@suse.de>
9383 * tree-ssa-sccvn.c (vuse_ssa_val): Return NULL for unvisited
9384 virtual operands that are not default defs to honor region
9386 (rpo_vn_valueize): Remove ineffective code here.
9388 2018-08-29 Richard Biener <rguenther@suse.de>
9390 PR tree-optimization/87132
9391 * tree-ssa-alias.c (get_continuation_for_phi): Do not translate
9392 when skipping defs reachable over backedges.
9394 2018-08-29 Richard Biener <rguenther@suse.de>
9396 * tree-core.h: Document use of deprecated_flag in SSA_NAME.
9397 * tree.h (SSA_NAME_POINTS_TO_READONLY_MEMORY): Define.
9398 * tree-into-ssa.c (pass_build_ssa::execute): Initialize
9399 function parameters SSA_NAME_POINTS_TO_READONLY_MEMORY from fnspec.
9400 * tree-ssa-sccvn.c (const_parms, init_const_parms): Remove.
9401 (vn_reference_lookup_3): Remove use of const_parms.
9402 (free_rpo_vn): Do not free const_parms.
9403 (do_rpo_vn): Do not call init_const_parms.
9404 * tree-ssa-alias.c (refs_may_alias_p_1): Honor
9405 SSA_NAME_POINTS_TO_READONLY_MEMORY.
9406 (call_may_clobber_ref_p_1): Likewise.
9408 2018-08-29 Alexander Monakov <amonakov@ispras.ru>
9411 * invoke.texi (Optimization Options): List -ftree-scev-cprop.
9413 (-ftree-scev-cprop): Document.
9415 2018-08-29 Jan Hubicka <jh@suse.cz>
9417 * sreal.h (normalize, normalize_up, normalize_down): Add new_sig/new_exp
9419 (sreal constructor): Update.
9420 * sreal.c (sreal:operator+, sreal:operator-, sreal:operator*,
9421 sreal:operator/): Update.
9423 2018-08-29 Martin Liska <mliska@suse.cz>
9425 * tree-switch-conversion.c (switch_conversion::expand):
9426 Strenghten assumption about gswitch statements.
9428 2018-08-29 Richard Biener <rguenther@suse.de>
9430 PR tree-optimization/87117
9431 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_stmt): Only
9432 re-value-number released SSA VDEFs.
9434 2018-08-29 Richard Biener <rguenther@suse.de>
9436 PR tree-optimization/87126
9437 * tree-ssa-sccvn.c (vn_reference_insert): Remove assert.
9439 2018-08-28 Jim Wilson <jimw@sifive.com>
9441 * config/riscv/pic.md: Rewrite.
9442 * config/riscv/riscv.c (riscv_address_insns): Return cost of 3 for
9444 * config/riscv/riscv.md (ZERO_EXTEND_LOAD): Delete.
9445 (SOFTF, default_load, softload, softstore): New.
9447 2018-08-28 Jeff Law <law@redhat.com>
9449 * fold-const.c (fold_binary_loc): Remove recently added assert.
9451 2018-08-28 Joern Rennecke <joern.rennecke@riscy-ip.com>
9453 * genpreds.c (write_predicate_subfunction): Also add ATTRIBUTE_UNUSED
9454 to OP parmeter of generated function.
9456 2018-08-28 MCC CS <deswurstes@users.noreply.github.com>
9458 PR tree-optimization/87009
9459 * match.pd: Add boolean optimizations.
9461 2018-08-28 Martin Sebor <msebor@redhat.com>
9464 * calls.c (alloc_max_size): Treat HOST_WIDE_INT special.
9465 * gimple-ssa-warn-alloca.c (adjusted_warn_limit): New function.
9466 (pass_walloca::gate): Use it.
9467 (alloca_call_type): Same.
9468 (pass_walloca::execute): Same.
9469 * stor-layout.c (layout_decl): Treat HOST_WIDE_INT special.
9471 2018-08-28 David Malcolm <dmalcolm@redhat.com>
9473 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): Change version check on
9474 GCC_VERSION for usage of "__gcc_dump_printf__" format from
9477 2018-08-28 Richard Biener <rguenther@suse.de>
9479 PR tree-optimization/87124
9480 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Guard against
9481 constants before looking up avail.
9483 2018-08-28 Jakub Jelinek <jakub@redhat.com>
9486 * calls.c (maybe_warn_nonstring_arg): Punt early if
9487 warn_stringop_overflow is zero. Don't call get_range_strlen
9488 on 3rd argument, keep iterating until lenrng[1] is INTEGER_CST.
9489 Swap comparison operands to have constants on rhs. Only use
9490 lenrng[1] if non-NULL and INTEGER_CST. Don't uselessly
9491 increment lenrng[0].
9493 2018-08-28 Richard Sandiford <richard.sandiford@arm.com>
9495 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Fix unguarded
9496 use of tree_to_shwi. Remove duplicated test for the size being
9497 a whole number of bytes.
9499 2018-08-28 Richard Biener <rguenther@suse.de>
9501 PR tree-optimization/87117
9502 * tree-ssa-sccvn.c (eliminate_dom_walker::eliminate_cleanup):
9503 Handle removed stmt without LHS (GIMPLE_NOP).
9505 2018-08-28 Richard Biener <rguenther@suse.de>
9507 PR tree-optimization/87117
9508 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Exclude
9509 void which is is_gimple_reg_type by checking for COMPLETE_TYPE_P.
9511 2018-08-28 Richard Biener <rguenther@suse.de>
9513 PR tree-optimization/87117
9514 * tree-ssa-pre.c (compute_avail): Do not make expressions
9515 with predicated values available.
9516 (get_expr_value_id): Assert we do not run into predicated value
9519 2018-08-28 Richard Biener <rguenther@suse.de>
9521 PR tree-optimization/87117
9522 * tree-ssa-operands.c (add_stmt_operand): STRING_CST may
9523 get virtual operands.
9524 (get_expr_operands): Handle STRING_CST like other decls.
9526 2018-08-28 Martin Liska <mliska@suse.cz>
9528 * tree.h: Update documentation of fndecl_built_in_p
9532 2018-08-27 Jeff Law <law@redhat.com>
9533 PR tree-optimization/87110
9534 * tree-ssa-dse.c (compute_trims): Handle non-constant
9537 2018-08-27 Martin Sebor <msebor@redhat.com>
9539 PR tree-optimization/86914
9540 * tree-ssa-strlen.c (maybe_set_strlen_range): Avoid MEM_REF.
9542 2018-08-27 Martin Sebor <msebor@redhat.com>
9544 PR tree-optimization/87112
9545 * builtins.c (expand_builtin_strnlen): Convert c_strlen result to
9546 the type of the bound argument.
9548 2018-08-27 Jeff Law <law@redhat.com>
9550 * tree-ssa-dse.c (compute_trims): Handle case where the reference's
9551 type does not have a TYPE_SIZE_UNIT.
9553 2018-08-27 Steve Ellcey <sellcey@cavium.com>
9555 * config/aarch64/aarch64-speculation.cc: Replace include of cfg.h
9556 with include of backend.h.
9558 2018-08-27 Richard Biener <rguenther@suse.de>
9560 PR tree-optimization/86927
9561 * tree-vect-loop.c (vect_create_epilog_for_reduction): Properly
9562 use const cond reduction code.
9564 2018-08-27 Alexander Monakov <amonakov@ispras.ru>
9566 PR tree-optimization/85758
9567 * match.pd ((X & Y) ^ Y): Add :s qualifier to inner expression.
9569 2018-08-27 David Malcolm <dmalcolm@redhat.com>
9572 * diagnostic-show-locus.c (class layout_range): Update for
9573 conversion of show_caret_p to a tri-state.
9574 (layout_range::layout_range): Likewise.
9575 (make_range): Likewise.
9576 (layout::maybe_add_location_range): Likewise.
9577 (layout::should_print_annotation_line_p): Don't show annotation
9578 lines for ranges that are SHOW_LINES_WITHOUT_RANGE.
9579 (layout::get_state_at_point): Update for conversion of
9580 show_caret_p to a tri-state. Bail out early for
9581 SHOW_LINES_WITHOUT_RANGE, so that such ranges don't affect
9582 underlining or source colorization.
9583 (gcc_rich_location::add_location_if_nearby): Update for conversion
9584 of show_caret_p to a tri-state.
9585 (selftest::test_one_liner_multiple_carets_and_ranges): Likewise.
9586 (selftest::test_one_liner_fixit_replace_equal_secondary_range):
9588 (selftest::test_one_liner_labels): Likewise.
9589 * gcc-rich-location.c (gcc_rich_location::add_expr): Update for
9590 conversion of show_caret_p to a tri-state.
9591 * pretty-print.c (text_info::set_location): Likewise.
9592 * pretty-print.h (text_info::set_location): Likewise.
9593 * substring-locations.c (format_warning_n_va): Likewise.
9594 * tree-diagnostic.c (default_tree_printer): Likewise.
9595 * tree-pretty-print.c (newline_and_indent): Likewise.
9597 2018-08-27 David Malcolm <dmalcolm@redhat.com>
9600 * diagnostic-show-locus.c (get_line_span_for_fixit_hint): Show the
9601 line above for line-insertion fix-it hints.
9602 (selftest::test_fixit_insert_containing_newline): Update the
9603 expected results, and add a test with line-numbering enabled.
9605 2018-08-27 Martin Liska <mliska@suse.cz>
9608 * sanopt.c (sanitize_rewrite_addressable_params): Ignore
9609 params with DECL_HAS_VALUE_EXPR_P.
9611 2018-08-27 Martin Liska <mliska@suse.cz>
9613 * config/i386/i386.c (ix86_expand_set_or_movmem): Dump
9614 selected expansion strategy.
9616 2018-08-27 Martin Liska <mliska@suse.cz>
9618 * builtins.h (is_builtin_fn): Remove and fndecl_built_in_p.
9619 * builtins.c (is_builtin_fn): Likewise.
9620 * attribs.c (diag_attr_exclusions): Use new function
9621 fndecl_built_in_p and remove check for FUNCTION_DECL if
9623 (builtin_mathfn_code): Likewise.
9624 (fold_builtin_expect): Likewise.
9625 (fold_call_expr): Likewise.
9626 (fold_builtin_call_array): Likewise.
9627 (fold_call_stmt): Likewise.
9628 (set_builtin_user_assembler_name): Likewise.
9629 (is_simple_builtin): Likewise.
9630 * calls.c (gimple_alloca_call_p): Likewise.
9631 (maybe_warn_nonstring_arg): Likewise.
9632 * cfgexpand.c (expand_call_stmt): Likewise.
9633 * cgraph.c (cgraph_update_edges_for_call_stmt_node): Likewise.
9634 (cgraph_edge::verify_corresponds_to_fndecl): Likewise.
9635 (cgraph_node::verify_node): Likewise.
9636 * cgraphclones.c (build_function_decl_skip_args): Likewise.
9637 (cgraph_node::create_clone): Likewise.
9638 * config/arm/arm.c (arm_insert_attributes): Likewise.
9639 * config/i386/i386.c (ix86_gimple_fold_builtin): Likewise.
9640 * dse.c (scan_insn): Likewise.
9641 * expr.c (expand_expr_real_1): Likewise.
9642 * fold-const.c (operand_equal_p): Likewise.
9643 (fold_binary_loc): Likewise.
9644 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
9645 * gimple-low.c (lower_stmt): Likewise.
9646 * gimple-pretty-print.c (dump_gimple_call): Likewise.
9647 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
9649 * gimple.c (gimple_build_call_from_tree): Likewise.
9650 (gimple_call_builtin_p): Likewise.
9651 (gimple_call_combined_fn): Likewise.
9652 * gimplify.c (gimplify_call_expr): Likewise.
9653 (gimple_boolify): Likewise.
9654 (gimplify_modify_expr): Likewise.
9655 (gimplify_addr_expr): Likewise.
9656 * hsa-gen.c (gen_hsa_insns_for_call): Likewise.
9657 * ipa-cp.c (determine_versionability): Likewise.
9658 * ipa-fnsummary.c (compute_fn_summary): Likewise.
9659 * ipa-param-manipulation.c (ipa_modify_formal_parameters): Likewise.
9660 * ipa-split.c (visit_bb): Likewise.
9661 (split_function): Likewise.
9662 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
9663 * lto-cgraph.c (input_node): Likewise.
9664 * lto-streamer-out.c (write_symbol): Likewise.
9665 * omp-low.c (setjmp_or_longjmp_p): Likewise.
9666 (lower_omp_1): Likewise.
9667 * predict.c (strip_predict_hints): Likewise.
9668 * print-tree.c (print_node): Likewise.
9669 * symtab.c (symtab_node::output_to_lto_symbol_table_p): Likewise.
9670 * trans-mem.c (is_tm_irrevocable): Likewise.
9671 (is_tm_load): Likewise.
9672 (is_tm_simple_load): Likewise.
9673 (is_tm_store): Likewise.
9674 (is_tm_simple_store): Likewise.
9675 (is_tm_abort): Likewise.
9676 (tm_region_init_1): Likewise.
9677 * tree-call-cdce.c (gen_shrink_wrap_conditions): Likewise.
9678 * tree-cfg.c (verify_gimple_call): Likewise.
9679 (move_stmt_r): Likewise.
9680 (stmt_can_terminate_bb_p): Likewise.
9681 * tree-eh.c (lower_eh_constructs_2): Likewise.
9682 * tree-if-conv.c (if_convertible_stmt_p): Likewise.
9683 * tree-inline.c (remap_gimple_stmt): Likewise.
9684 (copy_bb): Likewise.
9685 (estimate_num_insns): Likewise.
9686 (fold_marked_statements): Likewise.
9687 * tree-sra.c (scan_function): Likewise.
9688 * tree-ssa-ccp.c (surely_varying_stmt_p): Likewise.
9689 (optimize_stack_restore): Likewise.
9690 (pass_fold_builtins::execute): Likewise.
9691 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Likewise.
9692 (mark_all_reaching_defs_necessary_1): Likewise.
9693 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Likewise.
9694 * tree-ssa-forwprop.c (simplify_builtin_call): Likewise.
9695 (pass_forwprop::execute): Likewise.
9696 * tree-ssa-loop-im.c (stmt_cost): Likewise.
9697 * tree-ssa-math-opts.c (pass_cse_reciprocals::execute): Likewise.
9698 * tree-ssa-sccvn.c (fully_constant_vn_reference_p): Likewise.
9699 * tree-ssa-strlen.c (get_string_length): Likewise.
9700 * tree-ssa-structalias.c (handle_lhs_call): Likewise.
9701 (find_func_aliases_for_call): Likewise.
9702 * tree-ssa-ter.c (find_replaceable_in_bb): Likewise.
9703 * tree-stdarg.c (optimize_va_list_gpr_fpr_size): Likewise.
9704 * tree-tailcall.c (find_tail_calls): Likewise.
9705 * tree.c (need_assembler_name_p): Likewise.
9706 (free_lang_data_in_decl): Likewise.
9707 (get_call_combined_fn): Likewise.
9708 * ubsan.c (is_ubsan_builtin_p): Likewise.
9709 * varasm.c (incorporeal_function_p): Likewise.
9710 * tree.h (DECL_BUILT_IN): Remove and replace with
9712 (DECL_BUILT_IN_P): Transfort to fndecl_built_in_p.
9713 (fndecl_built_in_p): New.
9715 2018-08-27 Martin Liska <mliska@suse.cz>
9717 PR tree-optimization/86847
9718 * tree-switch-conversion.c (switch_decision_tree::dump_case_nodes):
9719 Dump also subtree probability.
9720 (switch_decision_tree::do_jump_if_equal): New function.
9721 (switch_decision_tree::emit_case_nodes): Handle special
9722 situations in balanced tree that can be emitted much simpler.
9723 Fix calculation of probabilities that happen in tree expansion.
9724 * tree-switch-conversion.h (struct cluster): Add
9726 (struct simple_cluster): Likewise.
9727 (struct case_tree_node): Add new function has_child.
9728 (do_jump_if_equal): New.
9730 2018-08-27 Martin Liska <mliska@suse.cz>
9732 * tree-switch-conversion.c (bit_test_cluster::find_bit_tests):
9733 Add new argument to bit_test_cluster constructor.
9734 (bit_test_cluster::emit): Set bits really number of values
9736 (bit_test_cluster::hoist_edge_and_branch_if_true): Add
9737 probability argument.
9738 * tree-switch-conversion.h (struct bit_test_cluster):
9739 Add m_handles_entire_switch.
9741 2018-08-27 Martin Liska <mliska@suse.cz>
9743 PR tree-optimization/86702
9744 * tree-switch-conversion.c (jump_table_cluster::emit):
9745 Make probabilities even for values in jump table
9746 according to number of cases handled.
9747 (switch_decision_tree::compute_cases_per_edge): Pass
9748 argument to reset_out_edges_aux function.
9749 (switch_decision_tree::analyze_switch_statement): Likewise.
9750 * tree-switch-conversion.h (switch_decision_tree::reset_out_edges_aux):
9753 2018-08-27 Martin Liska <mliska@suse.cz>
9755 * cfgexpand.c (expand_asm_stmt): Use label_to_block and pass
9756 cfun argument explicitly.
9757 * gimple-pretty-print.c (dump_gimple_switch): Likewise.
9758 * hsa-gen.c (gen_hsa_insns_for_switch_stmt): Use new
9759 function gimple_switch_default_bb.
9760 (convert_switch_statements):
9762 * ipa-fnsummary.c (set_switch_stmt_execution_predicate):
9763 * stmt.c (label_to_block_fn): Use label_to_block and pass
9764 cfun argument explicitly and use gimple_switch_label_bb.
9765 (expand_case): Likewise.
9766 * tree-cfg.c (lower_phi_internal_fn): Use label_to_block and pass
9767 cfun argument explicitly. Likewise.
9768 (make_edges_bb): Likewise.
9769 (make_cond_expr_edges): Likewise.
9770 (get_cases_for_edge): Likewise.
9771 (make_gimple_switch_edges): Likewise.
9772 (label_to_block_fn): Likewise.
9773 (label_to_block): Likewise.
9774 (make_goto_expr_edges): Likewise.
9775 (make_gimple_asm_edges): Likewise.
9776 (main_block_label): Likewise.
9777 (group_case_labels_stmt): Likewise.
9778 (find_taken_edge_computed_goto): Likewise.
9779 (find_taken_edge_switch_expr): Likewise.
9780 (gimple_verify_flow_info): Likewise.
9781 (gimple_redirect_edge_and_branch): Likewise.
9782 (gimple_switch_label_bb): New function.
9783 (gimple_switch_default_bb): Likewise.
9784 (gimple_switch_edge): Likewise.
9785 (gimple_switch_default_edge): Likewise.
9786 * tree-cfg.h (label_to_block_fn): Remove and replace ...
9787 (label_to_block): ... with this.
9788 (gimple_switch_label_bb): New.
9789 (gimple_switch_default_bb): Likewise.
9790 (gimple_switch_edge): Likewise.
9791 (gimple_switch_default_edge): Likewise.
9792 * tree-cfgcleanup.c (convert_single_case_switch): Use
9793 new gimple functions and pass new argument to label_to_block.
9794 (cleanup_control_flow_bb):
9795 * tree-eh.c (make_eh_dispatch_edges): Use label_to_block and pass
9796 cfun argument explicitly.
9797 (make_eh_edges): Likewise.
9798 (redirect_eh_dispatch_edge): Likewise.
9799 (lower_resx): Likewise.
9800 (lower_eh_dispatch): Likewise.
9801 (maybe_remove_unreachable_handlers): Likewise.
9802 (unsplit_eh): Likewise.
9803 (cleanup_empty_eh): Likewise.
9804 (verify_eh_edges): Likewise.
9805 (verify_eh_dispatch_edge): Likewise.
9806 * tree-ssa-dom.c (record_edge_info): Likewise.
9807 * tree-ssa-forwprop.c (simplify_gimple_switch_label_vec): Likewise.
9808 * tree-ssa-threadedge.c (thread_around_empty_blocks): Likewise.
9809 (thread_through_normal_block): Likewise.
9810 * tree-ssa-uncprop.c (associate_equivalences_with_edges): Likewise.
9811 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
9812 * tree-switch-conversion.c (switch_conversion::collect): Use new
9814 (switch_conversion::check_final_bb): Likewise.
9815 (switch_conversion::gather_default_values): Pass new argument
9817 (switch_conversion::build_constructors): Likewise.
9818 (switch_decision_tree::compute_cases_per_edge): Use new
9819 gimple_switch_edge function.
9820 (switch_decision_tree::analyze_switch_statement): Pass new argument
9822 (switch_decision_tree::try_switch_expansion): Use
9823 gimple_switch_default_edge.
9824 * tree-vrp.c (find_switch_asserts): Pass new argument
9826 * vr-values.c (vr_values::vrp_visit_switch_stmt): Likewise.
9827 (vr_values::simplify_switch_using_ranges): Likewise.
9829 2018-08-27 Richard Biener <rguenther@suse.de>
9831 * cfganal.h (rev_post_order_and_mark_dfs_back_seme): Declare.
9832 * cfganal.c (rev_post_order_and_mark_dfs_back_seme): New function.
9834 * tree-ssa-sccvn.h (struct vn_pval): New structure.
9835 (struct vn_nary_op_s): Add unwind_to member. Add
9836 predicated_values flag and put result into a union together
9837 with a linked list of vn_pval.
9838 (struct vn_ssa_aux): Add name member to make maintaining
9839 a map of SSA name to vn_ssa_aux possible. Remove no longer
9840 needed info, dfsnum, low, visited, on_sccstack, use_processed
9841 and range_info_anti_range_p members.
9842 (run_scc_vn, vn_eliminate, free_scc_vn, vn_valueize): Remove.
9843 (do_rpo_vn, run_rpo_vn, eliminate_with_rpo_vn, free_rpo_vn):
9845 (vn_valueize): New global.
9846 (vn_context_bb): Likewise.
9847 (VN_INFO_RANGE_INFO, VN_INFO_ANTI_RANGE_P, VN_INFO_RANGE_TYPE,
9848 VN_INFO_PTR_INFO): Remove.
9849 * tree-ssa-sccvn.c: ... (rewrite)
9850 (pass_fre::execute): For -O2+ initialize loops and run
9851 RPO VN in optimistic mode (iterating). For -O1 and -Og
9852 run RPO VN in non-optimistic mode.
9853 * params.def (PARAM_SCCVN_MAX_SCC_SIZE): Remove.
9854 (PARAM_RPO_VN_MAX_LOOP_DEPTH): Add.
9855 * doc/invoke.texi (sccvn-max-scc-size): Remove.
9856 (rpo-vn-max-loop-depth): Document.
9857 * tree-ssa-alias.c (walk_non_aliased_vuses): Stop walking
9858 when valuezing the VUSE signals we walked out of the region.
9859 * tree-ssa-pre.c (phi_translate_1): Ignore predicated values.
9860 (phi_translate): Set VN context block to use for availability
9862 (compute_avail): Likewise.
9863 (pre_valueize): New function.
9864 (pass_pre::execute): Adjust to the RPO VN API.
9866 * tree-ssa-loop-ivcanon.c: Include tree-ssa-sccvn.h.
9867 (propagate_constants_for_unrolling): Remove.
9868 (tree_unroll_loops_completely): Perform value-numbering
9869 on the unrolled bodies loop parent.
9871 2018-08-27 Richard Biener <rguenther@suse.de>
9873 * tree-ssa-pre.c (compute_antic): Re-use inverted postorder
9874 for partial antic compute.
9876 2018-08-27 Jakub Jelinek <jakub@redhat.com>
9878 PR rtl-optimization/87065
9879 * combine.c (simplify_if_then_else): Formatting fix.
9880 (if_then_else_cond): Guard MULT optimization with SCALAR_INT_MODE_P
9882 (known_cond): Don't return const_true_rtx for vector modes. Use
9883 CONST0_RTX instead of const0_rtx. Formatting fixes.
9885 2018-08-27 Martin Liska <mliska@suse.cz>
9887 PR gcov-profile/87069
9888 * gcov.c (process_file): Record files already processed
9889 and warn about a file being processed multiple times.
9891 2018-08-27 Martin Liska <mliska@suse.cz>
9894 * config/aarch64/aarch64.c (aarch64_override_options_internal):
9895 Set default values for x_aarch64_*_string strings.
9896 * config/aarch64/aarch64.opt: Remove --{march,mcpu,mtune}==
9897 prefix. For -mabi do not print '=ABI' in help and use
9898 <option_value> format for -msve-vector-bits and -moverride
9901 2018-08-26 Jeff Law <law@redhat.com>
9903 * config/mips/frame-header-opt.c: Include "backend.h" rather than
9906 2018-08-26 Marek Polacek <polacek@redhat.com>
9908 PR c++/87029, Implement -Wredundant-move.
9909 * doc/invoke.texi: Document -Wredundant-move.
9911 2018-08-25 Martin Sebor <msebor@redhat.com>
9913 PR tree-optimization/87059
9914 * builtins.c (expand_builtin_strncmp): Convert MIN_EXPR operand
9915 to the same type as the other.
9916 * fold-const.c (fold_binary_loc): Assert expectation.
9918 2018-08-25 Iain Sandoe <iain@sandoe.co.uk>
9920 * config/darwin.c (machopic_legitimize_pic_address): Clean up
9921 extraneous parentheses, dead code section and formatting.
9923 2018-08-24 David Malcolm <dmalcolm@redhat.com>
9926 * diagnostic-show-locus.c (layout::layout): Ensure the margin is
9927 wide enough for jumps in the line-numbering to be visible.
9928 (layout::print_gap_in_line_numbering): New member function.
9929 (layout::calculate_line_spans): When using line numbering, merge
9930 line spans that are only 1 line apart.
9931 (diagnostic_show_locus): When printing line numbers, show gaps in
9932 line numbering directly, rather than printing headers.
9933 (selftest::test_diagnostic_show_locus_fixit_lines): Add test of
9934 line-numbering with multiple line spans.
9935 (selftest::test_fixit_insert_containing_newline_2): Add test of
9936 line-numbering, in which the spans are close enough to be merged.
9938 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
9940 * gimple-ssa-evrp-analyze.c (set_ssa_range_info): Pass value_range
9941 to range_includes_zero_p. Do not special case VR_ANTI_RANGE.
9942 * tree-vrp.c (range_is_nonnull): Remove.
9943 (range_includes_zero_p): Accept value_range instead of min/max.
9944 (extract_range_from_binary_expr_1): Do not early bail on
9946 Use range_includes_zero_p instead of range_is_nonnull.
9947 (extract_range_from_unary_expr): Use range_includes_zero_p instead
9948 of range_is_nonnull.
9949 (vrp_meet_1): Pass value_range to range_includes_zero_p. Do not
9950 special case VR_ANTI_RANGE.
9951 (vrp_finalize): Same.
9952 * tree-vrp.h (range_includes_zero_p): Pass value_range as argument
9954 (range_is_nonnull): Remove.
9955 * vr-values.c (vrp_stmt_computes_nonzero): Use
9956 range_includes_zero_p instead of range_is_nonnull.
9957 (extract_range_basic): Pass value_range to range_includes_zero_p
9958 instead of range_is_nonnull.
9960 2018-08-24 Uros Bizjak <ubizjak@gmail.com>
9962 * emit-rtl.c (init_emit_once): Do not emit MODE_POINTER_BOUNDS RTXes.
9963 * emit-rtl.h (rtl_data): Remove return_bnd.
9964 * explow.c (trunc_int_for_mode): Do not handle POINTER_BOUNDS_MODE_P.
9965 * function.c (diddle_return_value): Do not handle crtl->return_bnd.
9966 * genmodes.c (complete_mode): Do not handle MODE_POINTER_BOUNDS.
9967 (POINTER_BOUNDS_MODE): Remove definition.
9968 (make_pointer_bounds_mode): Remove.
9969 (get_mode_class): Do not handle MODE_POINTER_BOUNDS.
9970 * machmode.h (POINTER_BOUNDS_MODE_P): Remove definition.
9971 (scalare_mode::includes_p): Do not handle MODE_POINTER_BOUNDS.
9972 * mode-classes.def: Do not define MODE_POINTER_BOUNDS.
9973 * stor-layout.c (int_mode_for_mode): Do not handle MODE_POINTER_BOUNDS.
9974 * tree-core.h (enum tree_index): Remove TI_POINTER_BOUNDS_TYPE.
9975 * varasm.c (output_constant_pool_2): Do not handle MODE_POINTER_BOUNDS.
9977 * config/i386/i386-modes.def (BND32, BND64): Remove.
9978 * config/i386/i386.c (dbx_register_map): Remove bound registers.
9979 (dbx64_register_map): Ditto.
9980 (svr4_dbx_register_map): Ditto.
9981 (indirect_thunk_bnd_needed): Remove.
9982 (indirect_thunks_bnd_used): Ditto.
9983 (indirect_return_bnd_needed): Ditto.
9984 (indirect_return_via_cx_bnd): Ditto.
9985 (enum indirect_thunk_prefix): Remove indirect_thunk_prefix_bnd.
9986 (indirect_thunk_name): Remove handling of indirect_thunk_prefix_bnd.
9987 (output_indirect_thunk): Ditto. Remove need_prefix argument.
9988 (output_indirect_thunk_function): Remove handling of
9989 indirect_return_bnd_needed, indirect_return_via_cx_bnd,
9990 indirect_thunk_bnd_needed and indirect_thunks_bnd_used variables.
9991 (ix86_save_reg): Remove handling of crtl->return_bnd.
9992 (ix86_legitimate_constant_p): Remove handling of POINTER_BOUNDS_MODE_P.
9993 (ix86_print_operand_address_as): Remove handling of UNSPEC_BNDMK_ADDR
9994 and UNSPEC_BNDLX_ADDR.
9995 (ix86_output_indirect_branch_via_reg): Remove handling of
9996 indirect_thunk_prefix_bnd.
9997 (ix86_output_indirect_branch_via_push): Ditto.
9998 (ix86_output_function_return): Ditto.
9999 (ix86_output_indirect_function_return): Ditto.
10000 (avoid_func_arg_motion): Do not handle UNSPEC_BNDSTX.
10001 * config/i386/i386.h (FIXED_REGISTERS): Remove bound registers.
10002 (CALL_USED_REGISTERS): Ditto.
10003 (REG_ALLOC_ORDER): Update for removal of bound registers.
10004 (HI_REGISTER_NAMES): Ditto.
10005 * config/i386/i386.md (UNSPEC_BNDMK, UNSPEC_BNDMK_ADDR, UNSPEC_BNDSTX)
10006 (UNSPEC_BNDLDX, UNSPEC_BNDLDX_ADDR, UNSPEC_BNDCL, UNSPEC_BNDCU)
10007 (UNSPEC_BNDCN, UNSPEC_MPX_FENCE): Remove.
10008 (BND0_REG, BND1_REG, BND2_REG, BND3_REG): Remove
10009 (FIRST_PSEUDO_REG): Update.
10010 (BND): Remove mode iterator.
10011 * config/i386/predicates.md (bnd_mem_operator): Remove.
10013 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
10015 * tree-vect-stmts.c (vectorizable_bswap): Handle variable-length
10018 2018-08-24 Richard Sandiford <richard.sandiford@arm.com>
10020 * tree-vect-slp.c (vect_transform_slp_perm_load): Separate out
10021 the case in which the permute needs only a single element and
10022 repeats for every vector of the result. Extend that case to
10023 handle variable-length vectors.
10024 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
10026 2018-08-24 H.J. Lu <hongjiu.lu@intel.com>
10029 * dwarf2out.c (save_macinfo_strings): Call set_indirect_string
10030 on DW_MACINFO_start_file for -gsplit-dwarf -g3.
10032 2018-08-24 Richard Biener <rguenther@suse.de>
10034 * cfg.h (struct control_flow_graph): Add edge_flags_allocated and
10035 bb_flags_allocated members.
10036 (auto_flag): New RAII class for allocating flags.
10037 (auto_edge_flag): New RAII class for allocating edge flags.
10038 (auto_bb_flag): New RAII class for allocating bb flags.
10039 * cfgloop.c (verify_loop_structure): Allocate temporary edge
10041 * cfganal.c (dfs_enumerate_from): Remove use of visited sbitmap
10042 in favor of temporarily allocated BB flag.
10043 * hsa-brig.c: Re-order includes.
10044 * hsa-dump.c: Likewise.
10045 * hsa-regalloc.c: Likewise.
10046 * print-rtl.c: Likewise.
10047 * profile-count.c: Likewise.
10049 2018-08-24 Segher Boessenkool <segher@kernel.crashing.org>
10052 * config/rs6000/rs6000.c (toc_relative_expr_p): Check that the base is
10055 2018-08-24 Aldy Hernandez <aldyh@redhat.com>
10058 * wide-int-range.cc (wide_int_range_div): Do not ignore result
10059 from wide_int_range_multiplicative_op.
10061 2018-08-23 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
10063 * tree-vect-data-refs.c (vect_grouped_store_supported): Fix typo
10066 2018-08-23 Giuliano Belinassi <giuliano.belinassi@usp.br>
10068 * genmatch.c (parser::parse_operation): Fix typo 'exapnded'
10071 2018-08-23 Alexander Monakov <amonakov@ispras.ru>
10073 * tree-scalar-evolution.c (final_value_replacement_loop): Dump
10074 full GENERIC expression used for replacement.
10076 2018-08-23 Aldy Hernandez <aldyh@redhat.com>
10078 * tree-vrp.c (abs_extent_range): Remove.
10079 (extract_range_into_wide_ints): Pass wide ints by reference.
10080 (extract_range_from_binary_expr_1): Rewrite the *DIV_EXPR code.
10081 Pass wide ints by reference in all calls to
10082 extract_range_into_wide_ints.
10083 * wide-int-range.cc (wide_int_range_div): New.
10084 * wide-int-range.h (wide_int_range_div): New.
10085 (wide_int_range_includes_zero_p): New.
10086 (wide_int_range_zero_p): New.
10088 2018-08-23 Matthew Malcomson <matthew.malcomson@arm.com>
10090 * config/aarch64/aarch64.md (arches): New enum.
10091 (arch): New enum attr.
10092 (arch_enabled): New attr.
10093 (enabled): Now uses arch_enabled only.
10094 (simd, sve, fp16): Removed attribute.
10095 (fp): Attr now defined in terms of 'arch'.
10096 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64, *movti_aarch64,
10097 *movhf_aarch64, <optab><fcvt_target><GPF:mode>2,
10098 <FCVT_F2FIXED:fcvt_fixed_insn><GPF:mode>3,
10099 <FCVT_FIXED2F:fcvt_fixed_insn><GPI:mode>3): Merge 'fp' and 'simd'
10100 attributes into 'arch'.
10101 (*movsf_aarch64, *movdf_aarch64, *movtf_aarch64, *add<mode>3_aarch64,
10102 subdi3, neg<mode>2, <optab><mode>3, one_cmpl<mode>2,
10103 *<NLOGICAL:optab>_one_cmpl<mode>3, *xor_one_cmpl<mode>3,
10104 *aarch64_ashl_sisd_or_int_<mode>3, *aarch64_lshr_sisd_or_int_<mode>3,
10105 *aarch64_ashr_sisd_or_int_<mode>3, *aarch64_sisd_ushl): Convert use of
10106 'simd' attribute into 'arch'.
10107 (load_pair_sw_<SX:mode><SX2:mode>, load_pair_dw_<DX:mode><DX2:mode>,
10108 store_pair_sw_<SX:mode><SX2:mode>, store_pair_dw_<DX:mode><DX2:mode>):
10109 Convert use of 'fp' attribute to 'arch'.
10110 * config/aarch64/aarch64-simd.md (move_lo_quad_internal_<mode>,
10111 move_lo_quad_internal_<mode>): (different modes) Merge 'fp' and 'simd'
10113 (move_lo_quad_internal_be_<mode>, move_lo_quad_internal_be_<mode>):
10114 (different modes) Merge 'fp' and 'simd' into 'arch'.
10115 (*aarch64_combinez<mode>, *aarch64_combinez_be<mode>): Merge 'fp' and
10116 'simd' into 'arch'.
10118 2018-08-23 Segher Boessenkool <segher@kernel.crashing.org>
10120 PR rtl-optimization/87026
10121 * expmed.c (canonicalize_comparison): If we can no longer create
10122 pseudoregisters, don't.
10124 2018-08-23 Richard Earnshaw <rearnsha@arm.com>
10127 * config/arm/arm-protos.h (arm_emit_speculation_barrier): New
10129 * config/arm/arm.c (speculation_barrier_libfunc): New static
10131 (arm_init_libfuncs): Initialize it.
10132 (arm_emit_speculation_barrier): New function.
10133 * config/arm/arm.md (speculation_barrier): Call
10134 arm_emit_speculation_barrier for architectures that do not have
10136 (speculation_barrier_insn): Only match on Armv7 or later.
10138 2018-08-23 Richard Biener <rguenther@suse.de>
10140 PR middle-end/87024
10141 * tree-inline.c (copy_bb): Drop unused __builtin_va_arg_pack_len
10144 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
10146 * config/aarch64/aarch64.c (aarch64_evpc_sve_tbl): Fix handling
10147 of single-vector TBLs.
10148 (aarch64_vectorize_vec_perm_const): Set one_vector_p when only
10149 one input is given.
10151 2018-08-23 Richard Sandiford <richard.sandiford@arm.com>
10154 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const_1): Fix
10155 aarch64_evpc_tbl guard.
10157 2018-08-22 Bernd Edlinger <bernd.edlinger@hotmail.de>
10159 * tree-ssa-dse.c (compute_trims): Avoid folding away undefined
10162 2018-08-22 Martin Sebor <msebor@redhat.com>
10164 PR middle-end/87052
10165 * tree-pretty-print.c (pretty_print_string): Add argument.
10166 (dump_generic_node): Call to pretty_print_string with string size.
10168 2018-08-22 Segher Boessenkool <segher@kernel.crashing.org>
10170 PR rtl-optimization/86771
10171 * combine.c (try_combine): Do not allow splitting a resulting PARALLEL
10172 of two SETs into those two SETs, one to be placed at i2, if that SETs
10173 destination is modified between i2 and i3.
10175 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
10177 PR tree-optimization/86725
10178 * tree-vect-loop.c (vect_inner_phi_in_double_reduction_p): New
10180 (vect_analyze_scalar_cycles_1): Check it.
10182 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
10184 PR tree-optimization/86725
10185 * tree-vect-loop.c (vect_is_simple_reduction): When treating
10186 an outer loop phi as a double reduction, make sure that the
10187 single user of the phi result is an inner loop phi.
10189 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
10191 * tree-vect-data-refs.c (vect_analyze_group_access_1): Convert
10192 grouped stores with gaps to a strided group.
10194 2018-08-22 Richard Sandiford <richard.sandiford@arm.com>
10196 * tree-vect-stmts.c (get_group_load_store_type)
10197 (get_load_store_type): Only test STMT_VINFO_STRIDED_P for the
10198 first statement in a group.
10200 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
10202 * config/darwin.h (LINK_COMMAND_SPEC_A): Sync LTO options with
10203 the sequence used in gcc/gcc.c.
10205 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
10208 * gcc-ar.c (main): Don’t try to invoke the plug-in if we’re not
10211 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
10213 * config/darwin10.h (LINK_GCC_C_SEQUENCE_SPEC): Adjust to use the
10214 Darwin10-specific unwinder-shim.
10215 * config/darwin12.h (LINK_GCC_C_SEQUENCE_SPEC): Remove.
10216 * config/rs6000/darwin.h (DARWIN_CRT1_SPEC, DARWIN_DYLIB1_SPEC):
10217 New to cater for Darwin10 Rosetta.
10219 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
10221 * config/i386/i386.c (ix86_output_addr_diff_elt): Move the MACH-O
10222 specific test before the one for HAVE_AS_GOTOFF_IN_DATA.
10224 2018-08-22 Iain Sandoe <iain@sandoe.co.uk>
10229 * gcc/dwarf2out.c (FUNC_SECOND_SECT_LABEL): New.
10230 (dwarf2out_switch_text_section): Generate a local label for the
10231 second function sub-section and apply it as the second FDE start
10233 * gcc/final.c (final_scan_insn_1): Emit second FDE label after the
10234 second sub-section start.
10236 2018-08-22 Richard Biener <rguenther@suse.de>
10238 PR tree-optimization/86988
10239 * tree-vrp.c (vrp_prop::check_mem_ref): Bail out on VLAs.
10241 2018-08-22 Richard Biener <rguenther@suse.de>
10243 PR tree-optimization/86945
10244 * tree-cfg.c (generate_range_test): Use unsigned arithmetic.
10246 2018-08-22 Alexandre Oliva <oliva@adacore.com>
10248 * config/rs6000/rs6000.c (SMALL_DATA_RELOC, SMALL_DATA_REG): Add
10249 a comment about how uses of r2 for .sdata2 come about.
10251 2018-08-22 Alexandre Oliva <aoliva@redhat.com>
10253 * tree-ssa-reassoc.c (is_reassociable_op): Fix cut&pasto.
10255 2018-08-21 Marek Polacek <polacek@redhat.com>
10257 PR c++/86981, Implement -Wpessimizing-move.
10258 * doc/invoke.texi: Document -Wpessimizing-move.
10260 2018-08-21 Jan Hubicka <jh@suse.cz>
10262 * tree.c (find_decls_types_r): Do not check for redundant typedefs.
10263 * tree.h (is_redundant_typedef): Remove.
10264 * dwarf2out.c (is_redundant_typedef): Turn into static function.
10266 2018-08-21 Jan Hubicka <jh@suse.cz>
10268 * tree.c (free_lang_data_in_decl): Remove types from DECL_CONTEXT
10271 2018-08-21 Tamar Christina <tamar.christina@arm.com>
10273 * expmed.c (extract_low_bits): Reject invalid subregs early.
10275 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
10277 PR middle-end/86121
10278 * tree-ssa-strlen.c (adjust_last_stmt): Avoid folding away undefined
10281 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
10283 * config/vxworks.h: Guard vxworks_asm_out_constructor and
10284 vxworks_asm_out_destructor by !HAVE_INITFINI_ARRAY_SUPPORT
10285 * config/vxworks.c: Likewise.
10287 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
10289 * config/vxworks.c: Set targetm.have_ctors_dtors
10290 if HAVE_INITFINI_ARRAY_SUPPORT.
10291 * config/vxworks.h: Set SUPPORTS_INIT_PRIORITY
10292 if HAVE_INITFINI_ARRAY_SUPPORT.
10294 2018-08-21 Rasmus Villemoes <rv@rasmusvillemoes.dk>
10296 * config/vxworks.h: Add $(WIND_BASE)/target/h/wrn/coreip to
10297 default search path for VxWorks < 7.
10299 2018-08-21 Bernd Edlinger <bernd.edlinger@hotmail.de>
10301 * gimple-ssa-sprintf.c (decl_constant_value): Remove.
10302 (get_format_string): Refer to c_getstr.
10304 2018-08-21 Tom de Vries <tdevries@suse.de>
10306 * cgraph.h (debuginfo_early_init, debuginfo_init, debuginfo_fini)
10307 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
10308 (debuginfo_early_stop): Declare.
10309 * cgraphunit.c (debuginfo_early_init, debuginfo_init, debuginfo_fini)
10310 (debuginfo_start, debuginfo_stop, debuginfo_early_start)
10311 (debuginfo_early_stop): New function.
10312 (symbol_table::finalize_compilation_unit): Call debuginfo_early_start
10313 and debuginfo_early_stop.
10314 * dwarf2out.c (dwarf2out_finish, dwarf2out_early_finish): Dump dwarf.
10315 * toplev.c (compile_file): Call debuginfo_start and debuginfo_stop.
10316 (general_init): Call debuginfo_early_init.
10317 (finalize): Call debuginfo_fini.
10318 (do_compile): Call debuginfo_init.
10319 * doc/invoke.texi (@gccoptlist): Add -fdump-debug and
10320 -fdump-early-debug.
10321 (@item -fdump-debug, @item -fdump-earlydebug): Add.
10323 2018-08-21 Tom de Vries <tdevries@suse.de>
10325 * dwarf2out.c (print_dw_val, print_loc_descr, print_die): Handle
10326 flag_dump_noaddr and flag_dump_unnumbered.
10328 2018-08-21 Aldy Hernandez <aldyh@redhat.com>
10330 * wide-int-range.cc (wide_int_range_abs): New.
10331 (wide_int_range_order_set): Rename from wide_int_range_min_max.
10332 * wide-int-range.h (wide_int_range_abs): New.
10333 (wide_int_range_min_max): New.
10334 * tree-vrp.c (extract_range_from_unary_expr): Rewrite ABS_EXPR
10335 case to call wide_int_range_abs.
10336 Rewrite MIN/MAX_EXPR to call wide_int_range_min_max.
10337 (extract_range_from_abs_expr): Delete.
10339 2018-08-20 Michael Meissner <meissner@linux.ibm.com>
10342 * config/rs6000/rs6000.md (extendsi<mode>2): Change constraints
10343 from 'Y' to 'YZ' to enable the LWAX instruction to be generated
10346 2018-08-20 Nathan Sidwell <nathan@acm.org>
10347 Jeff Law <law@redhat.com>
10349 * config/s390/s390-c (s390_macro_to_expand): Use cpp_macro_p.
10350 * config/spu/spu-c.c (spu_macro_to_expand): Likewise.
10352 2018-08-20 David Malcolm <dmalcolm@redhat.com>
10355 * attribs.c (diag_attr_exclusions): Add auto_diagnostic_group instance.
10356 (decl_attributes): Likewise.
10357 * calls.c (maybe_warn_nonstring_arg): Add auto_diagnostic_group
10359 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
10360 * diagnostic-core.h (class auto_diagnostic_group): New class.
10361 * diagnostic.c (diagnostic_initialize): Initialize the new fields.
10362 (diagnostic_report_diagnostic): Handle the first diagnostics within
10364 (emit_diagnostic): Add auto_diagnostic_group instance.
10365 (inform): Likewise.
10366 (inform_n): Likewise.
10367 (warning): Likewise.
10368 (warning_at): Likewise.
10369 (warning_n): Likewise.
10370 (pedwarn): Likewise.
10371 (permerror): Likewise.
10373 (error_n): Likewise.
10374 (error_at): Likewise.
10376 (fatal_error): Likewise.
10377 (internal_error): Likewise.
10378 (internal_error_no_backtrace): Likewise.
10379 (auto_diagnostic_group::auto_diagnostic_group): New ctor.
10380 (auto_diagnostic_group::~auto_diagnostic_group): New dtor.
10381 * diagnostic.h (struct diagnostic_context): Add fields
10382 "diagnostic_group_nesting_depth",
10383 "diagnostic_group_emission_count", "begin_group_cb",
10385 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior):
10386 Add auto_diagnostic_group instance(s).
10387 (find_explicit_erroneous_behavior): Likewise.
10388 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Likewise.
10389 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Likewise.
10390 * gimplify.c (warn_implicit_fallthrough_r): Likewise.
10391 (gimplify_va_arg_expr): Likewise.
10392 * hsa-gen.c (HSA_SORRY_ATV): Likewise.
10393 (HSA_SORRY_AT): Likewise.
10394 * ipa-devirt.c (compare_virtual_tables): Likewise.
10395 (warn_odr): Likewise.
10396 * multiple_target.c (expand_target_clones): Likewise.
10397 * opts-common.c (cmdline_handle_error): Likewise.
10398 * reginfo.c (globalize_reg): Likewise.
10399 * substring-locations.c (format_warning_n_va): Likewise.
10400 * tree-inline.c (expand_call_inline): Likewise.
10401 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Likewise.
10402 * tree-ssa-loop-niter.c
10403 (do_warn_aggressive_loop_optimizations): Likewise.
10404 * tree-ssa-uninit.c (warn_uninit): Likewise.
10405 * tree.c (warn_deprecated_use): Likewise.
10407 2018-08-20 H.J. Lu <hongjiu.lu@intel.com>
10410 * config/i386/i386.md (eh_return): Always update EH return
10411 address in word_mode.
10413 2018-08-20 Chung-Lin Tang <cltang@codesourcery.com>
10415 * targhooks.c (std_gimplify_va_arg_expr): Properly handle case of when
10416 TARGET_SPLIT_COMPLEX_ARG is defined.
10418 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
10420 * expr.c (store_field): Change gcc_assert to gcc_checking_assert.
10422 2018-08-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
10425 * expr.c (expand_assignment): Assert that bitpos is positive.
10426 (store_field): Likewise
10427 (expand_expr_real_1): Make sure that bitpos is positive.
10428 * config/alpha/alpha.h (CONSTANT_ADDRESS_P): Avoid signed
10431 2018-08-20 Nathan Sidwell <nathan@acm.org>
10433 * Makefile.in (CPP_ID_DATA_H): Delete.
10434 (CPP_INTERNAL_H): Don't add it.
10435 (GTFILES): Replace CPP_ID_DATA_H with CPPLIB_H.
10436 * gengtype.c (open_base_files): Replace cpp-id-data.h with cpplib.h
10438 2018-08-20 Richard Biener <rguenther@suse.de>
10440 PR tree-optimization/78655
10441 * tree-vrp.c (extract_range_from_binary_expr_1): Make
10442 pointer + offset nonnull if either operand is nonnull work.
10444 2018-08-20 Tom de Vries <tdevries@suse.de>
10446 * dwarf2out.c (add_scalar_info): Don't add reference to existing die
10447 unless the referenced die describes the added property using
10448 DW_AT_location or DW_AT_const_value. Fall back to exprloc case.
10449 Otherwise, add a DW_AT_location to the referenced die.
10451 2018-08-19 Uros Bizjak <ubizjak@gmail.com>
10454 * config/i386/i386.c (ix86_rtx_costs) [case SET]: Check source for
10455 register_operand when calling ix86_set_reg_reg_cost.
10456 [case CONST_INT, case CONST, case LABEL_REF, case SYMBOL_REF]:
10457 Set *total to 0 for operands that satisfy x86_64_immediate_operand
10458 predicate and to 1 otherwise.
10460 2018-08-18 Iain Sandoe <iain@sandoe.co.uk>
10462 * config/darwin.c (darwin_override_options): If -gsplit-dwarf is set,
10463 emit a diagnostic that it is not supported and reset the option.
10464 * config/darwin.h (DRIVER_SELF_SPECS): Note that gsplit-dwarf is not
10465 supported and consume the option. (ASM_FINAL_SPEC): New.
10467 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
10469 * doc/md.texi (Patterns): Use @ref instead of @xref in the middle of
10472 2018-08-17 Sandra Loosemore <sandra@codesourcery.com>
10474 C-SKY port: Documentation
10476 * doc/extend.texi (C-SKY Function Attributes): New section.
10477 * doc/invoke.texi (Option Summary): Add C-SKY options.
10478 (C-SKY Options): New section.
10479 * doc/md.texi (Machine Constraints): Document C-SKY constraints.
10481 2018-08-17 Jojo <jijie_rong@c-sky.com>
10482 Huibin Wang <huibin_wang@c-sky.com>
10483 Sandra Loosemore <sandra@codesourcery.com>
10484 Chung-Lin Tang <cltang@codesourcery.com>
10486 C-SKY port: Backend implementation
10488 * config/csky/*: New.
10489 * common/config/csky/*: New.
10491 2018-08-17 Jojo <jijie_rong@c-sky.com>
10492 Huibin Wang <huibin_wang@c-sky.com>
10493 Sandra Loosemore <sandra@codesourcery.com>
10494 Chung-Lin Tang <cltang@codesourcery.com>
10495 Andrew Jenner <andrew@codesourcery.com>
10497 C-SKY port: Configury
10499 * config.gcc (csky-*-*): New.
10500 * configure.ac: Add csky to targets for dwarf2 debug_line support.
10501 * configure: Regenerated.
10503 2018-08-17 David Malcolm <dmalcolm@redhat.com>
10505 * dump-context.h: Include "dumpfile.h".
10506 (dump_context::dump_printf_va): Convert final param from va_list
10507 to va_list *. Convert from ATTRIBUTE_PRINTF to
10508 ATTRIBUTE_GCC_DUMP_PRINTF.
10509 (dump_context::dump_printf_loc_va): Likewise.
10510 * dumpfile.c: Include "stringpool.h".
10511 (make_item_for_dump_printf_va): Delete.
10512 (make_item_for_dump_printf): Delete.
10513 (class dump_pretty_printer): New class.
10514 (dump_pretty_printer::dump_pretty_printer): New ctor.
10515 (dump_pretty_printer::emit_items): New member function.
10516 (dump_pretty_printer::emit_any_pending_textual_chunks): New member
10518 (dump_pretty_printer::emit_item): New member function.
10519 (dump_pretty_printer::stash_item): New member function.
10520 (dump_pretty_printer::format_decoder_cb): New member function.
10521 (dump_pretty_printer::decode_format): New member function.
10522 (dump_context::dump_printf_va): Reimplement in terms of
10523 dump_pretty_printer.
10524 (dump_context::dump_printf_loc_va): Convert final param from va_list
10526 (dump_context::begin_scope): Reimplement call to
10527 make_item_for_dump_printf.
10528 (dump_printf): Update for change to dump_printf_va.
10529 (dump_printf_loc): Likewise.
10530 (selftest::test_capture_of_dump_calls): Convert "stmt" from
10531 greturn * to gimple *. Add a test_decl. Add tests of dump_printf
10532 with %T, %E, and %G.
10533 * dumpfile.h (ATTRIBUTE_GCC_DUMP_PRINTF): New macro.
10534 (dump_printf): Replace ATTRIBUTE_PRINTF_2 with
10535 ATTRIBUTE_GCC_DUMP_PRINTF (2, 3).
10536 (dump_printf_loc): Replace ATTRIBUTE_PRINTF_3 with
10537 ATTRIBUTE_GCC_DUMP_PRINTF (3, 0).
10538 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Convert
10539 use of HOST_WIDE_INT_PRINT_DEC on unsigned HOST_WIDE_INT "count"
10540 within a dump_printf_loc call to "%wu".
10541 (vector_alignment_reachable_p): Merge two dump_printf[_loc] calls,
10542 converting a use of HOST_WIDE_INT_PRINT_DEC to "%wd". Add a
10543 missing space after "=".
10544 * tree-vect-loop.c (vect_analyze_loop_2) Within a dump_printf
10545 call, convert use of HOST_WIDE_INT_PRINT_DEC to "%wd".
10546 * tree-vect-slp.c (vect_slp_bb): Within a dump_printf_loc call,
10547 convert use of HOST_WIDE_INT_PRINT_UNSIGNED to "%wu".
10548 * tree-vectorizer.c (try_vectorize_loop_1): Likewise. Remove
10549 duplicate "vectorized" from message.
10551 2018-08-17 Szabolcs Nagy <szabolcs.nagy@arm.com>
10553 * config/arm/arm-builtins.c (arm_init_simd_builtin_types): Clear
10554 polyNxK_t element's TYPE_STRING_FLAG.
10556 2018-08-17 Segher Boessenkool <segher@kernel.crashing.org>
10558 * config/rs6000/rs6000.md (*cbranch, *creturn): Name these patterns
10559 (they were unnamed before). Fix comments.
10561 2018-08-17 Nathan Sidwell <nathan@acm.org>
10563 * cppbuiltin.c: Include "cpplib.h", not "cpp-id-data.h".
10565 2018-08-17 Richard Biener <rguenther@suse.de>
10567 PR tree-optimization/86841
10568 * wide-int-range.cc (wide_int_range_lshift): Use to_uhwi.
10570 2018-08-17 Martin Liska <mliska@suse.cz>
10572 * common.opt: Remove Warn, Init and Report for options with
10573 Ignore/Deprecated flag. Warning is done automatically for
10575 * config/i386/i386.opt: Likewise.
10576 * config/ia64/ia64.opt: Likewise.
10577 * config/rs6000/rs6000.opt: Likewise.
10578 * cppbuiltin.c (define_builtin_macros_for_compilation_flags):
10579 Remove usage of flag_check_pointer_bounds.
10580 * lto-wrapper.c (merge_and_complain): Do not handle
10581 OPT_fcheck_pointer_bounds.
10582 (append_compiler_options): Likewise.
10583 * opt-functions.awk: Do not handle Deprecated.
10584 * optc-gen.awk: Check that Var, Report and Init are not
10585 used for an option with Ignore/Deprecated flag.
10586 * opts-common.c (decode_cmdline_option): Do not report
10588 (read_cmdline_option): Report warning for OPT_SPECIAL_deprecated
10590 * opts.h (struct cl_option): Remove cl_deprecated flag.
10591 (CL_ERR_DEPRECATED): Remove error enum value.
10593 2018-08-17 Richard Biener <rguenther@suse.de>
10595 PR middle-end/86505
10596 * tree-inline.c (copy_bb): When inlining __builtin_va_arg_pack_len ()
10597 across a va-arg-pack using call adjust its return value accordingly.
10599 2018-08-16 Martin Sebor <msebor@redhat.com>
10601 PR tree-optimization/86853
10602 * gimple-ssa-sprintf.c (struct format_result): Rename member.
10603 (struct fmtresult): Add member and initialize it in ctors.
10604 (format_character): Handle %C. Extend range to NUL. Set MAYFAIL.
10605 (format_string): Handle %S the same as %ls. Set MAYFAIL.
10606 (format_directive): Set POSUNDER4K when MAYFAIL is set.
10607 (parse_directive): Handle %C same as %c.
10608 (sprintf_dom_walker::compute_format_length): Adjust.
10609 (is_call_safe): Adjust.
10611 2018-08-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
10613 * builtins.c (c_strlen): Add new parameter eltsize. Use it
10614 for determining how to count the elements.
10615 * builtins.h (c_strlen): Adjust prototype.
10616 * expr.c (string_constant): Add new parameter mem_size.
10617 Set *mem_size appropriately.
10618 * expr.h (string_constant): Adjust protoype.
10619 * gimple-fold.c (get_range_strlen): Add new parameter eltsize.
10620 * gimple-fold.h (get_range_strlen): Adjust prototype.
10621 * gimple-ssa-sprintf.c (get_string_length): Add new parameter eltsize.
10622 (format_string): Call get_string_length with eltsize.
10624 2018-08-16 David Malcolm <dmalcolm@redhat.com>
10626 * diagnostic.c (default_diagnostic_start_span_fn): Call pp_string
10627 to emit the span, rather than setting it as the prefix.
10629 2018-08-16 David Malcolm <dmalcolm@redhat.com>
10631 * diagnostic-show-locus.c (layout::start_annotation_line): Add
10632 "margin_char" parameter, defaulting to space. Use it in place
10633 of pp_space for the initial part of the margin.
10634 (layout::print_leading_fixits): Use '+' when filling the margin
10635 of line-insertion fix-it hints.
10637 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
10639 * config/rs6000/rs6000.md (two unnamed define_insn and define_split):
10642 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
10644 * config/rs6000/altivec.md: Don't set length attribute to the default
10646 * config/rs6000/darwin.md: Ditto.
10647 * config/rs6000/dfp.md: Ditto.
10648 * config/rs6000/htm.md: Ditto.
10649 * config/rs6000/rs6000.md: Ditto.
10650 * config/rs6000/sync.md: Ditto.
10651 * config/rs6000/vsx.md: Ditto.
10653 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
10655 * config/rs6000/altivec.md: Don't set length attribute to the default
10656 value, for branch instructions.
10657 * config/rs6000/darwin.md: Ditto.
10658 * config/rs6000/rs6000.md: Ditto.
10660 2018-08-16 Segher Boessenkool <segher@kernel.crashing.org>
10662 * config/rs6000/rs6000.md (length): Always define as const_int 4.
10663 (unnamed conditional branch define_insn): Set length to 4 or 8
10664 depending on offset.
10665 (<bd>_<mode>): Similar, for alternative 0.
10666 (<bd>tf_<mode>): Ditto.
10668 2018-08-16 Tamar Christina <tamar.christina@arm.com>
10670 * expr.c (copy_blkmode_to_reg): Perform larger copies when safe.
10672 2018-08-16 Matthew Malcomson <matthew.malcomson@arm.com>
10674 * doc/rtl.texi: Replace old RTX class names with new names.
10677 2018-08-16 Vlad Lazar <vlad.lazar@arm.com>
10679 * expmed.h (canonicalize_comparison): New declaration.
10680 * expmed.c (canonicalize_comparison, equivalent_cmp_code): New function.
10681 * expmed.c (emit_store_flag_1): Add call to canonicalize_comparison.
10682 * optabs.c (prepare_cmp_insn): Likewise.
10683 * rtl.h (unsigned_condition_p): New function which checks if a
10684 comparison operator is unsigned.
10686 2018-08-16 Nathan Sidwell <nathan@acm.org>
10688 * config/rs6000/rs6000-c.c (rs6000_macro_to_expend): Use cpp_macro_p.
10689 * config/powerpcspc/powerpcspe-c.c (rs6000_macro_to_expend): Likewise.
10691 2018-08-16 Tamar Christina <tamar.christina@arm.com>
10694 * config/arm/arm.c (arm_can_change_mode_class): Disallow subreg.
10695 * config/arm/neon.md (movv4hf, movv8hf): Refactored to..
10696 (mov<mov>): ..this and enable unconditionally.
10698 2018-08-16 Tamar Christina <tamar.christina@arm.com>
10700 * config/arm/neon.md (*neon_mov<mode>): Remove reg-to-reg alternative.
10702 2018-08-16 Sam Tebbs <sam.tebbs@arm.com>
10704 * config/aarch64/aarch64.opt (mlow-precision-recip-sqrt)
10705 (mlow-precision-sqrt, mlow-precision-div, mverbose-cost-dump): Replace
10706 "Common" with "Target".
10708 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
10710 * config/i386/i386.opt (mmitigate-rop): Mark as deprecated.
10711 * doc/invoke.texi (mmitigate-rop): Remove.
10712 * config/i386/i386.c: Do not include "regrename.h".
10713 (ix86_rop_should_change_byte_p, reg_encoded_number)
10714 (ix86_get_modrm_for_rop, set_rop_modrm_reg_bits, ix86_mitigate_rop):
10716 (ix86_reorg): Remove call to ix86_mitigate_rop.
10717 * config/i386/i386.md (attr "modrm_class"): Remove.
10718 (cmp<mode>_ccno_1, mov<mode>_xor, movstrict<mode>_xor)
10719 (x86_mov<mode>cc_0_m1. x86_mov<mode>cc_0_m1_se)
10720 (x86_mov<mode>cc_0_m1_neg): Remove modrm_class attribute override.
10722 2018-08-15 Will Schmidt <will_schmidt@vnet.ibm.com>
10724 * config/rs6000/rs600.c (rs6000_gimple_fold_builtin): Add entries to
10725 allow folding of mergeh() and mergel() for the float and double types.
10726 (fold_mergehl_helper): Rework to handle building a permute tree
10729 2018-08-15 Uros Bizjak <ubizjak@gmail.com>
10731 * config/i386/i386.c (expand_vec_perm_movs): Enable V4SFmode
10734 2018-08-15 David Malcolm <dmalcolm@redhat.com>
10736 * common.opt (fdiagnostics-show-labels): New option.
10737 * diagnostic-show-locus.c (class layout_range): Add field
10739 (class layout): Add field "m_show_labels_p".
10740 (layout_range::layout_range): Add param "label" and use it to
10741 initialize m_label.
10742 (make_range): Pass in NULL for new "label" param of layout_range's
10744 (layout::layout): Initialize m_show_labels_p.
10745 (layout::maybe_add_location_range): Pass in loc_range->m_label
10746 when constructing layout_range instances.
10747 (struct line_label): New struct.
10748 (layout::print_any_labels): New member function.
10749 (layout::print_line): Call it if label-printing is enabled.
10750 (selftest::test_one_liner_labels): New test.
10751 (selftest::test_diagnostic_show_locus_one_liner): Call it.
10752 * diagnostic.c (diagnostic_initialize): Initialize
10753 context->show_labels_p.
10754 * diagnostic.h (struct diagnostic_context): Add field
10756 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
10757 -fno-diagnostics-show-labels.
10758 * dwarf2out.c (gen_producer_string): Add
10759 OPT_fdiagnostics_show_labels to the ignored options.
10760 * gcc-rich-location.c (gcc_rich_location::add_expr): Add "label"
10762 (gcc_rich_location::maybe_add_expr): Likewise.
10763 * gcc-rich-location.h (gcc_rich_location::gcc_rich_location): Add
10764 label" param, defaulting to NULL.
10765 (gcc_rich_location::add_expr): Add "label" param.
10766 (gcc_rich_location::maybe_add_expr): Likewise.
10767 (class text_range_label): New class.
10768 (class range_label_for_type_mismatch): New class.
10769 * gimple-ssa-sprintf.c (fmtwarn): Pass NULL for new label params
10770 of format_warning_va.
10771 (fmtwarn_n): Likewise for new params of format_warning_n_va.
10772 * lto-wrapper.c (merge_and_complain): Add
10773 OPT_fdiagnostics_show_labels to the "pick one setting" options.
10774 (append_compiler_options): Likewise to the dropped options.
10775 (append_diag_options): Likewise to the passed-on options.
10776 * opts.c (common_handle_option): Handle the new option.
10777 * selftest-diagnostic.c
10778 (test_diagnostic_context::test_diagnostic_context): Enable
10780 * substring-locations.c: Include "gcc-rich-location.h".
10781 (format_warning_n_va): Add "fmt_label" and "param_label" params
10782 and use them as appropriate.
10783 (format_warning_va): Add "fmt_label" and "param_label" params,
10784 passing them on to format_warning_n_va.
10785 (format_warning_at_substring): Likewise.
10786 (format_warning_at_substring_n): Likewise.
10787 * substring-locations.h (format_warning_va): Add "fmt_label" and
10788 "param_label" params.
10789 (format_warning_n_va): Likewise.
10790 (format_warning_at_substring): Likewise.
10791 (format_warning_at_substring_n): Likewise.
10792 * toplev.c (general_init): Initialize global_dc->show_labels_p.
10794 2018-08-15 Qing Zhao <qing.zhao@oracle.com>
10797 * builtins.c (expand_builtin_memcmp): Do not expand the call
10798 when overflow is detected.
10800 2018-08-15 Martin Sebor <msebor@redhat.com>
10802 PR tree-optimization/71625
10803 * config/aarch64/aarch64-builtins.c
10804 (aarch64_init_simd_builtin_types): Clear Poly8_t's TYPE_STRING_FLAG.
10806 2018-08-15 Ilya Leoshkevich <iii@linux.ibm.com>
10808 * config/s390/s390.c (s390_reorg): Remove loop.
10810 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
10813 (darwin_function_switched_text_sections): Delete.
10814 * gcc/config/darwin.h
10815 (TARGET_ASM_FUNCTION_SWITCHED_TEXT_SECTIONS): Likewise.
10817 2018-08-15 Iain Sandoe <iain@sandoe.co.uk>
10820 * config/darwin.h: (DEBUG_STR_OFFSETS_SECTION, DEBUG_LOCLISTS_SECTION,
10821 DEBUG_RNGLISTS_SECTION) new macros. (DEBUG_PUBNAMES_SECTION,
10822 DEBUG_PUBTYPES_SECTION) update to include GNU variant.
10824 2018-08-15 Martin Liska <mliska@suse.cz>
10826 PR tree-optimization/86925
10827 * predict.c (expr_expected_value_1): When taking
10828 later predictor, assign also probability.
10829 Use fold_build2_initializer_loc in order to fold
10830 the expression in -frounding-math.
10832 2018-08-14 Allan Sandfeld Jensen <allan.jensen@qt.io>
10834 * config/i386/i386.c (expand_vec_perm_movs): New method matching movs
10836 (expand_vec_perm_1): Try the new method.
10838 2018-08-14 Ilya Leoshkevich <iii@linux.ibm.com>
10841 * lra-lives.c (remove_some_program_points_and_update_live_ranges):
10842 Check whether lra_live_max_point is 0 before dividing.
10844 2018-08-14 Martin Sebor <msebor@redhat.com>
10846 PR tree-optimization/86650
10847 * tree-vrp.c (vrp_prop::check_array_ref): Print an inform message.
10848 (vrp_prop::check_mem_ref): Same.
10850 2018-08-13 Liu Hao <lh_mouse@126.com>
10852 * pretty-print.c (eat_esc_sequence): Swap the foreground and
10853 background colors if the COMMON_LVB_REVERSE_VIDEO flag is set,
10854 and clear it thereafter, as it only works for DBCS.
10856 2018-08-13 Liu Hao <lh_mouse@126.com>
10858 * pretty-print.c (mingw_ansi_fputs): Do not call _close() on the
10859 handle returned by _get_osf_handle().
10861 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
10863 * gcc/config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add support
10864 for folding vec_perm.
10866 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
10868 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin):
10869 Add support for gimple-folding of vec_pack() and vec_unpack()
10872 2018-08-13 Will Schmidt <will_schmidt@vnet.ibm.com>
10874 * config/rs6000/rs6000.c (rs6000_builtin_valid_without_lhs): Add
10875 vec_xst variants to the list.
10876 (rs6000_gimple_fold_builtin): Add support for folding unaligned
10877 vector loads and stores.
10879 2018-08-13 David Edelsohn <dje.gcc@gmail.com>
10881 * config.gcc (rs6000-ibm-aix4.x): Delete.
10882 (rs6000-ibm-aix5.1): Delete.
10883 (rs6000-ibm-aix5.2): Delete.
10884 (rs6000-ibm-aix5.3): Delete.
10885 * config/rs6000/aix43.h: Delete.
10886 * config/rs6000/aix51.h: Delete.
10887 * config/rs6000/aix52.h: Delete.
10888 * config/rs6000/t-aix43: Delete.
10890 2018-08-13 Ilya Leoshkevich <iii@linux.ibm.com>
10892 * config/s390/s390.c (s390_decompose_constant_pool_ref):
10894 (s390_decompose_address): Factor out constant pool ref
10897 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
10899 * config/nds32/nds32-predicates.c
10900 (nds32_can_use_bclr_p): Change return type as bool.
10901 (nds32_can_use_bset_p): Ditto.
10902 (nds32_can_use_btgl_p): Ditto.
10903 (nds32_can_use_bitci_p): Ditto.
10904 * config/nds32/nds32-protos.h
10905 (nds32_can_use_bclr_p): Change declaration.
10906 (nds32_can_use_bset_p): Ditto.
10907 (nds32_can_use_btgl_p): Ditto.
10908 (nds32_can_use_bitci_p): Ditto.
10910 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
10912 * config/nds32/nds32.c (nds32_expand_prologue, nds32_expand_epilogue):
10913 Support -msched-prolog-epilog option.
10914 * config/nds32/nds32.opt (msched-prolog-epilog): New option.
10916 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
10918 * common/config/nds32/nds32-common.c
10919 (nds32_option_optimization_table): Enalbe -malways-align.
10921 2018-08-12 Chung-Ju Wu <jasonwucj@gmail.com>
10923 * config.gcc (nds32*): Add nds32_isr.h and nds32_init.inc in
10925 * common/config/nds32/nds32-common.c (nds32_handle_option): Handle
10926 OPT_misr_secure_ case.
10927 * config/nds32/nds32-isr.c: Implementation of backward compatibility.
10928 * config/nds32/nds32-protos.h (nds32_isr_function_critical_p): New.
10929 * config/nds32/nds32.c (nds32_attribute_table): Add critical and
10931 * config/nds32/nds32.h (nds32_isr_nested_type): Add NDS32_CRITICAL.
10932 (nds32_isr_info): New field security_level.
10933 (TARGET_ISR_VECTOR_SIZE_4_BYTE): New macro.
10934 * config/nds32/nds32.md (return_internal): Consider critical attribute.
10935 * config/nds32/nds32.opt (misr-secure): New option.
10936 * config/nds32/nds32_init.inc: New file.
10937 * config/nds32/nds32_isr.h: New file.
10939 2018-08-11 John David Anglin <danglin@gcc.gnu.org>
10941 * config/pa/pa.md (UNSPEC_MEMORY_BARRIER): New unspec enum.
10942 Update comment for atomic instructions.
10943 (atomic_storeqi, atomic_storehi, atomic_storesi, atomic_storesf,
10944 atomic_loaddf, atomic_loaddf_1, atomic_storedf, atomic_storedf_1):
10946 (atomic_loaddi): Revise fence expansion to only emit fence prior to
10947 load for __ATOMIC_SEQ_CST model.
10948 (atomic_loaddi_1): Remove float register target.
10949 (atomic_storedi): Handle CONST_INT values.
10950 (atomic_storedi_1): Remove float register source. Add special case
10952 (memory_barrier): New expander and insn.
10954 2018-08-11 Jakub Jelinek <jakub@redhat.com>
10956 PR tree-optimization/86835
10957 * tree-ssa-math-opts.c (insert_reciprocals): Even when inserting
10958 new_stmt after def_gsi, make sure to insert new_square_stmt after
10959 that stmt, not 2 stmts before it.
10961 2018-08-10 Alexander Monakov <amonakov@ispras.ru>
10964 * config/i386/i386.md (<s>mul<mode>3_highpart): Use DWIH mode iterator
10967 2018-08-10 Martin Liska <mliska@suse.cz>
10970 * builtin-types.def (BT_FN_LONG_LONG_LONG_DOUBLE): Add new
10972 * builtins.c (expand_builtin_expect_with_probability):
10974 (expand_builtin_expect_with_probability): New function.
10975 (build_builtin_expect_predicate): Add new argumnet probability
10976 for BUILT_IN_EXPECT_WITH_PROBABILITY.
10977 (fold_builtin_expect):
10980 * builtins.def (BUILT_IN_EXPECT_WITH_PROBABILITY):
10981 * builtins.h (fold_builtin_expect): Set new argument.
10982 * doc/extend.texi: Document __builtin_expect_with_probability.
10983 * doc/invoke.texi: Likewise.
10984 * gimple-fold.c (gimple_fold_call): Pass new argument.
10985 * ipa-fnsummary.c (find_foldable_builtin_expect): Handle
10986 also BUILT_IN_EXPECT_WITH_PROBABILITY.
10987 * predict.c (get_predictor_value): New function.
10988 (expr_expected_value): Add new argument probability. Assume
10989 that predictor and probability are always non-null.
10990 (expr_expected_value_1): Likewise. For __builtin_expect and
10991 __builtin_expect_with_probability set probability. Handle
10992 combination in binary expressions.
10993 (tree_predict_by_opcode): Simplify code by simply calling
10994 get_predictor_value.
10995 (pass_strip_predict_hints::execute): Add handling of
10996 BUILT_IN_EXPECT_WITH_PROBABILITY.
10997 * predict.def (PRED_BUILTIN_EXPECT_WITH_PROBABILITY): Add
10999 * tree.h (DECL_BUILT_IN_P): New function.
11001 2018-08-10 Martin Liska <mliska@suse.cz>
11003 PR tree-optimization/85799
11004 * passes.def: Add argument for pass_strip_predict_hints.
11005 * predict.c (class pass_strip_predict_hints): Add new argument
11007 (strip_predictor_early): New function.
11008 (pass_strip_predict_hints::execute): Call the function to
11010 (strip_predict_hints): New function.
11011 * predict.def: Fix comment.
11013 2018-08-10 Thomas Preud'homme <thomas.preudhomme@linaro.org>
11015 * Makefile.in: Clarify which tm.texi to copy over to assert the
11016 right to grant a GFDL license for all.
11018 2018-08-09 Jeff Law <law@redhat.com>
11020 * config/m68k/m68k.c (m68k_adjust_decorated_operand): Remove
11023 2018-08-09 Andreas Schwab <schwab@linux-m68k.org>
11025 * config/m68k/m68k-protos.h (m68k_final_prescan_insn): Remove
11028 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
11030 * tree-vect-loop.c (vectorizable_reduction): Allow inner-loop
11031 reductions for variable-length vectors.
11033 2018-08-09 David Malcolm <dmalcolm@redhat.com>
11036 * common.opt (fdiagnostics-show-line-numbers): New option.
11037 * diagnostic-show-locus.c (class layout): Add fields
11038 "m_show_line_numbers_p" and "m_linenum_width";
11039 (num_digits): New function.
11040 (test_num_digits): New function.
11041 (layout::layout): Initialize new fields. Update m_x_offset
11042 logic to handle any left margin.
11043 (layout::print_source_line): Print line number when requested.
11044 (layout::start_annotation_line): New member function.
11045 (layout::print_annotation_line): Call it.
11046 (layout::print_leading_fixits): Likewise.
11047 (layout::print_trailing_fixits): Likewise. Update calls to
11048 move_to_column for new parameter.
11049 (layout::get_x_bound_for_row): Add "add_left_margin" param and use
11050 it to potentially call start_annotation_line.
11051 (layout::show_ruler): Call start_annotation_line.
11052 (selftest::test_line_numbers_multiline_range): New selftest.
11053 (selftest::diagnostic_show_locus_c_tests): Call test_num_digits
11054 and selftest::test_line_numbers_multiline_range.
11055 * diagnostic.c (diagnostic_initialize): Initialize
11056 show_line_numbers_p.
11057 * diagnostic.h (struct diagnostic_context): Add field
11058 "show_line_numbers_p".
11059 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
11060 -fno-diagnostics-show-line-numbers.
11061 * dwarf2out.c (gen_producer_string): Add
11062 OPT_fdiagnostics_show_line_numbers to the ignored options.
11063 * lto-wrapper.c (merge_and_complain): Likewise to the "pick
11064 one setting" options.
11065 (append_compiler_options): Likewise to the dropped options.
11066 (append_diag_options): Likewise to the passed-on options.
11067 * opts.c (common_handle_option): Handle the new option.
11068 * toplev.c (general_init): Set up global_dc->show_line_numbers_p.
11070 2018-08-09 Kelvin Nilsen <kelvin@gcc.gnu.org>
11072 * doc/extend.texi (PowerPC AltiVec Built-in Functions Available on
11073 ISA 2.07): Correct spelling of bcdsub to be __builtin_bcdsub. Add
11074 third argument of type "const signed char" to descriptions of
11075 __builtin_bcdadd, __builtin_bcdadd_lt, __builtin_bcdadd_eq,
11076 __builtin_bcdadd_gt, __builtin_bcdadd_ov, __builtin_bcdsub,
11077 __builtin_bcdsub_lt, __builtin_bcdsub_eq, __builtin_bcdsub_gt,
11078 __builtin_bcdsub_ov functions.
11080 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
11082 PR tree-optimization/86858
11083 * tree-vect-loop.c (vect_is_simple_reduction): Restore
11084 flow_bb_inside_loop_p calls.
11086 2018-08-09 Richard Sandiford <richard.sandiford@arm.com>
11088 PR tree-optimization/86871
11089 * tree-vect-stmts.c (vect_transform_stmt): Use gimple_get_lhs
11090 instead of gimple_assign_lhs.
11092 2018-08-09 Richard Earnshaw <rearnsha@arm.com>
11095 * config/aarch64/aarch64.md (add<mode>3_carryinC_zero): Add missing
11096 register constraint to operand 0.
11097 (add<mode>3_carryinC): Likewise.
11098 (add<mode>3_carryinV_zero, add<mode>3_carryinV): Likewise.
11100 2018-08-09 Martin Liska <mliska@suse.cz>
11103 * common.opt: Remove extra line.
11105 2018-08-09 Martin Liska <mliska@suse.cz>
11107 * params.def (PARAM_ALIGN_LOOP_ITERATIONS): Remove double dots
11108 at the end of a line, make first letter capital and end up
11109 a sentence with a dot.
11110 (PARAM_LOOP_INTERCHANGE_STRIDE_RATIO): Likewise.
11111 (PARAM_LOOP_BLOCK_TILE_SIZE): Likewise.
11112 (PARAM_GRAPHITE_MAX_NB_SCOP_PARAMS): Likewise.
11113 (PARAM_GRAPHITE_MAX_ARRAYS_PER_SCOP): Likewise.
11114 (PARAM_MAX_ISL_OPERATIONS): Likewise.
11115 (PARAM_GRAPHITE_ALLOW_CODEGEN_ERRORS): Likewise.
11116 (PARAM_PROFILE_FUNC_INTERNAL_ID): Likewise.
11117 (PARAM_INDIR_CALL_TOPN_PROFILE): Likewise.
11118 (PARAM_SLP_MAX_INSNS_IN_BB): Likewise.
11119 (PARAM_IPA_CP_EVAL_THRESHOLD): Likewise.
11120 (PARAM_IPA_CP_RECURSION_PENALTY): Likewise.
11121 (PARAM_IPA_CP_SINGLE_CALL_PENALTY): Likewise.
11122 (PARAM_IPA_CP_LOOP_HINT_BONUS): Likewise.
11123 (PARAM_IPA_CP_ARRAY_INDEX_HINT_BONUS): Likewise.
11124 (PARAM_TREE_REASSOC_WIDTH): Likewise.
11125 (PARAM_HSA_GEN_DEBUG_STORES): Likewise.
11126 (PARAM_MAX_SPECULATIVE_DEVIRT_MAYDEFS): Likewise.
11127 (PARAM_MAX_VRP_SWITCH_ASSERTIONS): Likewise.
11129 2018-08-09 Andreas Krebbel <krebbel@linux.ibm.com>
11132 * config/s390/s390.c (s390_option_override_internal): Reduce the
11133 stack-clash-protection-probe-interval param if it would be too big
11136 2018-08-08 Andreas Schwab <schwab@linux-m68k.org>
11139 * config/m68k/m68k.h (FINAL_PRESCAN_INSN): Don't define.
11140 * config/m68k/m68k.c (handle_move_double): Don't call
11141 m68k_final_prescan_insn.
11142 (m68k_adjust_decorated_operand): Renamed from
11143 m68k_final_prescan_insn, remove first and third operand and
11145 (print_operand): Call it.
11146 (print_operand_address): Call it.
11148 2018-08-08 Nathan Sidwell <nathan@acm.org>
11150 * diagnostic.c (diagnostic_report_current_module): Use
11151 linemap_included_from & linemap_included_from_linemap.
11153 2018-08-08 Hongbo Zhang <hongbo.zhang@linaro.org>
11155 * config/aarch64/aarch64-cores.def: Add phecda core.
11156 * config/aarch64/aarch64-tune.md: Regenerate.
11157 * doc/invoke.texi: Add phecda core.
11159 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
11162 * config/s390/constraints.md ("NxHD0", "NxSD0"): New constraint
11164 * config/s390/s390.md ("movti"): Add more alternatives for
11165 constant to GPR copies.
11167 2018-08-08 Andreas Krebbel <krebbel@linux.ibm.com>
11169 * config/s390/s390.c: Fix whitespace damage throughout the file.
11170 * config/s390/s390.h: Likewise.
11171 * config/s390/tpf.h: Likewise.
11173 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
11175 * config/s390/s390.c (s390_loadrelative_operand_p):
11176 Remove TARGET_CPU_ZARCH usages.
11177 (s390_rtx_costs): Likewise.
11178 (s390_legitimate_constant_p): Likewise.
11179 (s390_cannot_force_const_mem): Likewise.
11180 (legitimate_reload_constant_p): Likewise.
11181 (s390_preferred_reload_class): Likewise.
11182 (legitimize_pic_address): Likewise.
11183 (legitimize_tls_address): Likewise.
11184 (s390_split_branches): Removed.
11185 (s390_add_execute): Removed.
11186 (s390_dump_pool): Remove TARGET_CPU_ZARCH usages.
11187 (s390_mainpool_start): Likewise.
11188 (s390_mainpool_finish): Likewise.
11189 (s390_mainpool_cancel): Removed.
11190 (s390_chunkify_start): Remove TARGET_CPU_ZARCH usages.
11191 (s390_chunkify_cancel): Likewise.
11192 (s390_return_addr_rtx): Likewise.
11193 (s390_register_info): Remove split_branches_pending_p uages.
11194 (s390_optimize_register_info): Likewise.
11195 (s390_init_frame_layout): Remove TARGET_CPU_ZARCH and
11196 split_branches_pending_p usages.
11197 (s390_can_eliminate): Remove TARGET_CPU_ZARCH usages.
11198 (s390_load_got): Likewise.
11199 (s390_expand_split_stack_prologue): Likewise.
11200 (output_asm_nops): Likewise.
11201 (s390_function_profiler): Likewise.
11202 (s390_emit_call): Likewise.
11203 (s390_conditional_register_usage): Likewise.
11204 (s390_optimize_prologue): Likewise.
11205 (s390_reorg): Remove TARGET_CPU_ZARCH and
11206 split_branches_pending_p usages.
11207 (s390_option_override_internal): Remove TARGET_CPU_ZARCH
11209 (s390_output_indirect_thunk_function): Likewise.
11210 * config/s390/s390.h (TARGET_CPU_ZARCH): Removed.
11211 (TARGET_CPU_ZARCH_P): Removed.
11212 (struct machine_function): Remove split_branches_pending_p.
11213 * config/s390/s390.md: Remove TARGET_CPU_ZARCH usages.
11215 2018-08-08 Ilya Leoshkevich <iii@linux.ibm.com>
11217 * common/config/s390/s390-common.c (processor_flags_table):
11219 * config.gcc: Remove with_arch/with_tune support.
11220 * config/s390/2064.md: Remove cpu attribute comparisons.
11221 * config/s390/driver-native.c (s390_host_detect_local_cpu):
11223 * config/s390/linux.h (ASM_SPEC):
11224 Remove -march support.
11225 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal):
11226 Use a table to get an arch level.
11227 * config/s390/s390-opts.h (enum processor_type):
11228 Remove enum values.
11229 * config/s390/s390.c
11230 (processor_table): Remove entries, add arch_level values.
11231 (s390_issue_rate): Remove cases.
11232 (s390_option_override): Adjust
11233 s390_option_override_internal() call.
11234 (s390_option_override_internal): Remove deprecation warning.
11235 (s390_valid_target_attribute_tree): Adjust
11236 s390_option_override_internal() call.
11237 * config/s390/s390.h (struct s390_processor):
11238 Share with s390-c.c, add arch_level field.
11239 * config/s390/s390.md:
11240 Remove occurrences in cpu attribute.
11241 * config/s390/s390.opt: Remove -march/-mtune support.
11242 * config/s390/tpf.h (ASM_SPEC): Remove -march support.
11243 * doc/invoke.texi: Remove deprecation warning.
11245 2018-08-08 Luis Machado <luis.machado@linaro.org>
11247 * config/aarch64/aarch64.c (qdf24xx_vector_cost): New static
11249 (qdf24xx_tunings): Set vector cost structure to
11250 qdf24xx_vector_cost.
11252 * config/aarch64/aarch64.c (qdf24xx_addrcost_table)
11253 <register_sextend>: Set to 3.
11255 2018-08-07 Richard Sandiford <richard.sandiford@arm.com>
11258 * config/aarch64/iterators.md (FRECP, frecp_suffix): Delete.
11259 * config/aarch64/aarch64-simd.md
11260 (aarch64_frecp<FRECP:frecp_suffix><mode>): Fold FRECPE into...
11261 (@aarch64_frecpe<mode>): ...here and the move FRECPX to...
11262 (aarch64_frecpx<mode>): ...this new pattern.
11263 * config/aarch64/aarch64-simd-builtins.def: Remove comment
11264 about aarch64_frecp<FRECP:frecp_suffix><mode>.
11266 2018-08-07 Martin Liska <mliska@suse.cz>
11268 PR middle-end/83023
11269 * predict.c (expr_expected_value_1): Handle DECL_IS_MALLOC,
11270 BUILT_IN_REALLOC and DECL_IS_OPERATOR_NEW.
11271 * predict.def (PRED_MALLOC_NONNULL): New predictor.
11272 * doc/extend.texi: Document that malloc attribute adds
11275 2018-08-06 John David Anglin <danglin@gcc.gnu.org>
11278 * config/pa/pa.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11279 Define to speculation_safe_value_not_needed.
11281 2018-08-06 Jeff Law <law@redhat.com>
11283 * tree-ssa-dom.c (dom_opt_dom_walker::optimize_stmt): Pass down
11284 the vr_values instance to cprop_into_stmt.
11285 (cprop_into_stmt): Pass vr_values instance down to cprop_operand.
11286 (cprop_operand): Also query EVRP to determine if OP is a constant.
11288 2018-08-06 Nathan Sidwell <nathan@acm.org>
11290 * diagnostic.c (diagnostic_report_current_module): Reroll
11291 included-at loop. Translate text.
11293 2018-08-06 David Malcolm <dmalcolm@redhat.com>
11295 * function-tests.c (selftest::test_expansion_to_rtl): Call
11296 free_after_compilation.
11298 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11300 * config/aarch64/aarch64.md: Add clobber highs to tls_desc.
11302 2018-08-06 Andreas Krebbel <krebbel@linux.ibm.com>
11304 * config/s390/s390.c (s390_loop_unroll_adjust): Prevent small
11305 loops with memory block operations from getting unrolled.
11307 2018-08-06 Ulrich Weigand <uweigand@de.ibm.com>
11310 * config/spu/spu.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11311 Define to speculation_safe_value_not_needed.
11313 2018-08-06 Jeff Law <law@redhat.com>
11315 * reload1.c (forget_old_reloads_1): Adjust CLOBBER_HIGH
11318 2018-08-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
11321 * gcc/tree.c (build_common_tree_nodes): Initialize integer_types array
11322 with all enabled __intN types.
11324 * gcc/testsuite/gcc.target/msp430/pr86662.c: New test.
11326 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11328 * alias.c (record_set): Check for clobber high.
11329 * cfgexpand.c (expand_gimple_stmt): Likewise.
11330 * combine-stack-adj.c (single_set_for_csa): Likewise.
11331 * combine.c (find_single_use_1): Likewise.
11332 (set_nonzero_bits_and_sign_copies): Likewise.
11333 (get_combine_src_dest): Likewise.
11334 (is_parallel_of_n_reg_sets): Likewise.
11335 (try_combine): Likewise.
11336 (record_dead_and_set_regs_1): Likewise.
11337 (reg_dead_at_p_1): Likewise.
11338 (reg_dead_at_p): Likewise.
11339 * dce.c (deletable_insn_p): Likewise.
11340 (mark_nonreg_stores_1): Likewise.
11341 (mark_nonreg_stores_2): Likewise.
11342 * df-scan.c (df_find_hard_reg_defs): Likewise.
11343 (df_uses_record): Likewise.
11344 (df_get_call_refs): Likewise.
11345 * dwarf2out.c (mem_loc_descriptor): Likewise.
11346 * haifa-sched.c (haifa_classify_rtx): Likewise.
11347 * ira-build.c (create_insn_allocnos): Likewise.
11348 * ira-costs.c (scan_one_insn): Likewise.
11349 * ira.c (equiv_init_movable_p): Likewise.
11350 (rtx_moveable_p): Likewise.
11351 (interesting_dest_for_shprep): Likewise.
11352 * jump.c (mark_jump_label_1): Likewise.
11353 * postreload-gcse.c (record_opr_changes): Likewise.
11354 * postreload.c (reload_cse_simplify): Likewise.
11355 (struct reg_use): Add source expr.
11356 (reload_combine): Check for clobber high.
11357 (reload_combine_note_use): Likewise.
11358 (reload_cse_move2add): Likewise.
11359 (move2add_note_store): Likewise.
11360 * print-rtl.c (print_pattern): Likewise.
11361 * recog.c (decode_asm_operands): Likewise.
11362 (store_data_bypass_p): Likewise.
11363 (if_test_bypass_p): Likewise.
11364 * regcprop.c (kill_clobbered_value): Likewise.
11365 (kill_set_value): Likewise.
11366 * reginfo.c (reg_scan_mark_refs): Likewise.
11367 * reload1.c (maybe_fix_stack_asms): Likewise.
11368 (eliminate_regs_1): Likewise.
11369 (elimination_effects): Likewise.
11370 (mark_not_eliminable): Likewise.
11371 (scan_paradoxical_subregs): Likewise.
11372 (forget_old_reloads_1): Likewise.
11373 * reorg.c (find_end_label): Likewise.
11374 (try_merge_delay_insns): Likewise.
11375 (redundant_insn): Likewise.
11376 (own_thread_p): Likewise.
11377 (fill_simple_delay_slots): Likewise.
11378 (fill_slots_from_thread): Likewise.
11379 (dbr_schedule): Likewise.
11380 * resource.c (update_live_status): Likewise.
11381 (mark_referenced_resources): Likewise.
11382 (mark_set_resources): Likewise.
11383 * rtl.c (copy_rtx): Likewise.
11384 * rtlanal.c (reg_referenced_p): Likewise.
11385 (single_set_2): Likewise.
11386 (noop_move_p): Likewise.
11387 (note_stores): Likewise.
11388 * sched-deps.c (sched_analyze_reg): Likewise.
11389 (sched_analyze_insn): Likewise.
11391 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11393 * cse.c (invalidate_reg): New function extracted from...
11394 (invalidate): ...here.
11395 (canonicalize_insn): Check for clobber high.
11396 (invalidate_from_clobbers): invalidate clobber highs.
11397 (invalidate_from_sets_and_clobbers): Likewise.
11398 (count_reg_usage): Check for clobber high.
11399 (insn_live_p): Likewise.
11400 * cselib.c (cselib_expand_value_rtx_1):Likewise.
11401 (cselib_invalidate_regno): Check for clobber in setter.
11402 (cselib_invalidate_rtx): Pass through setter.
11403 (cselib_invalidate_rtx_note_stores):
11404 (cselib_process_insn): Check for clobber high.
11405 * cselib.h (cselib_invalidate_rtx): Add operand.
11407 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11409 * lra-eliminations.c (lra_eliminate_regs_1): Check for clobber high.
11410 (mark_not_eliminable): Likewise.
11411 * lra-int.h (struct lra_insn_reg): Add clobber high marker.
11412 * lra-lives.c (process_bb_lives): Check for clobber high.
11413 * lra.c (new_insn_reg): Remember clobber highs.
11414 (collect_non_operand_hard_regs): Check for clobber high.
11415 (lra_set_insn_recog_data): Likewise.
11416 (add_regs_to_insn_regno_info): Likewise.
11417 (lra_update_insn_regno_info): Likewise.
11419 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11421 * rtl.h (reg_is_clobbered_by_clobber_high): Add declarations.
11422 * rtlanal.c (reg_is_clobbered_by_clobber_high): Add function.
11424 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11426 * emit-rtl.c (verify_rtx_sharing): Check for CLOBBER_HIGH.
11427 (copy_insn_1): Likewise.
11428 (gen_hard_reg_clobber_high): New gen function.
11429 * genconfig.c (walk_insn_part): Check for CLOBBER_HIGH.
11430 * genemit.c (gen_exp): Likewise.
11431 (gen_emit_seq): Pass through info.
11432 (gen_insn): Check for CLOBBER_HIGH.
11433 (gen_expand): Pass through info.
11434 (gen_split): Likewise.
11435 (output_add_clobbers): Likewise.
11436 * genrecog.c (validate_pattern): Check for CLOBBER_HIGH.
11437 (remove_clobbers): Likewise.
11438 * rtl.h (gen_hard_reg_clobber_high): New declaration.
11440 2018-08-06 Alan Hayward <alan.hayward@arm.com>
11442 * doc/rtl.texi (clobber_high): Add.
11443 (parallel): Add in clobber high
11444 * rtl.c (rtl_check_failed_code3): Add function.
11445 * rtl.def (CLOBBER_HIGH): Add expression.
11446 * rtl.h (RTL_CHECKC3): Add macro.
11447 (rtl_check_failed_code3): Add declaration.
11448 (XC3EXP): Add macro.
11450 2018-08-05 H.J. Lu <hongjiu.lu@intel.com>
11453 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Set
11454 cfun->machine->max_used_stack_alignment if needed.
11456 2018-08-04 Martin Sebor <msebor@redhat.com>
11458 PR tree-optimization/86571
11459 * gimple-ssa-sprintf.c (format_floating): Extend upper bound of
11462 2018-08-03 Sandra Loosemore <sandra@codesourcery.com>
11465 * config/nios2/nios2.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11468 2018-08-03 Jeff Law <law@redhat.com>
11471 * config/mn10300/mn10300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11472 Define to speculation_safe_value_not_needed.
11474 2018-08-03 David Malcolm <dmalcolm@redhat.com>
11476 * doc/gcov.texi (-x): Remove duplicate "to".
11477 * doc/invoke.texi (-Wnoexcept-type): Remove duplicate "calls".
11478 (-Wif-not-aligned): Remove duplicate "is".
11479 (-flto): Remove duplicate "the".
11480 (MicroBlaze Options): In examples of "-mcpu=cpu-type", remove
11481 duplicate "v5.00.b".
11482 (MSP430 Options): Remove duplicate "and" from the description
11483 of "-mgprel-sec=regexp".
11484 (x86 Options): Remove duplicate copies of "vmldLog102" and
11485 vmlsLog104 from description of "-mveclibabi=type".
11487 2018-08-03 Richard Sandiford <richard.sandiford@arm.com>
11489 * internal-fn.h (first_commutative_argument): Declare.
11490 * internal-fn.c (first_commutative_argument): New function.
11491 * tree-vect-slp.c (vect_get_and_check_slp_defs): Remove extra
11492 restrictions for pattern statements. Use first_commutative_argument
11493 to look for commutative operands in calls to internal functions.
11495 2018-08-03 Aldy Hernandez <aldyh@redhat.com>
11497 * Makefile.in (wide-int-range.o): New.
11498 * tree-vrp.c: Move all the wide_int_* functions to...
11499 * wide-int-range.cc: ...here.
11500 * tree-vrp.h: Move all the wide_int_* prototypes to...
11501 * wide-int-range.h: ...here.
11503 2018-08-03 Tom de Vries <tdevries@suse.de>
11505 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): Return
11507 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Remove define.
11508 * except.c (output_function_exception_table): Do early exit if
11509 targetm_common.except_unwind_info (&global_options) == UI_NONE.
11511 2018-08-03 Martin Liska <mliska@suse.cz>
11513 * predict.c (dump_prediction): Change to 2 digits
11516 2018-08-03 Siddhesh Poyarekar <siddhesh@sourceware.org>
11518 * config/aarch64/falkor.md (falkor_am_1_vxvy_vxvy): Move
11520 (falkor_am_1_gtov_gtov): ... a new insn reservation.
11522 2018-07-19 Ilya Leoshkevich <iii@linux.ibm.com>
11524 * config/nds32/nds32.c (nds32_hard_regno_mode_ok): Replace > with >=.
11525 * df-problems.c (df_remove_dead_eq_notes): Replace > with >=.
11526 * dwarf2out.c (mem_loc_descriptor): Replace > with >=.
11527 * lra-constraints.c (spill_hard_reg_in_range): Replace <= with <.
11528 * lra-remat.c (call_used_input_regno_present_p): Replace <= with <.
11530 2018-08-02 David Malcolm <dmalcolm@redhat.com>
11532 * diagnostic-show-locus.c (diagnostic_show_locus): Use
11533 pp_take_prefix when saving the existing prefix.
11534 * diagnostic.c (diagnostic_append_note): Likewise.
11535 * langhooks.c (lhd_print_error_function): Likewise.
11536 * pretty-print.c (pp_set_prefix): Drop the "const" from "prefix"
11537 param's type. Free the existing prefix.
11538 (pp_take_prefix): New function.
11539 (pretty_printer::pretty_printer): Drop the prefix parameter.
11540 Rename the length parameter to match the comment.
11541 (pretty_printer::~pretty_printer): Free the prefix.
11542 * pretty-print.h (pretty_printer::pretty_printer): Drop the prefix
11544 (struct pretty_printer): Drop the "const" from "prefix" field's
11545 type and clarify memory management.
11546 (pp_set_prefix): Drop the "const" from the 2nd param.
11547 (pp_take_prefix): New decl.
11549 2018-08-02 Aaron Sawdey <acsawdey@linux.ibm.com>
11551 * config/rs6000/rs6000-string.c (select_block_compare_mode): Move test
11552 for word_mode_ok here instead of passing as argument.
11553 (expand_block_compare): Change select_block_compare_mode() call.
11554 (expand_strncmp_gpr_sequence): New function.
11555 (expand_strn_compare): Make use of expand_strncmp_gpr_sequence.
11557 2018-08-02 Jeff Law <law@redhat.com>
11560 * config/m68k/m68k.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11561 Define to speculation_safe_value_not_needed.
11564 * config/h8300/h8300.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11565 Define to speculation_safe_value_not_needed.
11567 2018-08-02 Tom de Vries <tdevries@suse.de>
11570 * common/config/nvptx/nvptx-common.c (nvptx_except_unwind_info): New
11571 function. Return UI_TARGET unconditionally.
11572 (TARGET_EXCEPT_UNWIND_INFO): Redefine to nvptx_except_unwind_info.
11573 * config/nvptx/nvptx.c (TARGET_ASM_BYTE_OP): Emit commented out '.byte'.
11575 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
11577 * genemit.c (print_overload_test): Fix typo.
11579 2018-08-02 Richard Biener <rguenther@suse.de>
11581 PR tree-optimization/86816
11582 * tree-ssa-tail-merge.c (tail_merge_valueize): New function
11583 which checks for value availability before querying it.
11584 (gvn_uses_equal): Use it.
11585 (same_succ_hash): Likewise.
11586 (gimple_equal_p): Likewise.
11588 2018-08-02 Nick Clifton <nickc@redhat.com>
11591 * config/stormy16/stormy16.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11592 Define to speculation_safe_value_not_needed.
11595 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11596 Define to speculation_safe_value_not_needed.
11599 * config/v850/v850.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11600 Define to speculation_safe_value_not_needed.
11603 * config/rx/rx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11604 Define to speculation_safe_value_not_needed.
11607 * config/msp430/msp430.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11608 Define to speculation_safe_value_not_needed.
11611 * config/mcore/mcore.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11612 Define to speculation_safe_value_not_needed.
11615 * config/m32r/m32r.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11616 Define to speculation_safe_value_not_needed.
11619 * config/iq2000/iq2000.c (TARGET_HAVE_SPECULATION_SAFE_VALUE):
11620 Define to speculation_safe_value_not_needed.
11623 * config/frv/frv.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
11624 speculation_safe_value_not_needed.
11627 * config/fr30/fr30.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define
11628 to speculation_safe_value_not_needed.
11630 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
11632 * doc/md.texi: Expand the documentation of instruction names
11633 to mention port-local uses. Document '@' in pattern names.
11634 * read-md.h (overloaded_instance, overloaded_name): New structs.
11635 (mapping): Declare.
11636 (md_reader::handle_overloaded_name): New member function.
11637 (md_reader::get_overloads): Likewise.
11638 (md_reader::m_first_overload): New member variable.
11639 (md_reader::m_next_overload_ptr): Likewise.
11640 (md_reader::m_overloads_htab): Likewise.
11641 * read-md.c (md_reader::md_reader): Initialize m_first_overload,
11642 m_next_overload_ptr and m_overloads_htab.
11643 * read-rtl.c (iterator_group): Add "type" and "get_c_token" fields.
11644 (get_mode_token, get_code_token, get_int_token): New functions.
11645 (map_attr_string): Add an optional argument that passes back
11646 the associated iterator.
11647 (overloaded_name_hash, overloaded_name_eq_p, named_rtx_p):
11648 (md_reader::handle_overloaded_name, add_overload_instance): New
11650 (apply_iterators): Handle '@' names. Report an error if '@'
11651 is used without iterators.
11652 (initialize_iterators): Initialize the new iterator_group fields.
11653 * genopinit.c (handle_overloaded_code_for)
11654 (handle_overloaded_gen): New functions.
11655 (main): Use them to print declarations of maybe_code_for_* and
11656 maybe_gen_* functions, and inline definitions of code_for_* and gen_*.
11657 * genemit.c (print_overload_arguments, print_overload_test)
11658 (handle_overloaded_code_for, handle_overloaded_gen): New functions.
11659 (main): Use it to print definitions of maybe_code_for_* and
11660 maybe_gen_* functions.
11661 * config/aarch64/aarch64.c (aarch64_split_128bit_move): Use
11662 gen_aarch64_mov{low,high}_di and gen_aarch64_movdi_{low,high}
11663 instead of explicit mode checks.
11664 (aarch64_split_simd_combine): Likewise gen_aarch64_simd_combine.
11665 (aarch64_split_simd_move): Likewise gen_aarch64_split_simd_mov.
11666 (aarch64_emit_load_exclusive): Likewise gen_aarch64_load_exclusive.
11667 (aarch64_emit_store_exclusive): Likewise gen_aarch64_store_exclusive.
11668 (aarch64_expand_compare_and_swap): Likewise
11669 gen_aarch64_compare_and_swap and gen_aarch64_compare_and_swap_lse
11670 (aarch64_gen_atomic_cas): Likewise gen_aarch64_atomic_cas.
11671 (aarch64_emit_atomic_swap): Likewise gen_aarch64_atomic_swp.
11672 (aarch64_constant_pool_reload_icode): Delete.
11673 (aarch64_secondary_reload): Use code_for_aarch64_reload_movcp
11674 instead of aarch64_constant_pool_reload_icode. Use
11675 code_for_aarch64_reload_mov instead of explicit mode checks.
11676 (rsqrte_type, get_rsqrte_type, rsqrts_type, get_rsqrts_type): Delete.
11677 (aarch64_emit_approx_sqrt): Use gen_aarch64_rsqrte instead of
11678 get_rsqrte_type and gen_aarch64_rsqrts instead of gen_rqrts_type.
11679 (recpe_type, get_recpe_type, recps_type, get_recps_type): Delete.
11680 (aarch64_emit_approx_div): Use gen_aarch64_frecpe instead of
11681 get_recpe_type and gen_aarch64_frecps instead of get_recps_type.
11682 (aarch64_atomic_load_op_code): Delete.
11683 (aarch64_emit_atomic_load_op): Likewise.
11684 (aarch64_gen_atomic_ldop): Use UNSPECV_ATOMIC_* instead of
11685 aarch64_atomic_load_op_code. Use gen_aarch64_atomic_load
11686 instead of aarch64_emit_atomic_load_op.
11687 * config/aarch64/aarch64.md (aarch64_reload_movcp<GPF_TF:mode><P:mode>)
11688 (aarch64_reload_movcp<VALL:mode><P:mode>, aarch64_reload_mov<mode>)
11689 (aarch64_movdi_<mode>low, aarch64_movdi_<mode>high)
11690 (aarch64_mov<mode>high_di, aarch64_mov<mode>low_di): Add a '@'
11691 character before the pattern name.
11692 * config/aarch64/aarch64-simd.md (aarch64_split_simd_mov<mode>)
11693 (aarch64_rsqrte<mode>, aarch64_rsqrts<mode>)
11694 (aarch64_simd_combine<mode>, aarch64_frecpe<mode>)
11695 (aarch64_frecps<mode>): Likewise.
11696 * config/aarch64/atomics.md (atomic_compare_and_swap<mode>)
11697 (aarch64_compare_and_swap<mode>, aarch64_compare_and_swap<mode>_lse)
11698 (aarch64_load_exclusive<mode>, aarch64_store_exclusive<mode>)
11699 (aarch64_atomic_swp<mode>, aarch64_atomic_cas<mode>)
11700 (aarch64_atomic_load<atomic_ldop><mode>): Likewise.
11702 2018-08-02 Richard Sandiford <richard.sandiford@arm.com>
11704 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
11705 Allow HFmode constants if TARGET_FP_F16INST.
11707 2018-08-02 Jackson Woodruff <jackson.woodruff@arm.com>
11710 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp):
11711 No longer check last store for clobber of address register.
11713 2018-08-02 Martin Liska <mliska@suse.cz>
11715 PR gcov-profile/86817
11716 * gcov.c (process_all_functions): New function.
11718 (process_file): Move functions processing to
11719 process_all_functions.
11721 2018-08-02 David Malcolm <dmalcolm@redhat.com>
11723 * dumpfile.c (dump_user_location_t::dump_user_location_t): Add
11724 "const" to the "gimple *" and "rtx_insn *" parameters.
11725 * dumpfile.h (dump_user_location_t::dump_user_location_t):
11727 (dump_location_t::dump_location_t): Likewise.
11729 2018-08-01 Martin Sebor <msebor@redhat.com>
11731 PR tree-optimization/86650
11732 * gimple-pretty-print.c (percent_G_format): Accept a "gimple *"
11733 rather than a "gcall *". Directly pass the data of interest
11734 to percent_K_format, rather than building a temporary CALL_EXPR
11736 * gimple-fold.c (gimple_fold_builtin_strncpy): Adjust.
11737 (gimple_fold_builtin_strncat): Adjust.
11738 * gimple-ssa-warn-restrict.h (check_bounds_or_overlap): Replace
11739 gcall* argument with gimple*.
11740 * gimple-ssa-warn-restrict.c (check_call): Same.
11741 (wrestrict_dom_walker::before_dom_children): Same.
11742 (builtin_access::builtin_access): Same.
11743 (check_bounds_or_overlap): Same
11744 (maybe_diag_overlap): Same.
11745 (maybe_diag_offset_bounds): Same.
11746 * tree-diagnostic.c (default_tree_printer): Move usage of
11747 EXPR_LOCATION (t) and TREE_BLOCK (t) from within percent_K_format
11749 * tree-pretty-print.c (percent_K_format): Add argument.
11750 * tree-pretty-print.h: Add argument.
11751 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Adjust.
11752 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Adjust.
11753 (maybe_diag_stxncpy_trunc): Same.
11754 (handle_builtin_stxncpy): Same.
11755 (handle_builtin_strcat): Same.
11757 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11759 * match.pd: Optimise pointer range checks.
11761 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11763 PR tree-optimization/86758
11764 * tree-vect-stmts.c (vectorizable_simd_clone_call): Don't try
11765 to remove pattern statements.
11767 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11769 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Use the
11770 result of dfs_enumerate_from when constructing stmt_vec_infos,
11771 instead of additionally calling get_loop_body.
11773 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11775 * tree-vectorizer.h (vect_create_data_ref_ptr): Remove inv_p
11777 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Likewise.
11778 When creating an iv, assert that the step is not known to be zero.
11779 (vect_setup_realignment): Update call accordingly.
11780 * tree-vect-stmts.c (vectorizable_store): Likewise.
11781 (vectorizable_load): Likewise. Handle VMAT_INVARIANT separately.
11783 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11785 * tree-vectorizer.h (vect_stmt_to_vectorize): New function.
11786 * tree-vect-loop.c (vect_update_vf_for_slp): Use it.
11787 (vectorizable_reduction): Likewise.
11788 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
11789 (vect_detect_hybrid_slp_stmts): Likewise.
11790 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
11792 2018-08-01 Aldy Hernandez <aldyh@redhat.com>
11794 * tree-vrp (zero_nonzero_bits_from_bounds): Rename to...
11795 (wide_int_set_zero_nonzero_bits): ...this.
11796 (zero_nonzero_bits_from_vr): Rename to...
11797 (vrp_set_zero_nonzero_bits): ...this.
11798 (extract_range_from_multiplicative_op_1): Abstract wide int
11800 (wide_int_range_multiplicative_op): ...here.
11801 (extract_range_from_binary_expr_1): Extract wide int binary
11802 operations into their own functions.
11803 (wide_int_range_lshift): New.
11804 (wide_int_range_can_optimize_bit_op): New.
11805 (wide_int_range_shift_undefined_p): New.
11806 (wide_int_range_bit_xor): New.
11807 (wide_int_range_bit_ior): New.
11808 (wide_int_range_bit_and): New.
11809 (wide_int_range_trunc_mod): New.
11810 (extract_range_into_wide_ints): New.
11811 (vrp_shift_undefined_p): New.
11812 (extract_range_from_multiplicative_op): New.
11813 (vrp_can_optimize_bit_op): New.
11814 * tree-vrp.h (value_range::dump): New.
11815 (wide_int_range_multiplicative_op): New.
11816 (wide_int_range_lshift):New.
11817 (wide_int_range_shift_undefined_p): New.
11818 (wide_int_range_bit_xor): New.
11819 (wide_int_range_bit_ior): New.
11820 (wide_int_range_bit_and): New.
11821 (wide_int_range_trunc_mod): New.
11822 (zero_nonzero_bits_from_bounds): Rename to...
11823 (wide_int_set_zero_nonzero_bits): ...this.
11824 (zero_nonzero_bits_from_vr): Rename to...
11825 (vrp_set_zero_nonzero_bits): ...this.
11826 (range_easy_mask_min_max): Rename to...
11827 (wide_int_range_can_optimize_bit_op): this.
11828 * vr-values.c (simplify_bit_ops_using_ranges): Rename
11829 zero_nonzero_bits_from_vr into vrp_set_zero_nonzero_bits.
11831 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11833 * tree-vectorizer.h (vect_orig_stmt): New function.
11834 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Use it.
11835 * tree-vect-loop.c (vect_model_reduction_cost): Likewise.
11836 (vect_create_epilog_for_reduction): Likewise.
11837 (vectorizable_live_operation): Likewise.
11838 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
11839 (vect_detect_hybrid_slp_stmts, vect_schedule_slp): Likewise.
11840 * tree-vect-stmts.c (vectorizable_call): Likewise.
11841 (vectorizable_simd_clone_call, vect_remove_stores): Likewise.
11843 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11845 * tree-vectorizer.h (vect_transform_stmt): Remove grouped_store
11847 * tree-vect-stmts.c (vect_transform_stmt): Likewise.
11848 * tree-vect-loop.c (vect_transform_loop_stmt): Update call accordingly.
11849 (vect_transform_loop): Likewise.
11850 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11852 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11854 * tree-vectorizer.h (vect_schedule_slp): Return void.
11855 * tree-vect-slp.c (vect_schedule_slp_instance): Likewise.
11856 (vect_schedule_slp): Likewise.
11858 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11860 * tree-vect-loop.c (vect_transform_loop_stmt): Remove slp_scheduled
11862 (vect_transform_loop): Update calls accordingly. Schedule SLP
11863 instances before the main loop, if any exist.
11865 2018-08-01 Richard Sandiford <richard.sandiford@arm.com>
11867 PR tree-optimization/86749
11868 * tree-vect-patterns.c (vect_determine_min_output_precision_1):
11869 If the lhs is used in a COND_EXPR, check that it is being used
11870 as the "then" or "else" value.
11872 2018-08-01 Tom de Vries <tdevries@suse.de>
11875 * config/nvptx/nvptx.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Define to
11876 speculation_safe_value_not_needed.
11878 2018-08-01 Richard Biener <rguenther@suse.de>
11880 * tree-ssa-sccvn.c (visit_phi): Compare invariant addresses
11881 as base and offset.
11883 2018-08-01 Martin Liska <mliska@suse.cz>
11885 * value-prof.c (gimple_divmod_fixed_value_transform): Unify
11886 format how successful transformation is dumped.
11887 (gimple_mod_pow2_value_transform): Likewise.
11888 (gimple_mod_subtract_transform): Likewise.
11889 (gimple_stringops_transform): Likewise.
11891 2018-08-01 Martin Liska <mliska@suse.cz>
11893 PR value-prof/35543
11894 * value-prof.c (interesting_stringop_to_profile_p):
11895 Simplify the code and add BUILT_IN_MEMMOVE.
11896 (gimple_stringops_transform): Likewise.
11898 2018-08-01 Sam Tebbs <sam.tebbs@arm.com>
11900 * config/aarch64/aarch64-simd.md
11901 (*aarch64_get_lane_zero_extendsi<mode>): Rename to...
11902 (*aarch64_get_lane_zero_extend<GPI:mode><VDQQH:mode>): ... This and
11903 use GPI iterator instead of SI mode.
11905 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
11907 * config/rs6000/rs6000.md (speculation_barrier): Renamed from
11908 rs6000_speculation_barrier.
11909 * config/rs6000/rs6000.c (rs6000_expand_builtin): Adjust for
11910 new barrier pattern name.
11912 2018-08-01 Richard Earnshaw <rearnsha@arm.com>
11914 * config/i386/i386.md (unspecv): Add UNSPECV_SPECULATION_BARRIER.
11915 (speculation_barrier): New insn.
11917 2018-08-01 Richard Biener <rguenther@suse.de>
11920 * graphite.h: Include isl/id.h and isl/space.h to allow build
11923 2018-08-01 Jan Willem Jagersma <jwjagersma@gmail.com>
11926 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
11927 mode for COFF targets.
11928 * defaults.h (TARGET_COFF): Define.
11929 * config/i386/djgpp.h (TARGET_ASM_LTO_START, TARGET_ASM_LTO_END,
11930 TARGET_COFF): Define.
11931 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): Declare.
11932 * config/i386/djgpp.c (saved_debug_info_level): New static variable.
11933 (i386_djgpp_asm_lto_start, i386_djgpp_asm_lto_end): New functions.
11935 2018-07-31 Alexandre Oliva <oliva@adacore.com>
11937 * gimple-streamer-in.c (input_bb): Restore BB discriminator.
11938 * gimple-streamer-out.c (output_bb): Save it.
11939 * lto-streamer-in.c (input_struct_function_base): Restore
11940 instance discriminator if available. Create map on demand.
11941 * lto-streamer-out.c (output_struct_function_base): Save it if
11943 * final.c (decl_to_instance_map): Document LTO strategy.
11945 2018-07-31 Alexandre Oliva <oliva@adacore.com>
11946 Olivier Hainque <hainque@adacore.com>
11948 * debug.h (decl_to_instance_map_t): New type.
11949 (decl_to_instance_map): Declare.
11950 (maybe_create_decl_to_instance_map): New inline function.
11951 * final.c (bb_discriminator, last_bb_discriminator): New statics,
11952 to track basic block discriminators.
11953 (final_start_function_1): Initialize them.
11954 (final_scan_insn_1): On NOTE_INSN_BASIC_BLOCK, track
11956 (decl_to_instance_map): New variable.
11957 (map_decl_to_instance, maybe_set_discriminator): New functions.
11958 (notice_source_line): Set discriminator.
11960 2018-07-31 Ian Lance Taylor <iant@golang.org>
11962 * targhooks.c (default_have_speculation_safe_value): Add
11965 2018-07-31 David Malcolm <dmalcolm@redhat.com>
11967 * dump-context.h: Include "pretty-print.h".
11968 (dump_context::refresh_dumps_are_enabled): New decl.
11969 (dump_context::emit_item): New decl.
11970 (class dump_context): Add fields "m_test_pp" and
11972 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
11973 (temp_dump_context::get_dumped_text): New decl.
11974 (class temp_dump_context): Add field "m_pp".
11975 * dumpfile.c (refresh_dumps_are_enabled): Convert to...
11976 (dump_context::refresh_dumps_are_enabled): ...and add a test for
11978 (set_dump_file): Update for above change.
11979 (set_alt_dump_file): Likewise.
11980 (dump_loc): New overload, taking a pretty_printer *.
11981 (dump_context::dump_loc): Call end_any_optinfo. Dump the location
11982 to any test pretty-printer.
11983 (make_item_for_dump_gimple_stmt): New function, adapted from
11984 optinfo::add_gimple_stmt.
11985 (dump_context::dump_gimple_stmt): Call it, and use the result,
11986 eliminating the direct usage of dump_file and alt_dump_file in
11987 favor of indirectly using them via emit_item.
11988 (make_item_for_dump_gimple_expr): New function, adapted from
11989 optinfo::add_gimple_expr.
11990 (dump_context::dump_gimple_expr): Call it, and use the result,
11991 eliminating the direct usage of dump_file and alt_dump_file in
11992 favor of indirectly using them via emit_item.
11993 (make_item_for_dump_generic_expr): New function, adapted from
11995 (dump_context::dump_generic_expr): Call it, and use the result,
11996 eliminating the direct usage of dump_file and alt_dump_file in
11997 favor of indirectly using them via emit_item.
11998 (make_item_for_dump_printf_va): New function, adapted from
11999 optinfo::add_printf_va.
12000 (make_item_for_dump_printf): New function.
12001 (dump_context::dump_printf_va): Call make_item_for_dump_printf_va,
12002 and use the result, eliminating the direct usage of dump_file and
12003 alt_dump_file in favor of indirectly using them via emit_item.
12004 (make_item_for_dump_dec): New function.
12005 (dump_context::dump_dec): Call it, and use the result,
12006 eliminating the direct usage of dump_file and alt_dump_file in
12007 favor of indirectly using them via emit_item.
12008 (make_item_for_dump_symtab_node): New function, adapted from
12009 optinfo::add_symtab_node.
12010 (dump_context::dump_symtab_node): Call it, and use the result,
12011 eliminating the direct usage of dump_file and alt_dump_file in
12012 favor of indirectly using them via emit_item.
12013 (dump_context::begin_scope): Reimplement, avoiding direct usage
12014 of dump_file and alt_dump_file in favor of indirectly using them
12016 (dump_context::emit_item): New member function.
12017 (temp_dump_context::temp_dump_context): Add param "test_pp_flags".
12018 Set up test pretty-printer on the underlying context. Call
12019 refresh_dumps_are_enabled.
12020 (temp_dump_context::~temp_dump_context): Call
12021 refresh_dumps_are_enabled.
12022 (temp_dump_context::get_dumped_text): New member function.
12023 (selftest::verify_dumped_text): New function.
12024 (ASSERT_DUMPED_TEXT_EQ): New macro.
12025 (selftest::test_capture_of_dump_calls): Run all tests twice, with
12026 and then without optinfo enabled. Add uses of
12027 ASSERT_DUMPED_TEXT_EQ to all tests. Add test of nested scopes.
12028 * dumpfile.h: Update comment for the dump_* API.
12029 * optinfo-emit-json.cc
12030 (selftest::test_building_json_from_dump_calls): Update for new
12031 param for temp_dump_context ctor.
12032 * optinfo.cc (optinfo_item::optinfo_item): Remove "owned" param
12033 and "m_owned" field.
12034 (optinfo_item::~optinfo_item): Likewise.
12035 (optinfo::add_item): New member function.
12036 (optinfo::emit): Update comment.
12037 (optinfo::add_string): Delete.
12038 (optinfo::add_printf): Delete.
12039 (optinfo::add_printf_va): Delete.
12040 (optinfo::add_gimple_stmt): Delete.
12041 (optinfo::add_gimple_expr): Delete.
12042 (optinfo::add_tree): Delete.
12043 (optinfo::add_symtab_node): Delete.
12044 (optinfo::add_dec): Delete.
12045 * optinfo.h (class dump_context): New forward decl.
12046 (optinfo::add_item): New decl.
12047 (optinfo::add_string): Delete.
12048 (optinfo::add_printf): Delete.
12049 (optinfo::add_printf_va): Delete.
12050 (optinfo::add_gimple_stmt): Delete.
12051 (optinfo::add_gimple_expr): Delete.
12052 (optinfo::add_tree): Delete.
12053 (optinfo::add_symtab_node): Delete.
12054 (optinfo::add_dec): Delete.
12055 (optinfo::add_poly_int): Delete.
12056 (optinfo_item::optinfo_item): Remove "owned" param.
12057 (class optinfo_item): Remove field "m_owned".
12059 2018-07-31 Jozef Lawrynowicz <jozef.l@mittosystems.com>
12061 PR middle-end/86705
12062 * gcc/cfgexpand.c (set_parm_rtl): Use the alignment of Pmode when
12063 MAX_SUPPORTED_STACK_ALIGNMENT would otherwise be exceeded by the
12064 requested variable alignment.
12065 (expand_one_ssa_partition): Likewise.
12066 (expand_one_var): Likewise.
12068 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12070 * config/pdp11/pdp11.c (TARGET_HAVE_SPECULATION_SAFE_VALUE): Redefine
12071 to speculation_safe_value_not_needed.
12073 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12075 * targhooks.h (speculation_safe_value_not_needed): New prototype.
12076 * targhooks.c (speculation_safe_value_not_needed): New function.
12077 * target.def (have_speculation_safe_value): Update documentation.
12078 * doc/tm.texi: Regenerated.
12080 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12082 * config/aarch64/iterators.md (ALLI_TI): New iterator.
12083 * config/aarch64/aarch64.md (despeculate_copy<ALLI_TI:mode>): New
12085 (despeculate_copy<ALLI:mode>_insn): New insn.
12086 (despeculate_copyti_insn): New insn.
12087 (despeculate_simple<ALLI:mode>): New insn
12088 (despeculate_simpleti): New insn.
12089 * config/aarch64/aarch64.c (aarch64_speculation_safe_value): New
12091 (TARGET_SPECULATION_SAFE_VALUE): Redefine to
12092 aarch64_speculation_safe_value.
12093 (aarch64_print_operand): Handle const0_rtx in modifier 'H'.
12095 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12097 * config/aarch64/aarch64-speculation.cc: New file.
12098 * config/aarch64/aarch64-passes.def (pass_track_speculation): Add
12099 before pass_reorder_blocks.
12100 * config/aarch64/aarch64-protos.h (make_pass_track_speculation): Add
12102 * config/aarch64/aarch64.c (aarch64_conditional_register_usage): Fix
12103 X14 and X15 when tracking speculation.
12104 * config/aarch64/aarch64.md (register name constants): Add
12105 SPECULATION_TRACKER_REGNUM and SPECULATION_SCRATCH_REGNUM.
12106 (unspec): Add UNSPEC_SPECULATION_TRACKER.
12107 (speculation_barrier): New insn attribute.
12108 (cmp<mode>): Allow SP in comparisons.
12109 (speculation_tracker): New insn.
12110 (speculation_barrier): Add speculation_barrier attribute.
12111 * config/aarch64/t-aarch64: Add make rule for aarch64-speculation.o.
12112 * config.gcc (aarch64*-*-*): Add aarch64-speculation.o to extra_objs.
12113 * doc/invoke.texi (AArch64 Options): Document -mtrack-speculation.
12115 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12117 * config/aarch64/aarch64.md (cb<optab><mode>1): Disable when
12118 aarch64_track_speculation is true.
12119 (tb<optab><mode>1): Likewise.
12120 * config/aarch64/aarch64.c (aarch64_split_compare_regs): Do not
12121 generate CB[N]Z when tracking speculation.
12122 (aarch64_split_compare_and_swap): Likewise.
12123 (aarch64_split_atomic_op): Likewise.
12125 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12127 * config/aarch64/aarch64.opt (mtrack-speculation): New target option.
12129 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12131 * config/aarch64.md (unspecv): Add UNSPECV_SPECULAION_BARRIER.
12132 (speculation_barrier): New insn.
12134 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12136 * config/arm/unspecs.md (unspecv): Add VUNSPEC_SPECULATION_BARRIER.
12137 * config/arm/arm.md (speculation_barrier): New expand.
12138 (speculation_barrier_insn): New pattern.
12140 2018-07-31 Richard Earnshaw <rearnsha@arm.com>
12142 * builtin-types.def (BT_FN_PTR_PTR_VAR): New function type.
12143 (BT_FN_I1_I1_VAR, BT_FN_I2_I2_VAR, BT_FN_I4_I4_VAR): Likewise.
12144 (BT_FN_I8_I8_VAR, BT_FN_I16_I16_VAR): Likewise.
12145 * builtin-attrs.def (ATTR_NOVOPS_NOTHROW_LEAF_LIST): New attribute
12147 * builtins.def (BUILT_IN_SPECULATION_SAFE_VALUE_N): New builtin.
12148 (BUILT_IN_SPECULATION_SAFE_VALUE_PTR): New internal builtin.
12149 (BUILT_IN_SPECULATION_SAFE_VALUE_1): Likewise.
12150 (BUILT_IN_SPECULATION_SAFE_VALUE_2): Likewise.
12151 (BUILT_IN_SPECULATION_SAFE_VALUE_4): Likewise.
12152 (BUILT_IN_SPECULATION_SAFE_VALUE_8): Likewise.
12153 (BUILT_IN_SPECULATION_SAFE_VALUE_16): Likewise.
12154 * builtins.c (expand_speculation_safe_value): New function.
12155 (expand_builtin): Call it.
12156 * doc/cpp.texi: Document predefine __HAVE_SPECULATION_SAFE_VALUE.
12157 * doc/extend.texi: Document __builtin_speculation_safe_value.
12158 * doc/md.texi: Document "speculation_barrier" pattern.
12159 * doc/tm.texi.in: Pull in TARGET_SPECULATION_SAFE_VALUE and
12160 TARGET_HAVE_SPECULATION_SAFE_VALUE.
12161 * doc/tm.texi: Regenerated.
12162 * target.def (have_speculation_safe_value, speculation_safe_value): New
12164 * targhooks.c (default_have_speculation_safe_value): New function.
12165 (default_speculation_safe_value): New function.
12166 * targhooks.h (default_have_speculation_safe_value): Add prototype.
12167 (default_speculation_safe_value): Add prototype.
12169 2018-07-31 David Malcolm <dmalcolm@redhat.com>
12171 * dump-context.h (dump_context::dump_loc): New decl.
12172 * dumpfile.c (dump_context::dump_loc): New member function.
12173 (dump_context::dump_gimple_stmt_loc): Reimplement using dump_loc
12174 and dump_gimple_stmt.
12175 (dump_context::dump_gimple_expr_loc): Likewise, using
12177 (dump_context::dump_generic_expr_loc): Likewise, using
12179 (dump_context::dump_printf_loc_va): Likewise, using
12181 (dump_context::begin_scope): Explicitly using the global function
12182 "dump_loc", rather than the member function.
12184 2018-07-31 Martin Sebor <msebor@redhat.com>
12186 PR tree-optimization/86741
12187 * tree-vrp.c (vrp_prop::check_mem_ref): Avoid incomplete types.
12189 2018-07-31 Andreas Krebbel <krebbel@linux.ibm.com>
12191 * config/s390/s390.c (s390_expand_setmem): Make the unrolling to
12192 depend on whether prefetch instructions will be emitted or not.
12193 Use TARGET_SETMEM_PFD for checking whether prefetch instructions
12194 will be emitted or not.
12195 * config/s390/s390.h (TARGET_SETMEM_PREFETCH_DISTANCE)
12196 (TARGET_SETMEM_PFD): New macros.
12198 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12200 * tree-vectorizer.h (stmt_vec_info): Turn back into a typedef.
12201 (NULL_STMT_VEC_INFO): Delete.
12202 (stmt_vec_info::operator*): Likewise.
12203 (stmt_vec_info::operator gimple *): Likewise.
12204 * tree-vect-loop.c (vectorizable_reduction): Use NULL instead
12205 of NULL_STMT_VEC_INFO.
12206 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
12207 (vect_reassociating_reduction_p): Likewise.
12208 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
12209 (vectorizable_store): Likewise.
12210 * tree-vectorizer.c (vec_info::set_vinfo_for_stmt): Likewise.
12211 (vec_info::free_stmt_vec_infos): Likewise.
12213 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12215 * tree-vectorizer.h (vect_stmt_in_region_p): Delete.
12216 * tree-vectorizer.c (vect_stmt_in_region_p): Likewise.
12218 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12220 * tree-vectorizer.h (vec_info::new_vinfo_for_stmt)
12221 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
12222 (vec_info::free_stmt_vec_info): New private member functions.
12223 (set_stmt_vec_info_vec, free_stmt_vec_infos, vinfo_for_stmt)
12224 (set_vinfo_for_stmt, new_stmt_vec_info, free_stmt_vec_info): Delete.
12225 * tree-parloops.c (gather_scalar_reductions): Remove calls to
12226 set_stmt_vec_info_vec and free_stmt_vec_infos.
12227 * tree-vect-loop.c (_loop_vec_info): Remove call to
12228 set_stmt_vec_info_vec.
12229 * tree-vect-stmts.c (new_stmt_vec_info, set_stmt_vec_info_vec)
12230 (free_stmt_vec_infos, free_stmt_vec_info): Delete in favor of...
12231 * tree-vectorizer.c (vec_info::new_stmt_vec_info)
12232 (vec_info::set_vinfo_for_stmt, vec_info::free_stmt_vec_infos)
12233 (vec_info::free_stmt_vec_info): ...these new functions. Remove
12234 assignments in {vec_info::,}new_stmt_vec_info that are redundant
12235 with the clearing in the xcalloc.
12236 (stmt_vec_info_vec): Delete.
12237 (vec_info::vec_info): Don't call set_stmt_vec_info_vec.
12238 (vectorize_loops): Likewise.
12239 (vec_info::~vec_info): Remove argument from call to
12240 free_stmt_vec_infos.
12241 (vec_info::add_stmt): Remove vinfo argument from call to
12244 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12246 * tree-vectorizer.h (free_stmt_vec_info): Take a stmt_vec_info
12247 rather than a gimple stmt.
12248 * tree-vect-stmts.c (free_stmt_vec_info): Likewise. Don't free
12249 information for pattern statements when passed the original
12250 statement; instead wait to be passed the pattern statement itself.
12251 Don't call set_vinfo_for_stmt here.
12252 (free_stmt_vec_infos): Update call to free_stmt_vec_info.
12253 * tree-vect-loop.c (_loop_vec_info::~loop_vec_info): Don't free
12254 stmt_vec_infos here.
12255 * tree-vect-slp.c (_bb_vec_info::~bb_vec_info): Likewise.
12256 * tree-vectorizer.c (vec_info::remove_stmt): Nullify the statement's
12257 stmt_vec_infos entry.
12259 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12261 * tree-vectorizer.h (vec_info::replace_stmt): Declare.
12262 * tree-vectorizer.c (vec_info::replace_stmt): New function.
12263 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Use it.
12264 * tree-vect-stmts.c (vectorizable_call): Likewise.
12265 (vectorizable_simd_clone_call): Likewise.
12267 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12269 * tree-vectorizer.h (vec_info::remove_stmt): Declare.
12270 * tree-vectorizer.c (vec_info::remove_stmt): New function.
12271 * tree-vect-loop-manip.c (vect_set_loop_condition): Use it.
12272 * tree-vect-loop.c (vect_transform_loop): Likewise.
12273 * tree-vect-slp.c (vect_schedule_slp): Likewise.
12274 * tree-vect-stmts.c (vect_remove_stores): Likewise.
12276 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12278 * tree-vectorizer.h (vec_info::lookup_dr): New member function.
12279 (vect_dr_stmt): Delete.
12280 * tree-vectorizer.c (vec_info::lookup_dr): New function.
12281 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Use it instead
12283 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
12284 (vect_analyze_data_ref_dependence, vect_record_base_alignments)
12285 (vect_verify_datarefs_alignment, vect_peeling_supportable)
12286 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
12287 (vect_analyze_data_refs): Likewise.
12288 (vect_slp_analyze_data_ref_dependence): Likewise. Take a vec_info
12290 (vect_find_same_alignment_drs): Likewise.
12291 (vect_slp_analyze_node_dependences): Update calls accordingly.
12292 (vect_analyze_data_refs_alignment): Likewise. Use vec_info::lookup_dr
12293 instead of DR_VECT_AUX.
12294 (vect_get_peeling_costs_all_drs): Take a loop_vec_info instead
12295 of a vector data references. Use vec_info::lookup_dr instead of
12297 (vect_peeling_hash_get_lowest_cost): Update calls accordingly.
12298 (vect_enhance_data_refs_alignment): Likewise. Use vec_info::lookup_dr
12299 instead of DR_VECT_AUX.
12301 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12303 * tree-vectorizer.h (_loop_vec_info::unaligned_dr): Change to
12305 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
12307 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
12308 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
12309 (vect_gen_prolog_loop_niters): Likewise.
12311 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12313 * tree-vectorizer.h (set_dr_misalignment, dr_misalignment)
12314 (DR_TARGET_ALIGNMENT, aligned_access_p, known_alignment_for_access_p)
12315 (vect_known_alignment_in_bytes, vect_dr_behavior)
12316 (vect_get_scalar_dr_size): Take references as dr_vec_infos
12317 instead of data_references. Update calls to other routines for
12318 which the same change has been made.
12319 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Take
12320 dr_vec_infos instead of stmt_vec_infos.
12321 (vect_analyze_data_ref_dependence): Update call accordingly.
12322 (vect_slp_analyze_data_ref_dependence)
12323 (vect_record_base_alignments): Use DR_VECT_AUX.
12324 (vect_calculate_target_alignment, vect_compute_data_ref_alignment)
12325 (vect_update_misalignment_for_peel, verify_data_ref_alignment)
12326 (vector_alignment_reachable_p, vect_get_data_access_cost)
12327 (vect_peeling_supportable, vect_analyze_group_access_1)
12328 (vect_analyze_group_access, vect_analyze_data_ref_access)
12329 (vect_vfa_segment_size, vect_vfa_access_size, vect_vfa_align)
12330 (vect_compile_time_alias, vect_small_gap_p)
12331 (vectorizable_with_step_bound_p, vect_duplicate_ssa_name_ptr_info):
12332 (vect_supportable_dr_alignment): Take references as dr_vec_infos
12333 instead of data_references. Update calls to other routines for
12334 which the same change has been made.
12335 (vect_verify_datarefs_alignment, vect_get_peeling_costs_all_drs)
12336 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
12337 (vect_slp_analyze_and_verify_node_alignment)
12338 (vect_analyze_data_ref_accesses, vect_prune_runtime_alias_test_list)
12339 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
12340 (vect_setup_realignment): Use dr_vec_infos. Update calls after
12342 (_vect_peel_info::dr): Replace with...
12343 (_vect_peel_info::dr_info): ...this new field.
12344 (vect_peeling_hash_get_most_frequent)
12345 (vect_peeling_hash_choose_best_peeling): Update accordingly.
12346 (vect_peeling_hash_get_lowest_cost):
12347 (vect_enhance_data_refs_alignment): Likewise. Update calls to other
12348 routines for which the same change has been made.
12349 (vect_peeling_hash_insert): Likewise. Take a dr_vec_info instead of a
12351 * tree-vect-loop-manip.c (get_misalign_in_elems)
12352 (vect_gen_prolog_loop_niters): Use dr_vec_infos. Update calls after
12354 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
12355 * tree-vect-stmts.c (vect_get_store_cost, vect_get_load_cost)
12356 (vect_truncate_gather_scatter_offset, compare_step_with_zero)
12357 (get_group_load_store_type, get_negative_load_store_type)
12358 (vect_get_data_ptr_increment, vectorizable_store)
12359 (vectorizable_load): Likewise.
12360 (ensure_base_align): Take a dr_vec_info instead of a data_reference.
12361 Update calls to other routines for which the same change has been made.
12363 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12365 * tree-vectorizer.h (vec_info::move_dr): New member function.
12366 (dataref_aux): Rename to...
12367 (dr_vec_info): ...this and add "dr" and "stmt" fields.
12368 (_stmt_vec_info::dr_aux): Update accordingly.
12369 (_stmt_vec_info::data_ref_info): Delete.
12370 (STMT_VINFO_GROUPED_ACCESS, DR_GROUP_FIRST_ELEMENT)
12371 (DR_GROUP_NEXT_ELEMENT, DR_GROUP_SIZE, DR_GROUP_STORE_COUNT)
12372 (DR_GROUP_GAP, DR_GROUP_SAME_DR_STMT, REDUC_GROUP_FIRST_ELEMENT):
12373 (REDUC_GROUP_NEXT_ELEMENT, REDUC_GROUP_SIZE): Use dr_aux.dr instead
12375 (STMT_VINFO_DATA_REF): Likewise. Turn into an lvalue.
12376 (STMT_VINFO_DR_INFO): New macro.
12377 (DR_VECT_AUX): Use STMT_VINFO_DR_INKFO and vect_dr_stmt.
12378 (set_dr_misalignment): Update after rename of dataref_aux.
12379 (vect_dr_stmt): Move earlier in file. Return dr_aux.stmt.
12380 * tree-vect-stmts.c (new_stmt_vec_info): Remove redundant
12381 initialization of STMT_VINFO_DATA_REF.
12382 * tree-vectorizer.c (vec_info::move_dr): New function.
12383 * tree-vect-patterns.c (vect_recog_bool_pattern)
12384 (vect_recog_mask_conversion_pattern)
12385 (vect_recog_gather_scatter_pattern): Use it.
12386 * tree-vect-data-refs.c (vect_analyze_data_refs): Initialize
12387 the "dr" and "stmt" fields of dr_vec_info instead of
12388 STMT_VINFO_DATA_REF.
12390 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12392 * tree-vectorizer.h (_stmt_vec_info::pattern_stmt_p): New field.
12393 (is_pattern_stmt_p): Use it.
12394 * tree-vect-patterns.c (vect_init_pattern_stmt): Set pattern_stmt_p
12395 on pattern statements.
12397 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12399 * tree-vect-patterns.c (vect_mark_pattern_stmts): Take the
12400 original stmt as a stmt_vec_info rather than a gimple stmt.
12401 (vect_pattern_recog_1): Take the statement directly as a
12402 stmt_vec_info, rather than via a gimple_stmt_iterator.
12403 Update call to vect_mark_pattern_stmts.
12404 (vect_pattern_recog): Update calls accordingly.
12406 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12408 * tree-vectorizer.h (vect_get_vec_defs_for_stmt_copy)
12409 (vect_get_vec_def_for_stmt_copy): Take a vec_info rather than
12410 a vect_def_type for the first argument.
12411 * tree-vect-stmts.c (vect_get_vec_defs_for_stmt_copy): Likewise.
12412 (vect_get_vec_def_for_stmt_copy): Likewise. Return the original
12413 operand if it isn't defined by a vectorized statement.
12414 (vect_build_gather_load_calls): Remove the mask_dt argument and
12415 update calls to vect_get_vec_def_for_stmt_copy.
12416 (vectorizable_bswap): Likewise the dt argument.
12417 (vectorizable_call): Update calls to vectorizable_bswap and
12418 vect_get_vec_def_for_stmt_copy.
12419 (vectorizable_simd_clone_call, vectorizable_assignment)
12420 (vectorizable_shift, vectorizable_operation, vectorizable_condition)
12421 (vectorizable_comparison): Update calls to
12422 vect_get_vec_def_for_stmt_copy.
12423 (vectorizable_store): Likewise. Remove now-unnecessary calls to
12424 vect_is_simple_use.
12425 (vect_get_loop_based_defs): Remove dt argument and update call
12426 to vect_get_vec_def_for_stmt_copy.
12427 (vectorizable_conversion): Update calls to vect_get_loop_based_defs
12428 and vect_get_vec_def_for_stmt_copy.
12429 (vectorizable_load): Update calls to vect_build_gather_load_calls
12430 and vect_get_vec_def_for_stmt_copy.
12431 * tree-vect-loop.c (vect_create_epilog_for_reduction)
12432 (vectorizable_reduction, vectorizable_live_operation): Update calls
12433 to vect_get_vec_def_for_stmt_copy.
12435 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12437 * tree-vect-data-refs.c (vect_record_base_alignment): Replace vec_info
12438 and gimple stmt arguments with a stmt_vec_info.
12439 (vect_record_base_alignments): Update calls accordingly.
12440 * tree-vect-slp.c (vect_record_max_nunits): Replace vec_info
12441 and gimple stmt arguments with a stmt_vec_info.
12442 (vect_build_slp_tree_1): Remove vinfo argument and update call
12443 to vect_record_max_nunits.
12444 (vect_build_slp_tree_2): Update calls to vect_build_slp_tree_1
12445 and vect_record_max_nunits.
12447 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12449 * tree-vectorizer.h (nested_in_vect_loop_p): Move further down
12450 file and take a stmt_vec_info instead of a gimple stmt.
12451 (supportable_widening_operation, vect_finish_replace_stmt)
12452 (vect_finish_stmt_generation, vect_get_store_rhs)
12453 (vect_get_vec_def_for_operand_1, vect_get_vec_def_for_operand)
12454 (vect_get_vec_defs, vect_init_vector, vect_transform_stmt)
12455 (vect_remove_stores, vect_analyze_stmt, vectorizable_condition)
12456 (vect_get_smallest_scalar_type, vect_check_gather_scatter)
12457 (vect_create_data_ref_ptr, bump_vector_ptr)
12458 (vect_permute_store_chain, vect_setup_realignment)
12459 (vect_transform_grouped_load, vect_record_grouped_load_vectors)
12460 (vect_create_addr_base_for_vector_ref, vectorizable_live_operation)
12461 (vectorizable_reduction, vectorizable_induction)
12462 (get_initial_def_for_reduction, is_simple_and_all_uses_invariant)
12463 (vect_get_place_in_interleaving_chain): Take stmt_vec_infos rather
12464 than gimple stmts as arguments.
12465 * tree-vect-data-refs.c (vect_get_smallest_scalar_type)
12466 (vect_preserves_scalar_order_p, vect_slp_analyze_node_dependences)
12467 (can_group_stmts_p, vect_check_gather_scatter)
12468 (vect_create_addr_base_for_vector_ref, vect_create_data_ref_ptr)
12469 (bump_vector_ptr, vect_permute_store_chain, vect_setup_realignment)
12470 (vect_permute_load_chain, vect_shift_permute_load_chain)
12471 (vect_transform_grouped_load)
12472 (vect_record_grouped_load_vectors): Likewise.
12473 * tree-vect-loop.c (vect_fixup_reduc_chain)
12474 (get_initial_def_for_reduction, vect_create_epilog_for_reduction)
12475 (vectorize_fold_left_reduction, is_nonwrapping_integer_induction)
12476 (vectorizable_reduction, vectorizable_induction)
12477 (vectorizable_live_operation, vect_loop_kill_debug_uses): Likewise.
12478 * tree-vect-patterns.c (type_conversion_p, adjust_bool_stmts)
12479 (vect_get_load_store_mask): Likewise.
12480 * tree-vect-slp.c (vect_get_place_in_interleaving_chain)
12481 (vect_analyze_slp_instance, vect_mask_constant_operand_p): Likewise.
12482 * tree-vect-stmts.c (vect_mark_relevant)
12483 (is_simple_and_all_uses_invariant)
12484 (exist_non_indexing_operands_for_use_p, process_use)
12485 (vect_init_vector_1, vect_init_vector, vect_get_vec_def_for_operand_1)
12486 (vect_get_vec_def_for_operand, vect_get_vec_defs)
12487 (vect_finish_stmt_generation_1, vect_finish_replace_stmt)
12488 (vect_finish_stmt_generation, vect_truncate_gather_scatter_offset)
12489 (compare_step_with_zero, vect_get_store_rhs, get_group_load_store_type)
12490 (get_negative_load_store_type, get_load_store_type)
12491 (vect_check_load_store_mask, vect_check_store_rhs)
12492 (vect_build_gather_load_calls, vect_get_strided_load_store_ops)
12493 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
12494 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
12495 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
12496 (get_group_alias_ptr_type, vectorizable_store, hoist_defs_of_uses)
12497 (vectorizable_load, vectorizable_condition, vectorizable_comparison)
12498 (vect_analyze_stmt, vect_transform_stmt, vect_remove_stores)
12499 (supportable_widening_operation): Likewise.
12501 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12503 * tree-vect-data-refs.c (vect_describe_gather_scatter_call): Take
12504 a stmt_vec_info instead of a gcall.
12505 (vect_check_gather_scatter): Update call accordingly.
12506 * tree-vect-loop-manip.c (iv_phi_p): Take a stmt_vec_info instead
12508 (vect_can_advance_ivs_p, vect_update_ivs_after_vectorizer)
12509 (slpeel_update_phi_nodes_for_loops):): Update calls accordingly.
12510 * tree-vect-loop.c (vect_transform_loop_stmt): Take a stmt_vec_info
12511 instead of a gimple stmt.
12512 (vect_transform_loop): Update calls accordingly.
12513 * tree-vect-slp.c (vect_split_slp_store_group): Take and return
12514 stmt_vec_infos instead of gimple stmts.
12515 (vect_analyze_slp_instance): Update use accordingly.
12516 * tree-vect-stmts.c (read_vector_array, write_vector_array)
12517 (vect_clobber_variable, vect_stmt_relevant_p, permute_vec_elements)
12518 (vect_use_strided_gather_scatters_p, vect_build_all_ones_mask)
12519 (vect_build_zero_merge_argument, vect_get_gather_scatter_ops)
12520 (vect_gen_widened_results_half, vect_get_loop_based_defs)
12521 (vect_create_vectorized_promotion_stmts, can_vectorize_live_stmts):
12522 Take a stmt_vec_info instead of a gimple stmt and pass stmt_vec_infos
12523 down to subroutines.
12525 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12527 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Change the type
12528 of the worklist from a vector of gimple stmts to a vector of
12530 * tree-vect-stmts.c (vect_mark_relevant, process_use)
12531 (vect_mark_stmts_to_be_vectorized): Likewise
12533 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12535 * tree-vect-loop.c (vect_analyze_loop_operations): Look up the
12536 statement before passing it to vect_analyze_stmt.
12537 (vect_create_epilog_for_reduction): Use a stmt_vec_info to walk
12538 the chain of phi vector definitions. Track the exit phi via its
12540 (vectorizable_reduction): Set cond_stmt_vinfo directly from the
12541 STMT_VINFO_REDUC_DEF.
12542 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Use
12543 stmt_vec_infos to handle the statement chains.
12544 (vect_get_slp_defs): Record the first statement in the node
12545 using a stmt_vec_info.
12546 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Look up
12547 statements here and pass their stmt_vec_info down to subroutines.
12548 (vect_init_vector_1): Hoist call to vinfo_for_stmt and pass it
12549 down to vect_finish_stmt_generation.
12550 (vect_init_vector, vect_get_vec_defs, vect_finish_replace_stmt)
12551 (vect_finish_stmt_generation): Call vinfo_for_stmt and pass
12552 stmt_vec_infos to subroutines.
12553 (vect_remove_stores): Use stmt_vec_infos to handle the statement
12556 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12558 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences):
12559 (vect_check_gather_scatter, vect_create_data_ref_ptr, bump_vector_ptr)
12560 (vect_permute_store_chain, vect_setup_realignment)
12561 (vect_permute_load_chain, vect_shift_permute_load_chain)
12562 (vect_transform_grouped_load): Use stmt_vec_info rather than gimple
12563 stmts internally, and when passing values to other vectorizer routines.
12564 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
12565 * tree-vect-loop.c (vect_analyze_scalar_cycles_1)
12566 (vect_analyze_loop_operations, get_initial_def_for_reduction)
12567 (vect_create_epilog_for_reduction, vectorize_fold_left_reduction)
12568 (vectorizable_reduction, vectorizable_induction)
12569 (vectorizable_live_operation, vect_transform_loop_stmt)
12570 (vect_transform_loop): Likewise.
12571 * tree-vect-patterns.c (vect_reassociating_reduction_p)
12572 (vect_recog_widen_op_pattern, vect_recog_mixed_size_cond_pattern)
12573 (vect_recog_bool_pattern, vect_recog_gather_scatter_pattern): Likewise.
12574 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
12575 (vect_slp_analyze_node_operations_1): Likewise.
12576 * tree-vect-stmts.c (vect_mark_relevant, process_use)
12577 (exist_non_indexing_operands_for_use_p, vect_init_vector_1)
12578 (vect_mark_stmts_to_be_vectorized, vect_get_vec_def_for_operand)
12579 (vect_finish_stmt_generation_1, get_group_load_store_type)
12580 (get_load_store_type, vect_build_gather_load_calls)
12581 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
12582 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
12583 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
12584 (vectorizable_store, vectorizable_load, vectorizable_condition)
12585 (vectorizable_comparison, vect_analyze_stmt, vect_transform_stmt)
12586 (supportable_widening_operation): Likewise.
12587 (vect_get_vector_types_for_stmt): Likewise.
12588 * tree-vectorizer.h (vect_dr_behavior): Likewise.
12590 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12592 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
12593 (vect_slp_analyze_node_dependences, vect_analyze_data_ref_accesses)
12594 (vect_permute_store_chain, vect_permute_load_chain)
12595 (vect_shift_permute_load_chain, vect_transform_grouped_load): Avoid
12596 repeated stmt_vec_info lookups.
12597 * tree-vect-loop-manip.c (vect_can_advance_ivs_p): Likewise.
12598 (vect_update_ivs_after_vectorizer): Likewise.
12599 * tree-vect-loop.c (vect_is_simple_reduction): Likewise.
12600 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
12601 * tree-vect-patterns.c (adjust_bool_stmts): Likewise.
12602 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
12603 (vect_bb_slp_scalar_cost): Likewise.
12604 * tree-vect-stmts.c (get_group_alias_ptr_type): Likewise.
12606 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12608 * tree-vect-data-refs.c (vect_check_gather_scatter): Pass the
12609 gcall rather than the generic gimple stmt to gimple_call_internal_fn.
12610 (vect_get_smallest_scalar_type, can_group_stmts_p): Use dyn_cast
12611 to get gassigns and gcalls, rather than operating on generc gimple
12613 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p)
12614 (vect_mark_stmts_to_be_vectorized, vectorizable_store)
12615 (vectorizable_load, vect_analyze_stmt): Likewise.
12616 * tree-vect-loop.c (vectorizable_reduction): Likewise gphi.
12618 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12620 * tree-vectorizer.h (get_earlier_stmt, get_later_stmt): Take and
12621 return stmt_vec_infos rather than gimple stmts. Do not accept
12623 (vect_find_last_scalar_stmt_in_slp): Return a stmt_vec_info instead
12625 * tree-vect-slp.c (vect_find_last_scalar_stmt_in_slp): Likewise.
12626 Update use of get_later_stmt.
12627 (vect_get_constant_vectors): Update call accordingly.
12628 (vect_schedule_slp_instance): Likewise
12629 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Likewise.
12630 (vect_slp_analyze_instance_dependence): Likewise.
12631 (vect_preserves_scalar_order_p): Update use of get_earlier_stmt.
12633 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12635 * tree-vectorizer.h (stmt_info_for_cost::stmt): Replace with...
12636 (stmt_info_for_cost::stmt_info): ...this new field.
12637 (add_stmt_costs): Update accordingly.
12638 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
12639 (vect_get_known_peeling_cost): Likewise.
12640 (vect_estimate_min_profitable_iters): Likewise.
12641 * tree-vect-stmts.c (record_stmt_cost): Likewise.
12643 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12645 * tree-vectorizer.h (_loop_vec_info::may_misalign_stmts): Change
12646 from an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
12647 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Update
12649 * tree-vect-loop-manip.c (vect_create_cond_for_align_checks): Likewise.
12651 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12653 * tree-vectorizer.h (_stmt_vec_info::same_dr_stmt): Change from
12654 a gimple stmt to a stmt_vec_info.
12655 * tree-vect-stmts.c (vectorizable_load): Update accordingly.
12657 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12659 * tree-vectorizer.h (vec_info::grouped_stores): Change from
12660 an auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
12661 (_loop_vec_info::reduction_chains): Likewise.
12662 * tree-vect-loop.c (vect_fixup_scalar_cycles_with_patterns): Update
12664 * tree-vect-slp.c (vect_analyze_slp): Likewise.
12666 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12668 * tree-vectorizer.h (_stmt_vec_info::first_element): Change from
12669 a gimple stmt to a stmt_vec_info.
12670 (_stmt_vec_info::next_element): Likewise.
12671 * tree-vect-data-refs.c (vect_update_misalignment_for_peel)
12672 (vect_slp_analyze_and_verify_node_alignment)
12673 (vect_analyze_group_access_1, vect_analyze_group_access)
12674 (vect_small_gap_p, vect_prune_runtime_alias_test_list)
12675 (vect_create_data_ref_ptr, vect_record_grouped_load_vectors)
12676 (vect_supportable_dr_alignment): Update accordingly.
12677 * tree-vect-loop.c (vect_fixup_reduc_chain): Likewise.
12678 (vect_fixup_scalar_cycles_with_patterns, vect_is_slp_reduction)
12679 (vect_is_simple_reduction, vectorizable_reduction): Likewise.
12680 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
12681 * tree-vect-slp.c (vect_build_slp_tree_1)
12682 (vect_attempt_slp_rearrange_stmts, vect_supported_load_permutation_p)
12683 (vect_split_slp_store_group, vect_analyze_slp_instance)
12684 (vect_analyze_slp, vect_transform_slp_perm_load): Likewise.
12685 * tree-vect-stmts.c (vect_model_store_cost, vect_model_load_cost)
12686 (get_group_load_store_type, get_load_store_type)
12687 (get_group_alias_ptr_type, vectorizable_store, vectorizable_load)
12688 (vect_transform_stmt, vect_remove_stores): Likewise.
12690 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12692 * tree-vectorizer.h (vect_dr_stmt): Return a stmt_vec_info rather
12693 than a gimple stmt.
12694 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
12695 (vect_slp_analyze_data_ref_dependence, vect_record_base_alignments)
12696 (vect_calculate_target_alignmentm, vect_compute_data_ref_alignment)
12697 (vect_update_misalignment_for_peel, vect_verify_datarefs_alignment)
12698 (vector_alignment_reachable_p, vect_get_data_access_cost)
12699 (vect_get_peeling_costs_all_drs, vect_peeling_hash_get_lowest_cost)
12700 (vect_peeling_supportable, vect_enhance_data_refs_alignment)
12701 (vect_find_same_alignment_drs, vect_analyze_data_refs_alignment)
12702 (vect_analyze_group_access_1, vect_analyze_group_access)
12703 (vect_analyze_data_ref_access, vect_analyze_data_ref_accesses)
12704 (vect_vfa_access_size, vect_small_gap_p, vect_analyze_data_refs)
12705 (vect_supportable_dr_alignment): Remove vinfo_for_stmt from the
12706 result of vect_dr_stmt and use the stmt_vec_info instead of
12707 the associated gimple stmt.
12708 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
12709 (vect_gen_prolog_loop_niters): Likewise.
12710 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
12712 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12714 * tree-vectorizer.h (_slp_tree::stmts): Change from a vec<gimple *>
12715 to a vec<stmt_vec_info>.
12716 * tree-vect-slp.c (vect_free_slp_tree): Update accordingly.
12717 (vect_create_new_slp_node): Take a vec<gimple *> instead of a
12718 vec<stmt_vec_info>.
12719 (_slp_oprnd_info::def_stmts): Change from a vec<gimple *>
12720 to a vec<stmt_vec_info>.
12721 (bst_traits::value_type, bst_traits::value_type): Likewise.
12722 (bst_traits::hash): Update accordingly.
12723 (vect_get_and_check_slp_defs): Change the stmts parameter from
12724 a vec<gimple *> to a vec<stmt_vec_info>.
12725 (vect_two_operations_perm_ok_p, vect_build_slp_tree_1): Likewise.
12726 (vect_build_slp_tree): Likewise.
12727 (vect_build_slp_tree_2): Likewise. Update uses of
12728 SLP_TREE_SCALAR_STMTS.
12729 (vect_print_slp_tree): Update uses of SLP_TREE_SCALAR_STMTS.
12730 (vect_mark_slp_stmts, vect_mark_slp_stmts_relevant)
12731 (vect_slp_rearrange_stmts, vect_attempt_slp_rearrange_stmts)
12732 (vect_supported_load_permutation_p, vect_find_last_scalar_stmt_in_slp)
12733 (vect_detect_hybrid_slp_stmts, vect_slp_analyze_node_operations_1)
12734 (vect_slp_analyze_node_operations, vect_slp_analyze_operations)
12735 (vect_bb_slp_scalar_cost, vect_slp_analyze_bb_1)
12736 (vect_get_constant_vectors, vect_get_slp_defs)
12737 (vect_transform_slp_perm_load, vect_schedule_slp_instance)
12738 (vect_remove_slp_scalar_calls, vect_schedule_slp): Likewise.
12739 (vect_analyze_slp_instance): Build up a vec of stmt_vec_infos
12740 instead of gimple stmts.
12741 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Change
12742 the stores parameter for a vec<gimple *> to a vec<stmt_vec_info>.
12743 (vect_slp_analyze_instance_dependence): Update uses of
12744 SLP_TREE_SCALAR_STMTS.
12745 (vect_slp_analyze_and_verify_node_alignment): Likewise.
12746 (vect_slp_analyze_and_verify_instance_alignment): Likewise.
12747 * tree-vect-loop.c (neutral_op_for_slp_reduction): Likewise.
12748 (get_initial_defs_for_reduction): Likewise.
12749 (vect_create_epilog_for_reduction): Likewise.
12750 (vectorize_fold_left_reduction): Likewise.
12751 * tree-vect-stmts.c (vect_prologue_cost_for_slp_op): Likewise.
12752 (vect_model_simple_cost, vectorizable_shift, vectorizable_load)
12753 (can_vectorize_live_stmts): Likewise.
12755 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12757 * tree-vectorizer.h (_loop_vec_info::reductions): Change from an
12758 auto_vec<gimple *> to an auto_vec<stmt_vec_info>.
12759 (vect_force_simple_reduction): Take and return stmt_vec_infos rather
12761 * tree-parloops.c (valid_reduction_p): Take a stmt_vec_info instead
12763 (gather_scalar_reductions): Update after above interface changes.
12764 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Likewise.
12765 (vect_is_simple_reduction): Take and return stmt_vec_infos rather
12767 (vect_force_simple_reduction): Likewise.
12768 * tree-vect-patterns.c (vect_pattern_recog_1): Update use of
12769 LOOP_VINFO_REDUCTIONS.
12770 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
12772 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12774 * tree-vectorizer.h (_stmt_vec_info::reduc_def): Change from
12775 a gimple stmt to a stmt_vec_info.
12776 * tree-vect-loop.c (vect_active_double_reduction_p)
12777 (vect_force_simple_reduction, vectorizable_reduction): Update
12780 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12782 * tree-vectorizer.h (_slp_tree::vec_stmts): Change from a
12783 vec<gimple *> to a vec<stmt_vec_info>.
12784 * tree-vect-loop.c (vect_create_epilog_for_reduction): Change
12785 the reduction_phis argument from a vec<gimple *> to a
12786 vec<stmt_vec_info>.
12787 (vectorizable_reduction): Likewise the phis local variable that
12788 is passed to vect_create_epilog_for_reduction. Update for new type
12789 of SLP_TREE_VEC_STMTS.
12790 (vectorizable_induction): Update for new type of SLP_TREE_VEC_STMTS.
12791 (vectorizable_live_operation): Likewise.
12792 * tree-vect-slp.c (vect_get_slp_vect_defs): Likewise.
12793 (vect_transform_slp_perm_load, vect_schedule_slp_instance): Likewise.
12795 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12797 * tree-vectorizer.h (_stmt_vec_info::vectorized_stmt): Change from
12798 a gimple stmt to a stmt_vec_info.
12799 (vectorizable_condition, vectorizable_live_operation)
12800 (vectorizable_reduction, vectorizable_induction): Pass back the
12801 vectorized statement as a stmt_vec_info.
12802 * tree-vect-data-refs.c (vect_record_grouped_load_vectors): Update
12803 use of STMT_VINFO_VEC_STMT.
12804 * tree-vect-loop.c (vect_create_epilog_for_reduction): Likewise,
12805 accumulating the inner phis that feed the STMT_VINFO_VEC_STMT
12806 as stmt_vec_infos rather than gimple stmts.
12807 (vectorize_fold_left_reduction): Change vec_stmt from a gimple stmt
12808 to a stmt_vec_info.
12809 (vectorizable_live_operation): Likewise.
12810 (vectorizable_reduction, vectorizable_induction): Likewise,
12811 updating use of STMT_VINFO_VEC_STMT.
12812 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Update use
12813 of STMT_VINFO_VEC_STMT.
12814 (vect_build_gather_load_calls, vectorizable_bswap, vectorizable_call)
12815 (vectorizable_simd_clone_call, vectorizable_conversion)
12816 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
12817 (vectorizable_store, vectorizable_load, vectorizable_condition)
12818 (vectorizable_comparison, can_vectorize_live_stmts): Change vec_stmt
12819 from a gimple stmt to a stmt_vec_info.
12820 (vect_transform_stmt): Update use of STMT_VINFO_VEC_STMT. Pass a
12821 pointer to a stmt_vec_info to the vectorizable_* routines.
12823 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12825 * tree-vectorizer.h (_stmt_vec_info::related_stmt): Change from
12826 a gimple stmt to a stmt_vec_info.
12827 (is_pattern_stmt_p): Update accordingly.
12828 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Likewise.
12829 (vect_record_grouped_load_vectors): Likewise.
12830 * tree-vect-loop.c (vect_determine_vf_for_stmt): Likewise.
12831 (vect_fixup_reduc_chain, vect_update_vf_for_slp): Likewise.
12832 (vect_model_reduction_cost): Likewise.
12833 (vect_create_epilog_for_reduction): Likewise.
12834 (vectorizable_reduction, vectorizable_induction): Likewise.
12835 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
12836 Return the stmt_vec_info for the pattern statement.
12837 (vect_set_pattern_stmt): Update use of STMT_VINFO_RELATED_STMT.
12838 (vect_split_statement, vect_mark_pattern_stmts): Likewise.
12839 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
12840 (vect_detect_hybrid_slp, vect_get_slp_defs): Likewise.
12841 * tree-vect-stmts.c (vect_mark_relevant): Likewise.
12842 (vect_get_vec_def_for_operand_1, vectorizable_call): Likewise.
12843 (vectorizable_simd_clone_call, vect_analyze_stmt, new_stmt_vec_info)
12844 (free_stmt_vec_info, vect_is_simple_use): Likewise.
12846 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12848 * tree-vectorizer.h (vect_finish_replace_stmt): Return a stmt_vec_info
12849 (vect_finish_stmt_generation): Likewise.
12850 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
12851 (vect_finish_replace_stmt, vect_finish_stmt_generation): Likewise.
12852 (vect_build_gather_load_calls): Use the return value of the above
12853 functions instead of a separate call to vinfo_for_stmt. Use narrow
12854 scopes for the input gimple stmt and wider scopes for the associated
12855 stmt_vec_info. Use vec_info::lookup_def when setting these
12856 stmt_vec_infos from an SSA_NAME definition.
12857 (vectorizable_bswap, vectorizable_call, vectorizable_simd_clone_call)
12858 (vect_create_vectorized_demotion_stmts, vectorizable_conversion)
12859 (vectorizable_assignment, vectorizable_shift, vectorizable_operation)
12860 (vectorizable_store, vectorizable_load, vectorizable_condition)
12861 (vectorizable_comparison): Likewise.
12862 * tree-vect-loop.c (vectorize_fold_left_reduction): Likewise.
12863 (vectorizable_reduction): Likewise.
12865 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12867 * tree-vectorizer.h (vect_is_simple_use): Add an optional
12868 stmt_vec_info * parameter before the optional gimple **.
12869 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
12870 (process_use, vect_get_vec_def_for_operand_1): Update callers.
12871 (vect_get_vec_def_for_operand, vectorizable_shift): Likewise.
12872 * tree-vect-loop.c (vectorizable_reduction): Likewise.
12873 (vectorizable_live_operation): Likewise.
12874 * tree-vect-patterns.c (type_conversion_p): Likewise.
12875 (vect_look_through_possible_promotion): Likewise.
12876 (vect_recog_rotate_pattern): Likewise.
12877 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
12879 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12881 * tree-vectorizer.h (stmt_vec_info): Temporarily change from
12882 a typedef to a wrapper class.
12883 (NULL_STMT_VEC_INFO): New macro.
12884 (vec_info::stmt_infos): Change to vec<stmt_vec_info>.
12885 (stmt_vec_info::operator*): New function.
12886 (stmt_vec_info::operator gimple *): Likewise.
12887 (set_vinfo_for_stmt): Use NULL_STMT_VEC_INFO.
12888 (add_stmt_costs): Likewise.
12889 * tree-vect-loop-manip.c (iv_phi_p): Likewise.
12890 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost)
12891 (vect_get_known_peeling_cost): Likewise.
12892 (vect_estimate_min_profitable_iters): Likewise.
12893 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
12894 * tree-vect-slp.c (vect_remove_slp_scalar_calls): Likewise.
12895 * tree-vect-stmts.c (vect_build_gather_load_calls): Likewise.
12896 (vectorizable_store, free_stmt_vec_infos): Likewise.
12897 (new_stmt_vec_info): Change return type of xcalloc to
12900 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12902 * tree-vectorizer.h (vec_info::lookup_single_use): Declare.
12903 * tree-vectorizer.c (vec_info::lookup_single_use): New function.
12904 * tree-vect-loop.c (vectorizable_reduction): Use it instead of
12905 a single_imm_use-based sequence.
12906 * tree-vect-stmts.c (supportable_widening_operation): Likewise.
12908 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12910 * tree-vectorizer.h (vec_info::lookup_def): Declare.
12911 * tree-vectorizer.c (vec_info::lookup_def): New function.
12912 * tree-vect-patterns.c (vect_get_internal_def): Use it.
12913 (vect_widened_op_tree): Likewise.
12914 * tree-vect-stmts.c (vect_is_simple_use): Likewise.
12915 * tree-vect-loop.c (vect_analyze_loop_operations): Likewise.
12916 (vectorizable_reduction): Likewise.
12917 (vect_valid_reduction_input_p): Take a stmt_vec_info instead
12919 (vect_is_slp_reduction): Update calls accordingly. Use
12920 vec_info::lookup_def.
12921 (vect_is_simple_reduction): Likewise
12922 * tree-vect-slp.c (vect_detect_hybrid_slp_1): Use vec_info::lookup_def.
12924 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12926 * tree-vectorizer.h (vec_info::lookup_stmt): Declare.
12927 * tree-vectorizer.c (vec_info::lookup_stmt): New function.
12928 * tree-vect-loop.c (vect_determine_vf_for_stmt): Use it instead
12930 (vect_determine_vectorization_factor, vect_analyze_scalar_cycles_1)
12931 (vect_compute_single_scalar_iteration_cost, vect_analyze_loop_form)
12932 (vect_update_vf_for_slp, vect_analyze_loop_operations)
12933 (vect_is_slp_reduction, vectorizable_induction)
12934 (vect_transform_loop_stmt, vect_transform_loop): Likewise.
12935 * tree-vect-patterns.c (vect_init_pattern_stmt):
12936 (vect_determine_min_output_precision_1, vect_determine_precisions)
12937 (vect_pattern_recog): Likewise.
12938 * tree-vect-stmts.c (vect_analyze_stmt, vect_transform_stmt): Likewise.
12939 * config/powerpcspe/powerpcspe.c (rs6000_density_test): Likewise.
12940 * config/rs6000/rs6000.c (rs6000_density_test): Likewise.
12941 * tree-vect-slp.c (vect_detect_hybrid_slp_stmts): Likewise.
12942 (vect_detect_hybrid_slp_1, vect_detect_hybrid_slp_2)
12943 (vect_detect_hybrid_slp): Likewise. Change the walk_stmt_info
12944 info field from a loop to a loop_vec_info.
12946 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12948 * tree-vectorizer.h (stmt_vec_info): Move typedef earlier in file.
12949 (vec_info::add_stmt): Declare.
12950 * tree-vectorizer.c (vec_info::add_stmt): New function.
12951 * tree-vect-data-refs.c (vect_create_data_ref_ptr): Use it.
12952 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Likewise.
12953 (vect_create_epilog_for_reduction, vectorizable_reduction): Likewise.
12954 (vectorizable_induction): Likewise.
12955 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Likewise.
12956 * tree-vect-stmts.c (vect_finish_stmt_generation_1): Likewise.
12957 (vectorizable_simd_clone_call, vectorizable_store): Likewise.
12958 (vectorizable_load): Likewise.
12959 * tree-vect-patterns.c (vect_init_pattern_stmt): Likewise.
12960 (vect_recog_bool_pattern, vect_recog_mask_conversion_pattern)
12961 (vect_recog_gather_scatter_pattern): Likewise.
12962 (append_pattern_def_seq): Likewise. Remove a check that is
12963 performed by add_stmt itself.
12965 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12967 * tree-vect-loop.c (vectorizable_reduction): Fix an instance in
12968 which make_ssa_name was called with new_stmt before new_stmt
12971 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12973 * tree-vect-loop.c (vect_valid_reduction_input_p): New function,
12975 (vect_is_slp_reduction): ...here...
12976 (vect_is_simple_reduction): ...and here. Remove repetition of tests
12977 that are already known to be false.
12979 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12981 * tree-vectorizer.h (vect_free_slp_instance): Add a final_p parameter.
12982 * tree-vect-slp.c (vect_free_slp_tree): Likewise. Don't update
12983 STMT_VINFO_NUM_SLP_USES when it's true.
12984 (vect_free_slp_instance): Add a final_p parameter and pass it to
12985 vect_free_slp_tree.
12986 (vect_build_slp_tree_2): Update call to vect_free_slp_instance.
12987 (vect_analyze_slp_instance): Likewise.
12988 (vect_slp_analyze_operations): Likewise.
12989 (vect_slp_analyze_bb_1): Likewise.
12990 * tree-vectorizer.c (vec_info): Likewise.
12991 * tree-vect-loop.c (vect_transform_loop): Likewise.
12993 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
12995 * tree-vect-loop.c (vectorizable_reduction): Assert that the
12996 function is not called for second and subsequent members of
12999 2018-07-31 Richard Sandiford <richard.sandiford@arm.com>
13001 * tree-vect-loop.c (get_initial_def_for_reduction): Move special
13002 cases for nested loops from here to ...
13003 (vect_create_epilog_for_reduction): ...here. Only call
13004 vect_is_simple_use for inner-loop reductions.
13006 2018-07-31 Martin Liska <mliska@suse.cz>
13008 PR gcov-profile/85338
13009 PR gcov-profile/85350
13010 PR gcov-profile/85372
13011 * profile.c (struct location_triplet): New.
13012 (struct location_triplet_hash): Likewise.
13013 (output_location): Do not output a BB that
13014 is already recorded for a line.
13015 (branch_prob): Use streamed_locations.
13017 2018-07-31 Martin Liska <mliska@suse.cz>
13019 PR gcov-profile/85370
13020 * coverage.c (coverage_begin_function): Do not mark target
13021 clones as artificial functions.
13023 2018-07-31 Martin Liska <mliska@suse.cz>
13025 PR gcov-profile/83813
13026 PR gcov-profile/84758
13027 PR gcov-profile/85217
13028 PR gcov-profile/85332
13029 * profile.c (branch_prob): Do not record GOTO expressions
13030 for GIMPLE statements which locations are already streamed.
13032 2018-07-31 Olivier Hainque <hainque@adacore.com>
13034 * gcc.c (handle_spec_function): Accept a soft_matched_part
13035 argument, as do_spec_1. Pass it down to ...
13036 (eval_spec_function): Accept a soft_matched_part argument,
13037 and pass it down to ...
13038 (do_spec_2): Accept a soft_matched_part argument, and pass
13039 it down to do_spec_1.
13040 (do_spec_1): Pass soft_matched_part to handle_spec_function.
13041 (handle_braces): Update call to handle_spec_function.
13042 (driver::set_up_specs): Update calls to do_spec_2.
13043 (compare_debug_dump_opt_spec_function): Likewise.
13044 (compare_debug_self_opt_spec_function): Likewise.
13046 2018-07-31 Olivier Hainque <hainque@adacore.com>
13048 * common.opt (nolibc): New option.
13049 * doc/invoke.texi (Link Options): Document it.
13050 * gcc.c (LINK_GCC_C_SEQUENCE_SPEC): Honor nolibc.
13051 * config/alpha/linux.h: Likewise.
13052 * config/arc/elf.h: Likewise.
13053 * config/arm/uclinux-elf.h: Likewise.
13054 * config/arm/unknown-elf.h: Likewise.
13055 * config/avr/avrlibc.h: Likewise.
13056 * config/bfin/bfin.h: Likewise.
13057 * config/bfin/linux.h: Likewise.
13058 * config/bfin/uclinux.h: Likewise.
13059 * config/darwin.h: Likewise.
13060 * config/darwin10.h: Likewise.
13061 * config/darwin12.h: Likewise.
13062 * config/gnu-user.h: Likewise.
13063 * config/lm32/uclinux-elf.h: Likewise.
13064 * config/pa/pa-hpux11.h: Likewise.
13065 * config/pa/pa64-hpux.h: Likewise.
13066 * config/sparc/sparc.h: Likewise.
13068 2018-07-31 Olivier Hainque <hainque@adacore.com>
13070 * gcc.c (getenv_spec_function): Prepend '/' to value for allowed
13071 undefined variables.
13073 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
13076 * config/arm/arm.c (arm_block_set_aligned_vect): Use gen_int_mode
13077 instead of GEN_INT.
13079 2018-07-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
13081 * tree-ssa-forwprop.c (simplify_builtin_call): Don't create a not NUL
13082 terminated string literal.
13084 2018-07-30 Segher Boessenkool <segher@kernel.crashing.org>
13086 PR rtl-optimization/85160
13087 * combine.c (is_just_move): New function.
13088 (try_combine): Allow combining two instructions into two if neither of
13089 the original instructions was a move.
13091 2018-07-30 Alexander Monakov <amonakov@ispras.ru>
13094 * doc/extend.texi (Global Register Variables): Discourage use of type
13096 (Local Register Variables): Likewise.
13098 2018-07-30 Richard Sandiford <richard.sandiford@arm.com>
13100 PR tree-optimization/86506
13101 * hwint.h (ceil_log2): Resync with hwint.c implementation.
13103 2018-07-30 Ilya Leoshkevich <iii@linux.ibm.com>
13106 * lra-constraints.c (spill_hard_reg_in_range): When selecting the
13107 hard_regno, make sure no insn between `from` and `to` clobbers it.
13109 2018-07-30 Cesar Philippidis <cesar@codesourcery.com>
13110 Tom de Vries <tdevries@suse.de>
13112 * config/nvptx/nvptx.c (PTX_GANG_DEFAULT): Rename to ...
13113 (PTX_DEFAULT_RUNTIME_DIM): ... this.
13114 (nvptx_goacc_validate_dims): Set default worker and gang dims to
13115 PTX_DEFAULT_RUNTIME_DIM.
13116 (nvptx_dim_limit): Ignore GOMP_DIM_WORKER.
13118 2018-07-29 John David Anglin <danglin@gcc.gnu.org>
13120 * config/pa/pa.c (pa_output_addr_vec): Align address table.
13121 * config/pa/pa.h (JUMP_TABLES_IN_TEXT_SECTION): Revise comment.
13122 * config/pa/pa32-linux.h (JUMP_TABLES_IN_TEXT_SECTION): Define.
13124 2018-07-27 Michael Meissner <meissner@linux.ibm.com>
13126 * config/rs6000/constraints.md (wG constraint): Delete, no longer
13128 * config/rs6000/predicates.md (p9_fusion_reg_operand): Rename
13129 predicate to reflect toc fusion has been deleted.
13130 (toc_fusion_mem_raw): Delete, no longer used.
13131 (toc_fusion_mem_wrapped): Likewise.
13132 * config/rs6000/rs6000-cpus.def (POWERPC_MASKS): Delete toc
13134 * config/rs6000/rs6000-protos.h (fusion_wrap_memory_address):
13135 Delete, no longer used.
13136 * config/rs6000/rs6000.c (struct rs6000_reg_addr): Delete fields
13137 meant to be used for toc fusion.
13138 (rs6000_debug_print_mode): Delete toc fusion debugging.
13139 (rs6000_debug_reg_global): Likewise.
13140 (rs6000_init_hard_regno_mode_ok): Delete setting up fields for toc
13141 fusion and secondary reload support that were never used.
13142 (rs6000_option_override_internal): Delete TOC fusion, that was only
13143 partially defined, and it did not work unless you also used the
13145 (rs6000_legitimate_address_p): Delete TOC fusion support.
13146 (rs6000_opt_masks): Likewise.
13147 (fusion_wrap_memory_address): Delete function, no longer used.
13148 (fusion_split_address); Delete TOC fusion support.
13149 * config/rs6000/rs6000.h (TARGET_TOC_FUSION_INT): Delete, no
13150 longer used with toc fusion being deleted.
13151 (TARGET_TOC_FUSION_FP): Likewise.
13152 * config/rs6000/rs6000.md (UNSPEC_FUSION_ADDIS): Delete TOC fusion
13154 (toc fusion spliter): Delete TOC fusion support.
13155 (toc_fusionload_<mode>): Likewise.
13156 (toc_fusionload_di): Likewise.
13157 (fusion_gpr_load_<mode>): Delete generator function, this insn no
13158 longer needs to be named. Rename predicate to delete TOC fusion.
13159 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
13160 (fusion_gpr_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
13161 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_load): Likewise.
13162 (fusion_vsx_<P:mode>_<GPR_FUSION:mode>_store): Likewise.
13163 (p9 fusion peephole2s): Rename predicate to delete TOC fusion.
13165 2018-07-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
13167 * doc/extend.texi (Basic PowerPC Built-in Functions Available on
13168 ISA 2.05): Replace __uint128_t with __uint128 and __int128_t with
13169 __int128 in built-in function prototypes.
13170 (PowerPC AltiVec Built-in Functions on ISA 2.07): Likewise.
13171 (PowerPC AltiVec Built-in Functions on ISA 3.0): Likewise.
13173 2018-07-27 Martin Sebor <msebor@redhat.com>
13175 PR tree-optimization/86696
13176 * tree-ssa-strlen.c (get_min_string_length): Handle all integer
13177 types, including enums.
13178 (handle_char_store): Be prepared for the above function to fail.
13180 2018-07-26 Qing Zhao <qing.zhao@oracle.com>
13182 * builtins.c (inline_expand_builtin_string_cmp): Disable inlining
13183 when optimization level is lower than 2 or optimize for size.
13185 2018-07-26 Martin Sebor <msebor@redhat.com>
13187 PR tree-optimization/86043
13188 PR tree-optimization/86042
13189 * tree-ssa-strlen.c (handle_builtin_memcpy): Handle strict overlaps.
13190 (get_string_cst_length): Rename...
13191 (get_min_string_length): ...to this. Add argument.
13192 (handle_char_store): Extend to handle multi-character stores by
13194 * tree.c (initializer_zerop): Use new argument. Handle MEM_REF.
13195 * tree.h (initializer_zerop): Add argument.
13197 2018-07-26 Jakub Jelinek <jakub@redhat.com>
13199 PR middle-end/86660
13200 * omp-low.c (scan_sharing_clauses): Don't ignore map clauses for
13201 declare target to variables if they have always,{to,from,tofrom} map
13204 2018-07-26 Martin Liska <mliska@suse.cz>
13207 * lto-wrapper.c: Add linker_output as prefix
13208 for ltrans_output_file.
13210 2018-07-26 Segher Boessenkool <segher@kernel.crashing.org>
13212 PR rtl-optimization/85805
13213 * combine.c (reg_nonzero_bits_for_combine): Only use the last set
13214 value for hard registers if that was written in the same mode.
13216 2018-07-26 Martin Liska <mliska@suse.cz>
13218 PR gcov-profile/86536
13219 * gcov.c (format_gcov): Use printf format %.*f directly
13220 and do not handle special values.
13222 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
13224 * common/config/arc/arc-common.c (arc_option_optimization_table):
13225 Update default optimizations for size.
13227 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
13229 * config/arc/arc.md (movsf_insn): Add short instruction selection.
13230 * config/arc/constraints.md (CfZ): New constraint.
13231 * config/arc/fpu.md (addssf3_fpu): Use CfZ constraint.
13232 (subsf3_fpu): Likewise.
13233 (cmpsf_fpu): Likewise.
13234 (cmpsf_fpu_uneq): Likewise.
13236 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
13238 * config/arc/arc.c (compact_memory_operand_p): Check for uncached
13240 (arc_is_uncached_mem_p): uncached applies to both the variable and
13243 2018-07-25 Claudiu Zissulescu <claziss@synopsys.com>
13245 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Add additional
13248 2018-07-25 David Malcolm <dmalcolm@redhat.com>
13250 * optinfo-emit-json.cc (class optrecord_json_writer): Convert
13251 field "m_scopes" from vec to auto_vec.
13253 2018-07-25 Martin Liska <mliska@suse.cz>
13255 * config/powerpcspe/powerpcspe-protos.h (rs6000_loop_align): Fix
13258 2018-07-25 Richard Biener <rguenther@suse.de>
13261 * dwarf2out.c (dwarf2out_decl): Do not handle nested functions
13262 special wrt context_die late.
13263 (gen_subprogram_die): Re-use DIEs in local scope.
13265 2018-07-25 Richard Sandiford <richard.sandiford@arm.com>
13267 PR tree-optimization/86644
13268 * hwint.c (ceil_log2): Fix comment. Return 0 for 0.
13270 2018-07-25 Martin Liska <mliska@suse.cz>
13272 PR middle-end/86645
13273 * dumpfile.c: And excluded values with TDF_ALL_VALUES.
13274 * dumpfile.h (enum dump_flag): Defince TDF_ALL_VALUES.
13276 2018-07-25 Martin Liska <mliska@suse.cz>
13279 * params.def: Explain ASan abbreviation and provide
13280 a documentation link.
13282 2018-07-24 Martin Sebor <msebor@redhat.com>
13284 PR tree-optimization/86622
13285 PR tree-optimization/86532
13286 * builtins.h (string_length): Declare.
13287 * builtins.c (c_strlen): Correct handling of non-constant offsets.
13288 (check_access): Be prepared for non-constant length ranges.
13289 (string_length): Make extern.
13290 * expr.c (string_constant): Only handle the minor non-constant
13291 array index. Use string_constant to compute the length of
13292 a generic string constant.
13294 2018-07-24 Richard Sandiford <richard.sandiford@arm.com>
13296 PR tree-optimization/86618
13297 * tree-vect-stmts.c (vectorizable_call): Don't take the address
13298 of LOOP_VINFO_MASKS (loop_vinfo) when loop_vinfo is null.
13300 2018-07-24 David Malcolm <dmalcolm@redhat.com>
13302 PR tree-optimization/86636
13303 * json.cc (json::object::set): Fix comment. Add assertions.
13304 (json::array::append): Move here from json.h. Add comment and an
13306 (json::string::string): Likewise.
13307 * json.h (json::array::append): Move to json.cc.
13308 (json::string::string): Likewise.
13309 * optinfo-emit-json.cc
13310 (optrecord_json_writer::impl_location_to_json): Assert that we
13311 aren't attempting to write out UNKNOWN_LOCATION, or an ad-hoc
13312 wrapper around it. Expand the location once, rather than three
13314 (optrecord_json_writer::inlining_chain_to_json): Fix the check for
13315 UNKNOWN_LOCATION, to use LOCATION_LOCUS to look through ad-hoc
13317 (optrecord_json_writer::optinfo_to_json): Likewise, in four
13318 places. Fix some overlong lines.
13320 2018-07-24 Matthew Malcomson <matthew.malcomson@arm.com>
13322 * config/aarch64/aarch64-simd.md
13323 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>): Split into...
13324 (aarch64_<ANY_EXTEND:su>subw<mode>): ... This...
13325 (aarch64_<ANY_EXTEND:su>addw<mode>): ... And this.
13326 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w<mode>_internal): Split into...
13327 (aarch64_<ANY_EXTEND:su>subw<mode>_internal): ... This...
13328 (aarch64_<ANY_EXTEND:su>addw<mode>_internal): ... And this.
13329 (aarch64_<ANY_EXTEND:su><ADDSUB:optab>w2<mode>_internal): Split into...
13330 (aarch64_<ANY_EXTEND:su>subw2<mode>_internal): ... This...
13331 (aarch64_<ANY_EXTEND:su>addw2<mode>_internal): ... And this.
13333 2018-07-24 Jakub Jelinek <jakub@redhat.com>
13335 PR middle-end/86627
13336 * expmed.c (expand_divmod): Punt if d == HOST_WIDE_INT_MIN
13337 and size > HOST_BITS_PER_WIDE_INT. For size > HOST_BITS_PER_WIDE_INT
13338 and abs_d == d, do the power of two handling if profitable.
13340 2018-07-24 Richard Biener <rguenther@suse.de>
13342 * match.pd: Add BIT_FIELD_REF canonicalizations.
13344 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13347 * genmatch.c (dt_operand::gen_match_op): Avoid folding volatile values.
13349 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13351 * gimple-fold.c (gimple_fold_builtin_printf): Don't create a not NUL
13352 terminated STRING_CST object.
13354 2018-07-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
13356 hsa-dump.c (dump_hsa_symbol): Avoid out of scope access to buf.
13358 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
13360 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Adjust.
13361 * config/rs6000/rs6000-protos.h (rs6000_split_v4si_init): Delete.
13362 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Always force
13363 the elements into a register.
13364 (rs6000_split_v4si_init_di_reg): Delete.
13365 (rs6000_split_v4si_init): Delete.
13366 * config/rs6000/vsx.md (unspec): Delete UNSPEC_VSX_VEC_INIT.
13367 (vsx_init_v4si): Rewrite as a define_expand.
13369 2018-07-23 Segher Boessenkool <segher@kernel.crashing.org>
13371 * config/rs6000/rs6000.md (splitters for rldimi and rlwimi with the
13372 zero_extend argument from memory): New.
13374 2018-07-22 Martin Sebor <msebor@redhat.com>
13377 * gimple-ssa-warn-alloca.c (alloca_call_type_by_arg): Avoid
13378 diagnosing calls with unknown arguments unless -Walloca-larger-than
13379 is restricted to less than PTRDIFF_MAX bytes.
13381 2018-07-22 Gerald Pfeifer <gerald@pfeifer.com>
13383 * doc/gcov.texi (Invoking Gcov): Editorial changes.
13385 2018-07-20 David Malcolm <dmalcolm@redhat.com>
13387 * pretty-print.c (text_info::set_location): Remove redundant
13388 "line_table" parameter from call to rich_location::set_range.
13390 2018-07-20 Martin Sebor <msebor@redhat.com>
13392 PR middle-end/82063
13393 * builtins.c (expand_builtin_alloca): Adjust.
13394 * calls.c (alloc_max_size): Simplify.
13395 * cgraphunit.c (cgraph_node::expand): Adjust.
13396 * common.opt (larger_than_size, warn_frame_larger_than): Remove
13398 (frame_larger_than_size): Same.
13399 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Change options
13400 to take a HOST_WIDE_INT argument and accept a byte-size suffix.
13402 * doc/invoke.texi (GCC Command Options): Document option arguments.
13403 Explain byte-size arguments and suffixes.
13404 (-Wvla-larger-than, -Wno-alloc-size-larger-than): Update.
13405 (-Wno-alloca-larger-than, -Wno-vla-larger-than): Same.
13406 (-Wframe-larger-than, -Wlarger-than, -Wstack-usage): Same.
13407 * doc/options.texi (UInteger): Expand.
13408 (Host_Wide_Int, ByteSize): Document new properties.
13409 * final.c (final_start_function_1): Include sizes in an error message.
13410 * function.c (frame_offset_overflow): Same.
13411 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Adjust.
13412 (alloca_call_type_by_arg): Change function argument to HOST_WIDE_INT.
13413 Diagnose unbounded alloca calls only for limits of less than
13415 (alloca_call_type): Adjust. Diagnose possibly out-of-bounds alloca
13416 calls and VLA size only for limits of less than PTRDIFF_MAX. Same
13418 (pass_walloca::execute): Adjust. Diagnose alloca calls in loops
13419 only for limits of less than PTRDIFF_MAX.
13420 * langhooks-def.h (lhd_handle_option): Change function argument
13422 * langhooks.c (lhd_handle_option): Same.
13423 * langhooks.h (handle_option): Same.
13424 * opt-functions.awk (switch_bit_fields): Handle Host_Wide_Int and
13426 (var_type, var_type_struct): Same.
13427 (var_set): Handle ByteSize flag.
13428 * optc-gen.awk: Add comments to output to ease debugging. Make
13429 use of HOST_WIDE_INT where appropriate.
13430 * opts-gen-save.awk: Use %lx to format unsigned long.
13431 * opth-gen.awk: Change function argument to HOST_WIDE_INT.
13432 * opts-common.c (integral_argument): Return HOST_WIDE_INT and add
13433 arguments. Parse bytes-size suffixes.
13434 (enum_arg_to_value): Change function argument to HOST_WIDE_INT.
13435 (enum_value_to_arg): Same.
13436 (decode_cmdline_option): Handle cl_host_wide_int. Adjust.
13437 (handle_option): Adjust.
13438 (generate_option): Change function argument to HOST_WIDE_INT.
13439 (cmdline_handle_error): Adjust.
13440 (read_cmdline_option): Change function argument to HOST_WIDE_INT.
13441 (set_option): Change function argument to HOST_WIDE_INT.
13442 (option_enabled): Handle cl_host_wide_int.
13443 (get_option_state): Handle CLVC_SIZE.
13444 (control_warning_option): Same.
13445 * opts.c (common_handle_option): Change function argument to
13446 HOST_WIDE_INT. Remove handling of OPT_Walloca_larger_than_ and
13447 OPT_Wvla_larger_than_.
13448 * opts.h (enum cl_var_type): Add an enumerator.
13449 * stor-layout.c (layout_decl): Print a more meaningful warning.
13450 * toplev.c (output_stack_usage): Adjust.
13452 2018-07-20 Qing Zhao <qing.zhao@oracle.com>
13454 * builtins.c (expand_builtin_memcmp): Delete the last parameter for
13455 call to inline_expand_builtin_string_cmp.
13456 (expand_builtin_strcmp): Likewise.
13457 (expand_builtin_strncmp): Likewise.
13458 (inline_string_cmp): Delete the last parameter, change char_type_node
13459 to unsigned_char_type_node for strcmp/strncmp, add conversions to the
13461 (inline_expand_builtin_string_cmp): Delete the last parameter, give up
13462 the inlining expansion on target where the type of the call has same or
13463 narrower precision than unsigned char.
13465 2018-07-20 David Malcolm <dmalcolm@redhat.com>
13467 * Makefile.in (OBJS): Add json.o and optinfo-emit-json.o.
13468 (CFLAGS-optinfo-emit-json.o): Define TARGET_NAME.
13469 * common.opt (fsave-optimization-record): New option.
13470 * coretypes.h (struct kv_pair): Move here from dumpfile.c.
13471 * doc/invoke.texi (-fsave-optimization-record): New option.
13472 * dumpfile.c: Include "optinfo-emit-json.h".
13473 (struct kv_pair): Move to coretypes.h.
13474 (optgroup_options): Make non-static.
13475 (dump_context::end_scope): Call
13476 optimization_records_maybe_pop_dump_scope.
13477 * dumpfile.h (optgroup_options): New decl.
13478 * json.cc: New file.
13479 * json.h: New file.
13480 * optinfo-emit-json.cc: New file.
13481 * optinfo-emit-json.h: New file.
13482 * optinfo.cc: Include "optinfo-emit-json.h".
13483 (optinfo::emit): Call optimization_records_maybe_record_optinfo.
13484 (optinfo_enabled_p): Check optimization_records_enabled_p.
13485 (optinfo_wants_inlining_info_p): Likewise.
13486 * optinfo.h: Update comment.
13487 * profile-count.c (profile_quality_as_string): New function.
13488 * profile-count.h (profile_quality_as_string): New decl.
13489 (profile_count::quality): New accessor.
13490 * selftest-run-tests.c (selftest::run_tests): Call json_cc_tests
13491 and optinfo_emit_json_cc_tests.
13492 * selftest.h (selftest::json_cc_tests): New decl.
13493 (selftest::optinfo_emit_json_cc_tests): New decl.
13494 * toplev.c: Include "optinfo-emit-json.h".
13495 (compile_file): Call optimization_records_finish.
13496 (do_compile): Call optimization_records_start.
13497 * tree-ssa-live.c: Include optinfo.h.
13498 (remove_unused_scope_block_p): Retain inlining information if
13499 optinfo_wants_inlining_info_p returns true.
13501 2018-07-20 Richard Biener <rguenther@suse.de>
13504 * dwarf2out.c (dwarf2out_die_ref_for_decl): Test in_lto_p
13505 to cover -flto-partition=none.
13507 2018-07-20 Martin Liska <mliska@suse.cz>
13509 * tree.h (DECL_LOCATION_RANGE): Remove unused macro.
13510 (get_decl_source_range): Remove unused function.
13512 2018-07-20 Richard Biener <rguenther@suse.de>
13514 * tree-ssa-sccvn.h (struct vn_nary_op_s): Add next member.
13515 (struct vn_phi_s): Likewise.
13516 (struct vn_reference_s): Likewise.
13517 * tree-ssa-sccvn.c (vn_nary_op_hasher::equal): Add shortcut
13518 for searching the slot of an entry known to be in the hash itself.
13519 (vn_phi_hasher::equal): Likewise.
13520 (vn_reference_hasher::equal): Likewise.
13521 (last_inserted_ref, last_inserted_phi, last_inserted_nary): New
13523 (optimistic_info, current_info): Remove, keeping only valid_info.
13524 (vn_reference_lookup_1): Remove fallback lookup.
13525 (vn_reference_lookup_2): Likewise.
13526 (vn_nary_op_lookup_1): Likewise.
13527 (vn_phi_lookup): Likewise.
13528 (vn_nary_build_or_lookup_1): Make sure to not chain the built
13530 (vn_reference_insert): Adjust, chain the inserted hash element
13531 at last_inserted_ref.
13532 (vn_reference_insert_pieces): Likewise.
13533 (visit_reference_op_call): Likewise.
13534 (vn_nary_op_insert_into): Chain the inserted hash element at
13535 last_inserted_nary.
13536 (vn_nary_op_insert_pieces): Adjust.
13537 (vn_nary_op_insert): Likewise.
13538 (vn_nary_op_insert_stmt): Likewise.
13539 (vn_phi_insert): Adjust, chain the inserted hash element at
13541 (process_scc): Remove clearing and copying the optimistic
13542 table. Instead remove elements inserted during an optimistic
13543 iteration from the single table we maintain.
13544 (init_scc_vn): Adjust.
13545 (free_scc_vn): Likewise.
13546 (sccvn_dom_walker::record_cond): Likewise.
13547 (sccvn_dom_walker::after_dom_children): Likewise.
13549 2018-07-19 Martin Sebor <msebor@redhat.com>
13551 PR tree-optimization/84047
13552 PR tree-optimization/83776
13553 * tree-vrp.c (vrp_prop::check_mem_ref): New function.
13554 (check_array_bounds): Call it.
13556 2018-07-19 Martin Sebor <msebor@redhat.com>
13558 * align.h (align_flags): Use member initialization.
13560 2018-07-19 David Malcolm <dmalcolm@redhat.com>
13562 * Makefile.in (OBJS): Add optinfo.o.
13563 * coretypes.h (class symtab_node): New forward decl.
13564 (struct cgraph_node): New forward decl.
13565 (class varpool_node): New forward decl.
13566 * dump-context.h: New file.
13567 * dumpfile.c: Include "optinfo.h", "dump-context.h", "cgraph.h",
13569 (refresh_dumps_are_enabled): Use optinfo_enabled_p.
13570 (set_dump_file): Call dumpfile_ensure_any_optinfo_are_flushed.
13571 (set_alt_dump_file): Likewise.
13572 (dump_context::~dump_context): New dtor.
13573 (dump_gimple_stmt): Move implementation to...
13574 (dump_context::dump_gimple_stmt): ...this new member function.
13575 Add the stmt to any pending optinfo, creating one if need be.
13576 (dump_gimple_stmt_loc): Move implementation to...
13577 (dump_context::dump_gimple_stmt_loc): ...this new member function.
13578 Start a new optinfo and add the stmt to it.
13579 (dump_gimple_expr): Move implementation to...
13580 (dump_context::dump_gimple_expr): ...this new member function.
13581 Add the stmt to any pending optinfo, creating one if need be.
13582 (dump_gimple_expr_loc): Move implementation to...
13583 (dump_context::dump_gimple_expr_loc): ...this new member function.
13584 Start a new optinfo and add the stmt to it.
13585 (dump_generic_expr): Move implementation to...
13586 (dump_context::dump_generic_expr): ...this new member function.
13587 Add the tree to any pending optinfo, creating one if need be.
13588 (dump_generic_expr_loc): Move implementation to...
13589 (dump_context::dump_generic_expr_loc): ...this new member
13590 function. Add the tree to any pending optinfo, creating one if
13592 (dump_printf): Move implementation to...
13593 (dump_context::dump_printf_va): ...this new member function. Add
13594 the text to any pending optinfo, creating one if need be.
13595 (dump_printf_loc): Move implementation to...
13596 (dump_context::dump_printf_loc_va): ...this new member function.
13597 Start a new optinfo and add the stmt to it.
13598 (dump_dec): Move implementation to...
13599 (dump_context::dump_dec): ...this new member function. Add the
13600 value to any pending optinfo, creating one if need be.
13601 (dump_context::dump_symtab_node): New member function.
13602 (dump_context::get_scope_depth): New member function.
13603 (dump_context::begin_scope): New member function.
13604 (dump_context::end_scope): New member function.
13605 (dump_context::ensure_pending_optinfo): New member function.
13606 (dump_context::begin_next_optinfo): New member function.
13607 (dump_context::end_any_optinfo): New member function.
13608 (dump_context::s_current): New global.
13609 (dump_context::s_default): New global.
13610 (dump_scope_depth): Delete global.
13611 (dumpfile_ensure_any_optinfo_are_flushed): New function.
13612 (dump_symtab_node): New function.
13613 (get_dump_scope_depth): Reimplement in terms of dump_context.
13614 (dump_begin_scope): Likewise.
13615 (dump_end_scope): Likewise.
13616 (selftest::temp_dump_context::temp_dump_context): New ctor.
13617 (selftest::temp_dump_context::~temp_dump_context): New dtor.
13618 (selftest::verify_item): New function.
13619 (ASSERT_IS_TEXT): New macro.
13620 (ASSERT_IS_TREE): New macro.
13621 (ASSERT_IS_GIMPLE): New macro.
13622 (selftest::test_capture_of_dump_calls): New test.
13623 (selftest::dumpfile_c_tests): Call it.
13624 * dumpfile.h (dump_printf, dump_printf_loc, dump_basic_block)
13625 (dump_generic_expr_loc, dump_generic_expr, dump_gimple_stmt_loc)
13626 (dump_gimple_stmt, dump_dec): Gather these related decls and add a
13627 descriptive comment.
13628 (dump_function, print_combine_total_stats, enable_rtl_dump_file)
13629 (dump_node, dump_bb): Move these unrelated decls.
13630 (class dump_manager): Add leading comment.
13631 * optinfo.cc: New file.
13632 * optinfo.h: New file.
13634 2018-07-19 Michael Collison <michael.collison@arm.com>
13635 Richard Henderson <rth@redhat.com>
13637 * config/aarch64/aarch64.md (subv<GPI>4, usubv<GPI>4): New patterns.
13638 (subti): Handle op1 zero.
13639 (subvti4, usub4ti4): New.
13640 (*sub<GPI>3_compare1_imm): New.
13641 (sub<GPI>3_carryinCV): New.
13642 (*sub<GPI>3_carryinCV_z1_z2, *sub<GPI>3_carryinCV_z1): New.
13643 (*sub<GPI>3_carryinCV_z2, *sub<GPI>3_carryinCV): New.
13645 2018-07-19 Michael Collison <michael.collison@arm.com>
13646 Richard Henderson <rth@redhat.com>
13648 * config/aarch64/aarch64.md: (addv<GPI>4, uaddv<GPI>4): New.
13649 (addti3): Create simpler code if low part is already known to be 0.
13650 (addvti4, uaddvti4): New.
13651 (*add<GPI>3_compareC_cconly_imm): New.
13652 (*add<GPI>3_compareC_cconly): New.
13653 (*add<GPI>3_compareC_imm): New.
13654 (*add<GPI>3_compareC): Rename from add<GPI>3_compare1; do not
13655 handle constants within this pattern..
13656 (*add<GPI>3_compareV_cconly_imm): New.
13657 (*add<GPI>3_compareV_cconly): New.
13658 (*add<GPI>3_compareV_imm): New.
13659 (add<GPI>3_compareV): New.
13660 (add<GPI>3_carryinC, add<GPI>3_carryinV): New.
13661 (*add<GPI>3_carryinC_zero, *add<GPI>3_carryinV_zero): New.
13662 (*add<GPI>3_carryinC, *add<GPI>3_carryinV): New.
13663 ((*add<GPI>3_compareC_cconly_imm): Replace 'ne' operator
13664 with 'comparison' operator.
13665 (*add<GPI>3_compareV_cconly_imm): Ditto.
13666 (*add<GPI>3_compareV_cconly): Ditto.
13667 (*add<GPI>3_compareV_imm): Ditto.
13668 (add<GPI>3_compareV): Ditto.
13669 (add<mode>3_carryinC): Ditto.
13670 (*add<mode>3_carryinC_zero): Ditto.
13671 (*add<mode>3_carryinC): Ditto.
13672 (add<mode>3_carryinV): Ditto.
13673 (*add<mode>3_carryinV_zero): Ditto.
13674 (*add<mode>3_carryinV): Ditto.
13676 2018-07-19 Michael Collison <michael.collison@arm.com>
13677 Richard Henderson <rth@redhat.com>
13679 * config/aarch64/aarch64-modes.def (CC_V): New.
13680 * config/aarch64/aarch64-protos.h
13681 (aarch64_addti_scratch_regs): Declare
13682 (aarch64_subvti_scratch_regs): Declare.
13683 (aarch64_expand_subvti): Declare.
13684 (aarch64_gen_unlikely_cbranch): Declare
13685 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Test
13686 for signed overflow using CC_Vmode.
13687 (aarch64_get_condition_code_1): Handle CC_Vmode.
13688 (aarch64_gen_unlikely_cbranch): New function.
13689 (aarch64_addti_scratch_regs): New function.
13690 (aarch64_subvti_scratch_regs): New function.
13691 (aarch64_expand_subvti): New function.
13693 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
13695 * config/aarch64/aarch64-option-extensions.def: New entry for profile
13697 * config/aarch64/aarch64.h (AARCH64_FL_PROFILE): New.
13698 * doc/invoke.texi (aarch64-feature-modifiers): New entry for profile
13701 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
13704 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
13705 address check not strict.
13707 2018-07-19 Andre Vieira <andre.simoesdiasvieira@arm.com>
13709 * config/aarch64/aarch64-simd.md (aarch64_simd_mov<VQ:mode>): Replace
13711 (store_pair_lanes<mode>): Likewise.
13712 * config/aarch64/aarch64-protos.h (aarch64_addr_query_type): Add new
13713 enum value 'ADDR_QUERY_LDP_STP_N'.
13714 * config/aarch64/aarch64.c (aarch64_addr_query_type): Likewise.
13715 (aarch64_print_address_internal): Add declaration.
13716 (aarch64_print_ldpstp_address): Remove.
13717 (aarch64_classify_address): Adapt mode for 'ADDR_QUERY_LDP_STP_N'.
13718 (aarch64_print_operand): Change printing of 'y'.
13719 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Use
13720 new enum value 'ADDR_QUERY_LDP_STP_N', don't hardcode mode and use
13721 'true' rather than '1'.
13722 * config/aarch64/constraints.md (Uml): Likewise.
13723 (Uml): Rename to Umn.
13726 2018-07-19 Richard Biener <rguenther@suse.de>
13728 * tree-ssa-sccvn.h (struct vn_phi_s): Make phiargs member
13730 * tree-ssa-sccvn.c: Remove alloc-pool.h use.
13731 (vn_phi_hasher): Derive from nofree_ptr_hash and remove remove method.
13732 (vn_reference_hasher): Likewise.
13733 (struct vn_tables_s): Remove obstack and alloc-pool members.
13734 (vn_tables_obstack, vn_tables_insert_obstack): New global obstacks.
13735 (vn_nary_build_or_lookup_1): Manually build in vn_tables_insert_obstack.
13736 (vn_reference_insert): Allocate from obstack instead of from alloc-pool.
13737 (vn_reference_insert_pieces): Likewise.
13738 (alloc_vn_nary_op_noinit): Adjust.
13739 (vn_nary_op_insert_stmt): Allocate phiargs in-place.
13740 (vn_phi_eq): Adjust.
13741 (shared_lookup_phiargs): Remove.
13742 (vn_phi_lookup): Allocate temporary vn_phi_s on the stack.
13743 (vn_phi_insert): Allocate from obstack instead of from alloc-pool.
13744 (visit_reference_op_call): Likewise.
13745 (copy_nary, copy_phi, copy_reference): Remove.
13746 (process_scc): Rewind the obstack when iterating. Do not
13747 copy the elements to valid_info but just move them from one
13748 hashtable to the other.
13749 (allocate_vn_table): Adjust.
13750 (free_vn_table): Likewise.
13751 (init_scc_vn): Likewise.
13752 (free_scc_vn): Likewise.
13754 2018-07-19 H.J. Lu <hongjiu.lu@intel.com>
13757 * config/i386/i386.c (rest_of_insert_endbranch): Lookup
13758 indirect_return as function type attribute.
13759 (ix86_attribute_table): Change indirect_return to function
13761 * doc/extend.texi: Update indirect_return attribute.
13763 2018-07-19 Aldy Hernandez <aldyh@redhat.com>
13765 * wide-int.h (widest2_int): New.
13766 * gimple-fold.c (arith_overflowed_p): Use it.
13767 * tree.h (widest2_int_cst): New.
13768 * tree-vrp.c (wide_int_binop_overflow): Rename from
13769 vrp_int_const_binop.
13770 Rewrite to work on trees.
13771 (extract_range_from_multiplicative_op_1): Abstract code to...
13772 (wide_int_range_min_max): ...here.
13773 (wide_int_range_cross_product): ...and here.
13774 (extract_range_from_binary_expr_1): Abstract overflow code to...
13775 (wide_int_range_mult_wrapping): ...here.
13776 * tree-vrp.h (wide_int_range_cross_product): New.
13777 (wide_int_range_mult_wrapping): New.
13779 2018-07-19 Andrew Senkevich <andrew.senkevich@intel.com>
13780 Julia Koval <julia.koval@intel.com>
13782 * config/i386/x86-tune-costs.h (skylake_memcpy,
13783 skylake_memset): Replace rep_prefix with unrolling for size 512.
13785 2018-07-18 Kugan Vivekanandarajah <kuganv@linaro.org>
13787 PR middle-end/86544
13788 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): Handle
13789 comparision with EQ_EXPR in last stmt.
13791 2018-07-18 Kelvin Nilsen <kelvin@gcc.gnu.org>
13793 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Rename
13794 this subsection to "PowerPC AltiVec/VSX Built-in Functions".
13795 (PowerPC AltiVec/VSX Built-in Functions): New name for subsection
13796 previously known as "PowerPC AltiVec Built-in Functions". Move
13797 some material to new subsubsections "PowerPC AltiVec Built-in
13798 Functions on ISA 2.06" and "PowerPC AltiVec Built-in Functions on
13800 (PowerPC Altivec Built-in Functions on ISA 2.05): New subsubsection.
13801 (PowerPC Altivec Built-in Functions on ISA 2.06): Likewise.
13802 (PowerPC Altivec Built-in Functions on ISA 2.07): Likewise.
13803 (PowerPC Altivec Built-in Functions on ISA 3.0): Likewise.
13805 2018-07-18 Richard Biener <rguenther@suse.de>
13807 PR tree-optimization/86557
13808 * tree-vect-patterns.c (vect_recog_divmod_pattern): Also handle
13811 2018-07-18 Ilya Leoshkevich <iii@linux.ibm.com>
13813 * config/s390/s390.c (s390_function_profiler): Generate CFI.
13815 2018-07-17 Jeff Law <law@redhat.com>
13817 * config/arm/arm.c (get_label_padding): Update for recent
13818 changes to label_to_alignment.
13820 PR tree-optimization/86010
13821 * tree-ssa-dse.c (compute_trims): Fix typo/thinko.
13823 * config/mips/mips.c (vr4130_align_insns): Update for recent
13824 changes to label_to_alignment.
13826 * config/frv/frv.c (frv_label_align): Update for recent changes
13827 to label_to_alignment.
13829 * config/nios2/nios2.c (nios2_label_align): Update for recent
13830 changes which dropped ALIGN_LABELS_LOG.
13832 2018-07-17 Andreas Schwab <schwab@linux-m68k.org>
13834 * config/m68k/m68k.md (umulsi3_highpart+1, const_umulsi3_highpart)
13835 (smulsi3_highpart+1, const_smulsi3_highpart): Add CC_STATUS_INIT.
13837 2018-07-17 Claudiu Zissulescu <claziss@synopsys.com>
13839 * config/arc/arc.c (arc_label_align): Use align_labels instead of
13840 deprecated align_labels_log.
13842 2018-07-17 Richard Biener <rguenther@suse.de>
13845 * dwarf2out.c (init_sections_and_labels): Always generate
13846 a debug_line_str_section for early LTO debug.
13847 (dwarf2out_finish): Reset debug_line_str_hash output early.
13848 Bump counter for extra dwarf5 .debug_loc labels to not conflict
13850 (dwarf2out_early_finish): Output debug_line_str.
13852 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
13854 * config/s390/s390.c (preferred_la_operand_p): Do not use LA with
13855 index register on z196 or later.
13857 2018-07-17 Robin Dapp <rdapp@linux.ibm.com>
13859 * config/s390/s390.c (s390_default_align): Set default function
13861 (s390_override_options_after_change): Call s390_default align.
13862 (s390_option_override_internal): Call s390_default_align.
13863 (TARGET_OVERRIDE_OPTIONS_AFTER_CHANGE): Define.
13865 2018-07-17 Jakub Jelinek <jakub@redhat.com>
13867 PR middle-end/86542
13868 * omp-low.c (create_task_copyfn): Copy over also fields corresponding
13869 to _looptemp_ clauses, other than the first two.
13871 2018-07-17 Martin Liska <mliska@suse.cz>
13873 * opts.c: Do not enable OPT_falign_* for -Os.
13875 2018-07-17 Martin Liska <mliska@suse.cz>
13877 * align.h (MAX_CODE_ALIGN): New.
13878 (MAX_CODE_ALIGN_VALUE): New.
13879 * common/config/i386/i386-common.c (ix86_handle_option):
13880 (MAX_CODE_ALIGN): Moved to align.h.
13881 * final.c (MAX_CODE_ALIGN): Likewise.
13882 * opts.c (parse_and_check_align_values):
13883 (MAX_CODE_ALIGN): Likewise.
13884 (MAX_CODE_ALIGN_VALUE): Likewise.
13886 2018-07-17 Martin Liska <mliska@suse.cz>
13888 * config/i386/att.h (ASM_OUTPUT_ALIGN): Fix spacing
13889 in order to fulfil coding style.
13890 * config/i386/cygming.h (ASM_OUTPUT_ALIGN): Likewise.
13891 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13892 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
13893 * config/iq2000/iq2000.h (ASM_OUTPUT_ALIGN): Likewise.
13894 * config/pa/pa.h (ASM_OUTPUT_ALIGN): Likewise.
13895 * config/sparc/sol2.h (ASM_OUTPUT_ALIGN_WITH_NOP): Likewise.
13896 * config/sparc/sparc.h (ASM_OUTPUT_ALIGN): Likewise.
13897 * config/visium/visium.h (ASM_OUTPUT_ALIGN): Likewise.
13898 (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
13900 2018-07-17 Martin Liska <mliska@suse.cz>
13902 * align.h: New file.
13903 * config/alpha/alpha.c (alpha_align_insns_1): Use align_functions
13905 * config/i386/i386.c (ix86_avoid_jump_mispredicts): Use new return type
13906 align_flags of label_to_alignment.
13907 * config/m32r/m32r.h (LOOP_ALIGN): Wrap returned values into
13909 * config/m68k/m68k.c: Do not use removed align_labels_value and
13911 * config/nds32/nds32.h (JUMP_ALIGN): Wrap result into align_flags class.
13912 (LOOP_ALIGN): Likewise.
13913 (LABEL_ALIGN): Likewise.
13914 * config/powerpcspe/powerpcspe.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
13915 Remove not used macro.
13916 (rs6000_loop_align): Change return type to align_flags.
13917 (rs6000_loop_align_max_skip): Remove.
13918 * config/rs6000/rs6000-protos.h (rs6000_loop_align):
13919 Change return type to align_flags.
13920 * config/rs6000/rs6000.c (TARGET_ASM_LOOP_ALIGN_MAX_SKIP):
13921 Remove not used macro.
13922 (rs6000_loop_align): Change return type to align_flags.
13923 (rs6000_loop_align_max_skip): Remove.
13924 * config/rx/rx.h (JUMP_ALIGN): Wrap integer values
13925 * config/rx/rx-protos.h (rx_align_for_label): Make it
13927 * config/rx/rx.c (rx_align_for_label): Change return type
13929 (rx_max_skip_for_label): Remove TARGET_ASM_*_ALIGN_MAX_SKIP
13931 into align_flags class.
13932 (LABEL_ALIGN): Likewise.
13933 (LOOP_ALIGN): Likewise.
13934 * config/s390/s390.c (s390_label_align): Use align_flags
13936 (s390_asm_output_function_label): Likewise.
13937 * config/sh/sh.c (sh_override_options_after_change):
13938 Use align_flags class directly without macros.
13939 (find_barrier): Likewise.
13940 (barrier_align): Likewise.
13941 (sh_loop_align): Likewise.
13942 * config/spu/spu.c (spu_option_override):
13943 Use align_flags_tuple::get_value instead of removed macros.
13944 (spu_sched_init): Likewise.
13945 * config/spu/spu.h (GTY): Likewise.
13946 * config/visium/visium.c (visium_option_override):
13947 Set "8" as default secondary alignment.
13948 * config/visium/visium.h (SUBALIGN_LOG): Define to 3
13949 in order to guarantee secondary alignment of 8.
13950 * coretypes.h: Include align.h header file.
13951 * doc/tm.texi: Remove TARGET_ASM_JUMP_ALIGN_MAX_SKIP,
13952 TARGET_ASM_LOOP_ALIGN_MAX_SKIP, TARGET_ASM_LABEL_ALIGN_MAX_SKIP
13953 and TARGET_ASM_LABEL_ALIGN_AFTER_BARRIER_MAX_SKIP macros.
13954 * doc/tm.texi.in: Likewise.
13955 * final.c (struct label_alignment): Remove not used structure.
13956 (LABEL_ALIGN): Change type to align_flags.
13957 (LOOP_ALIGN): Likewise.
13958 (JUMP_ALIGN): Likewise.
13959 (default_loop_align_max_skip): Remove.
13960 (default_label_align_max_skip): Likewise.
13961 (default_jump_align_max_skip): Likewise.
13962 (default_label_align_after_barrier_max_skip):
13963 (LABEL_TO_ALIGNMENT): Change to access label_align vector.
13964 (LABEL_TO_MAX_SKIP): Remove.
13965 (label_to_alignment): Return align_flags type instead of integer.
13966 (label_to_max_skip): Remove.
13967 (align_fuzz): Use align_flags type.
13968 (compute_alignments): Use align_flags type and use align_flags::max
13969 to combine multiple alignments.
13970 (grow_label_align): Grow vec instead of C array.
13971 (update_alignments): Assign just LABEL_TO_ALIGNMENT.
13972 (shorten_branches): Use align_flags type and use align_flags::max
13973 to combine multiple alignments.
13974 (final_scan_insn_1): Remove usage of secondary alignment that comes
13975 from label alignment, but instead use proper secondary alignment
13976 which is computed in grow_label_align.
13977 * flags.h (struct align_flags_tuple): Move to align.h.
13978 (struct align_flags): Likewise.
13979 (state_align_loops): Rename to align_loops.
13980 (state_align_jumps): Rename to align_jumps.
13981 (state_align_labels): Rename to align_labels.
13982 (state_align_functions): Rename to align_functions.
13983 (align_loops_log): Remove.
13984 (align_jumps_log): Remove.
13985 (align_labels_log): Remove.
13986 (align_functions_log): Remove.
13987 (align_loops_max_skip): Remove.
13988 (align_jumps_max_skip): Remove.
13989 (align_labels_max_skip): Remove.
13990 (align_functions_max_skip): Remove.
13991 (align_loops_value): Remove.
13992 (align_jumps_value): Remove.
13993 (align_labels_value): Remove.
13994 (align_functions_value): Remove.
13995 * output.h (label_to_alignment): Change return type to align_flags.
13996 (label_to_max_skip): Remove.
13997 * target.def: Remove loop_align_max_skip, label_align_max_skip,
13998 jump_align_max_skip macros.
13999 * targhooks.h (default_loop_align_max_skip): Remove.
14000 (default_label_align_max_skip): Likewise.
14001 (default_jump_align_max_skip): Likewise.
14002 (default_label_align_after_barrier_max_skip): Remove.
14003 * toplev.c (read_log_maxskip): Use ::normalize function.
14004 (parse_N_M): Remove not used argument and also call ::normalize.
14005 (parse_alignment_opts): Do not pass unused arguments.
14006 * varasm.c (assemble_start_function): Use directly align_functions
14007 instead of removed macros.
14008 * system.h: Do not poison removed macros.
14010 2018-07-17 Jakub Jelinek <jakub@redhat.com>
14012 PR middle-end/86539
14013 * gimplify.c (gimplify_omp_for): Ensure taskloop firstprivatized init
14014 and cond temporaries don't have reference type if iterator has
14015 pointer type. For init use &for_pre_body instead of pre_p if
14016 for_pre_body is non-empty.
14018 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
14020 * config/rs6000/rs6000.md (trunc<mode>sf2): Expand truncates of
14021 double-double modes to SFmode directly directly.
14022 (trunc<mode>sf2_fprs): Delete.
14024 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
14026 * config/rs6000/rs6000.c (init_float128_ibm): Use the correct names
14027 for conversions between IFmode and the decimal floating point modes.
14028 (init_float128_ieee): Use the correct names for conversions between
14029 KFmode and the decimal floating point modes.
14031 2018-07-16 Segher Boessenkool <segher@kernel.crashing.org>
14033 * config/rs6000/rs6000.c (init_float128_ibm): Use more correct names
14034 for the conversions between TDmode and IFmode.
14035 (init_float128_ieee): Use more correct names for the conversions
14036 between TDmode and KFmode.
14038 2018-07-16 Jakub Jelinek <jakub@redhat.com>
14040 PR tree-optimization/86526
14041 * builtins.c (expand_builtin_memcmp): Formatting fixes.
14042 (inline_expand_builtin_string_cmp): Likewise.
14043 (inline_string_cmp): Likewise. Use c_readstr instead of
14044 builtin_memcpy_read_str. Add unit_mode temporary.
14046 2018-07-16 Bernd Edlinger <bernd.edlinger@hotmail.de>
14048 PR middle-end/86528
14049 * builtins.c (check_access): Bail out if range[0] is no INTEGER_CST.
14050 * expr.c (string_constant): Fix the element size of ARRAY_TYPE.
14052 2018-07-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
14054 * doc/extend.texi (PowerPC AltiVec Built-in Functions):
14055 Alphabetize prototypes of built-in functions, separating out
14056 built-in functions that are listed in this section but should be
14057 described elsewhere.
14059 2018-07-16 Uros Bizjak <ubizjak@gmail.com>
14062 * expmed.c (emit_store_flag): Do not emit setcc followed by a
14063 conditional move when trapping comparison was split to a
14064 non-trapping one (and vice versa).
14066 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
14068 * config/s390/s390.c (s390_function_profiler): Generate nops
14069 instead of profiler call sequences.
14070 * config/s390/s390.opt: Add the new option.
14072 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
14074 * config/s390/s390.c (s390_function_profiler): Generate
14075 __mcount_loc section.
14076 * config/s390/s390.opt: Add the new option.
14078 2018-07-16 Ilya Leoshkevich <iii@linux.ibm.com>
14080 * common.opt: Add the new warning.
14081 * config/s390/s390.c (s390_function_profiler): Emit "brasl
14082 %r0,__fentry__" when -mfentry is specified.
14083 (s390_option_override_internal): Disallow -mfentry for 31-bit
14085 * config/s390/s390.opt: Add the new option.
14087 2018-07-16 Richard Biener <rguenther@suse.de>
14090 * dwarf2out.c (dwarf2out_register_external_die): Assign DIE parents
14091 for function-local FUNCTION_DECL and RESULT_DECL immediately.
14093 2018-07-16 Martin Liska <mliska@suse.cz>
14096 * ipa-pure-const.c (malloc_candidate_p): Revert ::get
14099 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
14101 * config/arc/arcHS.md: Update ARCHS scheduling rules.
14103 2017-07-16 Claudiu Zissulescu <claziss@synopsys.com>
14105 * config/arc/arc-arch.h (arc_tune_attr): Add new tune parameters
14107 * config/arc/arc-cpus.def (hs4x): New cpu.
14109 * config/arc/arc-tables.opt: Regenerate.
14110 * config/arc/arc.c (arc_sched_issue_rate): New function.
14111 (TARGET_SCHED_ISSUE_RATE): Define.
14112 (TARGET_SCHED_EXPOSED_PIPELINE): Likewise.
14113 * config/arc/arc.md (attr type): Add fpu_fuse, fpu_sdiv, fpu_ddiv,
14115 (attr tune): Add ARCHS4x tune values.
14116 (attr tune_dspmpy): Define.
14117 (*tst): Correct instruction type.
14118 * config/arc/arcHS.md: Don't use this automaton for ARCHS4x cpus.
14119 * config/arc/arcHS4x.md: New file.
14120 * config/arc/fpu.md: Update instruction type attributes.
14121 * config/arc/t-multilib: Regenerate.
14123 2018-07-16 Tom de Vries <tdevries@suse.de>
14126 * var-tracking.c (vt_initialize): Fix pre_dec handling.
14128 2018-07-16 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
14130 * config/aarch64/atomics.md (aarch64_store_execlusive<mode>): Add
14133 2018-07-16 Eric Botcazou <ebotcazou@adacore.com>
14135 PR tree-optimization/86514
14136 * tree-ssa-reassoc.c (init_range_entry) <CASE_CONVERT>: Return for a
14137 conversion to a boolean type from a type with greater precision.
14139 2018-07-16 Tom de Vries <tdevries@suse.de>
14141 * var-tracking.c (vt_initialize): Print adjusted insn slim if
14142 dump_flags request TDF_SLIM.
14144 2018-07-16 Aldy Hernandez <aldyh@redhat.com>
14146 * fold-const.c (int_const_binop_1): Abstract...
14147 (wide_int_binop): ...wide int code here.
14148 (poly_int_binop): ...poly int code here.
14149 Abstract the rest of int_const_binop_1 into int_const_binop.
14150 * fold-const.h (wide_int_binop): New.
14151 * tree-vrp.c (vrp_int_const_binop): Call wide_int_binop.
14152 Remove useless PLUS/MINUS_EXPR case.
14153 (zero_nonzero_bits_from_vr): Move wide int code...
14154 (zero_nonzero_bits_from_bounds): ...here.
14155 (extract_range_from_binary_expr_1): Move mask optimization code...
14156 (range_easy_mask_min_max): ...here.
14157 * tree-vrp.h (zero_nonzero_bits_from_bounds): New.
14158 (range_easy_mask_min_max): New.
14160 2018-07-15 Jeff Law <law@redhat.com>
14163 * config/sh/sh.c (output_mi_thunk): Remove dead conditional
14166 2018-07-14 Jim Wilson <jimw@sifive.com>
14168 * config/riscv/linux.h (TARGET_ASM_FILE_END): New.
14170 2018-07-14 Paul Koning <ni1d@arrl.net>
14172 * config/pdp11/pdp11.c (pdp11_rtx_costs): Bugfixes.
14174 2018-07-13 Jan Hubicka <hubicka@ucw.cz>
14176 * lto-streamer-out.c (copy_function_or_variable): Dump info about
14179 2018-07-13 Bill Schmidt <wschmidt@linux.ibm.com>
14180 Steve Munroe <munroesj52@gmail.com>
14182 * config/rs6000/emmintrin.h (_mm_and_si128): New function.
14183 (_mm_andnot_si128): Likewise.
14184 (_mm_or_si128): Likewise.
14185 (_mm_xor_si128): Likewise.
14187 2018-07-13 Qing Zhao <qing.zhao@oracle.com>
14189 PR middle-end/78809
14190 * builtins.c (expand_builtin_memcmp): Inline the calls first
14191 when result_eq is false.
14192 (expand_builtin_strcmp): Inline the calls first.
14193 (expand_builtin_strncmp): Likewise.
14194 (inline_string_cmp): New routine. Expand a string compare
14195 call by using a sequence of char comparison.
14196 (inline_expand_builtin_string_cmp): New routine. Inline expansion
14197 a call to str(n)cmp/memcmp.
14198 * doc/invoke.texi (--param builtin-string-cmp-inline-length):
14200 * params.def (BUILTIN_STRING_CMP_INLINE_LENGTH): New.
14202 2018-07-13 Richard Earnshaw <rearnsha@arm.com>
14204 * config/arm/driver-arm.c: Include arm-native.h.
14205 (host_detect_local_cpu): Use auto-generated data tables.
14206 (vendors, arm_cpu_table): Delete. Move part information to ...
14207 * config/arm/arm-cpus.in: ... here.
14208 * config/arm/parsecpu.awk (gen_native): New function.
14209 (vendor, part): New CPU fields.
14210 (END): Add support for building the native CPU detection tables.
14211 * config/arm/t-arm (arm-native.h): Add build rule.
14212 (driver-arm.o): Add dependency on arm-native.h.
14214 2018-07-13 Richard Biener <rguenther@suse.de>
14216 PR middle-end/85974
14217 * match.pd (addr1 - addr2): Allow either of the operand to
14220 2018-07-13 Tom de Vries <tdevries@suse.de>
14222 * tree-inline.c (remap_ssa_name): Save and reuse debug exprs generated
14225 2018-07-13 Jackson Woodruff <jackson.woodruff@arm.com>
14227 * config/aarch64/aarch64.c (aarch64_operands_adjust_ok_for_ldpstp): Use
14228 arrays instead of numbered variables.
14230 2018-07-13 Eric Botcazou <ebotcazou@adacore.com>
14232 * config/sparc/sparc-protos.h (sparc_compute_frame_size): Delete.
14233 * config/sparc/sparc.c (sparc_compute_frame_size): Make static.
14235 2018-07-13 Richard Biener <rguenther@suse.de>
14238 * dwarf2out.c (gen_type_die_with_usage): Use scope_die_for
14239 instead of get_context_die.
14241 2018-07-13 Kugan Vivekanandarajah <kuganv@linaro.org>
14242 Richard Biener <rguenther@suse.de>
14244 PR middle-end/86489
14245 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Check
14246 that the loop latch destination where phi is defined.
14248 2018-07-12 Kito Cheng <kito.cheng@gmail.com>
14250 * config/riscv/riscv.c (enum riscv_privilege_levels): Add UNKNOWN_MODE.
14251 (riscv_expand_epilogue): Add assertion to check interrupt mode.
14252 (riscv_set_current_function): Extract getting interrupt type to new
14254 (riscv_get_interrupt_type): New function.
14255 (riscv_merge_decl_attributes): New function, checking interrupt type is
14257 (TARGET_MERGE_DECL_ATTRIBUTES): Define.
14259 2018-07-12 Paul Koning <ni1d@arrl.net>
14261 * config/pdp11/pdp11.c (pdp11_output_def): Fix typo in .set
14264 2018-07-12 Paul Koning <ni1d@arrl.net>
14266 * doc/rtl.texi (REG_NONNEG): Remove decrement and branch until
14267 zero reference, add doloop_end instead.
14268 * doc/md.texi (decrement_and_branch_until_zero): Remove.
14269 (Looping patterns): Remove decrement_and_branch_until_zero. Add
14270 detail for doloop_end.
14272 2018-07-12 Martin Sebor <msebor@redhat.com>
14275 * attribs.c (decl_attributes): Reject conflicting attributes before
14276 calling attribute handlers.
14278 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
14280 * dumpfile.c (gcc::dump_manager::get_dump_file_name): Add PART
14282 (gcc::dump_manager::get_dump_file_name): likewise.
14283 (dump_begin): Likewise.
14284 * dumpfile.h (dump_begin): Update prototype.
14285 (gcc::dump_manager::get_dump_file_name,
14286 gcc::dump_manager::get_dump_file_name): Update prototype.
14288 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
14290 * internal-fn.h (vectorizable_internal_fn_p): New function.
14291 * tree-vect-slp.c (compatible_calls_p): Likewise.
14292 (vect_build_slp_tree_1): Remove nops argument. Handle calls
14293 to internal functions.
14294 (vect_build_slp_tree_2): Update call to vect_build_slp_tree_1.
14296 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
14298 * fold-const.h (inverse_conditions_p): Declare.
14299 * fold-const.c (inverse_conditions_p): New function.
14300 * match.pd: Use inverse_conditions_p. Add folds of view_converts
14301 that test the inverse condition of a conditional internal function.
14302 * internal-fn.h (vectorized_internal_fn_supported_p): Declare.
14303 * internal-fn.c (internal_fn_mask_index): Handle conditional
14304 internal functions.
14305 (vectorized_internal_fn_supported_p): New function.
14306 * tree-if-conv.c: Include internal-fn.h and fold-const.h.
14307 (any_pred_load_store): Replace with...
14308 (need_to_predicate): ...this new variable.
14309 (redundant_ssa_names): New variable.
14310 (ifcvt_can_use_mask_load_store): Move initial checks to...
14311 (ifcvt_can_predicate): ...this new function. Handle tree codes
14312 for which a conditional internal function exists.
14313 (if_convertible_gimple_assign_stmt_p): Use ifcvt_can_predicate
14314 instead of ifcvt_can_use_mask_load_store. Update after variable
14316 (predicate_load_or_store): New function, split out from
14317 predicate_mem_writes.
14318 (check_redundant_cond_expr): New function.
14319 (value_available_p): Likewise.
14320 (predicate_rhs_code): Likewise.
14321 (predicate_mem_writes): Rename to...
14322 (predicate_statements): ...this. Use predicate_load_or_store
14323 and predicate_rhs_code.
14324 (combine_blocks, tree_if_conversion): Update after above name changes.
14325 (ifcvt_local_dce): Handle redundant_ssa_names.
14326 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): Handle
14327 general conditional functions.
14328 * tree-vect-stmts.c (vectorizable_call): Likewise.
14330 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
14331 Alan Hayward <alan.hayward@arm.com>
14332 David Sherwood <david.sherwood@arm.com>
14334 * internal-fn.h (can_interpret_as_conditional_op_p): Declare.
14335 * internal-fn.c (can_interpret_as_conditional_op_p): New function.
14336 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Handle conditional
14337 plus and minus and convert them into IFN_COND_FMA-based sequences.
14338 (convert_mult_to_fma): Handle conditional plus and minus.
14340 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
14342 * doc/md.texi (cond_fma, cond_fms, cond_fnma, cond_fnms): Document.
14343 * optabs.def (cond_fma_optab, cond_fms_optab, cond_fnma_optab)
14344 (cond_fnms_optab): New optabs.
14345 * internal-fn.def (COND_FMA, COND_FMS, COND_FNMA, COND_FNMS): New
14346 internal functions.
14347 (FMA): Use DEF_INTERNAL_FLT_FN rather than DEF_INTERNAL_FLT_FLOATN_FN.
14348 * internal-fn.h (get_conditional_internal_fn): Declare.
14349 (get_unconditional_internal_fn): Likewise.
14350 * internal-fn.c (cond_ternary_direct): New macro.
14351 (expand_cond_ternary_optab_fn): Likewise.
14352 (direct_cond_ternary_optab_supported_p): Likewise.
14353 (FOR_EACH_COND_FN_PAIR): Likewise.
14354 (get_conditional_internal_fn): New function.
14355 (get_unconditional_internal_fn): Likewise.
14356 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 5.
14357 (gimple_match_op::gimple_match_op): Add a new overload for 5
14359 (gimple_match_op::set_op): Likewise.
14360 (gimple_resimplify5): Declare.
14361 * genmatch.c (decision_tree::gen): Generate simplifications for
14363 * gimple-match-head.c (gimple_simplify): Define an overload for
14364 5 operands. Handle calls with 5 arguments in the top-level overload.
14365 (convert_conditional_op): Handle conversions from unconditional
14366 internal functions to conditional ones.
14367 (gimple_resimplify5): New function.
14368 (build_call_internal): Pass a fifth operand.
14369 (maybe_push_res_to_seq): Likewise.
14370 (try_conditional_simplification): Try converting conditional
14371 internal functions to unconditional internal functions.
14372 Handle 3-operand unconditional forms.
14373 * match.pd (UNCOND_TERNARY, COND_TERNARY): Operator lists.
14374 Define ternary equivalents of the current rules for binary conditional
14375 internal functions.
14376 * config/aarch64/aarch64.c (aarch64_preferred_else_value): Handle
14377 ternary operations.
14378 * config/aarch64/iterators.md (UNSPEC_COND_FMLA, UNSPEC_COND_FMLS)
14379 (UNSPEC_COND_FNMLA, UNSPEC_COND_FNMLS): New unspecs.
14380 (optab): Handle them.
14381 (SVE_COND_FP_TERNARY): New int iterator.
14382 (sve_fmla_op, sve_fmad_op): New int attributes.
14383 * config/aarch64/aarch64-sve.md (cond_<optab><mode>)
14384 (*cond_<optab><mode>_2, *cond_<optab><mode_4)
14385 (*cond_<optab><mode>_any): New SVE_COND_FP_TERNARY patterns.
14387 2018-07-12 Richard Sandiford <richard.sandiford@linaro.org>
14389 * target.def (preferred_else_value): New target hook.
14390 * doc/tm.texi.in (TARGET_PREFERRED_ELSE_VALUE): New hook.
14391 * doc/tm.texi: Regenerate.
14392 * targhooks.h (default_preferred_else_value): Declare.
14393 * targhooks.c (default_preferred_else_value): New function.
14394 * internal-fn.h (conditional_internal_fn_code): Declare.
14395 * internal-fn.c (FOR_EACH_CODE_MAPPING): New macro.
14396 (get_conditional_internal_fn): Use it.
14397 (conditional_internal_fn_code): New function.
14398 * gimple-match.h (gimple_match_cond): New struct.
14399 (gimple_match_op): Add a cond member function.
14400 (gimple_match_op::gimple_match_op): Update all forms to take a
14402 * genmatch.c (expr::gen_transform): Use the same condition as res_op
14403 for the suboperation, but don't specify a particular else_value.
14404 * tree-ssa-sccvn.c (vn_nary_simplify, vn_reference_lookup_3)
14405 (visit_nary_op, visit_reference_op_load): Pass
14406 gimple_match_cond::UNCOND to the gimple_match_op constructor.
14407 * gimple-match-head.c: Include tree-eh.h
14408 (convert_conditional_op): New function.
14409 (maybe_resimplify_conditional_op): Likewise.
14410 (gimple_resimplify1): Call maybe_resimplify_conditional_op.
14411 (gimple_resimplify2): Likewise.
14412 (gimple_resimplify3): Likewise.
14413 (gimple_resimplify4): Likewise.
14414 (maybe_push_res_to_seq): Return null for conditional operations.
14415 (try_conditional_simplification): New function.
14416 (gimple_simplify): Call it. Pass conditions to the gimple_match_op
14418 * match.pd: Fold VEC_COND_EXPRs of an IFN_COND_* call to a new
14420 * config/aarch64/aarch64.c (aarch64_preferred_else_value): New
14422 (TARGET_PREFERRED_ELSE_VALUE): Redefine.
14424 2018-07-12 Jan Hubicka <hubicka@ucw.cz>
14426 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
14428 (hash_tree): Do not hash DECL_FCONTEXT
14429 * tree-streamer-in.c (lto_input_ts_field_decl_tree_pointers):
14430 Do not stream DECL_FCONTEXT.
14431 * tree-streamer-out.c (write_ts_field_decl_tree_pointers): Likewise.
14432 * tree.c (free_lang_data_in_decl): Free DECL_FCONTEXT.
14434 2018-07-12 Richard Biener <rguenther@suse.de>
14437 * dwarf2out.c (gen_block_die): Only output blocks when they have
14438 at least one !DECL_IGNORED_P variable.
14440 2018-07-12 Richard Biener <rguenther@suse.de>
14443 * config/gnu-user.h (GNU_USER_TARGET_NO_PTHREADS_LIB_SPEC):
14444 Remove -mieee-fp handling.
14446 2018-07-12 Richard Biener <rguenther@suse.de>
14448 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Remove bogus
14449 left-over from last patch.
14451 2018-07-12 Jakub Jelinek <jakub@redhat.com>
14453 PR tree-optimization/86492
14454 * gimple-ssa-store-merging.c
14455 (imm_store_chain_info::coalesce_immediate_stores): Call
14456 check_no_overlap even for the merge_overlapping case. Formatting fix.
14458 2018-07-12 Richard Biener <rguenther@suse.de>
14460 PR middle-end/86479
14461 * fold-const.c (fold_binary_op_with_conditional_arg): Do not
14462 move possibly trapping operations into the conditional.
14464 2018-07-12 Richard Biener <rguenther@suse.de>
14466 * tree-ssa-sccvn.c (mprts_hook_cnt): Remove.
14467 (vn_lookup_simplify_result): Remove recursion limit applied
14469 (vn_nary_build_or_lookup_1): Adjust.
14470 (try_to_simplify): Likewise.
14471 * gimple-match-head.c (gimple_resimplify1): Instead apply one
14473 (gimple_resimplify2): Likewise.
14474 (gimple_resimplify3): Likewise.
14475 (gimple_resimplify4): Likewise.
14477 2018-07-11 Jakub Jelinek <jakub@redhat.com>
14479 * config/i386/avx512bitalgintrin.h (_mm512_mask_bitshuffle_epi64_mask):
14480 Use __mmask64 type instead of __mmask8 for __M argument.
14481 * config/i386/avx512fintrin.h (_mm512_mask_xor_epi64,
14482 _mm512_maskz_xor_epi64): Use __mmask8 type instead of __mmask16 for
14484 (_mm512_mask_cmpneq_epi64_mask): Use __mmask8 type instead of
14485 __mmask16 for __M argument.
14486 (_mm512_maskz_insertf32x4, _mm512_maskz_inserti32x4,
14487 _mm512_mask_insertf32x4, _mm512_mask_inserti32x4): Cast last argument
14488 to __mmask16 instead of __mmask8.
14489 * config/i386/avx512vlintrin.h (_mm_mask_add_ps, _mm_maskz_add_ps,
14490 _mm256_mask_add_ps, _mm256_maskz_add_ps, _mm_mask_sub_ps,
14491 _mm_maskz_sub_ps, _mm256_mask_sub_ps, _mm256_maskz_sub_ps,
14492 _mm256_maskz_cvtepi32_ps, _mm_maskz_cvtepi32_ps): Use __mmask8 type
14493 instead of __mmask16 for __U argument.
14494 * config/i386/avx512vlbwintrin.h (_mm_mask_cmp_epi8_mask): Use
14495 __mmask16 instead of __mmask8 for __U argument.
14496 (_mm256_mask_cmp_epi8_mask): Use __mmask32 instead of __mmask16 for
14498 (_mm256_cmp_epi8_mask): Use __mmask32 return type instead of
14500 (_mm_mask_cmp_epu8_mask): Use __mmask16 instead of __mmask8 for __U
14502 (_mm256_mask_cmp_epu8_mask): Use __mmask32 instead of __mmask16 for
14504 (_mm256_cmp_epu8_mask): Use __mmask32 return type instead of
14506 (_mm_mask_cmp_epi16_mask): Cast last argument to __mmask8 instead
14508 (_mm256_mask_cvtepi8_epi16): Use __mmask16 instead of __mmask32 for
14510 (_mm_mask_cvtepi8_epi16): Use __mmask8 instead of __mmask32 for
14512 (_mm256_mask_cvtepu8_epi16): Use __mmask16 instead of __mmask32 for
14514 (_mm_mask_cvtepu8_epi16): Use __mmask8 instead of __mmask32 for
14516 (_mm256_mask_cmpneq_epu8_mask, _mm256_mask_cmplt_epu8_mask,
14517 _mm256_mask_cmpge_epu8_mask, _mm256_mask_cmple_epu8_mask): Change
14518 return type as well as __M argument type and all casts from __mmask8
14520 (_mm256_mask_cmpneq_epu16_mask, _mm256_mask_cmplt_epu16_mask,
14521 _mm256_mask_cmpge_epu16_mask, _mm256_mask_cmple_epu16_mask): Change
14522 return type as well as __M argument type and all casts from __mmask8
14524 (_mm256_mask_cmpneq_epi8_mask, _mm256_mask_cmplt_epi8_mask,
14525 _mm256_mask_cmpge_epi8_mask, _mm256_mask_cmple_epi8_mask): Change
14526 return type as well as __M argument type and all casts from __mmask8
14528 (_mm256_mask_cmpneq_epi16_mask, _mm256_mask_cmplt_epi16_mask,
14529 _mm256_mask_cmpge_epi16_mask, _mm256_mask_cmple_epi16_mask): Change
14530 return type as well as __M argument type and all casts from __mmask8
14532 * config/i386/avx512vbmi2vlintrin.h (_mm_mask_shrdi_epi32,
14533 _mm_mask_shldi_epi32): Cast last argument to __mmask8 instead of
14536 2018-07-11 Grazvydas Ignotas <notasas@gmail.com>
14538 * config/i386/avx512bwintrin.h: (_mm512_mask_cmp_epi8_mask,
14539 _mm512_mask_cmp_epu8_mask): Use __mmask64 type instead of __mmask32
14542 2018-07-11 Paul Koning <ni1d@arrl.net>
14544 * doc/md.texi (define_subst): Document how multiple occurrences of
14545 the same argument in the replacement pattern are handled.
14547 2018-07-11 Paul Koning <ni1d@arrl.net>
14549 * doc/extend.texi (Common Variable Attributes): Move "mode" into
14550 alphabetical order.
14551 (Common Type Attributes): Add "mode" attribute.
14553 2018-07-11 Jan Hubicka <hubicka@ucw.cz>
14555 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not
14556 stream DECL_ORIGINAL_TYPE.
14557 (DFS::DFS_write_tree_body): Drop hack handling local external decls.
14558 (hash_tree): Do not walk DECL_ORIGINAL_TYPE.
14559 * tree-streamer-in.c (lto_input_ts_decl_non_common_tree_pointers):
14560 Do not walk original type.
14561 * tree-streamer-out.c (streamer_write_chain): Drop hack handling
14563 (write_ts_decl_non_common_tree_pointers): Do not stream
14565 * tree.c (free_lang_data_in_decl): Clear DECL_ORIGINAL_TYPE.
14566 (find_decls_types_r): Do not walk DEC_ORIGINAL_TYPE.
14568 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
14570 * tree-ssa-threadupdate.c (thread_through_all_blocks): Do not jump
14571 thread twice from the same starting edge.
14573 2018-07-11 Aldy Hernandez <aldyh@redhat.com>
14575 * vr-values.c (gimple_stmt_nonzero_p): Abstract common code to...
14576 * gimple.c (gimple_call_nonnull_result_p): ...here...
14577 (gimple_call_nonnull_arg): ...and here.
14578 * gimple.h (gimple_call_nonnull_result_p): New.
14579 (gimple_call_nonnull_arg): New.
14581 2018-07-11 Richard Earnshaw <rearnsha@arm.com>
14583 * config/arm/arm-cpus.in: Move information from fpu field of each
14584 cpu definition to the isa field.
14585 * config/arm/parsecpu.awk (fpu): Delete match rule.
14586 (gen_comm_data): Don't add bits from the CPU's FPU entry.
14588 2018-07-11 Richard Biener <rguenther@suse.de>
14591 * dwarf2out.c (init_sections_and_labels): Use
14592 output_asm_line_debug_info consistently.
14593 (dwarf2out_early_finish): Likewise.
14594 (dwarf2out_finish): Remove DW_AT_stmt_list from early generated
14597 2018-07-11 Richard Biener <rguenther@suse.de>
14599 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1):
14600 Rework father_bb setting in a way to avoid propagating constants
14601 multiple times on a loop body.
14603 2018-07-10 Mark Wielaard <mark@klomp.org>
14606 * dwarf2out.c (output_macinfo_op): Fix dwarf_FORM typo in gcc_assert.
14608 2018-07-10 Richard Biener <rguenther@suse.de>
14610 * hash-map.h (hash_map::iterator::operator*): Return
14611 references to key and value.
14613 2018-07-10 Jakub Jelinek <jakub@redhat.com>
14616 * gimplify.c (find_combined_omp_for): Add DATA argument, in addition
14617 to finding the inner OMP_FOR/OMP_SIMD stmt find non-trivial wrappers,
14618 BLOCKs with BLOCK_VARs, OMP_PARALLEL in between, OMP_FOR in between.
14619 (gimplify_omp_for): For composite loops, move outer
14620 OMP_{DISTRIBUTE,TASKLOOP,FOR,PARALLEL} right around innermost
14621 OMP_FOR/OMP_SIMD if there are any non-trivial wrappers. For class
14622 iterators add any needed clauses. Allow OMP_FOR_ORIG_DECLS to contain
14623 TREE_LIST for both the original class iterator and the "last" helper
14624 var. Gimplify OMP_FOR_PRE_BODY before the outermost composite
14625 loop, remember has_decl_expr from outer composite loops for the
14626 innermost OMP_SIMD in TREE_PRIVATE bit on OMP_FOR_INIT.
14628 2018-07-09 Martin Sebor <msebor@redhat.com>
14630 PR middle-end/77357
14631 PR middle-end/86428
14632 * builtins.c (c_strlen): Avoid out-of-bounds warnings when
14633 accessing implicitly initialized array elements.
14634 * expr.c (string_constant): Handle string initializers of
14635 character arrays within aggregates.
14636 * gimple-fold.c (fold_array_ctor_reference): Add argument.
14637 Store element offset. As a special case, handle zero size.
14638 (fold_nonarray_ctor_reference): Same.
14639 (fold_ctor_reference): Add argument. Store subobject offset.
14640 * gimple-fold.h (fold_ctor_reference): Add argument.
14642 2018-07-09 Paul Koning <ni1d@arrl.net>
14644 * config/pdp11/pdp11.c (pdp11_addr_cost): New function.
14645 (pdp11_insn_cost): New function.
14646 (pdp11_md_asm_adjust): New function.
14647 (TARGET_INVALID_WITHIN_DOLOOP): Define.
14648 (pdp11_rtx_costs): Update to match machine better.
14649 (output_addr_const_pdp11): Correct format mismatch warnings.
14650 * config/pdp11/pdp11.h (SLOW_BYTE_ACCESS): Correct definition.
14651 * config/pdp11/pdp11.md: General change to add base_cost and/or
14652 length attributes for use by new pdp11_insn_cost function.
14653 (MIN_BRANCH): Correct definition.
14655 (doloop_end): Use standard pattern name for looping pattern.
14656 (doloop_end_nocc): New.
14657 (movsf): Add another constraint alternative.
14658 (zero_extendqihi2): Add constraint alternatives for not in place
14660 (zero_extendhisi2): Remove.
14661 (shift patterns): Add CC handling variants.
14665 (define_peephole2): New peephole to recognize mov that sets CC for
14668 2018-07-09 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
14670 * config/sparc/sparc.c (sparc_fold_builtin) <SPARC_BUILTIN_PDIST,
14671 SPARC_BUILTIN_PDISTN>: Adapt for signature change of wi::neg,
14674 2018-07-09 Jakub Jelinek <jakub@redhat.com>
14677 * real.c (real_nextafter): Return true if result is denormal.
14679 2018-07-09 Martin Liska <mliska@suse.cz>
14681 * common.opt: Add back wrongly removed attribute.
14683 2018-07-09 Richard Biener <rguenther@suse.de>
14686 * dwarf2out.c (gen_block_die): For an early generated DIE
14687 always output high/low PC attributes.
14689 2018-07-09 Tom de Vries <tdevries@suse.de>
14691 * cfgexpand.c (expand_debug_source_expr): Handle VAR_DECL.
14692 * tree-inline.c (remap_ssa_name): Handle default def ssa_name mapping
14693 onto VAR_DECL with abstract origin.
14695 2018-07-07 Jim Wilson <jimw@sifive.com>
14697 * config/riscv/riscv.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): New.
14699 2018-07-07 Tom de Vries <tdevries@suse.de>
14701 * tree-dfa.c (dump_enumerated_decls): Handle cfun->cfg == NULL.
14703 2018-07-07 Aldy Hernandez <aldyh@redhat.com>
14705 * tree-vrp.c (vrp_int_const_binop): Change overflow type to
14707 (combine_bound): Use wide-int overflow calculation instead of
14709 * calls.c (maybe_warn_alloc_args_overflow): Change overflow type to
14711 * fold-const.c (int_const_binop_2): Same.
14712 (extract_muldiv_1): Same.
14713 (fold_div_compare): Same.
14714 (fold_abs_const): Same.
14716 * poly-int.h (add): Same.
14720 * predict.c (predict_iv_comparison): Same.
14721 * profile-count.c (slow_safe_scale_64bit): Same.
14722 * simplify-rtx.c (simplify_const_binary_operation): Same.
14723 * tree-chrec.c (tree_fold_binomial): Same.
14724 * tree-data-ref.c (split_constant_offset_1): Same.
14725 * tree-if-conv.c (idx_within_array_bound): Same.
14726 * tree-scalar-evolution.c (iv_can_overflow_p): Same.
14727 * tree-ssa-phiopt.c (minmax_replacement): Same.
14728 * tree-vect-loop.c (is_nonwrapping_integer_induction): Same.
14729 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Same.
14730 * vr-values.c (vr_values::adjust_range_with_scev): Same.
14731 * wide-int.cc (wi::add_large): Same.
14732 (wi::mul_internal): Same.
14733 (wi::sub_large): Same.
14734 (wi::divmod_internal): Same.
14735 * wide-int.h: Change overflow type to overflow_type for neg, add,
14736 mul, smul, umul, div_trunc, div_floor, div_ceil, div_round,
14737 mod_trunc, mod_ceil, mod_round, add_large, sub_large,
14738 mul_internal, divmod_internal.
14739 (overflow_type): New enum.
14740 (accumulate_overflow): New.
14742 2018-07-06 Kugan Vivekanandarajah <kugan.vivekanandarajah@linaro.org>
14744 * tree-ssa-phiopt.c (cond_removal_in_popcount_pattern): New.
14745 (tree_ssa_phiopt_worker): Call cond_removal_in_popcount_pattern.
14747 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
14749 * tree-ssa-loop-niter.c (number_of_iterations_popcount): If popcount
14750 argument is checked for zero before entering loop, avoid checking again.
14752 2018-07-06 Kugan Vivekanandarajah <kuganv@linaro.org>
14754 * gimplify.h (generic_expr_could_trap_p): Set as global function.
14755 * gimplify.h (generic_expr_could_trap_p): Likwise.
14756 * tree-scalar-evolution.c (expression_expensive_p): Handle COND_EXPR.
14758 2018-07-06 Jakub Jelinek <jakub@redhat.com>
14760 PR tree-optimization/86401
14761 * fold-const.c (fold_binary_loc) <case BIT_AND_EXPR>: Move the
14762 ((A & N) + B) & M -> (A + B) & M etc. optimization into ...
14763 (fold_bit_and_mask): ... here. New helper function for match.pd.
14764 * fold-const.h (fold_bit_and_mask): Declare.
14765 * match.pd (((A & N) + B) & M -> (A + B) & M): New optimization.
14767 2018-07-06 Peter Bergner <bergner@linux.ibm.com>
14770 * target.def (translate_mode_attribute): New hook.
14771 * targhooks.h (default_translate_mode_attribute): Declare.
14772 * targhooks.c (default_translate_mode_attribute): New function.
14773 * doc/tm.texi.in (TARGET_TRANSLATE_MODE_ATTRIBUTE): New hook.
14774 * doc/tm.texi: Regenerate.
14775 * config/rs6000/rs6000.c (TARGET_TRANSLATE_MODE_ATTRIBUTE): Define.
14776 (rs6000_translate_mode_attribute): New function.
14778 2018-07-06 Paul Koning <ni1d@arrl.net>
14780 * doc/md.texi (define_split): Document DONE and FAIL.
14781 (define_peephole2): Ditto.
14783 2018-07-05 Jeff Law <law@redhat.com>
14785 PR tree-optimization/86010
14786 * tree-ssa-dse.c (compute_trims): More aggressively trim at
14787 both the head and tail of mem* and str* calls.
14789 2018-07-05 Jim Wilson <jimw@sifive.com>
14791 * config.gcc (riscv*-*-*): When setting xlen, handle riscv-*.
14793 2018-07-05 Indu Bhagat <indu.bhagat@oracle.com>
14795 * config/aarch64/aarch64-simd.md: correct flags text for
14796 MIN_EXPR replacement.
14798 2018-07-05 James Clarke <jrtc27@jrtc27.com>
14800 * configure: Regenerated.
14802 2018-07-05 Carl Love <cel@us.ibm.com>
14804 * config/rs6000/rs6000-c.c: Map ALTIVEC_BUILTIN_VEC_UNPACKH for
14805 float argument to VSX_BUILTIN_DOUBLEH_V4SF.
14806 Map ALTIVEC_BUILTIN_VEC_UNPACKL for float argument to
14807 VSX_BUILTIN_DOUBLEL_V4SF.
14809 2018-07-05 Martin Sebor <msebor@redhat.com>
14812 * tree-ssa-strlen.c (maybe_set_strlen_range): Use type size rather
14813 than its domain to compute its the upper bound of a char array.
14815 2018-07-05 Nathan Sidwell <nathan@acm.org>
14817 Replace NO_IMPLICIT_EXTERN_C with SYSTEM_IMPLICIT_EXTERN_C.
14818 * doc/cpp.texi: Update comment.
14819 * doc/tm.texi: Rebuilt.
14820 * doc/tm.texi.in (NO_IMPLICIT_EXTERN_C): Replace with ...
14821 (SYSTEM_IMPLICIT_EXTERN_C): ... this, opposite sense.
14822 * doc/extend.texi (Backwards Compatibility): Clarify it is system
14823 headers affected by extern "C".
14824 * system.h: Poison NO_IMPLICIT_EXTERN_C.
14825 * config/alpha/alpha.h, config/arm/uclinux-elf.h,
14826 config/bfin/elf.h, config/cris/cris.h, config/darwin.h,
14827 config/dragonfly.h, config/freebsd.h, config/gnu-user.h,
14828 config/i386/cygming.h, config/i386/djgpp.h, config/i386/nto.h,
14829 config/ia64/hpux.h, config/lm32/lm32.h, config/lm32/uclinux-elf.h,
14830 config/lynx.h, config/mips/elf.h, config/mmix/mmix.h,
14831 config/netbsd.h, config/pa/pa-hpux.h, config/powerpcspe/sysv4.h,
14832 config/riscv/elf.h, config/rs6000/sysv4.h, config/rtems.h,
14833 config/s390/tpf.h, config/sh/newlib.h, config/sol2.h,
14834 config/sparc/openbsd64.h, config/sparc/sp-elf.h,
14835 config/sparc/sp64-elf.h, config/spu/spu.h,
14836 config/stormy16/stormy16.h, config/v850/v850.h,
14837 config/visium/visium.h, config/vx-common.h, config/xtensa/elf.h: Don't
14838 define NO_IMPLICIT_EXTERN_C.
14839 * config/rs6000/aix.h: Set SYSTEM_IMPLICIT_EXTERN_C.
14841 2018-07-05 Tamar Christina <tamar.christina@arm.com>
14844 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
14845 instead of GET_MODE_SIZE when comparing Units.
14847 2018-07-05 Tamar Christina <tamar.christina@arm.com>
14850 * rtlanal.c (set_noop_p): Constrain on mode change,
14851 include hard-reg-set.h
14853 2018-07-05 Tamar Christina <tamar.christina@arm.com>
14855 * config/aarch64/aarch64.c (aarch64_expand_movmem): Fix mode size.
14857 2018-07-05 Jakub Jelinek <jakub@redhat.com>
14860 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
14863 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
14865 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
14867 2018-07-04 Maxim Ostapenko <m.ostapenko@samsung.com>
14870 * config/gnu-user.h (LIBASAN_EARLY_SPEC): Pass -lstdc++ for static
14872 * gcc.c: Do not pass LIBUBSAN_SPEC if ASan is enabled with UBSan.
14874 2018-07-04 Eric Botcazou <ebotcazou@adacore.com>
14876 PR middle-end/86380
14877 * expmed.c (choose_multiplier): Fix incorrect comparison with mask.
14879 2018-07-04 Aldy Hernandez <aldyh@redhat.com>
14881 * tree-vrp.c (extract_range_from_binary_expr_1): Initialize
14882 neg_*_op* variables.
14884 2018-07-04 Martin Liska <mliska@suse.cz>
14886 * tree-switch-conversion.c: Define
14887 max_ratio_for_speed and max_ratio_for_size constants.
14889 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
14890 Martin Liska <mliska@suse.cz>
14892 PR middle-end/66240
14895 * common.opt: Rename align options with 'str_' prefix.
14896 * common/config/i386/i386-common.c (set_malign_value): New
14898 (ix86_handle_option): Use it to set -falign-* options/
14899 * config/aarch64/aarch64-protos.h (struct tune_params): Change
14900 type from int to string.
14901 * config/aarch64/aarch64.c: Update default values from int
14903 * config/alpha/alpha.c (alpha_override_options_after_change):
14905 * config/arm/arm.c (arm_override_options_after_change_1): Likewise.
14906 * config/i386/dragonfly.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14907 max skip conditionally.
14908 * config/i386/freebsd.h (SUBALIGN_LOG): New.
14909 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14910 max skip conditionally.
14911 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14912 max skip conditionally.
14913 * config/i386/gnu-user.h (SUBALIGN_LOG): New.
14914 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14915 max skip conditionally.
14916 * config/i386/i386.c (struct ptt): Change type from int to
14918 (ix86_default_align): Set default values.
14919 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Print
14920 max skip conditionally.
14921 * config/i386/iamcu.h (SUBALIGN_LOG): New.
14922 (ASM_OUTPUT_MAX_SKIP_ALIGN):
14923 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN):
14924 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14925 max skip conditionally.
14926 * config/i386/openbsdelf.h (SUBALIGN_LOG): New.
14927 (ASM_OUTPUT_MAX_SKIP_ALIGN) Print max skip conditionally.:
14928 * config/i386/x86-64.h (SUBALIGN_LOG): New.
14929 (ASM_OUTPUT_MAX_SKIP_ALIGN): Print
14930 max skip conditionally.
14931 (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
14932 * config/ia64/ia64.c (ia64_option_override): Set default values
14933 for alignment options.
14934 * config/m68k/m68k.c: Handle new str_align_* options.
14935 * config/mips/mips.c (mips_set_compression_mode): Change
14937 (mips_option_override): Set default values for options.
14938 * config/powerpcspe/powerpcspe.c (rs6000_option_override_internal):
14940 * config/rs6000/rs6000.c (rs6000_option_override_internal):
14942 * config/rx/rx.c (rx_option_override): Likewise.
14943 * config/rx/rx.h (JUMP_ALIGN): Use align_jumps_log.
14944 (LABEL_ALIGN): Use align_labels_log.
14945 (LOOP_ALIGN): Use align_loops_align.
14946 * config/s390/s390.c (s390_asm_output_function_label): Use new
14948 * config/sh/sh.c (sh_override_options_after_change):
14949 Change type of constants.
14950 * config/spu/spu.c (spu_sched_init): Likewise.
14951 * config/sparc/sparc.c (sparc_option_override): Set default
14952 values for options.
14953 * config/visium/visium.c (visium_option_override): Likewise.
14954 * config/visium/visium.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Do not
14955 emit p2align format with last argument if it's not needed.
14956 * doc/invoke.texi: Document extended format of -falign-*.
14957 * final.c: Use align_labels alignment.
14958 * flags.h (struct target_flag_state): Change type to use
14960 (struct align_flags_tuple): New.
14961 (struct align_flags): Likewise.
14962 (align_loops_log): Redefine macro to use new types.
14963 (align_loops_max_skip): Redefine macro to use new types.
14964 (align_jumps_log): Redefine macro to use new types.
14965 (align_jumps_max_skip): Redefine macro to use new types.
14966 (align_labels_log): Redefine macro to use new types.
14967 (align_labels_max_skip): Redefine macro to use new types.
14968 (align_functions_log): Redefine macro to use new types.
14969 (align_loops): Redefine macro to use new types.
14970 (align_jumps): Redefine macro to use new types.
14971 (align_labels): Redefine macro to use new types.
14972 (align_functions): Redefine macro to use new types.
14973 (align_functions_max_skip): Redefine macro to use new types.
14974 (align_loops_value): New macro.
14975 (align_jumps_value): New macro.
14976 (align_labels_value): New macro.
14977 (align_functions_value): New macro.
14978 * function.c (invoke_set_current_function_hook): Propagate
14979 alignment values from flags to global variables default in
14981 * ipa-icf.c (sem_function::equals_wpa): Use
14982 cl_optimization_option_eq instead of memcmp.
14983 * lto-streamer.h (cl_optimization_stream_out): Support streaming
14985 (cl_optimization_stream_in): Likewise.
14986 * optc-save-gen.awk: Support strings in cl_optimization.
14987 * opth-gen.awk: Likewise.
14988 * opts.c (finish_options): Remove error checking of invalid
14990 (MAX_CODE_ALIGN): Remove.
14991 (MAX_CODE_ALIGN_VALUE): Likewise.
14992 (parse_and_check_align_values): New function.
14993 (check_alignment_argument): Likewise.
14994 (common_handle_option): Use check_alignment_argument.
14995 * opts.h (parse_and_check_align_values): Declare.
14996 * toplev.c (init_alignments): Remove.
14997 (read_log_maxskip): New.
14998 (parse_N_M): Likewise.
14999 (parse_alignment_opts): Likewise.
15000 (backend_init_target): Remove usage of init_alignments.
15001 * toplev.h (parse_alignment_opts): Declare.
15002 * tree-streamer-in.c (streamer_read_tree_bitfields): Add new
15004 * tree-streamer-out.c (streamer_write_tree_bitfields): Likewise.
15005 * tree.c (cl_option_hasher::equal): New.
15006 * varasm.c: Use new global macros.
15008 2018-07-04 Denys Vlasenko <dvlasenk@redhat.com>
15010 * config/i386/dragonfly.h: (ASM_OUTPUT_MAX_SKIP_ALIGN):
15011 Use a simpler align directive also if MAXSKIP = ALIGN-1.
15012 * config/i386/gas.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15013 * config/i386/lynx.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15014 * config/i386/netbsd-elf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15015 * config/i386/i386.h (ASM_OUTPUT_MAX_SKIP_PAD): Likewise.
15016 * config/i386/freebsd.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Remove "If N
15017 is large, do at least 8 byte alignment" code. Add SUBALIGN_LOG
15018 define. Use a simpler align directive also if MAXSKIP = ALIGN-1.
15019 * config/i386/gnu-user.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15020 * config/i386/iamcu.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15021 * config/i386/openbsdelf.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15022 * config/i386/x86-64.h (ASM_OUTPUT_MAX_SKIP_ALIGN): Likewise.
15024 2018-07-04 Martin Liska <mliska@suse.cz>
15025 Jonathan Wakely <jwakely@redhat.com>
15027 * coverage.c: Use correct type.
15028 * doc/invoke.texi: Language correction.
15030 2018-07-03 H.J. Lu <hongjiu.lu@intel.com>
15033 * config/i386/i386.c (rest_of_insert_endbranch): Also generate
15034 ENDBRANCH for non-tail call which may return via indirect branch.
15035 * doc/extend.texi: Document indirect_return attribute.
15037 2018-07-03 Martin Sebor <msebor@redhat.com>
15039 PR tree-optimization/86274
15040 * gimple-ssa-sprintf.c (fmtresult::type_max_digits): Verify
15042 (format_floating): Correct handling of infinities and NaNs.
15044 2018-07-03 Martin Sebor <msebor@redhat.com>
15046 * print-tree.c (print_real_cst): New function.
15047 (print_node_brief): Call it.
15048 (print_node): Ditto.
15050 2018-07-03 Jeff Law <law@redhat.com>
15052 * config/h8300/h8300.md (logical<mode>3_sn, logical<mode>3): Merge
15053 into a single pattern.
15055 * config/h8300/h8300.md (ors code_iterator): New.
15056 (bsetqi_msx, bnotqi_msx patterns and splitters): Consolidate into
15057 a single pattern and single splitter.
15058 (bsethi_msx, bnothi_msx patterns): Consolidate into a single pattern.
15059 (iorqi3_1, xorqi3_1): Likewise.
15060 (iorqi3, xorqi3 expanders): Similarly.
15062 * config/h8300/h8300.md (movmd_internal_normal): Consolidated with
15063 (movmd_internal) into a single pattern using the P mode iterator.
15064 (movmd splitters): Similarly.
15065 (stpcpy_internal_normal, stpcpy_internal): Similarly for thes patterns.
15066 (movsd splitters): Similarly.
15068 * config/h8300/h8300.c (h8300_insn_length_from_table): Consolidate
15069 ADDB, ADDW and ADDL into a single ADD attribute which selects the
15070 right table based on the size of the operand.
15071 * config/h8300/h8300.md (length_table): Corresponding changes. All
15072 references to "addb", "addw" and "addl" changed to "add".
15073 (btst patterns): Merge two variants into a single pattern.
15074 (tstqi, tsthi): Likewise.
15075 (addhi3_incdec, addsi3_incdec): Likewise.
15076 (subhi3_h8300hs, subsi3_h8300hs): Likewise.
15077 (mulhi3, mulsi3): Likewise.
15078 (udivhi3, udivsi3): Likewise.
15079 (divhi3, divsi3): Likewise.
15080 (andorqi3, andorhi3, andorsi3): Likewise.
15082 2018-07-03 Uros Bizjak <ubizjak@gmail.com>
15085 * config/i386/sse.md (uavg<mode>3_ceil): New expander.
15086 (<sse2_avx2>_uavg<mode>3<mask_name>): Simplify expander.
15088 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15090 PR tree-optimization/85694
15091 * config/aarch64/iterators.md (HADD, RHADD): New int iterators.
15092 (u): Handle UNSPEC_SHADD, UNSPEC_UHADD, UNSPEC_SRHADD and
15094 * config/aarch64/aarch64-simd.md (<u>avg<mode>3_floor)
15095 (<u>avg<mode>3_ceil): New patterns.
15097 2018-07-03 David Malcolm <dmalcolm@redhat.com>
15099 * gcc.dg/vect/slp-perm-1.c: Remove "note: " prefix from
15100 scan-tree-dump directive.
15101 * gcc.dg/vect/slp-perm-2.c: Likewise.
15102 * gcc.dg/vect/slp-perm-3.c: Likewise.
15103 * gcc.dg/vect/slp-perm-5.c: Likewise.
15104 * gcc.dg/vect/slp-perm-6.c: Likewise.
15105 * gcc.dg/vect/slp-perm-7.c: Likewise.
15106 * gcc.dg/vect/slp-perm-8.c: Likewise.
15108 2018-07-03 Marek Polacek <polacek@redhat.com>
15110 PR middle-end/86202
15111 * gimple-fold.c (size_must_be_zero_p): Check the type of the size.
15113 2018-07-03 Richard Biener <rguenther@suse.de>
15116 * tree-ssa-structalias.c (find_func_clobbers): Properly
15117 handle indirect calls.
15119 2018-07-03 Jeff Law <law@redhat.com>
15121 * config/h8300/h8300.md (HSI, QHSI, QHSIF): New mode iterators.
15122 (shifts): New code iterator.
15123 (movqi, movhi, movsi, movsf expanders): Consolidate into a single
15124 expander. Fix HImode handling on H8/SX.
15125 (addqi3, addhi3, addsi3 expanders): Consolidate into a single expander.
15126 (subqi3, subhi3, subsi3 expanders): Likewise.
15127 (andqi3, andhi3, andsi3 expanders): Likewise.
15128 (iorqi3, iorhi3, iorsi3 expanders): Likewise.
15129 (xorqi3, xorhi3, xorsi3 expanders): Likewise.
15130 (negqi2, neghi2, negsi2, negsf2 expanders): Likewise.
15131 (one_cmplqi2, one_cmplhi2, one_cmplsi2): Likewise.
15132 (zero_extendqihi2, zero_extendqisi2): Likewise.
15133 (extendqihi2, extendqisi2): Likewise.
15134 (rotlqi3, rotlhi3, rotlsi3): Likewise.
15135 (neghi2_h8300, negsi2_h8300): Likewise for these patterns.
15136 (rotlqi3_1, rotlhi3_1): Likewise.
15137 (logicalhi3_sn, logicalsi3_sn): Likewise.
15138 (logicalhi3, logicalsi3): Likewise.
15140 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15142 * tree-vect-patterns.c (vect_recog_rotate_pattern)
15143 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
15144 (vect_recog_mixed_size_cond_pattern, adjust_bool_pattern_cast)
15145 (adjust_bool_pattern, vect_recog_bool_pattern): Pass the vector
15146 type to append_pattern_def_seq instead of creating a stmt_vec_info
15148 (build_mask_conversion): Likewise. Remove vinfo argument.
15149 (vect_add_conversion_to_patterm): Likewise, renaming to...
15150 (vect_add_conversion_to_pattern): ...this.
15151 (vect_recog_mask_conversion_pattern): Update call to
15152 build_mask_conversion. Pass the vector type to
15153 append_pattern_def_seq here too.
15154 (vect_recog_gather_scatter_pattern): Update call to
15155 vect_add_conversion_to_pattern.
15157 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15159 * tree-vect-patterns.c (new_pattern_def_seq): Delete.
15160 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
15161 (vect_recog_widen_op_pattern, vect_recog_over_widening_pattern)
15162 (vect_recog_rotate_pattern, vect_synth_mult_by_constant): Don't set
15163 STMT_VINFO_PATTERN_DEF_SEQ to null here.
15164 (vect_recog_pow_pattern, vect_recog_vector_vector_shift_pattern)
15165 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern): Use
15166 append_pattern_def_seq instead of new_pattern_def_seq.
15167 (vect_recog_divmod_pattern): Do both of the above.
15168 (vect_pattern_recog_1): Assert that STMT_VINO_PATTERN_DEF_SEQ
15171 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15173 * tree-vect-patterns.c (vect_recog_dot_prod_pattern):
15174 (vect_recog_sad_pattern, vect_recog_widen_op_pattern)
15175 (vect_recog_widen_mult_pattern, vect_recog_pow_pattern):
15176 (vect_recog_widen_sum_pattern, vect_recog_over_widening_pattern)
15177 (vect_recog_average_pattern, vect_recog_cast_forwprop_pattern)
15178 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
15179 (vect_recog_vector_vector_shift_pattern, vect_synth_mult_by_constant)
15180 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
15181 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
15182 (vect_recog_mask_conversion_pattern): Replace vec<gimple *>
15183 parameter with a single stmt_vec_info.
15184 (vect_recog_func_ptr): Likewise.
15185 (vect_recog_gather_scatter_pattern): Likewise, folding in...
15186 (vect_try_gather_scatter_pattern): ...this.
15187 (vect_pattern_recog_1): Remove stmts_to_replace and just pass
15188 the stmt_vec_info of the statement to be matched. Don't clear
15189 STMT_VINFO_RELATED_STMT.
15190 (vect_pattern_recog): Update call accordingly.
15192 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15194 PR tree-optimization/85694
15195 * doc/md.texi (avgM3_floor, uavgM3_floor, avgM3_ceil)
15196 (uavgM3_ceil): Document new optabs.
15197 * doc/sourcebuild.texi (vect_avg_qi): Document new target selector.
15198 * internal-fn.def (IFN_AVG_FLOOR, IFN_AVG_CEIL): New internal
15200 * optabs.def (savg_floor_optab, uavg_floor_optab, savg_ceil_optab)
15201 (savg_ceil_optab): New optabs.
15202 * tree-vect-patterns.c (vect_recog_average_pattern): New function.
15203 (vect_vect_recog_func_ptrs): Add it.
15204 * tree-vect-stmts.c (vectorizable_call): Get the type of the zero
15205 constant directly from the associated lhs.
15207 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15209 * tree-vect-patterns.c (vect_split_statement): New function.
15210 (vect_convert_input): Use it to try to split an existing cast.
15212 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15214 * poly-int.h (print_hex): New function.
15215 * dumpfile.h (dump_dec, dump_hex): Declare.
15216 * dumpfile.c (dump_dec, dump_hex): New poly_wide_int functions.
15217 * tree-vectorizer.h (_stmt_vec_info): Add min_output_precision,
15218 min_input_precision, operation_precision and operation_sign.
15219 * tree-vect-patterns.c (vect_get_range_info): New function.
15220 (vect_same_loop_or_bb_p, vect_single_imm_use)
15221 (vect_operation_fits_smaller_type): Delete.
15222 (vect_look_through_possible_promotion): Add an optional
15223 single_use_p parameter.
15224 (vect_recog_over_widening_pattern): Rewrite to use new
15225 stmt_vec_info infomration. Handle one operation at a time.
15226 (vect_recog_cast_forwprop_pattern, vect_narrowable_type_p)
15227 (vect_truncatable_operation_p, vect_set_operation_type)
15228 (vect_set_min_input_precision): New functions.
15229 (vect_determine_min_output_precision_1): Likewise.
15230 (vect_determine_min_output_precision): Likewise.
15231 (vect_determine_precisions_from_range): Likewise.
15232 (vect_determine_precisions_from_users): Likewise.
15233 (vect_determine_stmt_precisions, vect_determine_precisions): Likewise.
15234 (vect_vect_recog_func_ptrs): Put over_widening first.
15236 (vect_pattern_recog): Call vect_determine_precisions.
15238 2018-07-03 Richard Sandiford <richard.sandiford@arm.com>
15240 * tree-vect-patterns.c (vect_mark_pattern_stmts): Remove pattern
15241 statements that have been replaced by further pattern statements.
15242 (vect_pattern_recog_1): Clear STMT_VINFO_PATTERN_DEF_SEQ on failure.
15244 2018-07-03 Richard Biener <rguenther@suse.de>
15246 * tree-vect-stmts.c (vect_is_simple_use): Consolidate dumping,
15247 always set *dt. Dump vectype in vectype overload.
15248 * dumpfile.h (dump_gimple_expr): New function.
15249 (dump_gimple_expr_loc): Likewise.
15250 * dumpfile.c (dump_gimple_expr): New function.
15251 (dump_gimple_expr_loc): Likewise.
15253 2018-07-02 Jeff Law <law@redhat.com>
15255 * config/h8300/h8300.md (movqi_h8300, movqi_h8300hs): Consolidate
15256 the H8/300, H8/300H and H8/S variants into a single pattern.
15257 (movhi_h8300, movqi_h8300hs): Similarly.
15258 (pushqi_h8300hs, pushhi_h8300hs): Consolidate into a single pattern.
15259 (QHI mode iterator): New.
15261 * config/h8300/h8300.md: Remove trailing whitespace.
15263 2018-07-02 Jim Wilson <jimw@sifive.com>
15265 * config/riscv/riscv.c (riscv_expand_epilogue): Use emit_jump_insn
15266 instead of emit_insn for interrupt returns.
15267 * config/riscv/riscv.md (riscv_met): Add (return) to rtl.
15268 (riscv_sret, riscv_uret): Likewise.
15270 2018-07-02 David Malcolm <dmalcolm@redhat.com>
15272 * pretty-print.c (selftest::test_pp_format): Move save and restore
15273 of quotes to class auto_fix_quotes, and add an instance.
15274 * selftest.c: Include "intl.h".
15275 (selftest::auto_fix_quotes::auto_fix_quotes): New ctor.
15276 (selftest::auto_fix_quotes::~auto_fix_quotes): New dtor.
15277 * selftest.h (selftest::auto_fix_quotes): New class.
15279 2018-07-02 Richard Henderson <richard.henderson@linaro.org>
15281 * config/aarch64/aarch64-protos.h, config/aarch64/aarch64.c
15282 (aarch64_sve_prepare_conditional_op): Remove.
15283 * config/aarch64/aarch64-sve.md (cond_<SVE_INT_BINARY><SVE_I>):
15284 Allow aarch64_simd_reg_or_zero as select operand; remove
15285 the aarch64_sve_prepare_conditional_op call.
15286 (cond_<SVE_INT_BINARY_SD><SVE_SDI>): Likewise.
15287 (cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
15288 (*cond_<SVE_INT_BINARY><SVE_I>_z): New pattern.
15289 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_z): New pattern.
15290 (*cond_<SVE_COND_FP_BINARY><SVE_F>_z): New pattern.
15291 (*cond_<SVE_INT_BINARY><SVE_I>_any): New pattern.
15292 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_any): New pattern.
15293 (*cond_<SVE_COND_FP_BINARY><SVE_F>_any): New pattern
15294 and a splitters to match all of the *_any patterns.
15295 * config/aarch64/predicates.md (aarch64_sve_any_binary_operator): New.
15297 * config/aarch64/iterators.md (SVE_INT_BINARY_REV): Remove.
15298 (SVE_COND_FP_BINARY_REV): Remove.
15299 (sve_int_op_rev, sve_fp_op_rev): New.
15300 * config/aarch64/aarch64-sve.md (*cond_<SVE_INT_BINARY><SVE_I>_0): New.
15301 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_0): New.
15302 (*cond_<SVE_COND_FP_BINARY><SVE_F>_0): New.
15303 (*cond_<SVE_INT_BINARY><SVE_I>_2): Rename, add movprfx alternative.
15304 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_2): Similarly.
15305 (*cond_<SVE_COND_FP_BINARY><SVE_F>_2): Similarly.
15306 (*cond_<SVE_INT_BINARY><SVE_I>_3): Similarly; use sve_int_op_rev.
15307 (*cond_<SVE_INT_BINARY_SD><SVE_SDI>_3): Similarly.
15308 (*cond_<SVE_COND_FP_BINARY><SVE_F>_3): Similarly; use sve_fp_op_rev.
15310 * config/aarch64/aarch64-sve.md (cond_<SVE_COND_FP_BINARY><SVE_F>):
15311 Remove match_dup 1 from the inner unspec.
15312 (*cond_<SVE_COND_FP_BINARY><SVE_F>): Likewise.
15314 * config/aarch64/aarch64.md (movprfx): New attr.
15315 (length): Default movprfx to 8.
15316 * config/aarch64/aarch64-sve.md (*mul<SVE_I>3): Add movprfx alt.
15317 (*madd<SVE_I>, *msub<SVE_I): Likewise.
15318 (*<su>mul<SVE_I>3_highpart): Likewise.
15319 (*<SVE_INT_BINARY_SD><SVE_SDI>3): Likewise.
15320 (*v<ASHIFT><SVE_I>3): Likewise.
15321 (*<su><MAXMIN><SVE_I>3): Likewise.
15322 (*<su><MAXMIN><SVE_F>3): Likewise.
15323 (*fma<SVE_F>4, *fnma<SVE_F>4): Likewise.
15324 (*fms<SVE_F>4, *fnms<SVE_F>4): Likewise.
15325 (*div<SVE_F>4): Likewise.
15327 2018-07-02 Richard Sandiford <richard.sandiford@arm.com>
15329 * tree-vect-patterns.c (vect_recog_widen_shift_pattern): Fix typo
15332 2018-07-02 Richard Biener <rguenther@suse.de>
15334 PR tree-optimization/86363
15335 * tree-ssa-sccvn.c (vn_reference_lookup_3): Check the
15336 memset argument refers to a non-variable address.
15338 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
15340 * tree-vrp.c (extract_range_from_binary_expr_1): Abstract a lot of the
15341 {PLUS,MINUS}_EXPR code to...
15342 (adjust_symbolic_bound): ...here,
15343 (combine_bound): ...here,
15344 (set_value_range_with_overflow): ...and here.
15346 2018-07-02 Aldy Hernandez <aldyh@redhat.com>
15348 * tree-vrp.c (extract_range_from_unary_expr): Abstract ABS_EXPR
15350 (extract_range_from_abs_expr): ...here.
15352 2018-07-02 Eric Botcazou <ebotcazou@adacore.com>
15354 * config/i386/i386.c (ix86_finalize_stack_frame_flags): Do not overrule
15355 -fno-omit-frame-pointer when not optimizing.
15357 2018-07-02 Martin Liska <mliska@suse.cz>
15360 * ipa-pure-const.c (malloc_candidate_p): Revert usage of ::get.
15361 (propagate_nothrow): Likewise.
15363 2018-07-02 Martin Liska <mliska@suse.cz>
15366 * ipa-inline.c (early_inliner): Revert wrongly added ::get call.
15368 2018-07-02 David Malcolm <dmalcolm@redhat.com>
15370 * dumpfile.c (dump_generic_expr_loc): Undo removal of this
15371 function in r262149, changing "loc" param from source_location to
15372 const dump_location_t &.
15373 * dumpfile.h (dump_generic_expr_loc): Undo removal of this
15374 declaration, as above.
15376 2018-07-01 Paul Koning <ni1d@arrl.net>
15378 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
15379 -munit-asm, -mgnu-asm, -mdec-asm.
15380 * config/pdp11/pdp11-protos.h (pdp11_gen_int_label): New.
15381 (pdp11_output_labelref): New.
15382 (pdp11_output_def): New.
15383 (pdp11_output_addr_vec_elt): New.
15384 * config/pdp11/pdp11.c: Use tab between opcode and operands. Use
15385 %# and %@ format codes.
15386 (pdp11_option_override): New.
15387 (TARGET_ASM_FILE_START_FILE_DIRECTIVE): Define.
15388 (pdp11_output_ident): New.
15389 (pdp11_asm_named_section): New.
15390 (pdp11_asm_init_sections): New.
15391 (pdp11_file_start): New.
15392 (pdp11_file_end): New.
15393 (output_ascii): Use .ascii/.asciz for -mdec-asm.
15394 (pdp11_asm_print_operand): Update %# and %$ for -mdec-asm. Add
15395 %o, like %c but octal.
15396 (pdp11_option_override): New.
15397 * config/pdp11/pdp11.h (TEXT_SECTION_ASM_OP): Update for
15399 (DATA_SECTION_ASM_OP): Ditto.
15400 (READONLY_DATA_SECTION_ASM_OP): New.
15401 (IS_ASM_LOGICAL_LINE_SEPARATOR): New.
15402 (ASM_GENERATE_INTERNAL_LABEL): Use new function.
15403 (ASM_OUTPUT_LABELREF): Ditto.
15404 (ASM_OUTPUT_DEF): Ditto.
15405 (ASM_OUTPUT_EXTERNAL): New.
15406 (ASM_OUTPUT_SOURCE_FILENAME): New.
15407 (ASM_OUTPUT_ADDR_VEC_ELT): Use new function.
15408 (ASM_OUTPUT_SKIP): Update for -mdec-asm.
15409 * config/pdp11/pdp11.md: Use tab between opcode and operands. Use
15410 %# and %@ format codes.
15411 * config/pdp11/pdp11.opt (mgnu-asm): New.
15412 (mdec-asm): Conflicts with -mgnu-asm and -munix-asm.
15413 (munix-asm): Conflicts with -mdec-asm and -mgnu-asm.
15414 * doc/invoke.txt (PDP-11 Options): Add -mgnu-asm.
15416 2018-07-01 Aldy Hernandez <aldyh@redhat.com>
15418 * tree-ssa-threadupdate.c (mark_threaded_blocks): Avoid
15419 dereferencing path[] beyond its length.
15421 (debug_all_paths): New.
15422 (rewire_first_differing_edge): New.
15423 (adjust_paths_after_duplication): New.
15424 (duplicate_thread_path): Call adjust_paths_after_duplication.
15426 (thread_through_all_blocks): Add new argument to
15427 duplicate_thread_path.
15429 2018-06-30 Jim Wilson <jimw@sifive.com>
15431 * config/riscv/predicates.md (p2m1_shift_operand): New.
15432 (high_mask_shift_operand): New.
15433 * config/riscv/riscv.md (lshrsi3_zero_extend_3+1): New combiner
15434 pattern using p2m1_shift_operand.
15435 (lshsi3_zero_extend_3+2): New combiner pattern using
15436 high_mask_shift_operand.
15438 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15440 * tree-vect-patterns.c (vect_get_external_def_edge): New function,
15442 (vect_recog_rotate_pattern): ...here.
15443 (vect_convert_input): Try to insert casts of invariants in the
15445 * tree-vect-loop-manip.c (vect_loop_versioning): Don't require the
15446 preheader to be empty.
15448 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15450 * tree-vect-patterns.c (append_pattern_def_seq): Take an optional
15451 vector type. If given, install it in the new statement's
15452 STMT_VINFO_VECTYPE.
15453 (vect_element_precision): New function.
15454 (vect_unpromoted_value): New struct.
15455 (vect_unpromoted_value::vect_unpromoted_value): New function.
15456 (vect_unpromoted_value::set_op): Likewise.
15457 (vect_look_through_possible_promotion): Likewise.
15458 (vect_joust_widened_integer, vect_joust_widened_type): Likewise.
15459 (vect_widened_op_tree, vect_convert_input): Likewise.
15460 (vect_convert_inputs, vect_convert_output): Likewise.
15461 (vect_recog_dot_prod_pattern): Use vect_look_through_possible_promotion
15462 to handle the optional cast of the multiplication result and
15463 vect_widened_op_tree to detect the widened multiplication itself.
15464 Do not require the input and output of promotion casts to have
15465 the same sign, but base the signedness of the operation on the
15466 input rather than the result. If the pattern includes two
15467 promotions, check that those promotions have the same sign.
15468 Do not restrict the MULT_EXPR handling to a double-width result;
15469 handle quadruple-width results and wider. Use vect_convert_inputs
15470 to convert the inputs to the common type.
15471 (vect_recog_sad_pattern): Use vect_look_through_possible_promotion
15472 to handle the optional cast of the ABS result. Also allow a sign
15473 change or a sign extension between the ABS and MINUS.
15474 Use vect_widened_op_tree to detect the widened subtraction and use
15475 vect_convert_inputs to convert the inputs to the common type.
15476 (vect_handle_widen_op_by_const): Delete.
15477 (vect_recog_widen_op_pattern): New function.
15478 (vect_recog_widen_mult_pattern): Use it.
15479 (vect_recog_widen_shift_pattern): Likewise.
15480 (vect_recog_widen_sum_pattern): Use
15481 vect_look_through_possible_promotion to handle the promoted
15484 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15486 * gimple-iterator.c (gsi_for_stmt): Add a new overload that takes
15487 the containing gimple_seq *.
15488 * gimple-iterator.h (gsi_for_stmt): Declare it.
15489 * tree-vect-patterns.c (vect_recog_dot_prod_pattern)
15490 (vect_recog_sad_pattern, vect_recog_widen_sum_pattern)
15491 (vect_recog_widen_shift_pattern, vect_recog_rotate_pattern)
15492 (vect_recog_vector_vector_shift_pattern, vect_recog_divmod_pattern)
15493 (vect_recog_mask_conversion_pattern): Remove STMT_VINFO_IN_PATTERN_P
15495 (vect_init_pattern_stmt, vect_set_pattern_stmt): New functions,
15497 (vect_mark_pattern_stmts): ...here. Handle cases in which the
15498 statement being replaced is part of an existing pattern
15499 definition sequence, inserting the new pattern statements before
15501 (vect_pattern_recog_1): Don't return a bool. If the statement
15502 is already part of a pattern, instead apply pattern matching
15503 to the pattern definition statements. Don't clear the
15504 STMT_VINFO_RELATED_STMT if is_pattern_stmt_p.
15505 (vect_pattern_recog): Don't break after the first match;
15506 continue processing the pattern definition statements instead.
15507 Don't bail out for STMT_VINFO_IN_PATTERN_P here.
15509 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15511 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
15512 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
15513 (vect_recog_widen_sum_pattern): Use it.
15515 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15517 * tree-vect-loop.c (vectorizable_reduction): Assert that the
15518 phi is not a pattern statement and has not been replaced by
15519 a pattern statement.
15520 * tree-vect-patterns.c (type_conversion_p): Don't check
15521 STMT_VINFO_IN_PATTERN_P.
15522 (vect_recog_vector_vector_shift_pattern): Likewise.
15523 (vect_recog_dot_prod_pattern): Expect vect_is_simple_use to return
15524 the pattern statement rather than the original statement; check
15525 directly for a WIDEN_MULT_EXPR here.
15526 * tree-vect-slp.c (vect_get_and_check_slp_defs): Expect
15527 vect_is_simple_use to return the pattern statement rather
15528 than the original statement; use is_pattern_stmt_p to check
15529 for such a pattern statement.
15530 * tree-vect-stmts.c (process_use): Expect vect_is_simple_use
15531 to return the pattern statement rather than the original statement;
15532 don't do the same transformation here.
15533 (vect_is_simple_use): If the defining statement has been replaced
15534 by a pattern statement, return the pattern statement instead.
15535 Remove the corresponding (local) transformation from the vectype
15538 2018-06-30 Richard Sandiford <richard.sandiford@arm.com>
15540 * tree-vectorizer.h (vect_is_simple_use): Move the gimple ** to the
15541 end and default to null.
15542 * tree-vect-loop.c (vect_create_epilog_for_reduction)
15543 (vectorizable_reduction): Update calls accordingly, dropping the
15544 gimple ** argument if the passed-back statement isn't needed.
15545 * tree-vect-patterns.c (vect_get_internal_def, type_conversion_p)
15546 (vect_recog_rotate_pattern): Likewise.
15547 (vect_recog_mask_conversion_pattern): Likewise.
15548 * tree-vect-slp.c (vect_get_and_check_slp_defs): Likewise.
15549 (vect_mask_constant_operand_p): Likewise.
15550 * tree-vect-stmts.c (is_simple_and_all_uses_invariant, process_use):
15551 (vect_model_simple_cost, vect_get_vec_def_for_operand): Likewise.
15552 (get_group_load_store_type, get_load_store_type): Likewise.
15553 (vect_check_load_store_mask, vect_check_store_rhs): Likewise.
15554 (vectorizable_call, vectorizable_simd_clone_call): Likewise.
15555 (vectorizable_conversion, vectorizable_assignment): Likewise.
15556 (vectorizable_shift, vectorizable_operation): Likewise.
15557 (vectorizable_store, vect_is_simple_cond): Likewise.
15558 (vectorizable_condition, vectorizable_comparison): Likewise.
15559 (get_same_sized_vectype, vect_get_mask_type_for_stmt): Likewise.
15560 (vect_is_simple_use): Rename the def_stmt argument to def_stmt_out
15561 and move it to the end. Cope with null def_stmt_outs.
15563 2018-06-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
15565 * Makefile.in (FLAGS_TO_PASS): Add RANLIB_FOR_TARGET.
15567 2018-06-29 Jeff Law <law@redhat.com>
15569 * config/v850/v850.c (v850_legitimate_address_p): Handle large
15570 displacements for TARGET_V850E2V3 and newer.
15571 (TARGET_LRA_P): Remove. Defaults to LRA now.
15572 * config/v850/v850.md (sign23byte_load): Remove.
15573 (unsign23byte_load, sign23hword_load, unsign23hword_load): Likewise.
15574 (23word_load, 23byte_store, 23hword_store, 23word_store): Likewise.
15576 2018-06-29 Martin Liska <mliska@suse.cz>
15579 * coverage.c (coverage_init): Mangle full path name.
15580 * doc/invoke.texi: Document the change.
15581 * gcov-io.c (mangle_path): New.
15582 * gcov-io.h (mangle_path): Likewise.
15583 * gcov.c (mangle_name): Use mangle_path for path mangling.
15585 2018-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15587 * config/arm/arm.c (output_move_double): Don't allow STRD instructions
15588 if starting source register is not even.
15590 2018-06-29 Martin Liska <mliska@suse.cz>
15592 PR tree-optimization/86263
15593 * tree-switch-conversion.c (switch_decision_tree::try_switch_expansion):
15594 Make edge redirection.
15596 2018-06-29 David Malcolm <dmalcolm@redhat.com>
15598 * dumpfile.c (dump_loc): Add indentation based on scope depth.
15599 (dump_scope_depth): New variable.
15600 (get_dump_scope_depth): New function.
15601 (dump_begin_scope): New function.
15602 (dump_end_scope): New function.
15603 * dumpfile.h (get_dump_scope_depth): New declaration.
15604 (dump_begin_scope): New declaration.
15605 (dump_end_scope): New declaration.
15606 (class auto_dump_scope): New class.
15607 (AUTO_DUMP_SCOPE): New macro.
15608 * tree-vectorizer.h (DUMP_VECT_SCOPE): Reimplement in terms of
15611 2018-06-29 Richard Biener <rguenther@suse.de>
15613 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences): Assert
15614 compute_all_dependences succeeds.
15615 * tree-vect-loop.c (vect_get_datarefs_in_loop): Fail early if we
15616 exceed --param loop-max-datarefs-for-datadeps.
15618 2018-06-29 Jakub Jelinek <jakub@redhat.com>
15620 * config/rs6000/t-rs6000: Append rs6000-modes.h to TM_H.
15622 2018-06-28 Uros Bizjak <ubizjak@gmail.com>
15625 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): Use
15626 alternative 0 in preferred_for_speed attribute.
15628 2018-06-28 Paul Koning <ni1d@arrl.net>
15630 * config/pdp11/pdp11-protos.h (pdp11_shift_length): New function.
15631 * config/pdp11/pdp11.c (pdp11_shift_length): New function.
15632 * config/pdp11/pdp11.h (ADJUST_INSN_LENGTH): Remove.
15633 * config/pdp11/pdp11.md: Correct "length" attribute calculation
15634 for shift insn patterns.
15636 2018-06-28 David Malcolm <dmalcolm@redhat.com>
15638 * cgraph.c (cgraph_node::get_body): Replace assignments to
15639 "dump_file" with calls to set_dump_file.
15640 * dumpfile.c (alt_dump_file): Make static, and group with...
15641 (alt_flags): ...this definition.
15642 (dumps_are_enabled): New variable.
15643 (refresh_dumps_are_enabled): New function.
15644 (set_dump_file): New function.
15645 (set_alt_dump_file): New function.
15646 (gcc::dump_manager::dump_start): Replace assignments to
15647 "dump_file" and "alt_dump_file" with calls to set_dump_file and
15649 (gcc::dump_manager::dump_finish): Likewise.
15650 * dumpfile.h (alt_dump_file): Delete decl.
15651 (dumps_are_enabled): New variable decl.
15652 (set_dump_file): New function decl.
15653 (dump_enabled_p): Rewrite in terms of new "dumps_are_enabled"
15655 * tree-nested.c (lower_nested_functions): Replace assignments to
15656 "dump_file" with calls to set_dump_file.
15658 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
15660 * tree-cfg.c (verify_gimple_in_cfg): Call verify_location on the
15661 goto_locus of each outgoing edge of each basic block.
15663 2018-06-28 Richard Biener <rguenther@suse.de>
15665 * dwarf2out.c (decl_scope_table): Remove.
15666 (push_decl_scope): Likewise.
15667 (pop_decl_scope): Likewise.
15668 (gen_type_die_for_member): Do not call push/pop_decl_scope.
15669 (gen_struct_or_union_type_die): Likewise.
15670 (gen_tagged_type_die): Likewise.
15671 (dwarf2out_init): Do not initialize decl_scope_table.
15672 (dwarf2out_c_finalize): Do not free it.
15674 2018-06-28 Richard Biener <rguenther@suse.de>
15676 * dwarf2out.c (gen_subprogram_die): Use is_unit_die when
15677 deciding whether to not re-use a DIE.
15679 2018-06-28 Richard Biener <rguenther@suse.de>
15681 * dwarf2out.c (gen_subprogram_die): Always re-use DIEs with an
15682 DW_AT_abstract_origin attribute.
15684 2018-06-28 Martin Liska <mliska@suse.cz>
15686 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
15687 Use newly introduced constants.
15688 * tree-switch-conversion.h (struct jump_table_cluster):
15689 Define max_ratio_for_size and max_ratio_for_speed.
15691 2018-06-28 Martin Liska <mliska@suse.cz>
15693 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
15694 Add new checking assert to catch invalid state.
15695 (jump_table_cluster::can_be_handled): Handle single case
15697 (jump_table_cluster::is_beneficial): Bail out for such case.
15698 (bit_test_cluster::find_bit_tests):
15699 Add new checking assert to catch invalid state.
15700 (bit_test_cluster::can_be_handled): Handle single case
15702 (bit_test_cluster::is_beneficial): Bail out for such case.
15703 (switch_decision_tree::analyze_switch_statement):
15706 2018-06-28 Martin Liska <mliska@suse.cz>
15708 * common.opt: Introduce -completion option.
15709 * gcc.c (driver_handle_option): Handle it.
15710 (driver::main): Print completions if completion
15712 * opt-suggestions.c (option_proposer::get_completions):
15714 (option_proposer::suggest_completion): Likewise.
15715 (option_proposer::find_param_completions): Likewise.
15716 (verify_autocompletions): Likewise.
15717 (test_completion_valid_options): Likewise.
15718 (test_completion_valid_params): Likewise.
15719 (in_completion_p): Likewise.
15720 (empty_completion_p): Likewise.
15721 (test_completion_partial_match): Likewise.
15722 (test_completion_garbage): Likewise.
15723 (opt_proposer_c_tests): Likewise.
15724 * opt-suggestions.h: Declare new functions.
15725 * opts.c (common_handle_option): Handle OPT__completion_.
15726 * selftest-run-tests.c (selftest::run_tests): Add
15727 opt_proposer_c_tests.
15728 * selftest.c (assert_str_startswith): New.
15729 * selftest.h (assert_str_startswith): Likewise.
15730 (opt_proposer_c_tests): New.
15731 (ASSERT_STR_STARTSWITH): Likewise.
15733 2018-06-28 Martin Liska <mliska@suse.cz>
15735 * Makefile.in: Add opt-suggestions.o.
15736 * gcc-main.c: Include opt-suggestions.h.
15737 * gcc.c (driver::driver): Likewise.
15738 (driver::~driver): Remove m_option_suggestions.
15739 (driver::build_option_suggestions): Moved to option_proposer.
15740 (driver::suggest_option): Likewise.
15741 (driver::handle_unrecognized_options): Use option_proposer.
15742 * gcc.h (class driver): Add new memver m_option_proposer.
15743 * opt-suggestions.c: New file.
15744 * opt-suggestions.h: New file.
15746 2018-06-28 Martin Liska <mliska@suse.cz>
15748 * vec.h (class auto_string_vec): New (moved from auto_argvec).
15749 (auto_string_vec::~auto_string_vec): Likewise.
15751 2018-06-28 Eric Botcazou <ebotcazou@adacore.com>
15753 * tree-inline.h (struct copy_body_data): Move remapping_type_depth and
15754 prevent_decl_creation_for_types fields up and add reset_location field.
15755 * tree-inline.c (remap_gimple_stmt): Force input_location on the new
15756 statement if id->reset_location is true.
15757 (copy_edges_for_bb): Do not set goto_locus on the new edges if
15758 id->reset_location is true.
15759 (copy_phis_for_bb): Force input_location on the arguments if
15760 id->reset_location is true.
15761 (expand_call_inline): Set id->reset_location if DECL_IGNORED_P
15762 is set on the function to be inlined.
15764 2018-06-27 Stephan Bergmann <sbergman@redhat.com>
15766 * doc/invoke.texi (Debugging Options): Add -gsplit-dwarf.
15768 2018-06-27 Dimitar Dimitrov <dimitar@dinux.eu>
15770 * lra-eliminations.c (update_reg_eliminate): Mark all spanning hard
15771 registers for Pmode.
15772 * lra-lives.c (check_pseudos_live_through_calls): Mark all spanning
15773 hard registers for the clobbered pseudo.
15775 2018-06-27 Paul Koning <ni1d@arrl.net>
15777 * common/config/pdp11/pdp11-common.c (pdp11_handle_option): Handle
15778 mutually exclusive options.
15779 * config/pdp11/constraints.md (h): New constraint.
15780 (O): Update definition to match shift code generation.
15781 (D): New constraint.
15782 * config/pdp11/pdp11-modes.def (CCNZ): Define mode.
15784 * config/pdp11/pdp11-protos.h (int_no_side_effect_operand): New
15786 (output_jump): Change arguments.
15787 (pdp11_fixed_cc_regs): New function.
15788 (pdp11_cc_mode): Ditto.
15789 (pdp11_expand_shift): Ditto.
15790 (pdp11_assemble_shift): Ditto.
15791 (pdp11_small_shift): Ditto.
15792 (pdp11_branch_cost): Remove.
15793 * config/pdp11/pdp11.c (pdp11_assemble_integer): Remove comments
15795 (pdp11_register_move_cost): Update for CC registers.
15796 (pdp11_rtx_costs): Add case for LSHIFTRT.
15797 (pdp11_output_jump): Add CCNZ mode conditional branches.
15798 (notice_update_cc_on_set): Remove.
15799 (pdp11_cc_mode): New function.
15800 (simple_memory_operand): Correct pre/post decrement case.
15801 (no_side_effect_operand): New function.
15802 (pdp11_regno_reg_class): Add CC_REGS class.
15803 (pdp11_fixed_cc_regs): New function.
15804 (pdp11_small_shift): New function.
15805 (pdp11_expand_shift): New function to expand shift insns.
15806 (pdp11_assemble_shift): New function to output shifts.
15807 (pdp11_branch_cost): Remove.
15808 (pdp11_modes_tieable_p): Make QI/HI modes tieable.
15809 * config/pdp11/pdp11.h (SIZE_TYPE): Ensure 16-bit type.
15810 (WCHAR_TYPE): Ditto.
15811 (PTRDIFF_TYPE): Ditto.
15812 (ADJUST_INSN_LENGTH): New macro.
15813 (FIXED_REGISTERS): Add CC registers.
15814 (CALL_USED_REGISTERS): Ditto.
15815 (reg_class): Ditto.
15816 (REG_CLASS_NAMES): Ditto.
15817 (REG_CLASS_CONTENTS): Ditto.
15818 (SELECT_CC_MODE): Use new function.
15819 (TARGET_FLAGS_REGNUM): New macro.
15820 (TARGET_FIXED_CONDITION_CODE_REGS): Ditto.
15821 (cc0_reg_rtx): Remove.
15822 (CC_STATUS_MDEP): Remove.
15823 (CC_STATUS_MDEFP_INIT): Remove.
15824 (CC_IN_FPU): Remove.
15825 (NOTICE_UPDATE_CC): Remove.
15826 (REGISTER_NAMES): Add CC registers.
15827 (BRANCH_COST): Change to constant 1.
15828 * config/pdp11/pdp11.md: Rewrite for CCmode condition code
15830 * config/pdp11/pdp11.opt (mbcopy): Remove.
15831 (mbcopy-builtin): Remove.
15832 (mbranch-cheap): Remove.
15833 (mbranch-expensive): Remove.
15834 * config/pdp11/predicates.md (expand_shift_operand): Update to
15835 match shift code generation.
15836 (ccnz_operator): New predicate.
15837 * doc/invoke.texi (PDP-11 Options): Remove deleted options
15838 -mbcopy, -mbcopy-builtin, -mbranch-cheap, -mbranch-expensive.
15839 Remove non-existent option -mabshi, -mno-abshi. Document mutually
15841 * doc/md.texi (PDP-11): Document new D and h constraints. Update
15842 description of O constraint.
15844 2018-06-27 Jeff Law <law@redhat.com>
15845 Austin Law <austinklaw@gmail.com>
15847 * config/v850/v850.md (addsi3_set_flags): New pattern.
15848 (subsi3_set_flags, negsi2_set_flags, andsi3_set_flags): Likewise.
15849 (iorsi3_set_flags, xorsi3_set_flags, one_cmplsi2_set_flags): Likewise.
15850 (zero_extendhisi2_v850_set_flags): Likewise.
15851 (zero_extendqisi2_v850_set_flags): Likewise.
15852 (ashlsi3_set_flags, ashlsi3_v850e2_set_flags): Likewise.
15853 (lshrsi3_set_flags, lshrsi3_v850e2_set_flags): Likewise.
15854 (ashrsi3_set_flags, ashrsi3_v850e2_set_flags): Likewise.
15856 * config/v850/v850-protos.h (notice_update_cc): Remove.
15857 * config/v850/v850.c (v850_compare_op0, v850_compare_op1): Remove.
15858 (v850_print_operand): Handle 'D' and "d".
15859 (v850_select_cc_mode): Remove ATTRIBUTE_UNUSED for last argument.
15860 Add handling of arithmetic/logical operations compared against zero.
15861 (v850_gen_float_compare): Remove ATTRIBUTE_UNUSED for last argument.
15862 Do not look at v850_compare_op, instead get mode from last argument.
15863 (v850_gen_compare): Remove
15864 (increment_stack): Use addsi3_clobber_flags to avoid splitting failure
15865 after reload for prologue insns.
15866 (expand_prologue): Account for CLOBBER of CC_REGNUM in various
15868 (construct_save_jarl): Likewise.
15869 (TARGET_FLAGS_REGNUM): Define.
15870 * config/v850/v850.h (v850_compare_op0, v850_compare_op1): Remove.
15871 (NOTICE_UPDATE_CC): Remove.
15872 * config/v850/v850.md (v850_tst1): Use (reg:CCZ CC_REGNUM) rather
15873 than cc0. Conditionalize on reload_completed.
15874 (cmpsi_insn, setfcc_insn): Likewise.
15875 (tst1 splitter): Turn into define_and_split which sets the flags
15877 (cstoresi4, cbranchsf4, cbranchdf4, cbranchsi4_insn): Likewise.
15878 (cbranchsi4, branch_normal, branch_invert): Do not expose cc0 here.
15879 (cstoresf4, cstoredf4): Clobber the flags.
15880 (cmpsi, cmpsf, cmpdf): Remove expanders.
15881 (setf_insn): Remove pattern.
15882 (addsi3): Turn into define_and_split which clobbers the flags after
15883 reload and a suitable pattern (addsi3_clobber_flags) for use after
15885 (subsi3, negsi2, andsi3, iorsi3, xorsi3, one_cmplsi2) Likewise.
15886 (ashlsi3, ashlsi3_v850e2, lshrsi3, lsh4si3_v850e2): Likewise.
15887 (ashrsi3, ashrsi3_v850e2): Likewise.
15888 (bins): Clobber the flags.
15889 (movsicc_normal_cc, movsicc_normal, movsicc_tst1): Likewise.
15890 (movsicc_tst1_revesed, sasf, swap and rotate patterns): Likewise.
15891 (fix_loop_counter, call_internal_short, call_internal_long): Likewise.
15892 (call_value_internal_short, call_value_internal_long): Likewise.
15893 (callt_save_interrupt, callt_return_interrupt): Likewise.
15894 (save_interrupt, return_interrupt): Likewise.
15895 (callt_save_all_interrupt, save_all_interrupt): Likewise.
15896 (_save_all_interrupt, callt_restore_all_interrupt): Likewise.
15897 (restore_all_interrupt, _restore_all_interrupt): Likewise.
15898 (All FP comparisons): Only allow after reload has completed.
15900 (divh, divhu): Tweak output template.
15901 (branch_z_normal, branch_z_invert): Remove
15902 (branch_nz_normal, branch_nz_invert): Likewise.
15903 (extendhisi_insn, extendqisi_insn): Do not clobber flags.
15905 * config/v850/v850-modes.def (CCZ, CCNZ): Add new modes.
15906 * config/v850/v850.c (notice_update_cc): Remove.
15907 * config/v850/v850.h (CC_OVERFLOW_UNUSABLE): Remove
15908 (CC_NO_CARRY): Likewise.
15909 (NOTICE_UPDATE_CC): Define to nothing.
15910 * config/v850/v850.md: Remove block comment on cc0 handling
15911 Remove "cc" attribute from all patterns. Remove cc_status handling
15912 from all patterns. Minor formatting fixes.
15914 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15916 * config/aarch64/aarch64-cores.def (cortex-a76): New entry.
15917 (cortex-a76.cortex-a55): Likewise.
15918 * config/aarch64/aarch64-tune.md: Regenerate.
15919 * doc/invoke.texi (AArch64 Options): Document cortex-a76 and
15920 cortex-a76.cortex-a55.
15922 2018-06-27 Jeff Law <law@redhat.com>
15924 * config/v850/t-v850 (MULTILIB_OPTIONS): Remove 8byte-align.
15925 (MULTILIB_DIRNAMES): Similarly.
15927 2018-06-27 Eric Botcazou <ebotcazou@adacore.com>
15929 * gimple.h (gimple_return_retbnd): Delete.
15930 (gimple_return_set_retbnd): Likewise.
15931 * cgraphunit.c (cgraph_node::expand_thunk): Remove call to
15932 gimple_return_set_retbnd.
15933 * gimple-pretty-print.c (dump_gimple_return): Remove call to
15934 gimple_return_retbnd and adjust.
15935 * tree-inline.h (struct copy_body_data): Remove retbnd field.
15936 * tree-inline.c (remap_gimple_stmt): Remove handling of retbnd.
15937 Explicitly return NULL in a couple more cases. Move assertion
15938 on debug statements and remove unreachable code.
15939 (reset_debug_binding): Do not test id->retbnd.
15940 (expand_call_inline): Do not set it.
15942 2018-06-27 Rasmus Villemoes <rv@rasmusvillemoes.dk>
15944 * configure.ac: Add --disable-gcov option.
15945 * configure: Regenerate.
15946 * Makefile.in: Honour @enable_gcov@.
15947 * doc/install.texi: Document --disable-gcov.
15949 2018-06-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15951 * config/arm/arm-cpus.in (cortex-a76): New entry.
15952 (cortex-a76.cortex-a55): Likewise.
15953 * config/arm/arm-tables.opt: Regenerate.
15954 * config/arm/arm-tune.md: Likewise.
15955 * config/arm/driver-arm.c (arm_cpu_table): Add Cortex-A76 entry.
15956 * doc/invoke.texi (ARM Options): Document cortex-a76 and
15957 cortex-a76.cortex-a55.
15959 2018-06-27 Tamar Christina <tamar.christina@arm.com>
15962 * config/aarch64/aarch64.md (*movhf_aarch64): Add dup v0.4h pattern.
15964 2018-06-27 Siddhesh Poyarekar <siddhesh@sourceware.org>
15966 * config/aarch64/aarch64.h (CALL_USE_REGISTERS): Fix obsolete
15968 (EPILOGUE_USES): Likewise.
15970 2018-06-26 Eric Botcazou <ebotcazou@adacore.com>
15972 * tree-inline.c (remap_location): New function extracted from...
15973 (copy_edges_for_bb): Add ID parameter. Remap goto_locus.
15974 (copy_phis_for_bb): ...here. Call remap_location.
15975 (copy_cfg_body): Adjust call to copy_edges_for_bb.
15977 2018-06-26 Aaron Sawdey <acsawdey@linux.ibm.com>
15979 * config/rs6000/rs6000-string.c (expand_block_clear): Don't use
15980 unaligned vsx for 16B memset.
15982 2018-06-26 Segher Boessenkool <segher@kernel.crashing.org>
15985 * config/rs6000/rs6000.c (rs6000_init_builtins): Do not set
15986 ieee128_float_type_node to long_double_type_node unless
15987 TARGET_LONG_DOUBLE_128 is set.
15989 2018-06-26 David Malcolm <dmalcolm@redhat.com>
15991 * cfgloop.c (get_loop_location): Convert return type from
15992 location_t to dump_user_location_t, replacing INSN_LOCATION lookups
15993 by implicit construction from rtx_insn *, and using
15994 dump_user_location_t::from_function_decl for the fallback case.
15995 * cfgloop.h (get_loop_location): Convert return type from
15996 location_t to dump_user_location_t.
15997 * cgraphunit.c (walk_polymorphic_call_targets): Update call to
15998 dump_printf_loc to pass in a dump_location_t rather than a
15999 location_t, via the gimple stmt.
16000 * coverage.c (get_coverage_counts): Update calls to
16001 dump_printf_loc to pass in dump_location_t rather than a
16003 * doc/optinfo.texi (Dump types): Convert example of
16004 dump_printf_loc from taking "locus" to taking "insn". Update
16005 description of the "_loc" calls to cover dump_location_t.
16006 * dumpfile.c: Include "backend.h", "gimple.h", "rtl.h", and
16008 (dump_user_location_t::dump_user_location_t): New constructors,
16009 from gimple *stmt and rtx_insn *.
16010 (dump_user_location_t::from_function_decl): New function.
16011 (dump_loc): Make static.
16012 (dump_gimple_stmt_loc): Convert param "loc" from location_t to
16013 const dump_location_t &.
16014 (dump_generic_expr_loc): Delete.
16015 (dump_printf_loc): Convert param "loc" from location_t to
16016 const dump_location_t &.
16017 (selftest::test_impl_location): New function.
16018 (selftest::dumpfile_c_tests): New function.
16019 * dumpfile.h: Include "profile-count.h".
16020 (class dump_user_location_t): New class.
16021 (struct dump_impl_location_t): New struct.
16022 (class dump_location_t): New class.
16023 (dump_printf_loc): Convert 2nd param from source_location to
16024 const dump_location_t &.
16025 (dump_generic_expr_loc): Delete.
16026 (dump_gimple_stmt_loc): Convert 2nd param from source_location to
16027 const dump_location_t &.
16028 * gimple-fold.c (fold_gimple_assign): Update call to
16029 dump_printf_loc to pass in a dump_location_t rather than a
16030 location_t, via the gimple stmt.
16031 (gimple_fold_call): Likewise.
16032 * gimple-loop-interchange.cc
16033 (loop_cand::analyze_iloop_reduction_var): Update for change to
16034 check_reduction_path.
16035 (tree_loop_interchange::interchange): Update for change to
16036 find_loop_location.
16037 * graphite-isl-ast-to-gimple.c (scop_to_isl_ast): Update for
16038 change in return-type of find_loop_location.
16039 (graphite_regenerate_ast_isl): Likewise.
16040 * graphite-optimize-isl.c (optimize_isl): Likewise.
16041 * graphite.c (graphite_transform_loops): Likewise.
16042 * ipa-devirt.c (ipa_devirt): Update call to dump_printf_loc to
16043 pass in a dump_location_t rather than a location_t, via the
16045 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
16046 * ipa.c (walk_polymorphic_call_targets): Likewise.
16047 * loop-unroll.c (report_unroll): Convert "locus" param from
16048 location_t to dump_location_t.
16049 (decide_unrolling): Update for change to get_loop_location's
16051 * omp-grid.c (struct grid_prop): Convert field "target_loc" from
16052 location_t to dump_user_location_t.
16053 (grid_find_single_omp_among_assignments_1): Updates calls to
16054 dump_printf_loc to pass in a dump_location_t rather than a
16055 location_t, via the gimple stmt.
16056 (grid_parallel_clauses_gridifiable): Convert "tloc" from
16057 location_t to dump_location_t. Updates calls to dump_printf_loc
16058 to pass in a dump_location_t rather than a location_t, via the
16060 (grid_inner_loop_gridifiable_p): Likewise.
16061 (grid_dist_follows_simple_pattern): Likewise.
16062 (grid_gfor_follows_tiling_pattern): Likewise.
16063 (grid_target_follows_gridifiable_pattern): Likewise.
16064 (grid_attempt_target_gridification): Convert initialization
16065 of local "grid" from memset to zero-initialization; FIXME: does
16066 this require C++11? Update call to dump_printf_loc to pass in a
16067 optinfo_location rather than a location_t, via the gimple stmt.
16068 * profile.c (read_profile_edge_counts): Updates call to
16069 dump_printf_loc to pass in a dump_location_t rather than a
16071 (compute_branch_probabilities): Likewise.
16072 * selftest-run-tests.c (selftest::run_tests): Call
16074 * selftest.h (dumpfile_c_tests): New decl.
16075 * tree-loop-distribution.c (pass_loop_distribution::execute):
16076 Update for change in return type of find_loop_location.
16077 * tree-parloops.c (parallelize_loops): Likewise.
16078 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Convert
16079 "locus" from location_t to dump_user_location_t.
16080 (canonicalize_loop_induction_variables): Likewise.
16081 * tree-ssa-loop-ivopts.c (tree_ssa_iv_optimize_loop): Update
16082 for change in return type of find_loop_location.
16083 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update call
16084 to dump_printf_loc to pass in a dump_location_t rather than a
16085 location_t, via the stmt.
16086 * tree-ssa-sccvn.c (eliminate_dom_walker::before_dom_children):
16088 * tree-vect-loop-manip.c (find_loop_location): Convert return
16089 type from source_location to dump_user_location_t.
16090 (vect_do_peeling): Update for above change.
16091 (vect_loop_versioning): Update for change in type of
16093 * tree-vect-loop.c (check_reduction_path): Convert "loc" param
16094 from location_t to dump_user_location_t.
16095 (vect_estimate_min_profitable_iters): Update for change in type
16097 * tree-vect-slp.c (vect_print_slp_tree): Convert param "loc" from
16098 location_t to dump_location_t.
16099 (vect_slp_bb): Update for change in type of vect_location.
16100 * tree-vectorizer.c (vect_location): Convert from source_location
16101 to dump_user_location_t.
16102 (try_vectorize_loop_1): Update for change in vect_location's type.
16103 (vectorize_loops): Likewise.
16104 (increase_alignment): Likewise.
16105 * tree-vectorizer.h (vect_location): Convert from source_location
16106 to dump_user_location_t.
16107 (find_loop_location): Convert return type from source_location to
16108 dump_user_location_t.
16109 (check_reduction_path): Convert 1st param from location_t to
16110 dump_user_location_t.
16111 * value-prof.c (check_counter): Update call to dump_printf_loc to
16112 pass in a dump_user_location_t rather than a location_t; update
16113 call to error_at for change in type of "locus".
16114 (check_ic_target): Update call to dump_printf_loc to
16115 pass in a dump_user_location_t rather than a location_t, via the
16118 2018-06-26 Robin Dapp <rdapp@linux.vnet.ibm.com>
16120 * config/s390/s390.h (enum processor_flags): Do not use
16121 default tune parameter when -march was specified.
16123 2018-06-26 Jakub Jelinek <jakub@redhat.com>
16126 * config/i386/i386.md (setcc + movzbl to xor + setcc peephole2s):
16127 Check reg_overlap_mentioned_p in addition to reg_set_p with the same
16130 2018-06-26 Richard Biener <rguenther@suse.de>
16132 PR tree-optimization/86287
16134 * tree-vect-loop.c (vect_transform_loop_stmt): Fix read-after-free.
16135 (vect_analyze_loop): Initialize n_stmts.
16137 2018-06-26 Richard Biener <rguenther@suse.de>
16139 PR middle-end/86271
16140 * fold-const.c (fold_convertible_p): Pointer extension
16143 2018-06-26 Alexandre Oliva <aoliva@redhat.com>
16146 * dwarf2out.c (loc_list_has_views): Adjust comments.
16147 (dw_loc_list): Split single cross-partition range with
16150 2018-06-25 Jeff Law <law@redhat.com>
16152 * common/config/v850/v850-common.c (TARGET_DEFAULT_TARGET_FLAGS): Turn
16153 on -mbig-switch by default.
16155 * config/v850/predicates.md (const_float_1_operand): Fix match_code
16157 (const_float_0_operand): Remove unused predicate.
16158 * config/v850/v850.md (define_constants): Remove UNSPEC_LOOP.
16159 (define_c_enum unspec): Add LOOP, RCP and RSQRT constants.
16160 (recipsf2): New expander. Original pattern now called
16162 (recipdf2, recipdf2_insn): Similarly.
16163 (rsqrtsf2, rsqrtsf2_insn): Similarly
16164 (rsqrtdf2, rsqrtdf2_insn): Similarly
16166 2018-06-26 Gerald Pfeifer <gerald@pfeifer.com>
16168 * ginclude/stddef.h: Remove an obsolete comment on FreeBSD 5.
16169 Simplify logic for FreeBSD (twice).
16171 2018-06-25 Martin Sebor <msebor@redhat.com>
16173 PR tree-optimization/86204
16174 * tree-ssa-strlen.c (handle_builtin_strlen): Avoid storing
16175 a strnlen result if it's less than the length of the string.
16177 2018-06-25 Martin Sebor <msebor@redhat.com>
16179 PR tree-optimization/85700
16180 * gimple-fold.c (gimple_fold_builtin_strncat): Adjust comment.
16181 * tree-ssa-strlen.c (is_strlen_related_p): Handle integer subtraction.
16182 (maybe_diag_stxncpy_trunc): Distinguish strncat from strncpy.
16184 2018-06-25 Martin Sebor <msebor@redhat.com>
16186 * doc/extend.texi (Zero-length arrays): Update and clarify.
16188 2018-06-25 Michael Meissner <meissner@linux.ibm.com>
16190 * config.gcc (powerpc64le*): Revert January 16th, 2018 patch that
16191 added IEEE/IBM long double multilib support on PowerPC little
16192 endian Linux systems.
16193 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Likewise.
16194 (MULTILIB_DEFAULTS): Likewise.
16195 * config/rs6000/rs6000.c (rs6000_option_override_internal):
16197 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Likewise.
16198 * config/rs6000/t-ldouble-linux64le-ibm: Delete, no longer used.
16199 * config/rs6000/t-ldouble-linux64le-ieee: Delete, no longer used.
16201 2018-06-25 Alexander Monakov <amonakov@ispras.ru>
16203 PR middle-end/86311
16204 * sort.cc (REORDER_23): Avoid memcpy with same destination and source.
16205 (REORDER_45): Likewise.
16207 2018-06-25 Jeff Law <law@redhat.com>
16209 * config/v850/v850.md (divmodhi4): Make sure to sign extend the
16210 dividend to 32 bits. Adjust length.
16211 (udivmodhi4): Cleanup output template. Fix length.
16213 2018-06-25 Carl Love <cel@us.ibm.com>
16215 * config/rs6000/vsx.md: Change word selector to prefered location.
16217 2018-06-25 Richard Biener <rguenther@suse.de>
16219 PR tree-optimization/86304
16220 * tree-vectorizer.c (vectorize_loops): Walk over new possibly
16221 epilogue-if-converted loops as well.
16223 2018-06-25 Jan Hubicka <hubicka@ucw.cz>
16225 * lto-section-out.c (lto_begin_section): Do not print section
16226 name for noaddr and unnumbered dumps.
16228 2018-06-25 Richard Biener <rguenther@suse.de>
16230 * tree-vectorizer.h (struct vec_info_shared): New structure
16231 with parts split out from struct vec_info and loop_nest from
16232 struct _loop_vec_info.
16233 (struct vec_info): Adjust accordingly.
16234 (struct _loop_vec_info): Likewise.
16235 (LOOP_VINFO_LOOP_NEST): Adjust.
16236 (LOOP_VINFO_DATAREFS): Likewise.
16237 (LOOP_VINFO_DDRS): Likewise.
16238 (struct _bb_vec_info): Likewise.
16239 (BB_VINFO_DATAREFS): Likewise.
16240 (BB_VINFO_DDRS): Likewise.
16241 (struct _stmt_vec_info): Add dr_aux member.
16242 (DR_VECT_AUX): Adjust to refer to member of DR_STMTs vinfo.
16243 (DR_MISALIGNMENT_UNINITIALIZED): New.
16244 (set_dr_misalignment): Adjust.
16245 (dr_misalignment): Assert misalign isn't DR_MISALIGNMENT_UNINITIALIZED.
16246 (vect_analyze_loop): Adjust prototype.
16247 (vect_analyze_loop_form): Likewise.
16248 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
16249 Compute dependences lazily.
16250 (vect_record_base_alignments): Use shared datarefs/ddrs.
16251 (vect_verify_datarefs_alignment): Likewise.
16252 (vect_analyze_data_refs_alignment): Likewise.
16253 (vect_analyze_data_ref_accesses): Likewise.
16254 (vect_analyze_data_refs): Likewise.
16255 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Add
16256 constructor parameter for shared part.
16257 (vect_analyze_loop_form): Pass in shared part and adjust.
16258 (vect_analyze_loop_2): Pass in storage for the number of
16259 stmts. Move loop nest finding to the caller. Compute
16261 (vect_analyze_loop): Pass in shared part.
16262 (vect_transform_loop): Verify shared datarefs are unchanged.
16263 * tree-vect-slp.c (_bb_vec_info::_bb_vec_info): Add
16264 constructor parameter for shared part.
16265 (vect_slp_analyze_bb_1): Pass in shared part and adjust.
16266 (vect_slp_bb): Verify shared datarefs are unchanged before
16268 * tree-vect-stmts.c (ensure_base_align): Adjust for DR_AUX
16270 (new_stmt_vec_info): Initialize DR_AUX misalignment to
16271 DR_MISALIGNMENT_UNINITIALIZED.
16272 * tree-vectorizer.c (vec_info::vec_info): Add constructor
16273 parameter for shared part.
16274 (vec_info::~vec_info): Adjust.
16275 (vec_info_shared::vec_info_shared): New.
16276 (vec_info_shared::~vec_info_shared): Likewise.
16277 (vec_info_shared::save_datarefs): Likewise.
16278 (vec_info_shared::check_datarefs): Likewise.
16279 (try_vectorize_loop_1): Construct shared part live for analyses
16280 of a single loop for multiple vector sizes.
16281 * tree-parloops.c (gather_scalar_reductions): Adjust.
16283 2018-06-25 Richard Biener <rguenther@suse.de>
16285 * tree-vect-data-refs.c (vect_find_stmt_data_reference): Modify
16286 DR for SIMD lane accesses here and mark DR with (void *)-1 aux.
16287 (vect_analyze_data_refs): Remove similar code from here and
16288 simplify accordingly.
16290 2018-06-25 Richard Biener <rguenther@suse.de>
16292 * tree-vect-data-refs.c (vect_check_gather_scatter): Fail
16293 for reverse storage order accesses rather than asserting
16294 they cannot happen here.
16296 2018-06-25 Tom de Vries <tdevries@suse.de>
16299 * config/i386/i386.md (define_insn "*tls_global_dynamic_64_<mode>"):
16300 Use data16 instead of .byte for insn prefix.
16302 2018-06-25 Andreas Krebbel <krebbel@linux.ibm.com>
16305 * parser.c (make_char_string_pack): Pass this literal chars
16306 through cpp_interpret_string.
16307 (cp_parser_userdef_numeric_literal): Check the result of
16308 make_char_string_pack.
16310 2018-06-24 Maya Rashish <coypu@sdf.org>
16312 * ginclude/stddef.h: Simplify conditions around avoiding
16313 re-definition of __size_t.
16315 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
16317 * lto-streamer-out.c (tree_is_indexable): Make LABEL_DECL nonindexable
16318 unless it is forced or nonlocal; assert that we stream no IMPORTED_DECL.
16320 2018-06-22 Maya Rashish <coypu@sdf.org>
16322 * doc/invoke.texi (mno-fancy-math-387): Update for changes
16323 made to OpenBSD and NetBSD through the years.
16325 2018-06-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
16327 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
16328 behavior of vec_pack (vector double, vector double) to match
16329 behavior of vec_float2 (vector double, vector double).
16331 2018-06-22 Olivier Hainque <hainque@adacore.com>
16333 * gimplify.c (gimplify_function_tree): Prevent creation
16334 of a trampoline for the address of the current function
16335 passed to entry/exit instrumentation hooks.
16337 2018-06-22 Aaron Sawdey <acsawdey@linux.ibm.com>
16340 * config/rs6000/rs6000-string.c (expand_strn_compare): Handle -m32
16343 2018-06-22 Martin Liska <mliska@suse.cz>
16345 PR tree-optimization/86263
16346 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
16347 Bail out if is_enabled is false.
16348 * tree-switch-conversion.h (jump_table_cluster::is_enabled):
16350 (jump_table_cluster::is_enabled): New function.
16352 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
16354 * lto-streamer-out.c (DFS::DFS_write_tree_body): Do not stream
16355 BINFO_BASE_ACCESSES and BINFO_VPTR_FIELD.
16356 * tree-streamer-in.c (streamer_read_tree_bitfields): Likewise.
16357 (lto_input_ts_binfo_tree_pointers): Likewise.
16358 * tree-streamer-out.c (streamer_write_tree_bitfields,
16359 write_ts_binfo_tree_pointers): Likewise.
16360 * tree.c (free_lang_data_in_binfo): Clear BINFO_VPTR_FIELD.
16362 2018-06-22 Jan Hubicka <hubicka@ucw.cz>
16364 * tree.c (free_lang_data_in_type): Free all TYPE_VFIELDs.
16366 2018-06-22 Martin Liska <mliska@suse.cz>
16368 * symbol-summary.h (get): Make it pure and inline move
16369 functionality from ::get function.
16370 (get): Remove and inline into ::get and ::get_create.
16371 (get_create): Move code from ::get function.
16373 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16376 * config/i386/sol2.h (CPP_SPEC): Don't pass -P for
16377 -x assembler-with-cpp.
16379 2018-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16381 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define
16382 _FILE_OFFSET_BITS=64 for C++.
16384 2018-06-21 Michael Meissner <meissner@linux.ibm.com>
16386 * config/rs6000/rs6000.md (extendtfif2): Add missing 128-bit
16387 conversion insn that shows up when pr85657-3.c is compiled using
16388 IEEE 128-bit long double.
16389 (neg<mode>2_internal): Use the correct mode to check whether the
16390 mode is IBM extended.
16391 * config/rs6000/rs6000.c (init_float128_ieee): Prevent complex
16392 multiply and divide external functions from being created more
16395 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
16397 * cfgrtl.c (fixup_reorder_chain): Do not emit NOPs in DECL_IGNORED_P
16399 (rtl_merge_blocks): Likewise. Do not emit a NOP if the location of
16400 the edge can be forwarded.
16401 (cfg_layout_merge_blocks): Likewise.
16403 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
16405 * except.c (finish_eh_generation): Commit edge insertions only after
16406 the EH edges have been redirected from post-landing to landing pads.
16408 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
16410 * tree-nested.c (get_frame_type): Use create_tmp_var_raw instead of
16411 create_tmp_var_for to create the FRAME decl.
16412 (finalize_nesting_tree_1): Do not unchain the FRAME decl.
16414 2018-06-21 Eric Botcazou <ebotcazou@adacore.com>
16416 * tree-inline.c (copy_edges_for_bb): Minor tweak.
16417 (maybe_move_debug_stmts_to_successors): Also reset the locus of the
16418 debug statement when resetting its value.
16419 (expand_call_inline): Copy the locus of the call onto the assignment
16420 of the return value, if any. Use local variable in more cases.
16422 2018-06-21 Martin Liska <mliska@suse.cz>
16424 * ipa-pure-const.c (propagate_nothrow): Use
16425 funct_state_summaries->get.
16426 (dump_malloc_lattice): Likewise.
16427 (propagate_malloc): Likewise.
16429 2018-06-21 Richard Biener <rguenther@suse.de>
16431 * lto-streamer-out.c (DFS::DFS_write_tree_body): Update outdated
16432 comment. Follow BLOCK_ABSTRACT_ORIGIN unconditionally.
16433 * tree-streamer-in.c (lto_input_ts_block_tree_pointers): Update
16435 * tree-streamer-out.c (write_ts_block_tree_pointers): Stream
16436 BLOCK_ABSTRACT_ORIGIN unconditionally.
16438 2018-06-21 David Malcolm <dmalcolm@redhat.com>
16440 * ipa-cp.c (ipcp_driver): Set edge_clone_summaries to NULL after
16442 * ipa-reference.c (ipa_reference_c_finalize): Delete
16443 ipa_ref_opt_sum_summaries and set it to NULL.
16445 2018-06-21 Tom de Vries <tdevries@suse.de>
16447 PR tree-optimization/85859
16448 * tree-ssa-tail-merge.c (stmt_local_def): Copy gimple_is_call
16449 test with comment from bb_no_side_effects_p.
16451 2018-06-21 Richard Biener <rguenther@suse.de>
16453 PR tree-optimization/86232
16454 * tree-ssa-loop-niter.c (number_of_iterations_popcount): Adjust
16455 max for constant niter.
16457 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
16459 * config/aarch64/aarch64-simd.md
16460 (*aarch64_crypto_aes<aes_op>v16qi_xor_combine): New.
16462 2018-06-21 Andre Vieira <andre.simoesdiasvieira@arm.com>
16464 * config/aarch64/aarch64-simd.md (aarch64_crypto_aes<aes_op>v16qi):
16465 Make opernads of the unspec commutative.
16467 2018-06-21 Richard Biener <rguenther@suse.de>
16469 * tree-data-ref.c (dr_step_indicator): Handle NULL DR_STEP.
16470 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr):
16471 Avoid calling vect_mark_for_runtime_alias_test with gathers or scatters.
16472 (vect_analyze_data_ref_dependence): Re-order checks to deal with
16474 (vect_record_base_alignments): Do not record base alignment
16475 for gathers or scatters.
16476 (vect_compute_data_ref_alignment): Drop return value that is always
16477 true. Bail out early for gathers or scatters.
16478 (vect_enhance_data_refs_alignment): Bail out early for gathers
16480 (vect_find_same_alignment_drs): Likewise.
16481 (vect_analyze_data_refs_alignment): Remove dead code.
16482 (vect_slp_analyze_and_verify_node_alignment): Likewise.
16483 (vect_analyze_data_refs): For possible gathers or scatters do
16484 not create an alternate DR, just check their possible validity
16485 and mark them. Adjust DECL_NONALIASED handling to not rely
16486 on DR_BASE_ADDRESS.
16487 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Do not
16488 update inits of gathers or scatters.
16489 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern):
16490 Also copy gather/scatter flag to pattern vinfo.
16492 2018-06-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
16494 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Change
16495 behavior of vec_packsu (vector unsigned long long, vector unsigned
16496 long long) to match behavior of vec_packs with same signature.
16498 2018-06-20 Chung-Lin Tang <cltang@codesourcery.com>
16499 Thomas Schwinge <thomas@codesourcery.com>
16500 Cesar Philippidis <cesar@codesourcery.com>
16502 * gimplify.c (gimplify_scan_omp_clauses): Add support for
16503 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
16504 (gimplify_adjust_omp_clauses): Likewise.
16505 (gimplify_oacc_declare_1): Add support for GOMP_MAP_RELEASE, remove
16506 support for GOMP_MAP_FORCE_{ALLOC,TO,FROM,TOFROM}.
16507 (gimplify_omp_target_update): Update handling of acc update and
16509 * omp-low.c (install_var_field): Remove unused parameter
16510 base_pointers_restrict.
16511 (scan_sharing_clauses): Remove base_pointers_restrict parameter.
16512 Update call to install_var_field. Handle OMP_CLAUSE_{IF_PRESENT,
16514 (omp_target_base_pointers_restrict_p): Delete.
16515 (scan_omp_target): Update call to scan_sharing_clauses.
16516 * tree-core.h (enum omp_clause_code): Add OMP_CLAUSE_{IF_PRESENT,
16518 * tree-nested.c (convert_nonlocal_omp_clauses): Handle
16519 OMP_CLAUSE_{IF_PRESENT,FINALIZE}.
16520 (convert_local_omp_clauses): Likewise.
16521 * tree-pretty-print.c (dump_omp_clause): Likewise.
16522 * tree.c (omp_clause_num_ops): Add entries for OMP_CLAUSE_{IF_PRESENT,
16524 (omp_clause_code_name): Likewise.
16526 2018-06-20 Jakub Jelinek <jakub@redhat.com>
16529 * var-tracking.c (use_narrower_mode_test): Check if shift amount can
16532 PR tree-optimization/86231
16533 * tree-vrp.c (union_ranges): For ( [ ) ] or ( )[ ] range and
16534 anti-range don't overwrite *vr0min before using it to compute *vr0max.
16536 2018-06-20 Tom de Vries <tdevries@suse.de>
16538 PR tree-optimization/86097
16539 * tree-ssa-loop-manip.c (canonicalize_loop_ivs): Also convert *nit to
16540 iv type if signedness of iv type is not the same as that of *nit.
16542 2018-06-20 Jakub Jelinek <jakub@redhat.com>
16544 * cfgrtl.c (rtl_verify_edges): Formatting fix. If bb->preds has any
16545 EDGE_EH edges, verify they are all EDGE_EH.
16547 2018-06-20 Maya Rashish <coypu@sdf.org>
16549 * ginclude/stddef.h: Limit #include <machine/ansi.h> to NetBSD.
16551 2018-06-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16553 * config/aarch64/aarch64-tuning-flags.def (no_ldp_stp_qregs): New.
16554 * config/aarch64/aarch64.c (xgene1_tunings): Add
16555 AARCH64_EXTRA_TUNE_NO_LDP_STP_QREGS to tune_flags.
16556 (aarch64_mode_valid_for_sched_fusion_p):
16557 Allow 16-byte modes.
16558 (aarch64_classify_address): Allow 16-byte modes for load_store_pair_p.
16559 * config/aarch64/aarch64-ldpstp.md: Add peepholes for LDP STP of
16561 * config/aarch64/aarch64-simd.md (load_pair<VQ:mode><VQ2:mode>):
16563 (vec_store_pair<VQ:mode><VQ2:mode>): Likewise.
16564 * config/aarch64/iterators.md (VQ2): New mode iterator.
16566 2018-06-20 Martin Liska <mliska@suse.cz>
16568 * tree-switch-conversion.c (jump_table_cluster::can_be_handled):
16569 Change default ratio from 10 to 8.
16571 2018-06-20 Martin Liska <mliska@suse.cz>
16573 * tree-switch-conversion.c (jump_table_cluster::find_jump_tables):
16575 (bit_test_cluster::find_bit_tests): Likewise.
16576 (switch_decision_tree::analyze_switch_statement): Find clusters.
16577 * tree-switch-conversion.h (struct jump_table_cluster): Document
16580 2018-06-20 Martin Liska <mliska@suse.cz>
16582 * tree-switch-conversion.c (switch_conversion::collect):
16583 Record m_uniq property.
16584 (switch_conversion::expand): Bail out for special conditions.
16585 (group_cluster::~group_cluster): New.
16586 (group_cluster::group_cluster): Likewise.
16587 (group_cluster::dump): Likewise.
16588 (jump_table_cluster::emit): New.
16589 (switch_decision_tree::fix_phi_operands_for_edges): New.
16590 (struct case_node): Remove struct.
16591 (jump_table_cluster::can_be_handled): New.
16592 (case_values_threshold): Moved to header.
16593 (reset_out_edges_aux): Likewise.
16594 (jump_table_cluster::is_beneficial): New.
16595 (bit_test_cluster::can_be_handled): Likewise.
16596 (add_case_node): Remove.
16597 (bit_test_cluster::is_beneficial): New.
16598 (case_bit_test::cmp): New.
16599 (bit_test_cluster::emit): New.
16600 (expand_switch_as_decision_tree_p): Remove.
16601 (bit_test_cluster::hoist_edge_and_branch_if_true): New.
16602 (fix_phi_operands_for_edge): Likewise.
16603 (switch_decision_tree::analyze_switch_statement): New.
16604 (compute_cases_per_edge): Move ...
16605 (switch_decision_tree::compute_cases_per_edge): ... here.
16606 (try_switch_expansion): Likewise.
16607 (switch_decision_tree::try_switch_expansion): Likewise.
16608 (record_phi_operand_mapping): Likewise.
16609 (switch_decision_tree::record_phi_operand_mapping): Likewise.
16610 (emit_case_decision_tree): Likewise.
16611 (switch_decision_tree::emit): Likewise.
16612 (balance_case_nodes): Likewise.
16613 (switch_decision_tree::balance_case_nodes): Likewise.
16614 (dump_case_nodes): Likewise.
16615 (switch_decision_tree::dump_case_nodes): Likewise.
16616 (emit_jump): Likewise.
16617 (switch_decision_tree::emit_jump): Likewise.
16618 (emit_cmp_and_jump_insns): Likewise.
16619 (switch_decision_tree::emit_cmp_and_jump_insns): Likewise.
16620 (emit_case_nodes): Likewise.
16621 (switch_decision_tree::emit_case_nodes): Likewise.
16622 (conditional_probability): Remove.
16623 * tree-switch-conversion.h (enum cluster_type): New.
16625 (struct cluster): Likewise.
16626 (cluster::cluster): Likewise.
16627 (struct simple_cluster): Likewise.
16628 (simple_cluster::simple_cluster): Likewise.
16629 (struct group_cluster): Likewise.
16630 (struct jump_table_cluster): Likewise.
16631 (struct bit_test_cluster): Likewise.
16632 (struct min_cluster_item): Likewise.
16633 (struct case_tree_node): Likewise.
16634 (case_tree_node::case_tree_node): Likewise.
16635 (jump_table_cluster::case_values_threshold): Likewise.
16636 (struct case_bit_test): Likewise.
16637 (struct switch_decision_tree): Likewise.
16638 (struct switch_conversion): Likewise.
16639 (switch_decision_tree::reset_out_edges_aux): Likewise.
16641 2018-06-20 Martin Liska <mliska@suse.cz>
16643 * tree-switch-conversion.c (MAX_CASE_BIT_TESTS): Remove.
16644 (hoist_edge_and_branch_if_true): Likewise.
16645 (expand_switch_using_bit_tests_p): Likewise.
16646 (struct case_bit_test): Likewise.
16647 (case_bit_test_cmp): Likewise.
16648 (emit_case_bit_tests): Likewise.
16649 (switch_conversion::switch_conversion): New class.
16650 (struct switch_conv_info): Remove old struct.
16651 (collect_switch_conv_info): More to ...
16652 (switch_conversion::collect): ... this.
16653 (check_range): Likewise.
16654 (switch_conversion::check_range): Likewise.
16655 (check_all_empty_except_final): Likewise.
16656 (switch_conversion::check_all_empty_except_final): Likewise.
16657 (check_final_bb): Likewise.
16658 (switch_conversion::check_final_bb): Likewise.
16659 (create_temp_arrays): Likewise.
16660 (switch_conversion::create_temp_arrays): Likewise.
16661 (free_temp_arrays): Likewise.
16662 (gather_default_values): Likewise.
16663 (switch_conversion::gather_default_values): Likewise.
16664 (build_constructors): Likewise.
16665 (switch_conversion::build_constructors): Likewise.
16666 (constructor_contains_same_values_p): Likewise.
16667 (switch_conversion::contains_same_values_p): Likewise.
16668 (array_value_type): Likewise.
16669 (switch_conversion::array_value_type): Likewise.
16670 (build_one_array): Likewise.
16671 (switch_conversion::build_one_array): Likewise.
16672 (build_arrays): Likewise.
16673 (switch_conversion::build_arrays): Likewise.
16674 (gen_def_assigns): Likewise.
16675 (switch_conversion::gen_def_assigns): Likewise.
16676 (prune_bbs): Likewise.
16677 (switch_conversion::prune_bbs): Likewise.
16678 (fix_phi_nodes): Likewise.
16679 (switch_conversion::fix_phi_nodes): Likewise.
16680 (gen_inbound_check): Likewise.
16681 (switch_conversion::gen_inbound_check): Likewise.
16682 (process_switch): Use the newly created class.
16683 (switch_conversion::expand): New.
16684 (switch_conversion::~switch_conversion): New.
16685 * tree-switch-conversion.h: New file.
16687 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16689 * tree-vectorizer.h (NUM_PATTERNS, vect_recog_func_ptr): Move to
16690 tree-vect-patterns.c.
16691 * tree-vect-patterns.c (vect_supportable_direct_optab_p): New function.
16692 (vect_recog_dot_prod_pattern): Use it. Remove the type_in argument.
16693 (vect_recog_sad_pattern): Likewise.
16694 (vect_recog_widen_sum_pattern): Likewise.
16695 (vect_recog_pow_pattern): Likewise. Check for a null vectype.
16696 (vect_recog_widen_shift_pattern): Remove the type_in argument.
16697 (vect_recog_rotate_pattern): Likewise.
16698 (vect_recog_mult_pattern): Likewise.
16699 (vect_recog_vector_vector_shift_pattern): Likewise.
16700 (vect_recog_divmod_pattern): Likewise.
16701 (vect_recog_mixed_size_cond_pattern): Likewise.
16702 (vect_recog_bool_pattern): Likewise.
16703 (vect_recog_mask_conversion_pattern): Likewise.
16704 (vect_try_gather_scatter_pattern): Likewise.
16705 (vect_recog_widen_mult_pattern): Likewise. Check for a null vectype.
16706 (vect_recog_over_widening_pattern): Likewise.
16707 (vect_recog_gather_scatter_pattern): Likewise.
16708 (vect_recog_func_ptr): Move from tree-vectorizer.h
16709 (vect_vect_recog_func_ptrs): Move further down the file.
16710 (vect_recog_func): Likewise. Remove the third argument.
16711 (NUM_PATTERNS): Define based on vect_vect_recog_func_ptrs.
16712 (vect_pattern_recog_1): Expect the pattern function to do any
16713 necessary target tests. Also expect it to provide a vector type.
16714 Remove the type_in handling.
16716 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16718 * tree-vect-patterns.c (vect_pattern_detected): New function.
16719 (vect_recog_dot_prod_patternm, vect_recog_sad_pattern)
16720 (vect_recog_widen_mult_pattern, vect_recog_widen_sum_pattern)
16721 (vect_recog_over_widening_pattern, vect_recog_widen_shift_pattern
16722 (vect_recog_rotate_pattern, vect_recog_vector_vector_shift_pattern)
16723 (vect_recog_mult_pattern, vect_recog_divmod_pattern)
16724 (vect_recog_mixed_size_cond_pattern, vect_recog_bool_pattern)
16725 (vect_recog_mask_conversion_pattern)
16726 (vect_try_gather_scatter_pattern): Likewise.
16728 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16730 * tree-vect-patterns.c (vect_get_internal_def): New function.
16731 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern)
16732 (vect_recog_vector_vector_shift_pattern, check_bool_pattern)
16733 (search_type_for_mask_1): Use it.
16735 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16737 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
16738 redundant WIDEN_SUM_EXPR handling.
16739 (vect_recog_sad_pattern): Likewise.
16741 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16743 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Remove
16744 redundant check that the types of a PLUS_EXPR or MULT_EXPR agree.
16745 (vect_recog_sad_pattern): Likewise PLUS_EXPR, ABS_EXPR and MINUS_EXPR.
16746 (vect_recog_widen_mult_pattern): Likewise MULT_EXPR.
16747 (vect_recog_widen_sum_pattern): Likewise PLUS_EXPR.
16749 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16751 * tree-vect-stmts.c (vectorizable_call): Make sure that we
16752 use the stmt_vec_info of the original bb statement for the
16753 new zero assignment, even if the call is part of a pattern.
16755 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16757 * tree-vectorizer.h (_stmt_vec_info): Note above pattern_def_seq
16758 that the sequence is attached to the original statement rather
16759 than the pattern statement.
16760 * tree-vect-loop.c (vect_determine_vf_for_stmt): Take the
16761 PATTERN_DEF_SEQ from the original statement rather than
16762 the main pattern statement.
16763 * tree-vect-stmts.c (free_stmt_vec_info): Likewise.
16764 * tree-vect-patterns.c (vect_recog_dot_prod_pattern): Likewise.
16765 (vect_mark_pattern_stmts): Don't copy the PATTERN_DEF_SEQ.
16767 2018-06-20 Richard Sandiford <richard.sandiford@arm.com>
16769 * tree-vect-stmts.c (vect_analyze_stmt): Move the handling of pattern
16770 definition statements before the early exit for statements that aren't
16772 * tree-vect-loop.c (vect_transform_loop_stmt): New function,
16774 (vect_transform_loop): ...here. Process pattern definition
16775 statements without first checking whether the main pattern
16776 statement is live or relevant.
16778 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
16780 * tree-cfgcleanup.c (tree_forwarder_block_p): Do not return false at
16781 -O0 if the locus represent UNKNOWN_LOCATION but have different values.
16783 2018-06-19 Aaron Sawdey <acsawdey@linux.ibm.com>
16785 * config/rs6000/rs6000-string.c (select_block_compare_mode): Check
16786 TARGET_EFFICIENT_OVERLAPPING_UNALIGNED here instead of in caller.
16787 (do_and3, do_and3_mask, do_compb3, do_rotl3): New functions.
16788 (expand_block_compare): Change select_block_compare_mode call.
16789 (expand_strncmp_align_check): Use new functions, fix comment.
16790 (emit_final_str_compare_gpr): New function.
16791 (expand_strn_compare): Refactor and clean up code.
16792 * config/rs6000/vsx.md (vsx_mov<mode>_64bit): Remove *.
16794 2018-06-19 Tony Reix <tony.reix@atos.com>
16795 Damien Bergamini <damien.bergamini@atos.com>
16796 David Edelsohn <dje.gcc@gmail.com>
16798 * collect2.c (static_obj): New variable.
16799 (static_libs): New variable.
16800 (is_in_list): Uncomment declaration.
16801 (main): Track AIX libraries linked statically.
16802 (is_in_list): Uncomment definition.
16803 (scan_prog_file): Don't add AIX shared libraries initializer
16804 to constructor list if linking statically.
16806 2018-06-19 Max Filippov <jcmvbkbc@gmail.com>
16808 * config/xtensa/xtensa.md (UNSPEC_FRAME_BLOCKAGE): New unspec
16810 (allocate_stack, frame_blockage, *frame_blockage): New patterns.
16812 2018-06-19 Jan Hubicka <hubicka@ucw.cz>
16814 * tree.c (find_decls_types_r): Remove all non-VAR_DECLs from
16817 2018-06-19 Martin Liska <mliska@suse.cz>
16819 * config/i386/i386.c (ix86_can_inline_p): Do not use
16820 ipa_fn_summaries::get_create.
16821 * ipa-cp.c (ipcp_cloning_candidate_p): Replace get_create with
16823 (devirtualization_time_bonus): Likewise.
16824 (ipcp_propagate_stage): Likewise.
16825 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
16826 (edge_set_predicate): Likewise.
16827 (evaluate_conditions_for_known_args): Likewise.
16828 (evaluate_properties_for_edge): Likewise.
16829 (ipa_call_summary::reset): Tranform to ...
16830 (ipa_call_summary::~ipa_call_summary): ... this.
16831 (ipa_fn_summary::reset): Transform to ...
16832 (ipa_fn_summary::~ipa_fn_summary): ... this.
16833 (ipa_fn_summary_t::remove): Rename to ...
16834 (ipa_fn_summary_t::remove_callees): ... this.
16835 (ipa_fn_summary_t::duplicate): Use placement new
16836 instead of memory copy.
16837 (ipa_call_summary_t::duplicate): Likewise.
16838 (ipa_call_summary_t::remove): Remove.
16839 (dump_ipa_call_summary): Change get_create to get.
16840 (ipa_dump_fn_summary): Dump only when summary exists.
16841 (analyze_function_body): Use symbol_summary::get instead
16843 (compute_fn_summary): Likewise.
16844 (estimate_edge_devirt_benefit): Likewise.
16845 (estimate_edge_size_and_time): Likewise.
16846 (inline_update_callee_summaries): Likewise.
16847 (remap_edge_change_prob): Likewise.
16848 (remap_edge_summaries): Likewise.
16849 (ipa_merge_fn_summary_after_inlining): Likewise.
16850 (write_ipa_call_summary): Likewise.
16851 (ipa_fn_summary_write): Likewise.
16852 (ipa_free_fn_summary): Likewise.
16853 * ipa-fnsummary.h (struct GTY): Add new ctor and copy ctor.
16854 (struct ipa_call_summary): Likewise.
16855 * ipa-icf.c (sem_function::merge): Use symbol_summary::get instead
16857 * ipa-inline-analysis.c (do_estimate_edge_time): Likewise.
16858 (estimate_size_after_inlining): Likewise.
16859 (estimate_growth): Likewise.
16860 (growth_likely_positive): Likewise.
16861 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
16862 (inline_call): Likewise.
16863 * ipa-inline.c (caller_growth_limits): Likewise.
16864 (can_inline_edge_p): Likewise.
16865 (can_inline_edge_by_limits_p): Likewise.
16866 (compute_uninlined_call_time): Likewise.
16867 (compute_inlined_call_time): Likewise.
16868 (want_inline_small_function_p): Likewise.
16869 (edge_badness): Likewise.
16870 (update_caller_keys): Likewise.
16871 (update_callee_keys): Likewise.
16872 (inline_small_functions): Likewise.
16873 (inline_to_all_callers_1): Likewise.
16874 (dump_overall_stats): Likewise.
16875 (early_inline_small_functions): Likewise.
16876 (early_inliner): Likewise.
16877 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
16878 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
16879 * ipa-pure-const.c (malloc_candidate_p): Likewise.
16880 * ipa-split.c (execute_split_functions): Likewise.
16881 * symbol-summary.h: Likewise.
16882 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
16884 2018-06-19 Richard Biener <rguenther@suse.de>
16886 * tree-vectorizer.c (try_vectorize_loop_1): Split out of ...
16887 (vectorize_loops): ... here. Fix dbgcnt handling.
16888 (try_vectorize_loop): Wrap try_vectorize_loop_1.
16890 2018-06-19 Segher Boessenkool <segher@kernel.crashing.org>
16893 * config/rs6000/rs6000.md (rs6000_discover_homogeneous_aggregate): An
16894 ieee128 argument takes up only one (vector) register, not two (floating
16897 2018-06-19 Eric Botcazou <ebotcazou@adacore.com>
16899 * gimplify.c (gimplify_init_constructor): Really never clear for an
16900 incomplete constructor if CONSTRUCTOR_NO_CLEARING is set.
16902 2018-06-19 Richard Biener <rguenther@suse.de>
16904 PR tree-optimization/86179
16905 * tree-vect-patterns.c (vect_pattern_recog_1): Clean up
16906 after failed recognition.
16908 2018-06-18 Martin Sebor <msebor@redhat.com>
16910 PR middle-end/85602
16911 * calls.c (maybe_warn_nonstring_arg): Handle strncat.
16912 * tree-ssa-strlen.c (is_strlen_related_p): Make extern.
16913 Handle integer subtraction.
16914 (maybe_diag_stxncpy_trunc): Handle nonstring source arguments.
16915 * tree-ssa-strlen.h (is_strlen_related_p): Declare.
16917 2018-06-18 David Malcolm <dmalcolm@redhat.com>
16919 * config/frv/frv-protos.h (frv_ifcvt_modify_insn): Strengthen 3rd
16920 param from rtx to rtx_insn *.
16921 * config/frv/frv.c (frv_ifcvt_add_insn): Likewise for "insn"
16923 (frv_ifcvt_modify_insn): Likwise.
16924 (frv_ifcvt_modify_final): Likwise for local "existing_insn",
16925 adding an as_a <rtx_insn *> cast. Likewise for local "insn".
16926 * config/mips/mips.c (r10k_insert_cache_barriers): Add an
16927 as_a <rtx_insn *> cast to local "unprotected_region" once
16928 it's been established that it's not NULL or pc_rtx.
16929 * config/nds32/nds32-relax-opt.c (nds32_group_insns): Strengthen
16930 param "sethi" from rtx to rtx_insn *.
16931 (nds32_group_float_insns): Likewise for param "insn".
16932 * config/vax/vax-protos.h (vax_output_int_add): Likewise for 1st
16934 (vax_output_int_subtract): Likewise.
16935 * config/vax/vax.c (vax_output_int_add): Likewise for param
16937 (vax_output_int_subtract): Likewise.
16938 * emit-rtl.c (set_insn_deleted): Likewise, removing cast.
16939 (emit_pattern_after): Likewise for param "after".
16940 (emit_insn_after): Likewise.
16941 (emit_jump_insn_after): Likewise.
16942 (emit_call_insn_after): Likewise.
16943 (emit_debug_insn_after): Likewise.
16944 (emit_pattern_before): Likewise for param "before".
16945 (emit_insn_before): Likewise.
16946 (emit_jump_insn_before): Likewise.
16947 * final.c (get_insn_template): Likewise for param "insn", removing
16949 * output.h (get_insn_template): Likewise for 2nd param.
16950 * rtl.h (emit_insn_before): Likewise.
16951 (emit_jump_insn_before): Likewise.
16952 (emit_debug_insn_before_noloc): Likewise.
16953 (emit_insn_after): Likewise.
16954 (emit_jump_insn_after): Likewise.
16955 (emit_call_insn_after): Likewise.
16956 (emit_debug_insn_after): Likewise.
16957 (set_insn_deleted): Likewise for param.
16959 2018-06-18 Michael Meissner <meissner@linux.ibm.com>
16962 * config/rs6000/rs6000-modes.def (toplevel): Rework the 128-bit
16963 floating point modes, so that IFmode is numerically greater than
16964 TFmode, which is greater than KFmode using FRACTIONAL_FLOAT_MODE
16965 to declare the ordering. This prevents IFmode from being
16966 converted to TFmode when long double is IEEE 128-bit on an ISA 3.0
16967 machine. Include rs6000-modes.h to share the fractional values
16968 between genmodes* and the rest of the compiler.
16969 (IFmode): Likewise.
16970 (KFmode): Likewise.
16971 (TFmode): Likewise.
16972 * config/rs6000/rs6000-modes.h: New file.
16973 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Change the
16974 meaning of rs6000_long_double_size so that 126..128 selects an
16975 appropriate 128-bit floating point type.
16976 (rs6000_option_override_internal): Likewise.
16977 * config/rs6000/rs6000.h (toplevel): Include rs6000-modes.h.
16978 (TARGET_LONG_DOUBLE_128): Change the meaning of
16979 rs6000_long_double_size so that 126..128 selects an appropriate
16980 128-bit floating point type.
16981 (LONG_DOUBLE_TYPE_SIZE): Update comment.
16982 * config/rs6000/rs6000.md (trunciftf2): Correct the modes of the
16983 source and destination to match the standard usage.
16984 (truncifkf2): Likewise.
16985 (copysign<mode>3, IEEE iterator): Rework copysign of float128 on
16986 ISA 2.07 to use an explicit clobber, instead of passing in a
16988 (copysign<mode>3_soft): Likewise.
16990 2018-06-18 David Malcolm <dmalcolm@redhat.com>
16992 * tree-vect-data-refs.c (vect_analyze_data_ref_dependences):
16993 Replace dump_printf_loc call with DUMP_VECT_SCOPE.
16994 (vect_slp_analyze_instance_dependence): Likewise.
16995 (vect_enhance_data_refs_alignment): Likewise.
16996 (vect_analyze_data_refs_alignment): Likewise.
16997 (vect_slp_analyze_and_verify_instance_alignment
16998 (vect_analyze_data_ref_accesses): Likewise.
16999 (vect_prune_runtime_alias_test_list): Likewise.
17000 (vect_analyze_data_refs): Likewise.
17001 * tree-vect-loop-manip.c (vect_update_inits_of_drs): Likewise.
17002 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
17003 (vect_analyze_scalar_cycles_1): Likewise.
17004 (vect_get_loop_niters): Likewise.
17005 (vect_analyze_loop_form_1): Likewise.
17006 (vect_update_vf_for_slp): Likewise.
17007 (vect_analyze_loop_operations): Likewise.
17008 (vect_analyze_loop): Likewise.
17009 (vectorizable_induction): Likewise.
17010 (vect_transform_loop): Likewise.
17011 * tree-vect-patterns.c (vect_pattern_recog): Likewise.
17012 * tree-vect-slp.c (vect_analyze_slp): Likewise.
17013 (vect_make_slp_decision): Likewise.
17014 (vect_detect_hybrid_slp): Likewise.
17015 (vect_slp_analyze_operations): Likewise.
17016 (vect_slp_bb): Likewise.
17017 * tree-vect-stmts.c (vect_mark_stmts_to_be_vectorized): Likewise.
17018 (vectorizable_bswap): Likewise.
17019 (vectorizable_call): Likewise.
17020 (vectorizable_simd_clone_call): Likewise.
17021 (vectorizable_conversion): Likewise.
17022 (vectorizable_assignment): Likewise.
17023 (vectorizable_shift): Likewise.
17024 (vectorizable_operation): Likewise.
17025 * tree-vectorizer.h (DUMP_VECT_SCOPE): New macro.
17027 2018-06-18 Martin Sebor <msebor@redhat.com>
17029 PR tree-optimization/81384
17030 * builtin-types.def (BT_FN_SIZE_CONST_STRING_SIZE): New.
17031 * builtins.c (expand_builtin_strnlen): New function.
17032 (expand_builtin): Call it.
17033 (fold_builtin_n): Avoid setting TREE_NO_WARNING.
17034 * builtins.def (BUILT_IN_STRNLEN): New.
17035 * calls.c (maybe_warn_nonstring_arg): Handle BUILT_IN_STRNLEN.
17036 Warn for bounds in excess of maximum object size.
17037 * tree-ssa-strlen.c (maybe_set_strlen_range): Return tree representing
17038 single-value ranges. Handle strnlen.
17039 (handle_builtin_strlen): Handle strnlen.
17040 (strlen_check_and_optimize_stmt): Same.
17041 * doc/extend.texi (Other Builtins): Document strnlen.
17043 2018-06-18 Maya Rashish <coypu@sdf.org>
17045 * config/alpha/openbsd.h (TARGET_DEFAULT): Define.
17046 (LINK_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): Likewise.
17047 (INTMAX_TYPE, UINTMAX_TYPE, WINT_TYPE): Likewise.
17049 * config/alpha/elf.h (STARTFILE_SPEC, ENDFILE_SPEC): Move from
17051 * config/alpha/linux.h (STARTFILE_SPEC, ENDFILE_SPEC): Here.
17053 2018-06-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
17055 * tree.c (escaped_string::escape): Replace cast to char * by
17056 const_cast<char *> (unescaped).
17058 2018-06-18 Nick Clifton <nickc@redhat.com>
17061 * tree.c (escaped_string): New class. Converts an unescaped
17062 string into its escaped equivalent.
17063 (warn_deprecated_use): Use the new class to convert the
17064 deprecation message, if present.
17065 (test_escaped_strings): New self test.
17066 (test_c_tests): Add test_escaped_strings.
17067 * doc/extend.texi (deprecated): Add a note that the
17068 deprecation message is affected by the -fmessage-length
17069 option, and that control characters will be escaped.
17070 (#pragma GCC error): Document this pragma.
17071 (#pragma GCC warning): Likewise.
17072 * doc/invoke.texi (-fmessage-length): Document this option's
17073 effect on the #warning and #error preprocessor directives and
17074 the deprecated attribute.
17076 2018-06-18 Eric Botcazou <ebotcazou@adacore.com>
17078 * tree.c (decl_value_expr_lookup): Revert latest change.
17079 (decl_value_expr_insert): Likewise.
17081 2018-06-17 Eric Botcazou <ebotcazou@adacore.com>
17083 * gimplify.c (nonlocal_vlas): Delete.
17084 (nonlocal_vla_vars): Likewise.
17085 (gimplify_var_or_parm_decl): Do not add debug VAR_DECLs for non-local
17087 (gimplify_body): Do not create and destroy nonlocal_vlas.
17088 * tree-nested.c: Include diagnostic.h.
17089 (use_pointer_in_frame): Tweak.
17090 (lookup_field_for_decl): Add assertion and declare the transformation.
17091 (convert_nonlocal_reference_op) <PARM_DECL>: Rework and issue an
17092 internal error when the reference is in a wrong context. Do not
17093 create a debug decl by default.
17094 (note_nonlocal_block_vlas): Delete.
17095 (convert_nonlocal_reference_stmt) <GIMPLE_BIND>: Do not call it.
17096 (convert_local_reference_op) <PARM_DECL>: Skip the frame decl. Do not
17097 create a debug decl by default.
17098 (convert_gimple_call) <GIMPLE_CALL>: Issue an internal error when the
17099 call is in a wrong context.
17100 (fixup_vla_decls): New function.
17101 (finalize_nesting_tree_1): Adjust comment. Call fixup_vla_decls if no
17102 debug variables were created.
17103 * tree.c (decl_value_expr_lookup): Add checking assertion.
17104 (decl_value_expr_insert): Likewise.
17106 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
17108 PR middle-end/82479
17109 * ipa-fnsummary.c (will_be_nonconstant_expr_predicate): Handle CALL_EXPR.
17110 * tree-scalar-evolution.c (interpret_expr): Likewise.
17111 (expression_expensive_p): Likewise.
17112 * tree-ssa-loop-ivopts.c (contains_abnormal_ssa_name_p): Likewise.
17113 * tree-ssa-loop-niter.c (number_of_iterations_popcount): New.
17114 (number_of_iterations_exit_assumptions): Use number_of_iterations_popcount.
17115 (ssa_defined_by_minus_one_stmt_p): New.
17117 2018-06-16 Kugan Vivekanandarajah <kuganv@linaro.org>
17119 PR middle-end/64946
17120 * cfgexpand.c (expand_debug_expr): Hande ABSU_EXPR.
17121 * config/i386/i386.c (ix86_add_stmt_cost): Likewise.
17122 * dojump.c (do_jump): Likewise.
17123 * expr.c (expand_expr_real_2): Check operand type's sign.
17124 * fold-const.c (const_unop): Handle ABSU_EXPR.
17125 (fold_abs_const): Likewise.
17126 * gimple-pretty-print.c (dump_unary_rhs): Likewise.
17127 * gimple-ssa-backprop.c (backprop::process_assign_use): Likesie.
17128 (strip_sign_op_1): Likesise.
17129 * match.pd: Add new pattern to generate ABSU_EXPR.
17130 * optabs-tree.c (optab_for_tree_code): Handle ABSU_EXPR.
17131 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
17132 * tree-eh.c (operation_could_trap_helper_p): Likewise.
17133 * tree-inline.c (estimate_operator_cost): Likewise.
17134 * tree-pretty-print.c (dump_generic_node): Likewise.
17135 * tree-vect-patterns.c (vect_recog_sad_pattern): Likewise.
17136 * tree.def (ABSU_EXPR): New.
17138 2018-06-16 Jakub Jelinek <jakub@redhat.com>
17140 PR middle-end/86095
17141 * common.opt (Wunsafe-loop-optimizations): Add Ignore, remove Var,
17142 documented as preserved for backward compatibility only.
17143 * doc/invoke.texi: Remove -Wunsafe-loop-optimizations documentation.
17145 PR rtl-optimization/86108
17146 * bb-reorder.c (create_forwarder_block): Renamed to ...
17147 (create_eh_forwarder_block): ... this. Split OLD_BB after labels and
17148 jump from new landing pad to the second part.
17149 (sjlj_fix_up_crossing_landing_pad, dw2_fix_up_crossing_landing_pad):
17152 2018-06-15 Jakub Jelinek <jakub@redhat.com>
17154 PR middle-end/85878
17155 * expr.c (expand_assignment): Remove now redundant COMPLEX_MODE_P
17156 check from first store_expr, use to_mode instead of GET_MODE (to_rtx).
17157 Only call store_expr for halves if the mode is the same.
17159 PR middle-end/86123
17160 * match.pd ((X / Y) == 0 -> X < Y): Don't transform complex divisions.
17161 Fix up comment formatting.
17163 2018-06-15 Bernd Edlinger <bernd.edlinger@hotmail.de>
17165 * typed-splay-tree.h (typed_splay_tree::remove): New function.
17166 (typed_splay_tree::closure,
17167 typed_splay_tree::inner_foreach_fn, typed_splay_tree::m_inner): Deleted.
17168 (typed_splay_tree::typed_splay_tree,
17169 typed_splay_tree::operator =): Declared private.
17170 (typed_splay_tree::splay_tree_key, typed_splay_tree::splay_tree_value,
17171 typed_splay_tree::splay_tree_node_s, typed_splay_tree::KDEL,
17172 typed_splay_tree::VDEL, typed_splay_tree::splay_tree_delete_helper,
17173 typed_splay_tree::rotate_left, typed_splay_tree::rotate_right,
17174 typed_splay_tree::splay_tree_splay,
17175 typed_splay_tree::splay_tree_foreach_helper,
17176 typed_splay_tree::splay_tree_insert,
17177 typed_splay_tree::splay_tree_remove,
17178 typed_splay_tree::splay_tree_lookup,
17179 typed_splay_tree::splay_tree_predecessor,
17180 typed_splay_tree::splay_tree_successor,
17181 typed_splay_tree::splay_tree_min,
17182 typed_splay_tree::splay_tree_max): Took over from splay-tree.c/.h.
17183 (typed_splay_tree::root, typed_splay_tree::comp,
17184 typed_splay_tree::delete_key,
17185 typed_splay_tree::delete_value): New data members.
17186 * typed-splay-tree.c (selftest::test_str_to_int): Add a test for
17187 typed_splay_tree::remove.
17189 2018-06-15 Matthew Fortune <matthew.fortune@mips.com>
17191 * config/mips/mips.h (ASM_SPEC): Pass through -mcrc, -mno-crc,
17192 -mginv and -mno-ginv to the assembler.
17193 * config/mips/mips.opt (-mcrc): New option.
17194 (-mginv): Likewise.
17195 * doc/invoke.text (-mcrc): Document.
17196 (-mginv): Likewise.
17198 2018-06-15 Nick Clifton <nickc@redhat.com>
17201 * tree.c (escaped_string): New class. Converts an unescaped
17202 string into its escaped equivalent.
17203 (warn_deprecated_use): Use the new class to convert the
17204 deprecation message, if present.
17205 (test_escaped_strings): New self test.
17206 (test_c_tests): Add test_escaped_strings.
17207 * doc/extend.texi (deprecated): Add a note that the
17208 deprecation message is affected by the -fmessage-length
17209 option, and that control characters will be escaped.
17210 (#pragma GCC error): Document this pragma.
17211 (#pragma GCC warning): Likewise.
17212 * doc/invoke.texi (-fmessage-length): Document this option's
17213 effect on the #warning and #error preprocessor directives and
17214 the deprecated attribute.
17216 2018-06-15 Richard Biener <rguenther@suse.de>
17218 * tree-vect-slp.c (vect_slp_bb): Dump MSG_OPTIMIZED_LOCATIONS
17219 here, also noting vector size used.
17220 * tree-vectorizer.c (vectorize_loops): Adjust. Note vector
17221 size used in MSG_OPTIMIZED_LOCATIONS dump.
17222 (pass_slp_vectorize::execute): Adjust.
17224 2018-06-15 Claudiu Zissulescu <claziss@synopsys.com>
17227 * config/arc/arc.c (arc_return_address_register): Fix
17230 2018-06-15 Richard Biener <rguenther@suse.de>
17232 PR middle-end/86159
17233 * tree-cfg.c (gimplify_build3): Do not strip sign conversions,
17234 leave useless conversion stripping to force_gimple_operand_gsi.
17235 (gimplify_build2): Likewise.
17236 (gimplify_build1): Likewise.
17238 2018-06-15 Richard Biener <rguenther@suse.de>
17240 PR middle-end/86076
17241 * tree-cfg.c (move_stmt_op): unshare invariant addresses
17242 before adjusting their block.
17244 2018-06-15 Sebastian Huber <sebastian.huber@embedded-brains.de>
17246 * config.gcc (riscv*-*-elf* | riscv*-*-rtems*): Use custom
17247 multilibs for *-*-rtems*.
17248 * config/riscv/t-rtems: New file.
17250 2018-06-14 Jakub Jelinek <jakub@redhat.com>
17252 PR middle-end/86122
17253 * match.pd ((A +- CST1) +- CST2): Punt if last resort
17254 unsigned_type_for returns NULL.
17257 * lower-subreg.c (find_decomposable_subregs): Don't decompose float
17258 subregs of multi-word pseudos unless the float mode has word size.
17260 2018-06-14 Richard Biener <rguenther@suse.de>
17262 PR middle-end/86139
17263 * tree-vect-generic.c (build_word_mode_vector_type): Remove
17264 duplicate and harmful type_hash_canon.
17265 * tree.c (type_hash_canon): Assert we didn't find ourselves.
17267 2018-06-14 Richard Biener <rguenther@suse.de>
17270 * tree-ssa-struct-alias.c (create_variable_info_for): Handle
17273 2018-06-14 Sebastian Huber <sebastian.huber@embedded-brains.de>
17275 * config/rtems.h (STDINT_LONG32): Define.
17277 2018-06-13 Matthew Fortune <matthew.fortune@mips.com>
17278 Prachi Godbole <prachi.godbole@imgtec.com>
17280 * config/mips/mips-cpus.def: Define P6600.
17281 * config/mips/mips-tables.opt: Regenerate.
17282 * config/mips/mips.c (mips_ucbranch_type): New enum.
17283 (mips_rtx_cost_data): Add support for P6600.
17284 (mips_issue_rate): Likewise.
17285 (mips_multipass_dfa_lookahead): Likewise.
17286 (mips_avoid_hazard): Likewise.
17287 (mips_reorg_process_insns): Likewise.
17288 (mips_classify_branch_p6600): New function.
17289 * config/mips/mips.h (TUNE_P6600): New define.
17290 (MIPS_ISA_LEVEL_SPEC): Infer mips64r6 from p6600.
17291 (ENABLE_LD_ST_PAIRS): Enable load/store bonding for p6600.
17292 * config/mips/mips.md: Include p6600.md.
17293 (processor): Add p6600.
17294 * config/mips/p6600.md: New file.
17295 * doc/invoke.texi: Add p6600 to supported architectures.
17297 2018-06-13 Martin Sebor <msebor@redhat.com>
17299 PR tree-optimization/86114
17300 * gimple-fold.c (gimple_fold_builtin_strlen): Only handle LHS
17302 * tree-ssa-strlen.c (maybe_set_strlen_range): Same.
17304 2018-06-13 Richard Biener <rguenther@suse.de>
17306 * tree-vect-patterns.c (vect_recog_vector_vector_shift_pattern):
17307 Properly set vector type of the intermediate stmt.
17308 * tree-vect-stmts.c (vectorizable_operation): The destination
17309 var always has vectype_out type.
17311 2018-06-13 Jeff Law <law@redhat.com>
17313 * config/rl78/rl78.c (move_elim_pass): Use TDF_NONE rather than
17314 integer 0 for argument to print_rtl_with_bb.
17315 (rl78_reorg): Likewise.
17317 2018-06-13 David Malcolm <dmalcolm@redhat.com>
17319 * config/arc/arc.c (hwloop_optimize): Strengthen local "end_label"
17320 from rtx to rtx_insn *.
17321 * config/bfin/bfin.c (hwloop_optimize): Likewise for local
17323 (add_sched_insns_for_speculation): Likewise for local "target",
17324 converting usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
17325 * config/c6x/c6x.c (reorg_split_calls): Strengthen param "call_labels"
17326 from rtx_insn ** to rtx_code_label **.
17327 (reorg_emit_nops): Likewise.
17328 (c6x_reorg): Likewise for local "call_labels".
17329 * config/sh/sh-protos.h (get_dest_uid): Strengthen 1st param from
17331 * config/sh/sh.c (dump_table): Strengthen local "lab" from rtx to
17332 rtx_code_label *, adding safe_as_a <rtx_code_label *> casts to
17333 the loops over LABEL_REFS.
17334 (fixup_addr_diff_vecs): Add as_a <rtx_insn *> to usage of
17336 (barrier_align): Convert usage of JUMP_LABEL to JUMP_LABEL_AS_INSN.
17337 (get_dest_uid): Strengthen param "label" from rtx to rtx_insn *.
17338 (split_branches): Strengthen local "olabel" from rtx to
17339 rtx_insn *, adding a safe_as_a cast.
17340 * emit-rtl.c (next_real_insn): Strengthen param from "rtx"
17342 (add_insn_after): Likewise for first two params.
17343 (add_insn_before): Likewise.
17344 (remove_insn): Likewise for param.
17345 (emit_pattern_before_noloc): Likewise for second and third params.
17346 (emit_jump_insn_before_noloc): Convert NULL_RTX to NULL.
17347 (emit_call_insn_before_noloc): Likewise.
17348 (emit_debug_insn_before_noloc): Strengthen "before" param from "rtx"
17350 (emit_barrier_before): Likewise.
17351 (emit_label_before): Strengthen "label" param from "rtx" to
17352 "rtx_code_label *". Strengthen "before" param from "rtx" to
17354 (emit_insn_after_1): Strengthen "after" param from "rtx" to
17356 (emit_pattern_after_noloc): Likewise.
17357 (emit_insn_after_noloc): Likewise.
17358 (emit_jump_insn_after_noloc): Likewise.
17359 (emit_call_insn_after_noloc): Likewise.
17360 (emit_debug_insn_after_noloc): Likewise.
17361 (emit_barrier_after): Likewise.
17362 (emit_label_after): Likewise for both params.
17363 (emit_pattern_after_setloc): Likewise for "after" param. Convert
17364 "loc" param from "int" to "location_t".
17365 (emit_insn_after_setloc): Likewise.
17366 (emit_jump_insn_after_setloc): Likewise.
17367 (emit_call_insn_after_setloc): Likewise.
17368 (emit_debug_insn_after_setloc): Likewise.
17369 (emit_pattern_before_setloc): Likewise for "before" param. Convert
17370 "loc" param from "int" to "location_t".
17371 (emit_pattern_before): Convert NULL_RTX to NULL.
17372 (emit_insn_before_setloc): Convert "loc" param from "int" to
17374 (emit_jump_insn_before_setloc): Likewise.
17375 (emit_call_insn_before_setloc): Likewise.
17376 (emit_debug_insn_before_setloc): Strengthen "before" param from rtx to
17377 rtx_insn *. Convert "loc" param from "int" to "location_t".
17378 * rtl.h (emit_insn_before_setloc, emit_jump_insn_before_setloc,
17379 emit_call_insn_before_setloc, emit_debug_insn_before_setloc):
17380 Convert 3rd param from "int" to "location_t".
17381 (emit_barrier_before, emit_barrier_after, next_real_insn):
17382 Strengthen param from rtx to rtx_insn *.
17383 (emit_label_before): Strengthen 1st param from "rtx" to
17384 "rtx_code_label *". Strengthen 2nd param from "rtx" to
17386 (emit_insn_after_noloc, emit_jump_insn_after_noloc,
17387 emit_call_insn_after_noloc, emit_debug_insn_after_noloc):
17388 Strengthen 2nd param from "rtx" to "rtx_insn *".
17389 (emit_insn_after_setloc, emit_jump_insn_after_setloc)
17390 emit_call_insn_after_setloc, emit_debug_insn_after_setloc):
17391 Likewise. Convert 3rd param from "int" to "location_t".
17392 (emit_label_after): Strengthen 1st param from "rtx" to
17393 "rtx_code_label *".
17394 (next_real_insn, remove_insn): Strengthen param from "rtx" to
17396 (add_insn_before, add_insn_after): Strengthen 1st and 2nd params
17397 from "rtx" to "rtx_insn *".
17399 2018-06-13 Jan Hubicka <hubicka@gcc.gnu.org>
17401 * cgraph.c (cgraph_node::get_untransformed_body): Dump function
17402 bodies streamed in with -Q.
17403 * dumpfile.c (dump_files): Add lto-stream-out dump file.
17404 * dumpfile.h (tree_dump_index): Add lto_stream_out.
17405 * gimple-streamer-out.c: Include gimple-pretty-print.h
17406 (output_bb): Dump stmts streamed.
17407 * lto-section-out.c: Include print-tree.h
17408 (lto_begin_section): Dump sections created.
17409 (lto_output_decl_index): Dump decl encoded.
17410 * lto-streamer-out.c: Include print-tree.h
17411 (create_output_block): Dump output block created.
17412 (DFS::DFS_write_tree_body): Dump DFS SCCs streamed.
17413 (output_function): Dump function output.
17414 (output_constructor): Dump constructor streamed.
17415 (write_global_stream): Output indexes encoded.
17416 (produce_asm_for_decls): Dump streams encoded.
17417 * lto-streamer.c (streamer_dump_file): New global var.
17418 * lto-streamer.h (streamer_dump_file): Declare.
17419 * passes.c (ipa_write_summaries): Initialize streamer dump.
17420 * varpool.c (varpool_node::get_constructor): Dump constructors streamed
17423 2018-06-13 Eric Botcazou <ebotcazou@adacore.com>
17426 * config/i386/winnt.c (i386_pe_seh_cold_init): Do not emit negative
17427 offsets for register save directives. Emit a second batch of save
17428 directives, if need be, when the function accesses prior frames.
17430 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17432 * config/arc/fpu.md (fmasf4): Force operand to register.
17433 (fnmasf4): Likewise.
17435 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17437 * config/arc/arc-protos.h (arc_pad_return): Remove.
17438 * config/arc/arc.c (machine_function): Remove force_short_suffix
17440 (arc_print_operand): Adjust printing of '&'.
17441 (arc_verify_short): Remove conditional printing of short suffix.
17442 (arc_final_prescan_insn): Remove reference to size_reason.
17443 (pad_return): New function.
17444 (arc_reorg): Call pad_return.
17445 (arc_pad_return): Remove.
17446 (arc_init_machine_status): Remove reference to force_short_suffix.
17447 * config/arc/arc.md (vunspec): Add VUNSPEC_ARC_BLOCKAGE.
17448 (attr length): When attribute iscompact is true force to 2
17449 regardless; in the case of maybe check if we want to force the
17450 instruction to have 4 bytes length.
17451 (nopv): Change it to generate 4 byte long nop as well.
17452 (blockage): New pattern.
17453 (simple_return): Remove call to arc_pad_return.
17454 (p_return_i): Likewise.
17456 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17458 * config/arc/elf.h (LINK_GCC_C_SEQUENCE_SPEC): Define.
17460 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17462 * config/arc/builtins.def (SYNC): SYNC instruction is valid on all
17465 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17467 * config/arc/arc.c (atomic_exchangesi): EX instruction is default
17468 for ARC700 and ARCv2.
17470 2018-06-13 Chenghua Xu <paul.hua.gm@gmail.com>
17473 * config/mips/loongson.md (vec_setv4hi): Gen_lowpart for
17474 operands[2] instead of operands[1].
17477 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17479 * lra-constraints.c (simplify_operand_subreg): In the paradoxical
17480 case, check whether the outer register overlaps an unallocatable
17481 register, not just whether it fits the required class.
17483 2018-06-12 Richard Sandiford <richard.sandiford@linaro.org>
17485 * poly-int.h (can_div_trunc_p): Add new overload in which all values
17487 * alias.c (get_addr): Extend CONST_INT handling to poly_int_rtx_p.
17488 (memrefs_conflict_p): Likewise.
17489 (init_alias_analysis): Likewise.
17490 * cfgexpand.c (expand_debug_expr): Likewise.
17491 * combine.c (combine_simplify_rtx, force_int_to_mode): Likewise.
17492 * cse.c (fold_rtx): Likewise.
17493 * explow.c (adjust_stack, anti_adjust_stack): Likewise.
17494 * expr.c (emit_block_move_hints): Likewise.
17495 (clear_storage_hints, push_block, emit_push_insn): Likewise.
17496 (store_expr_with_bounds, reduce_to_bit_field_precision): Likewise.
17497 (emit_group_load_1): Use rtx_to_poly_int64 for group offsets.
17498 (emit_group_store): Likewise.
17499 (find_args_size_adjust): Use strip_offset. Use rtx_to_poly_int64
17500 to read the PRE/POST_MODIFY increment.
17501 * calls.c (store_one_arg): Use strip_offset.
17502 * rtlanal.c (rtx_addr_can_trap_p_1): Extend CONST_INT handling to
17504 (set_noop_p): Use rtx_to_poly_int64 for the elements selected
17506 * simplify-rtx.c (avoid_constant_pool_reference): Use strip_offset.
17507 (simplify_binary_operation_1): Extend CONST_INT handling to
17509 * var-tracking.c (compute_cfa_pointer): Take a poly_int64 rather
17510 than a HOST_WIDE_INT.
17511 (hard_frame_pointer_adjustment): Change from HOST_WIDE_INT to
17513 (adjust_mems, add_stores): Update accodingly.
17514 (vt_canonicalize_addr): Track polynomial offsets.
17515 (emit_note_insn_var_location): Likewise.
17516 (vt_add_function_parameter): Likewise.
17517 (vt_initialize): Likewise.
17519 2018-06-12 Jeff Law <law@redhat.com>
17521 * config.gcc (alpha*-*-freebsd*): Remove.
17522 * config/alpha/freebsd.h: Remove.
17524 2018-06-12 David Malcolm <dmalcolm@redhat.com>
17527 * spellcheck-tree.c (levenshtein_distance): Rename to...
17528 (get_edit_distance): ...this, and update for underlying renaming.
17529 * spellcheck-tree.h (levenshtein_distance): Rename to...
17530 (get_edit_distance): ...this.
17531 * spellcheck.c (levenshtein_distance): Rename to...
17532 (get_edit_distance): ...this. Convert from Levenshtein distance
17533 to Damerau-Levenshtein distance by supporting transpositions of
17534 adjacent characters. Rename "v1" to "v_next" and "v0" to
17536 (selftest::levenshtein_distance_unit_test_oneway): Rename to...
17537 (selftest::test_edit_distance_unit_test_oneway): ...this, and
17538 update for underlying renaming.
17539 (selftest::levenshtein_distance_unit_test): Rename to...
17540 (selftest::test_get_edit_distance_unit): ...this, and update for
17541 underlying renaming.
17542 (selftest::test_find_closest_string): Add example from PR 69968
17543 where transposition helps
17544 (selftest::test_metric_conditions): Update for renaming.
17545 (selftest::test_metric_conditions): Likewise.
17546 (selftest::spellcheck_c_tests): Likewise.
17547 * spellcheck.h (levenshtein_distance): Rename both overloads to...
17548 (get_edit_distance): ...this.
17549 (best_match::consider): Update for renaming.
17551 2018-06-12 Martin Sebor <msebor@redhat.com>
17553 PR tree-optimization/85259
17554 * builtins.c (compute_objsize): Handle constant offsets.
17555 * gimple-ssa-warn-restrict.c (maybe_diag_offset_bounds): Return
17556 true iff a warning has been issued.
17557 * gimple.h (gimple_nonartificial_location): New function.
17558 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Call
17559 gimple_nonartificial_location and handle -Wno-system-headers.
17560 (handle_builtin_stxncpy): Same.
17562 2018-06-12 Martin Sebor <msebor@redhat.com>
17565 * fold-const.c (operand_equal_p): Handle SAVE_EXPR.
17567 2018-06-12 Will Schmidt <will_schmidt@vnet.ibm.com>
17569 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
17570 BUILTIN_VEC_XST entries for pointer to double and long long.
17572 2018-06-12 H.J. Lu <hongjiu.lu@intel.com>
17575 * config/i386/gnu-user.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
17577 * config/i386/gnu-user64.h (TARGET_THREAD_SPLIT_STACK_OFFSET):
17580 2018-06-12 Martin Liska <mliska@suse.cz>
17582 * doc/options.texi: Document IntegerRange.
17584 2018-06-12 Martin Liska <mliska@suse.cz>
17586 * config/i386/i386.opt: Make MPX-related options as Deprecated.
17587 * opt-functions.awk: Handle Deprecated flag.
17588 * opts-common.c (decode_cmdline_option): Handle cl_deprecated
17590 (read_cmdline_option): Report warning for a deprecated option.
17591 * opts.h (struct cl_option): Add new field cl_deprecated.
17592 (CL_ERR_DEPRECATED): New.
17594 2018-06-12 Martin Liska <mliska@suse.cz>
17596 * doc/options.texi: Document Deprecated option flag.
17598 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17600 * config/arc/arc-arch.h (arc_extras): New enum.
17601 (arc_cpu_t):Add field extra.
17602 (arc_cpu_types): Consider the extras.
17603 * config/arc/arc-cpus.def: Add extras info.
17604 * config/arc/arc-opts.h (processor_type): Consider extra field.
17605 * config/arc/arc.c (arc_override_options): Handle extra field.
17607 2018-06-12 Claudiu Zissulescu <claziss@synopsys.com>
17609 * config/arc/arc-arch.h: Update ARC_OPTX macro.
17610 * config/arc/arc-options.def (ARC_OPTX): Introduce a new doc
17612 * config/arc/arc.c (arc_init): Update pic warning.
17613 (irq_range): Update irq range parsing warnings.
17614 (arc_override_options): Update various warning messages.
17615 (arc_handle_aux_attribute): Likewise.
17617 2018-06-12 Robert Suchanek <robert.suchanek@mips.com>
17619 * config/mips/i6400.md (i6400_fpu_fadd): Remove frint.
17621 2018-06-12 Jozef Lawrynowicz <jozef.l@mittosystems.com>
17623 * doc/sourcebuild.texi: Document usage of line number 0 in verify
17624 compiler messages directives.
17626 2018-06-12 Matthew Fortune <mfortune@gmail.com>
17628 * config/mips/mips-cpus.def: New MIPS_CPU for i6500.
17629 * config/mips/mips-tables.opt: Regenerate.
17630 * config/mips/mips.h (MIPS_ISA_LEVEL_SPEC): Mark i6500 as
17632 * doc/invoke.texi: Document -march=i6500.
17634 2018-06-12 Prachi Godbole <prachi.godbole@imgtec.com>
17636 * config/mips/i6400.md (i6400_gpmuldiv): Remove cpu_unit.
17637 (i6400_gpmul): Add cpu_unit.
17638 (i6400_gpdiv): Likewise.
17639 (i6400_msa_add_d): Update reservations.
17640 (i6400_msa_int_add) Likewise.
17641 (i6400_msa_short_logic3) Likewise.
17642 (i6400_msa_short_logic2) Likewise.
17643 (i6400_msa_short_logic) Likewise.
17644 (i6400_msa_move) Likewise.
17645 (i6400_msa_cmp) Likewise.
17646 (i6400_msa_short_float2) Likewise.
17647 (i6400_msa_div_d) Likewise.
17648 (i6400_msa_long_logic1) Likewise.
17649 (i6400_msa_long_logic2) Likewise.
17650 (i6400_msa_mult) Likewise.
17651 (i6400_msa_long_float2) Likewise.
17652 (i6400_msa_long_float4) Likewise.
17653 (i6400_msa_long_float5) Likewise.
17654 (i6400_msa_long_float8) Likewise.
17655 (i6400_fpu_fadd): Include frint type.
17656 (i6400_fpu_store): New define_insn_reservation.
17657 (i6400_fpu_load): Likewise.
17658 (i6400_fpu_move): Likewise.
17659 (i6400_fpu_fcmp): Likewise.
17660 (i6400_fpu_fmadd): Likewise.
17661 (i6400_int_mult): Include imul3nc type and update reservation.
17662 (i6400_int_div): Include idiv3 type and update reservation.
17663 (i6400_int_load): Update to check type not move_type.
17664 (i6400_int_store): Likewise.
17665 (i6400_int_prefetch): Set zero latency.
17667 2018-06-12 Eric Botcazou <ebotcazou@adacore.com>
17669 * gcc.c: Document new %@{...} sequence.
17670 (LINK_COMMAND_SPEC): Use it for the -L switches.
17671 (cpp_unique_options): Use it for the -I switches.
17672 (at_file_argbuf): New global variable.
17673 (in_at_file): Likewise.
17674 (alloc_args): Create at_file_argbuf.
17675 (clear_args): Truncate at_file_argbuf.
17676 (store_arg): If in_at_file, push the argument onto at_file_argbuf.
17677 (open_at_file): New function.
17678 (close_at_file): Likewise.
17679 (create_at_file): Delete.
17680 (do_spec_1) <'i'>: Use open_at_file/close_at_file.
17683 (validate_switches_from_spec): Deal with %@{...} sequence.
17684 (validate_switches): Likewise.
17685 (driver::finalize): Call clear_args.
17687 2018-06-11 Rasmus Villemoes <rasmus.villemoes@prevas.dk>
17689 * config/vx-common.h (USE_TM_CLONE_REGISTRY): #define to 0.
17691 2018-06-11 Martin Sebor <msebor@redhat.com>
17693 * doc/invoke.texi (-Wall): List -Wc++17-compat.
17694 (Wno-class-memaccess): Add @opindex.
17695 (Wno-templates, Wno-multiple-inheritance): Same.
17696 (Wno-virtual-inheritance, Wno-namespaces): Same.
17697 (Wno-pedantic, Wno-chkp, Wcoverage-mismatch): Same.
17698 (Wno-format-overflow, Wno-format-truncation): Same.
17699 (Wno-unused-local-typedefs, Walloc-size-larger-than): Same.
17700 (Wno-alloc-size-larger-than, Wframe-larger-than): Same
17701 (Wno-stack-usage, Wno-c++-compat, Wno-c++11-compat): Same.
17702 (Wno-c++14-compat, Wno-c++17-compat, Wno-openmp-simd): Same.
17703 (Wno-unsuffixed-float-constants, Wno-addr-space-convert): Same.
17704 (Wno-misspelled-isr): Same.
17706 2018-06-11 Martin Sebor <msebor@redhat.com>
17708 * PR tree-optimization/86083
17709 * tree-ssa-strlen.c (handle_char_store): Use tree_expr_nonzero_p.
17711 2018-06-11 Zhouyi Zhou <zhouzhouyi@gmail.com>
17713 * tree-eh.c (lower_eh_constructs_2): Add a comma to comment.
17715 2018-06-11 Segher Boessenkool <segher@kernel.crashing.org>
17718 * config/rs6000/rs6000.md (*movdi_internal32): Put constraint modifiers
17719 on the correct operand.
17720 (*movdi_internal64): Ditto.
17722 2018-06-11 Martin Liska <mliska@suse.cz>
17724 PR tree-optimization/86089
17725 * tree-ssa-strlen.c (get_string_length): Move back removed hunk.
17727 2018-06-11 Julia Koval <julia.koval@intel.com>
17729 * config/i386/vaesintrin.h (_mm_aesdec_epi128, _mm_aesdeclast_epi128)
17730 _mm_aesenc_epi128, _mm_aesenclast_epi128: Remove.
17731 * config/i386/vpclmulqdqintrin.h (_mm_clmulepi64_epi128): Remove.
17733 2018-06-11 Olivier Hainque <hainque@adacore.com>
17735 * dwarf2out.c (gen_compile_unit_die): Fallout to DW_LANG_Ada83
17736 for Ada with strict dwarf2.
17738 2018-06-08 Peter Bergner <bergner@vnet.ibm.com>
17741 * config/rs6000/rs6000.c (mem_operand_gpr): Enable PRE_INC and PRE_DEC
17744 2018-06-08 Jan Hubicka <hubicka@ucw.cz>
17746 * dumpfile.c (FIRST_ME_AUTO_NUMBERED_DUMP): Bump to 4.
17748 2018-06-08 David Edelsohn <dje.gcc@gmail.com>
17750 * config/rs6000/rs6000.c (rs6000_passes_ieee128): Protect with #if
17753 2018-06-08 Martin Liska <mliska@suse.cz>
17755 * tree-cfg.h (debug_function): Fix argument type to match
17758 2018-06-08 Martin Liska <mliska@suse.cz>
17760 * config/powerpcspe/powerpcspe.c (rs6000_xcoff_visibility):
17761 Remove usage of MPX-related (and removed) fields.
17762 * config/rs6000/rs6000.c (rs6000_xcoff_visibility): Likewise.
17764 2018-06-08 David Malcolm <dmalcolm@redhat.com>
17766 * cfg.c (debug): Use TDF_NONE rather than 0.
17767 * cfghooks.c (debug): Likewise.
17768 * dumpfile.c (DUMP_FILE_INFO): Likewise; also for OPTGROUP.
17769 (struct dump_option_value_info): Convert to...
17770 (struct kv_pair): ...this template type.
17771 (dump_options): Convert to kv_pair<dump_flags_t>; use TDF_NONE
17773 (optinfo_verbosity_options): Likewise.
17774 (optgroup_options): Convert to kv_pair<optgroup_flags_t>; use
17776 (gcc::dump_manager::dump_register): Use optgroup_flags_t rather
17777 than int for "optgroup_flags" param.
17778 (dump_generic_expr_loc): Use dump_flags_t rather than int for
17780 (dump_dec): Likewise.
17781 (dump_finish): Use TDF_NONE rather than 0.
17782 (gcc::dump_manager::opt_info_enable_passes): Use optgroup_flags_t
17783 rather than int for "optgroup_flags" param. Use TDF_NONE rather
17784 than 0. Update for change to option_ptr.
17785 (opt_info_switch_p_1): Convert "optgroup_flags" param from int *
17786 to optgroup_flags_t *. Use TDF_NONE and OPTGROUP_NONE rather than
17787 0. Update for changes to optinfo_verbosity_options and
17789 (opt_info_switch_p): Convert optgroup_flags from int to
17791 (dump_basic_block): Use dump_flags_t rather than int
17792 for "dump_kind" param.
17793 * dumpfile.h (TDF_ADDRESS, TDF_SLIM, TDF_RAW, TDF_DETAILS,
17794 TDF_STATS, TDF_BLOCKS, TDF_VOPS, TDF_LINENO, TDF_UID)
17795 TDF_STMTADDR, TDF_GRAPH, TDF_MEMSYMS, TDF_RHS_ONLY, TDF_ASMNAME,
17796 TDF_EH, TDF_NOUID, TDF_ALIAS, TDF_ENUMERATE_LOCALS, TDF_CSELIB,
17797 TDF_SCEV, TDF_GIMPLE, TDF_FOLDING, MSG_OPTIMIZED_LOCATIONS,
17798 MSG_MISSED_OPTIMIZATION, MSG_NOTE, MSG_ALL, TDF_COMPARE_DEBUG,
17799 TDF_NONE): Convert from macros to...
17800 (enum dump_flag): ...this new enum.
17801 (dump_flags_t): Update to use enum.
17802 (operator|, operator&, operator~, operator|=, operator&=):
17803 Implement for dump_flags_t.
17804 (OPTGROUP_NONE, OPTGROUP_IPA, OPTGROUP_LOOP, OPTGROUP_INLINE,
17805 OPTGROUP_OMP, OPTGROUP_VEC, OPTGROUP_OTHER, OPTGROUP_ALL):
17806 Convert from macros to...
17807 (enum optgroup_flag): ...this new enum.
17808 (optgroup_flags_t): New typedef.
17809 (operator|, operator|=): Implement for optgroup_flags_t.
17810 (struct dump_file_info): Convert field "alt_flags" to
17811 dump_flags_t. Convert field "optgroup_flags" to
17813 (dump_basic_block): Use dump_flags_t rather than int for param.
17814 (dump_generic_expr_loc): Likewise.
17815 (dump_dec): Likewise.
17816 (dump_register): Convert param "optgroup_flags" to
17818 (opt_info_enable_passes): Likewise.
17819 * early-remat.c (early_remat::dump_edge_list): Use TDF_NONE rather
17821 * gimple-pretty-print.c (debug): Likewise.
17822 * gimple-ssa-store-merging.c (bswap_replace): Likewise.
17823 (merged_store_group::apply_stores): Likewise.
17824 * gimple-ssa-strength-reduction.c (insert_initializers): Likewise.
17825 * gimple.c (verify_gimple_pp): Likewise.
17826 * graphite-poly.c (print_pbb_body): Likewise.
17827 * passes.c (pass_manager::register_one_dump_file): Convert
17828 local "optgroup_flags" to optgroup_flags_t.
17829 * print-tree.c (print_node): Use TDF_NONE rather than 0.
17831 (debug_body): Likewise.
17832 * tree-pass.h (struct pass_data): Convert field "optgroup_flags"
17833 to optgroup_flags_t.
17834 * tree-pretty-print.c (print_struct_decl): Use TDF_NONE rather
17836 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Likewise.
17837 (convert_mult_to_fma): Likewise.
17838 * tree-ssa-reassoc.c (undistribute_ops_list): Likewise.
17839 * tree-ssa-sccvn.c (vn_eliminate): Likewise.
17840 * tree-vect-data-refs.c (dump_lower_bound): Convert param
17841 "dump_kind" to dump_flags_t.
17843 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
17845 * config/rs6000/rs6000.c (min, max): Delete.
17847 2018-06-08 Segher Boessenkool <segher@kernel.crashing.org>
17849 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete mention of
17850 -mabi=spe and -mabi=no-spe.
17852 2018-06-08 Martin Liska <mliska@suse.cz>
17854 * ipa-pure-const.c (propagate_pure_const): Use ::get at places
17855 where we expect an existing summary.
17857 2018-06-08 Martin Liska <mliska@suse.cz>
17859 * ipa-inline-analysis.c (simple_edge_hints): Use ::get method.
17860 * ipa-inline.h (estimate_edge_growth): Likewise.
17862 2018-06-08 Martin Liska <mliska@suse.cz>
17864 * cgraph.c (function_version_hasher::hash): Use
17865 cgraph_node::get_uid ().
17866 (function_version_hasher::equal):
17867 * cgraph.h (cgraph_node::get_uid): New method.
17868 * ipa-inline.c (update_caller_keys): Use
17869 cgraph_node::get_uid ().
17870 (update_callee_keys): Likewise.
17871 * ipa-utils.c (searchc): Likewise.
17872 (ipa_reduced_postorder): Likewise.
17873 * lto-cgraph.c (input_node): Likewise.
17874 * passes.c (is_pass_explicitly_enabled_or_disabled): Likewise.
17875 * symbol-summary.h (symtab_insertion): Likewise.
17876 (symtab_removal): Likewise.
17877 (symtab_duplication): Likewise.
17878 * tree-pretty-print.c (dump_function_header): Likewise.
17879 * tree-sra.c (convert_callers_for_node): Likewise.
17881 2018-06-08 Martin Liska <mliska@suse.cz>
17883 * cgraph.c (symbol_table::create_edge): Always assign a new
17885 (symbol_table::free_edge): Do not recycle numbers.
17886 * cgraph.h (cgraph_edge::get): New method.
17887 * symbol-summary.h (symtab_removal): Use it.
17888 (symtab_duplication): Likewise.
17889 (call_summary::hashable_uid): Remove.
17891 2018-06-08 Martin Liska <mliska@suse.cz>
17893 * ipa-inline-analysis.c (inline_edge_removal_hook): Remove.
17894 (initialize_growth_caches): Remove.
17895 (free_growth_caches): Likewise.
17896 (do_estimate_edge_time): Use edge_growth_cache.
17897 (do_estimate_edge_size): Likewise.
17898 (do_estimate_edge_hints): Likewise.
17899 * ipa-inline.c (reset_edge_caches): Likewise.
17900 (recursive_inlining): Likewise.
17901 (inline_small_functions): Likewise.
17902 * ipa-inline.h (initialize_growth_caches): Remove.
17903 (estimate_edge_size): Likewise.
17904 (estimate_edge_time): Likewise.
17905 (estimate_edge_hints): Likewise.
17906 (reset_edge_growth_cache): Likewise.
17907 * symbol-summary.h (call_summary::remove): New method.
17909 2018-06-08 Martin Liska <mliska@suse.cz>
17911 * ipa-cp.c (class edge_clone_summary): New summary.
17912 (grow_edge_clone_vectors): Remove.
17913 (ipcp_edge_duplication_hook): Remove.
17914 (class edge_clone_summary_t): New call_summary class.
17915 (ipcp_edge_removal_hook): Remove.
17916 (edge_clone_summary_t::duplicate): New function.
17917 (get_next_cgraph_edge_clone): Use edge_clone_summaries.
17918 (create_specialized_node): Likewise.
17919 (ipcp_driver): Initialize edge_clone_summaries and do not
17922 2018-06-08 Martin Liska <mliska@suse.cz>
17924 * symbol-summary.h (get): New function.
17925 (call_summary::m_initialize_when_cloning): New class member.
17927 2018-06-08 Martin Liska <mliska@suse.cz>
17929 * cgraph.c (cgraph_node::remove): Do not recycle uid.
17930 * cgraph.h (symbol_table::release_symbol): Do not pass uid.
17931 (symbol_table::allocate_cgraph_symbol): Do not set uid.
17932 * passes.c (uid_hash_t): Record removed_nodes by their uids.
17933 (remove_cgraph_node_from_order): Use the removed_nodes set.
17934 (do_per_function_toporder): Likwise.
17935 * symbol-summary.h (symtab_insertion): Use cgraph_node::uid
17936 instead of summary_uid.
17937 (symtab_removal): Likewise.
17938 (symtab_duplication): Likewise.
17940 2018-06-08 Martin Liska <mliska@suse.cz>
17942 * ipa-cp.c (ipcp_store_bits_results): Use
17943 ipcp_transformation_sum.
17944 (ipcp_store_vr_results): Likewise.
17945 * ipa-prop.c (ipcp_grow_transformations_if_necessary): Renamed
17947 (ipcp_transformation_initialize): ... this.
17948 (ipa_set_node_agg_value_chain):
17949 (ipa_node_params_t::duplicate): Use ipcp_transformation_sum.
17950 (write_ipcp_transformation_info): Likewise.
17951 (read_ipcp_transformation_info): Likewise.
17952 (ipcp_update_bits): Likewise.
17953 (ipcp_update_vr): Likewise.
17954 (ipcp_transform_function): Likewise.
17955 * ipa-prop.h: Rename ipcp_transformation_summary to
17956 ipcp_transformation.
17957 (class ipcp_transformation_t): New function summary.
17958 (ipcp_get_transformation_summary): Use ipcp_transformation_sum.
17959 (ipa_get_agg_replacements_for_node): Likewise.
17961 2018-06-08 Martin Liska <mliska@suse.cz>
17963 * ipa-pure-const.c (struct funct_state_d): Do it class instead
17965 (class funct_state_summary_t): New function_summary class.
17966 (has_function_state): Remove.
17967 (get_function_state): Likewise.
17968 (set_function_state): Likewise.
17969 (add_new_function): Likewise.
17970 (funct_state_summary_t::insert): New function.
17971 (duplicate_node_data): Remove.
17972 (remove_node_data): Remove.
17973 (funct_state_summary_t::duplicate): New function.
17974 (register_hooks): Create new funct_state_summaries.
17975 (pure_const_generate_summary): Use it.
17976 (pure_const_write_summary): Likewise.
17977 (pure_const_read_summary): Likewise.
17978 (propagate_pure_const): Likewise.
17979 (propagate_nothrow): Likewise.
17980 (dump_malloc_lattice): Likewise.
17981 (propagate_malloc): Likewise.
17982 (execute): Do not register hooks, just remove summary
17984 (pass_ipa_pure_const::pass_ipa_pure_const): Simplify
17987 2018-06-08 Martin Liska <mliska@suse.cz>
17989 * ipa-reference.c (remove_node_data): Remove.
17990 (duplicate_node_data): Likewise.
17991 (class ipa_ref_var_info_summary_t): New class.
17992 (class ipa_ref_opt_summary_t): Likewise.
17993 (get_reference_vars_info): Use ipa_ref_var_info_summaries.
17994 (get_reference_optimization_summary): Use
17995 ipa_ref_opt_sum_summaries.
17996 (set_reference_vars_info): Remove.
17997 (set_reference_optimization_summary): Likewise.
17998 (ipa_init): Create summaries.
17999 (init_function_info): Use function summary.
18000 (ipa_ref_opt_summary_t::duplicate): New function.
18001 (ipa_ref_opt_summary_t::remove): New function.
18002 (get_read_write_all_from_node): Fix GNU coding style.
18003 (propagate): Use function summary.
18004 (write_node_summary_p): Fix GNU coding style.
18005 (stream_out_bitmap): Likewise.
18006 (ipa_reference_read_optimization_summary): Use function summary.
18007 (ipa_reference_c_finalize): Do not release hooks.
18009 2018-06-08 Martin Liska <mliska@suse.cz>
18011 * ipa-fnsummary.c (dump_ipa_call_summary): Use ::get method.
18012 (analyze_function_body): Extract multiple calls of get_create.
18013 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
18014 * ipa-inline.c (recursive_inlining): Use ::get method.
18015 * ipa-inline.h (estimate_edge_growth): Likewise.
18017 2018-06-08 Martin Liska <mliska@suse.cz>
18019 * hsa-common.h (enum hsa_function_kind): Rename HSA_NONE to
18021 (hsa_function_summary::hsa_function_summary): Use the new enum
18023 (hsa_gpu_implementation_p): Use hsa_summaries::get.
18024 * hsa-gen.c (hsa_get_host_function): Likewise.
18025 (get_brig_function_name): Likewise.
18026 * ipa-hsa.c (process_hsa_functions): Likewise.
18027 (ipa_hsa_write_summary): Likewise.
18028 * symbol-summary.h (symtab_duplication): Use ::get function/
18029 (get): New function.
18031 2018-06-08 Martin Liska <mliska@suse.cz>
18033 * config/i386/i386.c (ix86_can_inline_p): Use get_create instead
18035 * hsa-common.c (hsa_summary_t::link_functions): Likewise.
18036 (hsa_register_kernel): Likewise.
18037 * hsa-common.h (hsa_gpu_implementation_p): Likewise.
18038 * hsa-gen.c (hsa_get_host_function): Likewise.
18039 (get_brig_function_name): Likewise.
18040 (generate_hsa): Likewise.
18041 (pass_gen_hsail::execute): Likewise.
18042 * ipa-cp.c (ipcp_cloning_candidate_p): Likewise.
18043 (devirtualization_time_bonus): Likewise.
18044 (ipcp_propagate_stage): Likewise.
18045 * ipa-fnsummary.c (redirect_to_unreachable): Likewise.
18046 (edge_set_predicate): Likewise.
18047 (evaluate_conditions_for_known_args): Likewise.
18048 (evaluate_properties_for_edge): Likewise.
18049 (ipa_fn_summary::reset): Likewise.
18050 (ipa_fn_summary_t::duplicate): Likewise.
18051 (dump_ipa_call_summary): Likewise.
18052 (ipa_dump_fn_summary): Likewise.
18053 (analyze_function_body): Likewise.
18054 (compute_fn_summary): Likewise.
18055 (estimate_edge_devirt_benefit): Likewise.
18056 (estimate_edge_size_and_time): Likewise.
18057 (estimate_calls_size_and_time): Likewise.
18058 (estimate_node_size_and_time): Likewise.
18059 (inline_update_callee_summaries): Likewise.
18060 (remap_edge_change_prob): Likewise.
18061 (remap_edge_summaries): Likewise.
18062 (ipa_merge_fn_summary_after_inlining): Likewise.
18063 (ipa_update_overall_fn_summary): Likewise.
18064 (read_ipa_call_summary): Likewise.
18065 (inline_read_section): Likewise.
18066 (write_ipa_call_summary): Likewise.
18067 (ipa_fn_summary_write): Likewise.
18068 (ipa_free_fn_summary): Likewise.
18069 * ipa-hsa.c (process_hsa_functions): Likewise.
18070 (ipa_hsa_write_summary): Likewise.
18071 (ipa_hsa_read_section): Likewise.
18072 * ipa-icf.c (sem_function::merge): Likewise.
18073 * ipa-inline-analysis.c (simple_edge_hints): Likewise.
18074 (do_estimate_edge_time): Likewise.
18075 (estimate_size_after_inlining): Likewise.
18076 (estimate_growth): Likewise.
18077 (growth_likely_positive): Likewise.
18078 * ipa-inline-transform.c (clone_inlined_nodes): Likewise.
18079 (inline_call): Likewise.
18080 * ipa-inline.c (caller_growth_limits): Likewise.
18081 (can_inline_edge_p): Likewise.
18082 (can_inline_edge_by_limits_p): Likewise.
18083 (compute_uninlined_call_time): Likewise.
18084 (compute_inlined_call_time): Likewise.
18085 (want_inline_small_function_p): Likewise.
18086 (edge_badness): Likewise.
18087 (update_caller_keys): Likewise.
18088 (update_callee_keys): Likewise.
18089 (recursive_inlining): Likewise.
18090 (inline_small_functions): Likewise.
18091 (inline_to_all_callers_1): Likewise.
18092 (dump_overall_stats): Likewise.
18093 (early_inline_small_functions): Likewise.
18094 (early_inliner): Likewise.
18095 * ipa-inline.h (estimate_edge_growth): Likewise.
18096 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
18097 * ipa-prop.c (ipa_make_edge_direct_to_target): Likewise.
18098 * ipa-prop.h (IPA_NODE_REF): Likewise.
18099 (IPA_EDGE_REF): Likewise.
18100 * ipa-pure-const.c (malloc_candidate_p): Likewise.
18101 (propagate_malloc): Likewise.
18102 * ipa-split.c (execute_split_functions): Likewise.
18103 * symbol-summary.h: Rename get to get_create.
18105 (get_create): Likewise.
18106 * tree-sra.c (ipa_sra_preliminary_function_checks): Likewise.
18108 2018-06-08 Martin Liska <mliska@suse.cz>
18110 * symbol-summary.h (release): Move definition out of class
18112 (symtab_removal): Likewise.
18113 (symtab_duplication): Likewise.
18115 2018-06-08 Martin Liska <mliska@suse.cz>
18117 * symbol-summary.h (function_summary): Move constructor
18118 implementation out of class declaration.
18119 (release): Likewise.
18120 (symtab_insertion): Likewise.
18121 (symtab_removal): Likewise.
18122 (symtab_duplication): Likewise.
18125 2018-06-08 Martin Liska <mliska@suse.cz>
18127 * Makefile.in: Remove support for MPX (macros, related functions,
18128 fields in cgraph_node, ...).
18129 * builtin-types.def (BT_BND): Likewise.
18130 (BT_FN_BND_CONST_PTR): Likewise.
18131 (BT_FN_CONST_PTR_BND): Likewise.
18132 (BT_FN_VOID_PTR_BND): Likewise.
18133 (BT_FN_BND_CONST_PTR_SIZE): Likewise.
18134 (BT_FN_VOID_CONST_PTR_BND_CONST_PTR): Likewise.
18135 * builtins.c (expand_builtin_memcpy_with_bounds): Likewise.
18136 (expand_builtin_mempcpy_with_bounds): Likewise.
18137 (expand_builtin_memset_with_bounds): Likewise.
18138 (expand_builtin_memset_args): Likewise.
18139 (std_expand_builtin_va_start): Likewise.
18140 (expand_builtin): Likewise.
18141 (expand_builtin_with_bounds): Likewise.
18142 * builtins.def (DEF_BUILTIN_CHKP): Likewise.
18143 (DEF_LIB_BUILTIN_CHKP): Likewise.
18144 (DEF_EXT_LIB_BUILTIN_CHKP): Likewise.
18145 (DEF_CHKP_BUILTIN): Likewise.
18146 (BUILT_IN_MEMCPY): Likewise.
18147 (BUILT_IN_MEMMOVE): Likewise.
18148 (BUILT_IN_MEMPCPY): Likewise.
18149 (BUILT_IN_MEMSET): Likewise.
18150 (BUILT_IN_STPCPY): Likewise.
18151 (BUILT_IN_STRCAT): Likewise.
18152 (BUILT_IN_STRCHR): Likewise.
18153 (BUILT_IN_STRCPY): Likewise.
18154 (BUILT_IN_STRLEN): Likewise.
18155 (BUILT_IN_MEMCPY_CHK): Likewise.
18156 (BUILT_IN_MEMMOVE_CHK): Likewise.
18157 (BUILT_IN_MEMPCPY_CHK): Likewise.
18158 (BUILT_IN_MEMSET_CHK): Likewise.
18159 (BUILT_IN_STPCPY_CHK): Likewise.
18160 (BUILT_IN_STRCAT_CHK): Likewise.
18161 (BUILT_IN_STRCPY_CHK): Likewise.
18162 * calls.c (store_bounds): Likewise.
18163 (emit_call_1): Likewise.
18164 (special_function_p): Likewise.
18165 (maybe_warn_nonstring_arg): Likewise.
18166 (initialize_argument_information): Likewise.
18167 (finalize_must_preallocate): Likewise.
18168 (compute_argument_addresses): Likewise.
18169 (expand_call): Likewise.
18170 * cfgexpand.c (expand_call_stmt): Likewise.
18171 (expand_return): Likewise.
18172 (expand_gimple_stmt_1): Likewise.
18173 (pass_expand::execute): Likewise.
18174 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
18175 (cgraph_node::remove): Likewise.
18176 (cgraph_node::dump): Likewise.
18177 (cgraph_node::verify_node): Likewise.
18178 * cgraph.h (chkp_function_instrumented_p): Likewise.
18179 (symtab_node::get_alias_target): Likewise.
18180 (cgraph_node::can_remove_if_no_direct_calls_and_refs_p): Likewise.
18181 (cgraph_local_p): Likewise.
18182 * cgraphbuild.c (cgraph_edge::rebuild_edges): Likewise.
18183 (cgraph_edge::rebuild_references): Likewise.
18184 * cgraphunit.c (varpool_node::finalize_decl): Likewise.
18185 (walk_polymorphic_call_targets): Likewise.
18186 (cgraph_node::expand_thunk): Likewise.
18187 (symbol_table::output_weakrefs): Likewise.
18188 * common/config/i386/i386-common.c (OPTION_MASK_ISA2_GENERAL_REGS_ONLY_UNSET): Likewise.
18189 (ix86_handle_option): Likewise.
18190 * config/i386/constraints.md: Likewise.
18191 * config/i386/i386-builtin-types.def (BND): Likewise.
18195 * config/i386/i386-builtin.def (BDESC_END): Likewise.
18196 (BDESC_FIRST): Likewise.
18198 * config/i386/i386-c.c (ix86_target_macros_internal): Likewise.
18199 * config/i386/i386-protos.h (ix86_bnd_prefixed_insn_p): Likewise.
18200 * config/i386/i386.c (enum reg_class): Likewise.
18201 (ix86_target_string): Likewise.
18202 (ix86_option_override_internal): Likewise.
18203 (ix86_conditional_register_usage): Likewise.
18204 (ix86_valid_target_attribute_inner_p): Likewise.
18205 (ix86_set_indirect_branch_type): Likewise.
18206 (ix86_set_current_function): Likewise.
18207 (ix86_function_arg_regno_p): Likewise.
18208 (init_cumulative_args): Likewise.
18209 (ix86_function_arg_advance): Likewise.
18210 (ix86_function_arg): Likewise.
18211 (ix86_pass_by_reference): Likewise.
18212 (ix86_function_value_regno_p): Likewise.
18213 (ix86_function_value_1): Likewise.
18214 (ix86_function_value_bounds): Likewise.
18215 (ix86_return_in_memory): Likewise.
18216 (ix86_setup_incoming_vararg_bounds): Likewise.
18217 (ix86_va_start): Likewise.
18218 (indirect_thunk_need_prefix): Likewise.
18219 (print_reg): Likewise.
18220 (ix86_print_operand): Likewise.
18221 (ix86_expand_call): Likewise.
18222 (ix86_output_function_return): Likewise.
18223 (reg_encoded_number): Likewise.
18224 (BDESC_VERIFYS): Likewise.
18225 (ix86_init_mpx_builtins): Likewise.
18226 (ix86_init_builtins): Likewise.
18227 (ix86_emit_cmove): Likewise.
18228 (ix86_emit_move_max): Likewise.
18229 (ix86_expand_builtin): Likewise.
18230 (ix86_builtin_mpx_function): Likewise.
18231 (ix86_get_arg_address_for_bt): Likewise.
18232 (ix86_load_bounds): Likewise.
18233 (ix86_store_bounds): Likewise.
18234 (ix86_load_returned_bounds): Likewise.
18235 (ix86_store_returned_bounds): Likewise.
18236 (ix86_class_likely_spilled_p): Likewise.
18237 (ix86_hard_regno_mode_ok): Likewise.
18238 (x86_order_regs_for_local_alloc): Likewise.
18239 (ix86_mitigate_rop): Likewise.
18240 (ix86_bnd_prefixed_insn_p): Likewise.
18241 (ix86_mpx_bound_mode): Likewise.
18242 (ix86_make_bounds_constant): Likewise.
18243 (ix86_initialize_bounds): Likewise.
18244 (TARGET_LOAD_BOUNDS_FOR_ARG): Likewise.
18245 (TARGET_STORE_BOUNDS_FOR_ARG): Likewise.
18246 (TARGET_LOAD_RETURNED_BOUNDS): Likewise.
18247 (TARGET_STORE_RETURNED_BOUNDS): Likewise.
18248 (TARGET_CHKP_BOUND_MODE): Likewise.
18249 (TARGET_BUILTIN_CHKP_FUNCTION): Likewise.
18250 (TARGET_CHKP_FUNCTION_VALUE_BOUNDS): Likewise.
18251 (TARGET_CHKP_MAKE_BOUNDS_CONSTANT): Likewise.
18252 (TARGET_CHKP_INITIALIZE_BOUNDS): Likewise.
18253 * config/i386/i386.h (TARGET_MPX): Likewise.
18254 (TARGET_MPX_P): Likewise.
18255 (VALID_BND_REG_MODE): Likewise.
18256 (FIRST_BND_REG): Likewise.
18257 (LAST_BND_REG): Likewise.
18258 (enum reg_class): Likewise.
18259 (BND_REG_P): Likewise.
18260 (BND_REGNO_P): Likewise.
18261 (BNDmode): Likewise.
18262 (ADJUST_INSN_LENGTH): Likewise.
18263 * config/i386/i386.md: Likewise.
18264 * config/i386/i386.opt: Likewise.
18265 * config/i386/linux-common.h (LIBMPX_LIBS): Likewise.
18266 (defined): Likewise.
18267 (LINK_MPX): Likewise.
18268 (MPX_SPEC): Likewise.
18269 (LIBMPX_SPEC): Likewise.
18270 (LIBMPXWRAPPERS_SPEC): Likewise.
18271 (CHKP_SPEC): Likewise.
18272 * config/i386/predicates.md: Likewise.
18273 * dbxout.c (dbxout_type): Likewise.
18274 * doc/extend.texi: Likewise.
18275 * doc/invoke.texi: Likewise.
18276 * doc/md.texi: Likewise.
18277 * doc/tm.texi: Likewise.
18278 * doc/tm.texi.in: Likewise.
18279 * dwarf2out.c (is_base_type): Likewise.
18280 (gen_formal_types_die): Likewise.
18281 (gen_subprogram_die): Likewise.
18282 (gen_type_die_with_usage): Likewise.
18283 (gen_decl_die): Likewise.
18284 (dwarf2out_late_global_decl): Likewise.
18285 * expr.c (expand_assignment): Likewise.
18286 (emit_storent_insn): Likewise.
18287 (store_expr_with_bounds): Likewise.
18288 (store_expr): Likewise.
18289 (expand_expr_real_1): Likewise.
18290 * expr.h (store_expr_with_bounds): Likewise.
18291 * function.c (use_register_for_decl): Likewise.
18292 (struct bounds_parm_data): Likewise.
18293 (assign_parms_augmented_arg_list): Likewise.
18294 (assign_parm_find_entry_rtl): Likewise.
18295 (assign_parm_is_stack_parm): Likewise.
18296 (assign_parm_load_bounds): Likewise.
18297 (assign_bounds): Likewise.
18298 (assign_parms): Likewise.
18299 (expand_function_start): Likewise.
18300 * gcc.c (CHKP_SPEC): Likewise.
18301 * gimple-fold.c (gimple_fold_builtin_memory_op): Likewise.
18302 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Likewise.
18303 (wrestrict_dom_walker::check_call): Likewise.
18304 * gimple.c (gimple_build_call_from_tree): Likewise.
18305 * gimple.h (enum gf_mask): Likewise.
18306 (gimple_call_with_bounds_p): Likewise.
18307 (gimple_call_set_with_bounds): Likewise.
18308 * gimplify.c (gimplify_init_constructor): Likewise.
18309 * ipa-cp.c (initialize_node_lattices): Likewise.
18310 (propagate_constants_across_call): Likewise.
18311 (find_more_scalar_values_for_callers_subset): Likewise.
18312 * ipa-hsa.c (process_hsa_functions): Likewise.
18313 * ipa-icf-gimple.c (func_checker::compare_gimple_call): Likewise.
18314 * ipa-icf.c (sem_function::merge): Likewise.
18315 * ipa-inline.c (early_inliner): Likewise.
18316 * ipa-pure-const.c (warn_function_noreturn): Likewise.
18317 (warn_function_cold): Likewise.
18318 (propagate_pure_const): Likewise.
18319 * ipa-ref.h (enum GTY): Likewise.
18320 * ipa-split.c (find_retbnd): Likewise.
18321 (consider_split): Likewise.
18322 (split_function): Likewise.
18323 * ipa-visibility.c (cgraph_externally_visible_p): Likewise.
18324 * ipa.c (walk_polymorphic_call_targets): Likewise.
18325 (symbol_table::remove_unreachable_nodes): Likewise.
18326 (process_references): Likewise.
18327 (cgraph_build_static_cdtor_1): Likewise.
18328 * lto-cgraph.c (lto_output_node): Likewise.
18329 (output_refs): Likewise.
18330 (compute_ltrans_boundary): Likewise.
18331 (input_overwrite_node): Likewise.
18332 (input_node): Likewise.
18333 (input_cgraph_1): Likewise.
18334 * params.def (PARAM_CHKP_MAX_CTOR_SIZE): Likewise.
18335 * passes.c (pass_manager::execute_early_local_passes): Likewise.
18336 (class pass_chkp_instrumentation_passes): Likewise.
18337 (make_pass_chkp_instrumentation_passes): Likewise.
18338 * passes.def: Likewise.
18339 * rtl.h (struct GTY): Likewise.
18340 (CALL_EXPR_WITH_BOUNDS_P): Likewise.
18341 * stor-layout.c (layout_type): Likewise.
18342 * symtab.c: Likewise.
18343 * target.def: Likewise.
18344 * targhooks.c (default_chkp_bound_type): Likewise.
18345 (default_chkp_bound_mode): Likewise.
18346 (default_builtin_chkp_function): Likewise.
18347 (default_chkp_function_value_bounds): Likewise.
18348 (default_chkp_make_bounds_constant): Likewise.
18349 (default_chkp_initialize_bounds): Likewise.
18350 * targhooks.h (default_chkp_bound_type): Likewise.
18351 (default_chkp_bound_mode): Likewise.
18352 (default_builtin_chkp_function): Likewise.
18353 (default_chkp_function_value_bounds): Likewise.
18354 (default_chkp_make_bounds_constant): Likewise.
18355 (default_chkp_initialize_bounds): Likewise.
18356 * toplev.c (compile_file): Likewise.
18357 (process_options): Likewise.
18358 * tree-core.h (DEF_BUILTIN): Likewise.
18359 (DEF_BUILTIN_CHKP): Likewise.
18360 * tree-inline.c (declare_return_variable): Likewise.
18361 (remap_gimple_stmt): Likewise.
18362 (copy_bb): Likewise.
18363 (initialize_inlined_parameters): Likewise.
18364 (expand_call_inline): Likewise.
18365 * tree-pass.h (make_pass_ipa_chkp_versioning): Likewise.
18366 (make_pass_ipa_chkp_early_produce_thunks): Likewise.
18367 (make_pass_ipa_chkp_produce_thunks): Likewise.
18368 (make_pass_chkp): Likewise.
18369 (make_pass_chkp_opt): Likewise.
18370 (make_pass_chkp_instrumentation_passes): Likewise.
18371 * tree-pretty-print.c (dump_generic_node): Likewise.
18372 * tree-ssa-ccp.c (insert_clobber_before_stack_restore): Likewise.
18373 * tree-ssa-dce.c (propagate_necessity): Likewise.
18374 (eliminate_unnecessary_stmts): Likewise.
18375 * tree-ssa-pre.c (create_expression_by_pieces): Likewise.
18376 * tree-ssa-sccvn.c (copy_reference_ops_from_call): Likewise.
18377 * tree-ssa-sccvn.h: Likewise.
18378 * tree-ssa-strlen.c (get_string_length): Likewise.
18379 (valid_builtin_call): Likewise.
18380 (adjust_last_stmt): Likewise.
18381 (handle_builtin_strchr): Likewise.
18382 (handle_builtin_strcpy): Likewise.
18383 (handle_builtin_stxncpy): Likewise.
18384 (handle_builtin_memcpy): Likewise.
18385 (handle_builtin_strcat): Likewise.
18386 (strlen_check_and_optimize_stmt): Likewise.
18387 * tree-stdarg.c (expand_ifn_va_arg_1): Likewise.
18388 * tree-streamer-in.c: Likewise.
18389 * tree-streamer.c (record_common_node): Likewise.
18390 * tree.c (tree_code_size): Likewise.
18391 (wide_int_to_tree_1): Likewise.
18392 (type_contains_placeholder_1): Likewise.
18393 (build_common_tree_nodes): Likewise.
18394 * tree.def (POINTER_BOUNDS_TYPE): Likewise.
18395 * tree.h (POINTER_BOUNDS_TYPE_P): Likewise.
18396 (POINTER_BOUNDS_P): Likewise.
18397 (BOUNDED_TYPE_P): Likewise.
18398 (BOUNDED_P): Likewise.
18399 (CALL_WITH_BOUNDS_P): Likewise.
18400 (pointer_bounds_type_node): Likewise.
18401 * value-prof.c (gimple_ic): Likewise.
18402 * var-tracking.c (vt_add_function_parameters): Likewise.
18403 * varasm.c (make_decl_rtl): Likewise.
18404 (assemble_start_function): Likewise.
18405 (output_constant): Likewise.
18406 (maybe_assemble_visibility): Likewise.
18407 * varpool.c (ctor_for_folding): Likewise.
18408 * chkp-builtins.def: Remove.
18409 * ipa-chkp.c: Remove.
18410 * ipa-chkp.h: Remove.
18411 * rtl-chkp.c: Remove.
18412 * rtl-chkp.h: Remove.
18413 * tree-chkp-opt.c: Remove.
18414 * tree-chkp.c: Remove.
18415 * tree-chkp.h: Remove.
18417 2018-06-07 Carl Love <cel@us.ibm.com>
18419 * config/rs6000/vsx.md (vextract_fp_from_shorth,
18420 vextract_fp_from_shortl): Add BE support.
18422 2018-06-07 Paul Koning <ni1d@arrl.net>
18424 * compare-elim.c (try_merge_compare): Don't merge compare if
18425 address contains a side effect.
18426 (try_eliminate_compare): Likewise.
18428 2018-06-07 Olga Makhotina <olga.makhotina@intel.com>
18430 * config.gcc: Support "tremont".
18431 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "tremont".
18432 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
18434 * config/i386/i386.c (m_TREMONT): Define.
18435 (processor_target_table): Add "tremont".
18436 (PTA_TREMONT): Define.
18437 (ix86_lea_outperforms): Add TARGET_TREMONT.
18438 (get_builtin_code_for_version): Handle PROCESSOR_TREMONT.
18439 (fold_builtin_cpu): Add M_INTEL_TREMONT, replace M_INTEL_GOLDMONT
18440 and M_INTEL_GOLDMONT_PLUS.
18441 (fold_builtin_cpu): Add "tremont".
18442 (ix86_add_stmt_cost): Add TARGET_TREMONT.
18443 (ix86_option_override_internal): Add "tremont".
18444 * config/i386/i386.h (processor_costs): Define TARGET_TREMONT.
18445 (processor_type): Add PROCESSOR_TREMONT.
18446 * config/i386/x86-tune.def: Add m_TREMONT.
18447 * doc/invoke.texi: Add tremont as x86 -march=/-mtune= CPU type.
18449 2018-06-07 Jozef Lawrynowicz <jozef.l@mittosystems.com>
18451 * config/msp430/msp430.c (msp430_mcu_name): Set the "i" in the
18452 symbol defined for msp430i* devices to be lower case.
18454 2018-06-07 Richard Biener <rguenther@suse.de>
18456 * graphite-sese-to-poly.c (extract_affine): Avoid unneded
18457 wrapping. Properly wrap the result of a BIT_NOT_EXPR.
18458 Properly wrap signed arithmetic if overflow wraps.
18460 2018-06-07 Jakub Jelinek <jakub@redhat.com>
18462 PR tree-optimization/69615
18463 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If rhs2 is lhs
18464 of a cast from a same precision integral SSA_NAME in a bb dominated
18465 by first_bb, retry with rhs2 set to the rhs1 of the cast. Don't emit
18466 cast to utype if rhs2 has already a compatible type.
18468 2018-06-07 Richard Biener <rguenther@suse.de>
18470 PR tree-optimization/85935
18471 * graphite-scop-detection.c (find_params_in_bb): Analyze
18472 condition operands with respect to the correct loop. Assert
18473 the analysis doesn't fail.
18475 2018-06-04 Carl Love <cel@us.ibm.com>
18477 * config/rs6000/vsx.md (first_match_index_<mode>): Calculate index
18478 using natural element order. Use gen_lshrsi3 instead of gen_ashrsi3
18479 as it is slightly cheaper.
18480 (first_match_or_eos_index_<mode>):
18481 Calculate index using natural element order.
18482 (first_match_index_<mode>):
18483 Calculate index using natural element order.
18484 (first_match_or_eos_index_<mode>):
18485 Calculate index using natural order.
18486 (define_insn vclzlsbb): Change to define_insn vclzlsbb_<mode>.
18487 for BE and LE modes.
18488 * config/rs6000/rs6000-c.c: Rename P9V_BUILTIN_VCLZLSBB,
18489 P9V_BUILTIN_VCLZLSBB_V16QI.
18490 * config/rs6000/rs6000-builtin.def: Make VCLZLSBB mode
18493 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
18495 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
18496 indentation and line wrap for many prototypes. Add missing
18497 @smallexample directives around block of prototypes for vec_xl and
18500 2018-06-05 Michael Meissner <meissner@linux.ibm.com>
18502 * config/rs6000/rs6000.c (rs6000_passes_ieee128): New boolean to
18503 track if we pass or return IEEE 128-bit floating point.
18504 (ieee128_mangling_gcc_8_1): New boolean to say whether to generate
18505 C++ mangling that is compatible with GCC 8.1.
18506 (TARGET_ASM_GLOBALIZE_DECL_NAME): Override target hook.
18507 (init_cumulative_args): Note if we pass or return IEEE 128-bit
18508 floating point types.
18509 (rs6000_function_arg_advance_1): Likewise.
18510 (rs6000_mangle_type): Optionally generate mangled names that match
18511 what GCC 8.1 generated for IEEE 128-bit floating point types.
18512 (rs6000_globalize_decl_name): If we have an external function that
18513 passes or returns IEEE 128-bit types, generate a weak reference
18514 from the mangled name used in GCC 8.1 to the current mangled
18516 (rs6000_init_builtins): Make __ibm128 use the long double type if
18517 long double is IBM extended double. Make __float128 use the long
18518 double type if long double is IEEE 128-bit.
18521 * config/rs6000/rs6000-builtin.def (BU_IBM128_2): New helper
18522 macro for __ibm128 built-in functions.
18523 (PACK_IF): Add __ibm128 pack/unpack functions.
18524 (UNPACK_IF): Likewise.
18525 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Do not
18526 enable long double built-in functions if long double is IEEE
18527 128-bit floating point.
18528 (rs6000_invalid_builtin): Update long double built-in function
18530 (rs6000_expand_builtin): For PACK_IF and UNPACK_IF built-in
18531 functions, adjust the built-in function to use the long double
18532 built-in function if __ibm128 and long double are the same type.
18533 * doc/extend.texi (PowerPC builtins): Update documention for
18534 __builtin_{,un}pack_longdouble. Add documentation for
18535 __builtin_{,un}pack_ibm128.
18537 2018-06-06 Jim Wilson <jimw@sifive.com>
18539 * config/riscv/riscv.c (enum riscv_privilege_levels): New.
18540 (struct machine_function): New field interrupt_mode.
18541 (riscv_handle_type_attribute): New function. Add forward declaration.
18542 (riscv_attribute_table) <interrupt>: Use riscv_handle_type_attribute.
18543 (riscv_expand_epilogue): Check interrupt_mode field.
18544 (riscv_set_current_function): Check interrupt attribute args and
18545 set interrupt_mode field.
18546 * config/riscv/riscv.md (UNSPECV_SRET, UNSPECV_URET): New.
18547 (riscv_sret, riscv_uret): New.
18548 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: Document
18549 new arguments to interrupt attribute.
18551 2018-06-06 Peter Bergner <bergner@vnet.ibm.com>
18554 * /config/rs6000/rs6000.h (ASM_CPU_SPEC): Add support for -mpower9.
18555 Don't handle -mcpu=power8 if -mpower9-vector is also used.
18557 2018-06-06 Kelvin Nilsen <kelvin@gcc.gnu.org>
18559 * config/rs6000/rs6000-builtin.def (VSX_BUILTIN_VEC_LD,
18560 VSX_BUILTIN_VEC_ST): Add comment to explain non-traditional uses.
18561 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
18562 several redundant entries.
18564 2018-06-06 David Malcolm <dmalcolm@redhat.com>
18566 * config/i386/i386-protos.h (ix86_expand_call): Strengthen return
18567 type from "rtx" to "rtx_insn *".
18568 * config/i386/i386.c (ix86_expand_split_stack_prologue): Likewise
18569 for local "call_insn", removing cast.
18570 (ix86_expand_call): Likewise, introducing a "call_insn" local.
18572 2018-06-06 Eric Botcazou <ebotcazou@adacore.com>
18574 PR tree-optimization/86066
18575 * gimple-ssa-store-merging.c (process_store): Do not bypass BIT_NOT_EXPR
18576 for BIT_INSERT_EXPR stores.
18578 2018-06-06 Richard Biener <rguenther@suse.de>
18580 PR tree-optimization/86062
18581 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
18582 component refs ontop
18583 of to be offsetted base.
18585 2018-06-06 Jozef Lawrynowicz <jozef.l@mittosystems.com>
18587 * config/msp430/msp430.c (msp430_attr): Allow interrupt handlers
18588 to be static and remove check on interrupt attribute name.
18590 2018-06-05 Kelvin Nilsen <kelvin@gcc.gnu.org>
18592 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
18593 volatile qualifier from vec_lvsl and vec_lvsr argument prototypes.
18595 2018-06-05 Steve Ellcey <sellcey@cavium.com>
18598 * config/aarch64/aarch64-protos.h (aarch64_err_no_fpadvsimd): Remove
18600 * config/aarch64/aarch64-protos..c (aarch64_err_no_fpadvsimd):
18601 Remove second argument, change how error is called.
18602 (aarch64_layout_arg): Remove second argument from
18603 aarch64_err_no_fpadvsimd call.
18604 (aarch64_init_cumulative_args): Ditto.
18605 (aarch64_gimplify_va_arg_expr): Ditto.
18606 * config/aarch64/aarch64.md (mov<mode>): Ditto.
18608 2018-06-05 Uros Bizjak <ubizjak@gmail.com>
18610 * config/i386/i386.md (simple_return_indirect_internal): New expander.
18611 (*simple_return_indirect_internal<mode>): Rename from
18612 simple_return_indirect_internal. Use W mode iterator.
18613 (rstorssp): New expander.
18614 (*rstorssp<mode>): Rename from rstorssp. Use P mode iterator.
18615 (clrssbsy): New expander.
18616 (*clrssbsy<mode>): Rename from clrssbsy. Use P mode iterator.
18618 2018-06-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
18620 * config/arm/arm_cmse.h (cmse_nsfptr_create): Change typeof to
18622 (cmse_check_pointed_object): Likewise.
18624 2018-06-05 Martin Liska <mliska@suse.cz>
18626 PR gcov-profile/47618
18627 * doc/invoke.texi: Document how -fprofile-dir format
18630 2018-06-05 Richard Biener <rguenther@suse.de>
18632 * tree-cfgcleanup.c (cleanup_control_flow_pre): For edge
18633 removal pretend DOM info isn't available so we do not update
18634 it and only remove edges, not dominated blocks. Actually free
18635 DOM info in case we removed something. Remove unreachable blocks.
18636 (mfb_keep_latches): Work with either DOM info or marked backedges.
18637 (cleanup_tree_cfg_noloop): Do not remove unreachable blocks
18638 first. Mark backedges if DOM info isn't available.
18639 (Re-)compute DOM info after cleanup_control_flow_pre.
18641 2018-06-05 Richard Biener <rguenther@suse.de>
18643 * tree-cfg.c (struct locus_discrim_map): Store line, not location.
18644 (locus_discrim_hasher::hash): Adjust.
18645 (locus_discrim_hasher::equal): Likewise.
18646 (next_discriminator_for_locus): Work on line directly.
18647 (same_line_p): Pass in expanded locus1 as well.
18648 (assign_discriminators): Avoid redundant location expansions.
18650 2018-06-05 Richard Biener <rguenther@suse.de>
18652 PR tree-optimization/86046
18653 * tree-ssa.c (maybe_optimize_var): Clear DECL_GIMPLE_REG_P
18654 if required after clearing TREE_ADDRESSABLE.
18656 2018-06-05 Richard Biener <rguenther@suse.de>
18658 PR tree-optimization/86047
18659 * tree-ssa-loop.c (for_each_index): Glob handling of all
18660 decls and constants and really handle all of them.
18662 2018-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18665 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
18666 qualifier_void_pointer and qualifier_const_void_pointer.
18667 (arm_ldc_qualifiers, arm_stc_qualifiers): Use the above.
18668 (arm_init_builtins): Handle the above.
18669 * config/arm/arm_acle.h (__arm_cdp, __arm_ldc, __arm_ldcl, __arm_stc,
18670 __arm_stcl, __arm_mcr, __arm_cdp2, __arm_ldc2, __arm_ldcl2, __arm_stc2,
18671 __arm_stcl2,__arm_mcr2, __arm_mcrr, __arm_mcrr2): Remove return for
18674 2018-06-05 Martin Liska <mliska@suse.cz>
18676 * auto-profile.c (read_autofdo_file): Do not use
18677 gcov_ctr_summary struct.
18678 (afdo_callsite_hot_enough_for_early_inline): Likewise.
18679 * coverage.c (struct counts_entry): Likewise.
18680 (read_counts_file): Read just single summary entry.
18681 (get_coverage_counts): Use gcov_summary struct.
18682 * coverage.h (get_coverage_counts): Likewise.
18683 * gcov-dump.c (dump_working_sets): Likewise.
18684 (tag_summary): Dump just single summary.
18685 * gcov-io.c (gcov_write_summary): Write just histogram
18687 (gcov_read_summary): Read just single summary.
18688 (compute_working_sets): Use gcov_summary struct.
18689 * gcov-io.h (GCOV_TAG_SUMMARY_LENGTH): Remove usage
18690 of GCOV_COUNTERS_SUMMABLE.
18691 (GCOV_COUNTERS_SUMMABLE): Remove.
18692 (GCOV_FIRST_VALUE_COUNTER): Replace with
18693 GCOV_COUNTER_V_INTERVAL.
18694 (struct gcov_ctr_summary): Remove.
18695 (struct gcov_summary): Directly use fields of former
18697 (compute_working_sets): Use gcov_summary struct.
18698 * gcov.c (read_count_file): Do not use ctrs fields.
18699 * lto-cgraph.c (merge_profile_summaries): Use gcov_summary
18701 * lto-streamer.h (struct GTY): Make profile_info gcov_summary
18703 * profile.c: Likewise.
18704 * profile.h: Likewise.
18706 2018-06-05 Martin Liska <mliska@suse.cz>
18708 PR gcov-profile/84846
18709 * gcov.c (output_lines): Print working directory only
18710 in intermediate format.
18712 2018-06-05 Andreas Krebbel <krebbel@linux.ibm.com>
18714 * config/s390/s390-builtin-types.def: Add void function type.
18715 * config/s390/s390-builtins.def: Use the function type for the
18718 2018-06-04 Jim Wilson <jimw@sifive.com>
18720 * config/riscv/riscv-protos.h (riscv_expand_epilogue): Change bool arg
18722 * config/riscv/riscv.c (riscv_for_each_saved_reg): New args epilogue
18723 and maybe_eh_return. Change regno to unsigned int. Use new args to
18724 handle EH_RETURN_DATA_REGNO registers properly.
18725 (riscv_expand_prologue): Pass new args to riscv_for_each_saved_reg.
18726 (riscv_expand_epilogue): Update comment. Change argument name and
18727 type. Update code to use new name and type. Pass new args to
18728 riscv_for_each_saved_reg. Only use EH_RETURN_STACKADJ_RTX when
18730 * config/riscv/riscv.md (NORMAL_RETURN): New.
18731 (SIBCALL_RETURN, EXCEPTION_RETURN): New.
18732 (epilogue, sibcall_epilogue): Update riscv_expand_epilogue arg.
18733 (eh_return): Call gen_eh_return_internal and emit barrier.
18734 (eh_return_internal): Call riscv_expand_epilogue.
18736 2018-06-04 Eric Botcazou <ebotcazou@adacore.com>
18738 * gimple-ssa-store-merging.c (struct merged_store_group): Move up
18739 bit_insertion field and declare can_be_merged_into method.
18740 (merged_store_group::can_be_merged_into): New method.
18741 (imm_store_chain_info::coalesce_immediate): Call it to decide whether
18742 consecutive non-overlapping stores can be merged. Turn MEM_REF stores
18743 into BIT_INSERT_EXPR stores if the group contains a non-MEM_REF store.
18745 2018-06-04 Richard Biener <rguenther@suse.de>
18747 PR tree-optimization/85955
18748 * builtins.c (fold_builtin_sincos): Convert pointers to
18749 destination to appropriate type before dereferencing.
18751 2018-06-04 Segher Boessenkool <segher@kernel.crashing.org>
18753 * config/rs6000/rs6000.md (abs<mode>2 for FLOAT128): Handle IFmode.
18755 2018-06-04 Richard Sandiford <richard.sandiford@linaro.org>
18757 * expr.c (expand_expr_real_1): Force the operand into memory if
18758 its TYPE_MODE is BLKmode and if there is no integer mode for
18759 the number of bits being extracted.
18761 2018-06-04 Jakub Jelinek <jakub@redhat.com>
18765 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
18766 Use vptestnm rather than vptestm in (=Yc,v,C) variant.
18768 2018-06-04 Richard Biener <rguenther@suse.de>
18770 * tree-cfgcleanup.c (cleanup_tree_cfg_1): Fold into...
18771 (cleanup_tree_cfg_noloop): ... single caller. Do
18772 start_recording_case_labels later.
18774 2018-06-04 Sebastian Peryt <sebastian.peryt@intel.com>
18776 * config/i386/cldemoteintrin.h: Change define from _X86INTRIN_H_INCLUDED
18777 to _IMMINTRIN_H_INCLUDED.
18778 * config/i386/pconfigintrin.h: Ditto.
18779 * config/i386/waitpkgintrin.h: Ditto.
18780 * config/i386/immintrin.h: Add includes for sgxintrin.h,
18781 pconfigintrin.h, waitpkgintrin.h and cldemoteintrin.h.
18782 * config/i386/x86intrin.h: Remove includes for mintrin.h, xmmintrin.h,
18783 emmintrin.h, pmmintrin.h, tmmintrin.h, smmintrin.h, wmmintrin.h,
18784 bmiintrin.h, bmi2intrin.h, lzcntintrin.h, sgxintrin.h, pconfigintrin.h,
18785 waitpkgintrin.h and cldemoteintrin.h.
18787 2018-06-04 Richard Biener <rguenther@suse.de>
18789 PR tree-optimization/86038
18790 * tracer.c (find_best_successor): Check probability for
18791 being initialized, bail out if not.
18793 2018-06-04 Richard Earnshaw <rearnsha@arm.com>
18796 * config/arm/arm-cpus.in (ALL_QUIRKS): Add xscale feature to the list
18797 of bits to ignore when comparing architectures.
18799 2018-06-04 Jakub Jelinek <jakub@redhat.com>
18801 PR tree-optimization/69615
18802 * fold-const.c (merge_ranges): If range1 is - [x, x] and x is the
18803 maximum or minimum of the type, try to merge it also as if
18804 range1 is + [-, x - 1] or + [x + 1, -].
18807 * tree.c (inchash::add_expr): Handle IDENTIFIER_NODE.
18809 2018-06-03 Eric Botcazou <ebotcazou@adacore.com>
18811 PR tree-optimization/86034
18812 * gimple-ssa-store-merging.c (output_merged_store): Convert the RHS to
18813 the unsigned bitfield type in a bit insertion sequence if it does not
18814 have a larger precision than the bitfield size.
18815 (process_store): Also bypass widening conversions for BIT_INSERT_EXPR.
18817 2018-06-03 Kito Cheng <kito.cheng@gmail.com>
18819 * config/nds32/nds32-peephole2.md: Add new patterns for code size.
18821 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
18823 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3J.
18824 * config/nds32/nds32.c (nds32_option_override): Consider ARCH_V3J.
18825 * config/nds32/nds32.h (TARGET_ISA_V3): Add ARCH_V3J checking.
18826 * config/nds32/nds32.opt (march): Add enum value Value(ARCH_V3J).
18828 2018-06-03 Chung-Ju Wu <jasonwucj@gmail.com>
18830 * common/config/nds32/nds32-common.c (nds32_option_optimization_table):
18831 Disable -fdelete-null-pointer-checks for ELF toolchain.
18833 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
18834 Kito Cheng <kito.cheng@gmail.com>
18836 * config.gcc (nds32*): Use nds32-linux.opt and nds32-elf.opt.
18837 (nds32le-*-*, nds32be-*-*): Integrate checking process.
18838 (nds32*-*-*): Add glibc and uclibc conditions.
18839 * common/config/nds32/nds32-common.c (nds32_except_unwind_info): New.
18840 (TARGET_EXCEPT_UNWIND_INFO): Define.
18841 * config/nds32/elf.h: New file.
18842 * config/nds32/linux.h: New file.
18843 * config/nds32/nds32-elf.opt: New file.
18844 * config/nds32/nds32-linux.opt: New file.
18845 * config/nds32/nds32-fp-as-gp.c
18846 (pass_nds32_fp_as_gp::gate): Consider TARGET_LINUX_ABI.
18847 * config/nds32/nds32.c (nds32_conditional_register_usage): Consider
18849 (nds32_asm_file_end): Ditto.
18850 (nds32_print_operand): Ditto.
18851 (nds32_insert_attributes): Ditto.
18852 (nds32_init_libfuncs): New function.
18853 (TARGET_HAVE_TLS): Define.
18854 (TARGET_INIT_LIBFUNCS): Define.
18855 * config/nds32/nds32.h (TARGET_DEFAULT_RELAX): Apply different relax
18857 (TARGET_ELF): Apply different mcmodel setting.
18858 (LINK_SPEC, LIB_SPEC, STARTFILE_SPEC, ENDFILE_SPEC): The content has
18859 been migrated into elf.h and linux.h files.
18860 * config/nds32/nds32.md (add_pc): Consider TARGET_LINUX_ABI.
18861 * config/nds32/nds32.opt (mvh): Consider TARGET_LINUX_ABI.
18862 (mcmodel): The content has been migrated into nds32-elf.opt and
18863 nds32-linux.opt files.
18864 * config/nds32/t-elf: New file.
18865 * config/nds32/t-linux: New file.
18867 2018-06-02 Chung-Ju Wu <jasonwucj@gmail.com>
18868 Shiva Chen <shiva0217@gmail.com>
18870 * config/nds32/constants.md (unspec_volatile_element): Add
18871 UNSPEC_VOLATILE_OMIT_FP_BEGIN and UNSPEC_VOLATILE_OMIT_FP_END.
18872 * config/nds32/nds32-fp-as-gp.c: New implementation of fp_as_gp
18874 * config/nds32/nds32-protos.h (nds32_naked_function_p): Declare.
18875 (make_pass_nds32_fp_as_gp): Declare.
18876 * config/nds32/nds32.c (nds32_register_passes): Add fp_as_gp as one
18878 (nds32_asm_function_end_prologue): Remove unused asm output.
18879 (nds32_asm_function_begin_epilogue): Remove unused asm output.
18880 (nds32_asm_file_start): Output necessary fp_as_gp information.
18881 (nds32_option_override): Adjust register usage.
18882 (nds32_expand_prologue): Consider fp_as_gp situation.
18883 (nds32_expand_prologue_v3push): Consider fp_as_gp situation.
18884 * config/nds32/nds32.md (prologue): Check fp_as_gp_p and naked_p.
18887 (simple_return): Ditto.
18888 (omit_fp_begin): Output special directive for fp_as_gp.
18889 (omit_fp_end): Output special directive for fp_as_gp.
18890 * config/nds32/nds32.opt (mfp-as-gp, mno-fp-as-gp, mforce-fp-as-gp,
18891 mforbid-fp-as-gp): New options.
18893 2018-06-01 Mark Wielaard <mark@klomp.org>
18895 * dwarf2out.c (dwarf2out_finish): Remove generation of
18896 DW_AT_loclists_base.
18898 2018-06-01 Eric Botcazou <ebotcazou@adacore.com>
18900 * gimple-ssa-store-merging.c: Include gimple-fold.h.
18901 (struct store_immediate_info): Document BIT_INSERT_EXPR stores.
18902 (struct merged_store_group): Add bit_insertion field.
18903 (dump_char_array): Use standard hexadecimal format.
18904 (merged_store_group::merged_store_group): Set bit_insertion to false.
18905 (merged_store_group::apply_stores): Use optimal buffer size. Deal
18906 with BIT_INSERT_EXPR stores. Move up code updating the mask and
18907 also print the mask in the dump file.
18908 (pass_store_merging::gate): Minor tweak.
18909 (imm_store_chain_info::coalesce_immediate): Fix wrong association
18910 of stores with groups in dump. Allow coalescing of BIT_INSERT_EXPR
18911 stores with INTEGER_CST stores.
18912 (count_multiple_uses) <BIT_INSERT_EXPR>: New case.
18913 (imm_store_chain_info::output_merged_store): Add try_bitpos variable
18914 and use it throughout. Generate bit insertion sequences if need be.
18915 (pass_store_merging::process_store): Remove redundant condition.
18916 Record stores from a SSA name to a bit-field with BIT_INSERT_EXPR.
18918 2018-06-01 Segher Boessenkool <segher@kernel.crashing.org>
18920 * config/rs6000/rs6000.c (rs6000_mangle_type): Change the mangling of
18921 the 128-bit floating point types. Fix function comment.
18923 2018-06-01 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
18925 * config/aarch64/aarch64-simd.md
18926 (aarch64_simd_vec_unpack<su>_lo_<mode>): Use UXTL and SXTL assembler
18928 (aarch64_simd_vec_unpack<su>_hi_<mode>): Use UXTL2 and SXTL2 assembler
18931 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
18933 PR tree-optimization/85989
18934 * gimple-ssa-backprop.c (backprop::m_visited_phis): New member
18936 (backprop::intersect_uses): Check it when deciding whether this
18937 is a backedge reference.
18938 (backprop::process_block): Add each phi to m_visited_phis
18939 after visiting it, then clear it at the end.
18941 2018-06-01 Richard Biener <rguenther@suse.de>
18943 * tree-vectorizer.h (vect_dr_stmt): New function.
18944 (vect_get_load_cost): Adjust.
18945 (vect_get_store_cost): Likewise.
18946 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence):
18947 Use vect_dr_stmt instead of DR_SMTT.
18948 (vect_record_base_alignments): Likewise.
18949 (vect_calculate_target_alignment): Likewise.
18950 (vect_compute_data_ref_alignment): Likewise and make static.
18951 (vect_update_misalignment_for_peel): Likewise.
18952 (vect_verify_datarefs_alignment): Likewise.
18953 (vector_alignment_reachable_p): Likewise.
18954 (vect_get_data_access_cost): Likewise. Pass down
18955 vinfo to vect_get_load_cost/vect_get_store_cost instead of DR.
18956 (vect_get_peeling_costs_all_drs): Likewise.
18957 (vect_peeling_hash_get_lowest_cost): Likewise.
18958 (vect_enhance_data_refs_alignment): Likewise.
18959 (vect_find_same_alignment_drs): Likewise.
18960 (vect_analyze_data_refs_alignment): Likewise.
18961 (vect_analyze_group_access_1): Likewise.
18962 (vect_analyze_group_access): Likewise.
18963 (vect_analyze_data_ref_access): Likewise.
18964 (vect_analyze_data_ref_accesses): Likewise.
18965 (vect_vfa_segment_size): Likewise.
18966 (vect_small_gap_p): Likewise.
18967 (vectorizable_with_step_bound_p): Likewise.
18968 (vect_prune_runtime_alias_test_list): Likewise.
18969 (vect_analyze_data_refs): Likewise.
18970 (vect_supportable_dr_alignment): Likewise.
18971 * tree-vect-loop-manip.c (get_misalign_in_elems): Likewise.
18972 (vect_gen_prolog_loop_niters): Likewise.
18973 * tree-vect-loop.c (vect_analyze_loop_2): Likewise.
18974 * tree-vect-patterns.c (vect_recog_bool_pattern): Do not
18976 (vect_recog_mask_conversion_pattern): Likewise.
18977 (vect_try_gather_scatter_pattern): Likewise.
18978 * tree-vect-stmts.c (vect_model_store_cost): Pass stmt_info
18979 to vect_get_store_cost.
18980 (vect_get_store_cost): Get stmt_info instead of DR.
18981 (vect_model_load_cost): Pass stmt_info to vect_get_load_cost.
18982 (vect_get_load_cost): Get stmt_info instead of DR.
18984 2018-06-01 Richard Biener <rguenther@suse.de>
18986 PR middle-end/86017
18987 * gimple-fold.c (var_decl_component_p): Also allow offsetted
18988 vars wrapped in MEM_REFs.
18990 2018-06-01 Richard Sandiford <richard.sandiford@linaro.org>
18992 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
18993 Fix subreg tests so that we only return a choice between
18994 GENERAL_REGS and FP_REGS if the original classes included both.
18996 2018-06-01 Richard Biener <rguenther@suse.de>
18999 * tree-ssa-structalias.c (get_function_part_constraint):
19000 Handle NULL fi->decl.
19001 (find_func_aliases_for_call): Properly handle indirect
19002 fi from direct call.
19003 (find_func_clobbers): Likewise.
19004 (ipa_pta_execute): Likewise.
19005 (create_variable_info_for): For functions that are ifunc_resolver
19006 resolve to a varinfo that contains the result of the resolver call.
19007 (associate_varinfo_to_alias): Do not treat ifunc resolvers as
19010 2018-05-31 Michael Collison <michael.collison@arm.com>
19012 * config/aarch64/aarch64.md:
19013 (*fix_to_zero_extenddfdi2): New pattern.
19014 * gcc.target/aarch64/fix_extend1.c: New testcase.
19016 2018-05-31 Qing Zhao <qing.zhao@oracle.com>
19018 PR middle-end/78809
19019 PR middle-end/83026
19020 * builtins.c (expand_builtin): Add the handling of BUILT_IN_STRCMP_EQ
19021 and BUILT_IN_STRNCMP_EQ.
19022 * builtins.def: Add new builtins BUILT_IN_STRCMP_EQ and
19023 BUILT_IN_STRNCMP_EQ.
19024 * gimple-fold.c (gimple_fold_builtin_string_compare): Add the
19025 handling of BUILTIN_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
19026 (gimple_fold_builtin): Likewise.
19027 * tree-ssa-strlen.c (compute_string_length): New function.
19028 (determine_min_obsize): New function.
19029 (handle_builtin_string_cmp): New function to handle calls to
19030 string compare functions.
19031 (strlen_optimize_stmt): Add handling to builtin string compare
19033 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
19034 Add the handling of BUILT_IN_STRCMP_EQ and BUILT_IN_STRNCMP_EQ.
19035 * tree.c (build_common_builtin_nodes): Add new defines of
19036 BUILT_IN_STRNCMP_EQ and BUILT_IN_STRCMP_EQ.
19038 2018-05-31 Jakub Jelinek <jakub@redhat.com>
19041 * bb-reorder.c (pass_partition_blocks::gate): Return false for
19042 functions with naked attribute.
19044 2018-05-31 Uros Bizjak <ubizjak@gmail.com>
19046 * config/i386/sse.md (avx_vec_concat<mode>):
19047 Substitute concat_tg_mode mode attribute with xtg_mode.
19048 (<mask_codefor>avx512dq_broadcast<mode><mask_name>_1): Ditto.
19049 (concat_tg_mode): Remove mode attribute.
19051 2018-05-31 Martin Sebor <msebor@redhat.com>
19054 * calls.c (alloc_max_size): Correct a logic error/typo.
19055 Treat excessive arguments as infinite. Warn for invalid arguments.
19056 * doc/invoke.texi (-Walloc-size-larger-than): Update.
19058 2018-05-31 H.J. Lu <hongjiu.lu@intel.com>
19061 * config/i386/x86-tune.def: Re-enable partial_reg_dependency
19062 and movx for Haswell.
19064 2018-05-31 Chung-Lin Tang <cltang@codesourcery.com>
19065 Cesar Philippidis <cesar@codesourcery.com>
19067 PR middle-end/85879
19068 * gimplify.c (gimplify_adjust_omp_clauses): Add 'remove = true'
19069 when emitting error on private/firstprivate reductions.
19070 * omp-low.c (lower_omp_target): Avoid reference-type processing
19071 on pointers for firstprivate clause.
19073 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
19075 * config/aarch64/aarch64-simd-builtins.def (ld1x3): New.
19078 * config/aarch64/aarch64-simd.md
19079 (aarch64_ld1x3<VALLDIF:mode>): New pattern.
19080 (aarch64_ld1_x3_<mode>): Likewise
19081 (aarch64_st1x2<VALLDIF:mode>): Likewise
19082 (aarch64_st1_x2_<mode>): Likewise
19083 (aarch64_st1x3<VALLDIF:mode>): Likewise
19084 (aarch64_st1_x3_<mode>): Likewise
19085 * config/aarch64/arm_neon.h (vld1_u8_x3): New function.
19086 (vld1_s8_x3): Likewise.
19087 (vld1_u16_x3): Likewise.
19088 (vld1_s16_x3): Likewise.
19089 (vld1_u32_x3): Likewise.
19090 (vld1_s32_x3): Likewise.
19091 (vld1_u64_x3): Likewise.
19092 (vld1_s64_x3): Likewise.
19093 (vld1_f16_x3): Likewise.
19094 (vld1_f32_x3): Likewise.
19095 (vld1_f64_x3): Likewise.
19096 (vld1_p8_x3): Likewise.
19097 (vld1_p16_x3): Likewise.
19098 (vld1_p64_x3): Likewise.
19099 (vld1q_u8_x3): Likewise.
19100 (vld1q_s8_x3): Likewise.
19101 (vld1q_u16_x3): Likewise.
19102 (vld1q_s16_x3): Likewise.
19103 (vld1q_u32_x3): Likewise.
19104 (vld1q_s32_x3): Likewise.
19105 (vld1q_u64_x3): Likewise.
19106 (vld1q_s64_x3): Likewise.
19107 (vld1q_f16_x3): Likewise.
19108 (vld1q_f32_x3): Likewise.
19109 (vld1q_f64_x3): Likewise.
19110 (vld1q_p8_x3): Likewise.
19111 (vld1q_p16_x3): Likewise.
19112 (vld1q_p64_x3): Likewise.
19113 (vst1_s64_x2): Likewise.
19114 (vst1_u64_x2): Likewise.
19115 (vst1_f64_x2): Likewise.
19116 (vst1_s8_x2): Likewise.
19117 (vst1_p8_x2): Likewise.
19118 (vst1_s16_x2): Likewise.
19119 (vst1_p16_x2): Likewise.
19120 (vst1_s32_x2): Likewise.
19121 (vst1_u8_x2): Likewise.
19122 (vst1_u16_x2): Likewise.
19123 (vst1_u32_x2): Likewise.
19124 (vst1_f16_x2): Likewise.
19125 (vst1_f32_x2): Likewise.
19126 (vst1_p64_x2): Likewise.
19127 (vst1q_s8_x2): Likewise.
19128 (vst1q_p8_x2): Likewise.
19129 (vst1q_s16_x2): Likewise.
19130 (vst1q_p16_x2): Likewise.
19131 (vst1q_s32_x2): Likewise.
19132 (vst1q_s64_x2): Likewise.
19133 (vst1q_u8_x2): Likewise.
19134 (vst1q_u16_x2): Likewise.
19135 (vst1q_u32_x2): Likewise.
19136 (vst1q_u64_x2): Likewise.
19137 (vst1q_f16_x2): Likewise.
19138 (vst1q_f32_x2): Likewise.
19139 (vst1q_f64_x2): Likewise.
19140 (vst1q_p64_x2): Likewise.
19141 (vst1_s64_x3): Likewise.
19142 (vst1_u64_x3): Likewise.
19143 (vst1_f64_x3): Likewise.
19144 (vst1_s8_x3): Likewise.
19145 (vst1_p8_x3): Likewise.
19146 (vst1_s16_x3): Likewise.
19147 (vst1_p16_x3): Likewise.
19148 (vst1_s32_x3): Likewise.
19149 (vst1_u8_x3): Likewise.
19150 (vst1_u16_x3): Likewise.
19151 (vst1_u32_x3): Likewise.
19152 (vst1_f16_x3): Likewise.
19153 (vst1_f32_x3): Likewise.
19154 (vst1_p64_x3): Likewise.
19155 (vst1q_s8_x3): Likewise.
19156 (vst1q_p8_x3): Likewise.
19157 (vst1q_s16_x3): Likewise.
19158 (vst1q_p16_x3): Likewise.
19159 (vst1q_s32_x3): Likewise.
19160 (vst1q_s64_x3): Likewise.
19161 (vst1q_u8_x3): Likewise.
19162 (vst1q_u16_x3): Likewise.
19163 (vst1q_u32_x3): Likewise.
19164 (vst1q_u64_x3): Likewise.
19165 (vst1q_f16_x3): Likewise.
19166 (vst1q_f32_x3): Likewise.
19167 (vst1q_f64_x3): Likewise.
19168 (vst1q_p64_x3): Likewise.
19170 2018-05-30 Jozef Lawrynowicz <jozef.l@mittosystems.com>
19172 * config/msp430/msp430.c (msp430_output_labelref): Prepend
19173 user_label_prefix to name.
19175 * tree-core.h: Update comment about the format of NAME string
19176 passed to handler in attribute_spec.
19178 * config/msp430/msp430.md: Remove erroneous subreg expression from
19179 zero_extendqisi2 insn pattern. Remove msp430x ISA restriction on
19180 zero_extend{q,h}isi2.
19182 2018-05-30 Borislav Petkov <bp@suse.de>
19184 * doc/extend.texi: Document some architecture specific
19185 constraints and sort entries.
19187 2018-05-30 Martin Sebor <msebor@redhat.com>
19189 PR middle-end/85369
19190 * builtins.c (expand_builtin_stpcpy_1): New function.
19191 (expand_builtin_stpcpy): Call it, and call maybe_warn_nonstring_arg
19192 only if the former succeeds.
19194 2018-05-31 Sameera Deshpande <sameera.deshpande@linaro.org>
19196 * config/aarch64/aarch64-cores.def (saphira) : Add support for ARMv8.4
19199 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
19201 * doc/invoke.texi (-flinker-output): Document
19203 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
19205 * passes.c (ipa_write_summaries): Only modify statements if body
19207 * cgraphunit.c (ipa_passes): Also produce intermeidate code when
19208 incrementally linking.
19209 (ipa_passes): Likewise.
19210 * lto-cgraph.c (lto_output_node): When incrementally linking do not
19211 pass down resolution info.
19212 * common.opt (flag_incremental_link): Update info.
19213 * gcc.c (plugin specs): Turn flinker-output=* to
19214 -plugin-opt=-linker-output-known
19215 * toplev.c (compile_file): Also cut compilation when doing incremental
19217 * flag-types. (enum lto_partition_model): Add
19218 LTO_LINKER_OUTPUT_NOLTOREL.
19219 (invoke.texi): Add -flinker-output docs.
19220 * ipa.c (symbol_table::remove_unreachable_nodes): Handle LTO incremental
19221 link same way as WPA; do not stream in dead initializers.
19223 * dwarf2out.c (dwarf2out_die_ref_for_decl,
19224 darf2out_register_external_decl): Support incremental link.
19226 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
19228 * lto-opts.c (lto_write_options): Skip OPT_dumpdir, OPT_fresolution_.
19230 2018-05-30 Jan Hubicka <hubicka@ucw.cz>
19232 * lto-wrapper.c (debug_objcopy): Add rename parameter; pass
19233 it down to simple_object_copy_lto_debug_sections.
19234 (run_gcc): Determine incremental LTO link time and configure
19235 lto1 into non-wpa mode, disable renaming of debug sections.
19237 2018-05-30 Kelvin Nilsen <kelvin@gcc.gnu.org>
19239 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Remove
19240 descriptions of various incorrectly documented functions.
19242 2018-05-30 Andre Vieira <andre.simoesdiasvieira@arm.com>
19245 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
19246 address check not strict.
19248 2018-05-30 Richard Biener <rguenther@suse.de>
19250 PR tree-optimization/85964
19251 * tracer.c (better_p): Drop initialized count check, we only
19252 call the function with initialized counts now.
19253 (find_best_successor): Do find a best edge if one
19254 has uninitialized count.
19255 (find_best_predecessor): Likewise. Do BB frequency check only
19256 if count is initialized.
19258 2017-05-30 Jackson Woodruff <jackson.woodruff@arm.com>
19260 * config/aarch64/aarch64.c (aarch64_host_wide_int_compare): New.
19261 (aarch64_ldrstr_offset_compare): New.
19262 (aarch64_operands_adjust_ok_for_ldpstp): Update to consider all
19263 load/store orderings.
19264 (aarch64_gen_adjusted_ldpstp): Likewise.
19266 2018-05-30 Wilco Dijkstra <wdijkstr@arm.com>
19268 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
19269 Check for subset of GENERAL_REGS and FP_REGS.
19270 * config/aarch64/aarch64-simd.md (aarch64_get_lane): Increase cost of
19273 2018-05-30 Richard Sandiford <richard.sandiford@linaro.org>
19275 * alias.c (adjust_offset_for_component_ref): Use poly_int_tree_p
19276 and wi::to_poly_offset. Add the current offset and then check
19277 whether the sum fits, rather than using an unchecked addition of
19278 a checked term. Check for a shwi rather than a uhwi.
19279 * expr.c (get_bit_range): Use tree_to_poly_uint64.
19280 (store_constructor): Use poly_int_tree_p.
19281 (expand_expr_real_1): Likewise.
19282 * function.c (assign_temp): Likewise.
19283 * fold-const.c (const_binop): Use poly_int_tree_p and
19284 wi::to_poly_offset.
19285 (fold_indirect_ref_1): Likewise. Use multiple_p to attempt an exact
19287 * ipa-icf-gimple.c (func_checker::compare_operand): Use
19288 to_poly_offset for MEM offsets.
19289 * ipa-icf.c (sem_variable::equals): Likewise.
19290 * stor-layout.c (compute_record_mode): Use poly_int_tree_p.
19291 * tree-ssa-sccvn.c (ao_ref_init_from_vn_reference): Use
19292 wi::to_poly_offset for BIT_FIELD_REF offsets.
19293 (vn_reference_maybe_forwprop_address): Use poly_int_tree_p and
19294 wi::to_poly_offset.
19295 * var-tracking.c (emit_note_insn_var_location): Use
19296 tree_to_poly_uint64.
19298 2018-05-29 Jim Wilson <jimw@sifive.com>
19300 * config/riscv/riscv.c (riscv_interrupt_type): Fix comment typo.
19302 2018-05-29 Uros Bizjak <ubizjak@gmail.com>
19305 * config/i386/i386.md (l<rounding_insn><MODEF:mode><SWI48:mode>2):
19306 Enable for TARGET_SSE4_1 and generate rounds{s,d} and cvtts{s,d}2si{,q}
19308 (sse4_1_round<mode>2): Use nonimmediate_operand
19309 for operand 1 predicate.
19311 2018-05-29 Martin Sebor <msebor@redhat.com>
19312 Richard Biener <rguenther@suse.de>
19315 * calls.c (get_size_range): Call determine_value_range instead
19316 of get_value_range..
19317 * tree-vrp.h (determine_value_range): Declared new function.
19318 * tree-vrp.c (determine_value_range_1, determine_value_range): New.
19320 2018-05-29 Richard Biener <rguenther@suse.de>
19322 * tree-vect-data-refs.c (vect_preserves_scalar_order_p): Make
19323 sure to use non-pattern stmts for get_earlier_stmt arguments.
19324 * tree-vectorizer.h (get_earlier_stmt): Assert we do not get
19325 called on pattern stmts.
19326 (get_later_stmt): Likewise.
19328 2018-05-29 Martin Liska <mliska@suse.cz>
19330 PR gcov-profile/85759
19331 * doc/gcov.texi: Document GCOV_ERROR_FILE and GCOV_EXIT_AT_ERROR
19334 2018-05-29 Jakub Jelinek <jakub@redhat.com>
19336 * tree-cfg.c (verify_gimple_assign_unary): Add checking for
19338 (verify_gimple_assign_binary): Check TYPE_VECTOR_SUBPARTS for
19342 * tree.def (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
19343 VEC_PACK_FLOAT_EXPR): New tree codes.
19344 * tree-pretty-print.c (op_code_prio): Handle
19345 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR.
19346 (dump_generic_node): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
19347 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
19348 * tree-inline.c (estimate_operator_cost): Likewise.
19349 * gimple-pretty-print.c (dump_binary_rhs): Handle VEC_PACK_FLOAT_EXPR.
19350 * fold-const.c (const_binop): Likewise.
19351 (const_unop): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR and
19352 VEC_UNPACK_FIX_TRUNC_LO_EXPR.
19353 * tree-cfg.c (verify_gimple_assign_unary): Likewise.
19354 (verify_gimple_assign_binary): Handle VEC_PACK_FLOAT_EXPR.
19355 * cfgexpand.c (expand_debug_expr): Handle VEC_UNPACK_FIX_TRUNC_HI_EXPR,
19356 VEC_UNPACK_FIX_TRUNC_LO_EXPR and VEC_PACK_FLOAT_EXPR.
19357 * expr.c (expand_expr_real_2): Likewise.
19358 * optabs.def (vec_packs_float_optab, vec_packu_float_optab,
19359 vec_unpack_sfix_trunc_hi_optab, vec_unpack_sfix_trunc_lo_optab,
19360 vec_unpack_ufix_trunc_hi_optab, vec_unpack_ufix_trunc_lo_optab): New
19362 * optabs.c (expand_widen_pattern_expr): For
19363 VEC_UNPACK_FIX_TRUNC_HI_EXPR and VEC_UNPACK_FIX_TRUNC_LO_EXPR use
19364 sign from result type rather than operand's type.
19365 (expand_binop_directly): For vec_packu_float_optab and
19366 vec_packs_float_optab allow result type to be different from operand's
19368 * optabs-tree.c (optab_for_tree_code): Handle
19369 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
19370 VEC_PACK_FLOAT_EXPR. Formatting fixes.
19371 * tree-vect-generic.c (expand_vector_operations_1): Handle
19372 VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR and
19373 VEC_PACK_FLOAT_EXPR.
19374 * tree-vect-stmts.c (supportable_widening_operation): Handle
19376 (supportable_narrowing_operation): Handle FLOAT_EXPR.
19377 * config/i386/i386.md (fixprefix, floatprefix): New code attributes.
19378 * config/i386/sse.md (*float<floatunssuffix>v2div2sf2): Rename to ...
19379 (float<floatunssuffix>v2div2sf2): ... this. Formatting fix.
19380 (vpckfloat_concat_mode, vpckfloat_temp_mode, vpckfloat_op_mode): New
19382 (vec_pack<floatprefix>_float_<mode>): New expander.
19383 (vunpckfixt_mode, vunpckfixt_model, vunpckfixt_extract_mode): New mode
19385 (vec_unpack_<fixprefix>fix_trunc_lo_<mode>,
19386 vec_unpack_<fixprefix>fix_trunc_hi_<mode>): New expanders.
19387 * doc/md.texi (vec_packs_float_@var{m}, vec_packu_float_@var{m},
19388 vec_unpack_sfix_trunc_hi_@var{m}, vec_unpack_sfix_trunc_lo_@var{m},
19389 vec_unpack_ufix_trunc_hi_@var{m}, vec_unpack_ufix_trunc_lo_@var{m}):
19391 * doc/generic.texi (VEC_UNPACK_FLOAT_HI_EXPR,
19392 VEC_UNPACK_FLOAT_LO_EXPR): Fix pasto in description.
19393 (VEC_UNPACK_FIX_TRUNC_HI_EXPR, VEC_UNPACK_FIX_TRUNC_LO_EXPR,
19394 VEC_PACK_FLOAT_EXPR): Document.
19396 2018-05-29 Richard Biener <rguenther@suse.de>
19398 * tree-vectorizer.h (struct vec_info): Add stmt_vec_infos
19400 (stmt_vec_info_vec): Make pointer.
19401 (init_stmt_vec_info_vec): Remove.
19402 (free_stmt_vec_info_vec): Likewise.
19403 (set_stmt_vec_info_vec): New function.
19404 (free_stmt_vec_infos): Likewise.
19405 (vinfo_for_stmt): Adjust for stmt_vec_info_vec indirection.
19406 (set_vinfo_for_stmt): Likewise.
19407 (get_earlier_stmt): Likewise.
19408 (get_later_stmt): Likewise.
19409 * tree-vectorizer.c (stmt_vec_info_vec): Make pointer.
19410 (vec_info::vec_info): Allocate stmt_vec_infos and set the global.
19411 (vec_info::~vec_info): Free stmt_vec_infos.
19412 (vectorize_loops): Set the global stmt_vec_info_vec to NULL.
19413 Remove old init_stmt_vec_info_vec/free_stmt_vec_info_vec calls.
19414 (pass_slp_vectorize::execute): Likewise.
19415 * tree-vect-stmts.c (init_stmt_vec_info_vec): Remove.
19416 (free_stmt_vec_info_vec): Likewise.
19417 (set_stmt_vec_info_vec): New function.
19418 (free_stmt_vec_infos): Likewise.
19419 * tree-vect-loop.c (_loop_vec_info::~_loop_vec_info): Set
19420 the global stmt_vec_info_vec.
19421 * tree-parloops.c (gather_scalar_reductions): Use
19422 set_stmt_vec_info_vec/free_stmt_vec_infos and maintain a local
19425 2018-05-29 Richard Biener <rguenther@suse.de>
19427 * dominance.c (iterate_fix_dominators): Push/pop TV_DOMINANCE.
19429 2018-05-29 Martin Liska <mliska@suse.cz>
19430 David Malcolm <dmalcolm@redhat.com>
19432 * vec.c (test_reverse): New.
19433 (vec_c_tests): Add new test.
19434 * vec.h (vl_ptr>::reverse): New function.
19436 2018-05-29 Gerald Pfeifer <gerald@pfeifer.com>
19438 * config.gcc: Identify FreeBSD 3.x and 4.x as unsupported.
19440 * config/freebsd-spec.h (FBSD_LIB_SPEC): Only consider FreeBSD 5
19443 2018-05-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
19445 * tree-dump.c (dump_node): Use splay_tree_delete_pointers.
19447 2018-05-28 Richard Biener <rguenther@suse.de>
19449 PR tree-optimization/85933
19450 * tree-vect-data-refs.c (vect_record_base_alignments): Only
19451 look at stmts marked as vectorizable.
19453 2018-05-28 Richard Biener <rguenther@suse.de>
19455 PR tree-optimization/85934
19456 * tree-vect-generic.c (expand_vector_operations_1): Hoist
19457 vector boolean check before scalar optimization.
19459 2018-05-28 Jakub Jelinek <jakub@redhat.com>
19461 * doc/invoke.texi (ARM Options): Use @item instead of @itemx
19464 2018-05-28 Mark Wielaard <mark@klomp.org>
19466 * dwarf2asm.c (dw2_asm_output_delta_uleb128): Add brackets around lab2
19467 if it is an expression containing a minus sign.
19469 2018-05-27 John David Anglin <danglin@gcc.gnu.org>
19471 * config/pa/pa-linux.h (NEED_INDICATE_EXEC_STACK): Define to 0.
19473 2018-05-27 Paul Koning <ni1d@arrl.net>
19475 * config/pdp11/pdp11.md (truncsihi2): Remove.
19477 2018-05-27 Monk Chiang <sh.chiang04@gmail.com>
19478 Chung-Ju Wu <jasonwucj@gmail.com>
19480 * config/nds32/nds32-intrinsic.md (unaligned_storedi): Modify patterns
19482 (unaligned_store_dw): Ditto.
19483 * config/nds32/nds32-memory-manipulation.c
19484 (nds32_expand_movmemsi_loop_known_size): Refactoring implementation.
19485 (nds32_gen_dup_4_byte_to_word_value): Rename to ...
19486 (nds32_gen_dup_4_byte_to_word_value_aux): ... this.
19487 (emit_setmem_word_loop): Rename to ...
19488 (emit_setmem_doubleword_loop): ... this.
19489 (nds32_gen_dup_4_byte_to_word_value): New function.
19490 (nds32_gen_dup_8_byte_to_double_word_value): New function.
19491 (nds32_expand_setmem_loop): Refine implementation.
19492 (nds32_expand_setmem_loop_v3m): Ditto.
19493 * config/nds32/nds32-multiple.md (unaligned_store_update_base_dw): New
19496 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
19498 * config/nds32/nds32.md (bswapsi2, bswaphi2): New patterns.
19500 2018-05-27 Chung-Ju Wu <jasonwucj@gmail.com>
19502 * config/nds32/nds32.c (nds32_attribute_table): Add "no_prologue".
19503 (nds32_init_machine_status): Initialize machine->attr_naked_p and
19504 machine->attr_no_prologue_p.
19505 (nds32_compute_stack_frame): Check "naked" and "no_prologue" attributes.
19506 (nds32_naked_function_p): Handle "naked" and "no_prologue" attributes.
19507 (nds32_expand_epilogue): Consider attr_naked_p.
19508 (nds32_expand_epilogue_v3pop): Likewise.
19509 (nds32_can_use_return_insn): Likewise.
19510 * config/nds32/nds32.h (machine_function): Add attr_naked_p and
19511 attr_no_prologue_p fields.
19512 * config/nds32/nds32.opt (mret-in-naked-func): New option.
19514 2018-05-27 Jakub Jelinek <jakub@redhat.com>
19517 * config/i386/i386.md (fixunssuffix, floatunssuffix): New code
19519 * config/i386/sse.md
19520 (<floatsuffix>float<sseintvecmodelower><mode>2<mask_name><round_name>):
19522 (float<floatunssuffix><sseintvecmodelower><mode>2<mask_name><round_name>):
19524 (<floatsuffix>float<sselongvecmodelower><mode>2<mask_name><round_name>):
19526 (float<floatunssuffix><sselongvecmodelower><mode>2<mask_name><round_name>):
19528 (*<floatsuffix>floatv2div2sf2): Rename to ...
19529 (*float<floatunssuffix>v2div2sf2): ... this.
19530 (<floatsuffix>floatv2div2sf2_mask): Rename to ...
19531 (float<floatunssuffix>v2div2sf2_mask): ... this.
19532 (*<floatsuffix>floatv2div2sf2_mask_1): Rename to ...
19533 (*float<floatunssuffix>v2div2sf2_mask_1): ... this.
19534 (<fixsuffix>fix_truncv8dfv8si2<mask_name><round_saeonly_name>): Rename
19536 (fix<fixunssuffix>_truncv8dfv8si2<mask_name><round_saeonly_name>):
19538 (<fixsuffix>fix_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
19540 (fix<fixunssuffix>_trunc<mode><sseintvecmodelower>2<mask_name><round_saeonly_name>):
19542 (<fixsuffix>fix_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
19544 (fix<fixunssuffix>_trunc<mode><sselongvecmodelower>2<mask_name><round_saeonly_name>):
19546 (<fixsuffix>fix_truncv2sfv2di2<mask_name>): Rename to ...
19547 (fix<fixunssuffix>_truncv2sfv2di2<mask_name>): ... this.
19548 (vec_pack_ufix_trunc_<mode>): Use gen_fixuns_truncv8dfv8si2 instead of
19549 gen_ufix_truncv8dfv8si2.
19550 * config/i386/i386-builtin.def (__builtin_ia32_cvttpd2uqq256_mask,
19551 __builtin_ia32_cvttpd2uqq128_mask, __builtin_ia32_cvttps2uqq256_mask,
19552 __builtin_ia32_cvttps2uqq128_mask, __builtin_ia32_cvtuqq2ps256_mask,
19553 __builtin_ia32_cvtuqq2ps128_mask, __builtin_ia32_cvtuqq2pd256_mask,
19554 __builtin_ia32_cvtuqq2pd128_mask, __builtin_ia32_cvttpd2udq512_mask,
19555 __builtin_ia32_cvtuqq2ps512_mask, __builtin_ia32_cvtuqq2pd512_mask,
19556 __builtin_ia32_cvttps2uqq512_mask, __builtin_ia32_cvttpd2uqq512_mask):
19557 Use fixuns instead ufix or floatuns instead ufloat in CODE_FOR_ names.
19559 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
19563 * varasm.c (assemble_alias): Lookup ifunc attribute on error.
19565 2018-05-25 Jim Wilson <jimw@sifive.com>
19567 * config/riscv/riscv-protos.h (riscv_epilogue_uses): New.
19568 * config/riscv/riscv.c (struct machine_function): Add
19569 interrupt_handler_p and attribute_checked_p fields.
19570 (riscv_attribute_table): Add interrupt.
19571 (riscv_interrupt_type_p): New.
19572 (riscv_save_reg_p): Save extra regs for interrupt handler.
19573 (riscv_use_save_libcall): Return false for interrupt handler.
19574 (riscv_first_stack_step): Add forward declaration.
19575 (riscv_compute_frame_info): New local interrupt_save_t1. Set it
19576 for interrupt handler with large frame. Use it for saved reg list.
19577 (riscv_expand_prologue): Move flag_stack_usage_info support to
19578 eliminate duplication.
19579 (riscv_expand_epilogue): Generate mret for interrupt handler.
19580 (riscv_epilogue_uses): New.
19581 (riscv_can_use_return_insn): Return false for interrupt handler.
19582 (riscv_function_ok_for_sibcall): Likewise.
19583 (riscv_set_current_function): Add interrupt handler support.
19584 * config/riscv/riscv.h (EPILOGUE_USES): Call riscv_epilogue_uses.
19585 * config/riscv/riscv.md (UNSPECV_MRET): New.
19587 (riscv_frflags, riscv_fsflags): Use tab after opcode.
19589 * doc/extend.texi (RISC-V Function Attributes) <interrupt>: New.
19591 2018-05-25 Bill Schmidt <wschmidt@linux.ibm.com>
19593 PR tree-optimization/85712
19594 * gimple-ssa-strength-reduction.c (replace_one_candidate): Skip if
19595 this candidate has already been replaced in-situ by a copy.
19597 2018-05-25 Jason Merrill <jason@redhat.com>
19599 PR c++/80485 - inline function non-zero address.
19600 * symtab.c (nonzero_address): Check DECL_COMDAT.
19602 2018-05-25 Uros Bizjak <ubizjak@gmail.com>
19605 * config/alpha/alpha.md (ashlsi3): New insn pattern.
19606 (*ashlsi_se): Rename from *ashldi_se. Define as sign
19607 extension of SImode operation. Use const123_operand predicate.
19608 (*saddsi_1): Remove.
19609 (*saddl_se_1): Ditto.
19610 (*ssubsi_1): Ditto.
19611 (*ssubl_se_1): Ditto.
19612 * config/alpha/predicates.md (const123_operand): New predicate.
19613 * config/alpha/constraints.md (P): Use IN_RANGE.
19615 2018-05-25 Richard Biener <rguenther@suse.de>
19617 * tree-ssa-alias.h (refs_may_alias_p): Add tbaa_p bool parameter,
19619 (ref_maybe_used_by_stmt_p): Likewise.
19620 (stmt_may_clobber_ref_p): Likewise.
19621 (stmt_may_clobber_ref_p_1): Likewise.
19622 * tree-ssa-alias.c (refs_may_alias_p): Add tbaa_p bool parameter
19624 (ref_maybe_used_by_stmt_p): Likewise.
19625 (stmt_may_clobber_ref_p): Likewise.
19626 (stmt_may_clobber_ref_p_1): Likewise.
19627 * tree-vect-data-refs.c (vect_slp_analyze_node_dependences): Use
19628 the alias oracle to disambiguate DRs with stmts DR analysis
19630 (vect_analyze_data_refs): Do not give up on not analyzable
19631 DRs for BB vectorization. Remove code truncating the dataref
19634 2018-05-25 Jakub Jelinek <jakub@redhat.com>
19637 * config/i386/sse.md (<avx512>_eq<mode>3<mask_scalar_merge_name>_1):
19638 Add (=Yk,v,C) variant using vptestm insn. Use TARGET_AVX512BW
19639 in test instead of TARGET_AVX512F for VI12_AVX512VL iterator.
19641 2018-05-25 Richard Biener <rguenther@suse.de>
19643 * tree-vect-data-refs.c (vect_find_stmt_data_reference): New
19644 function, combining stmt data ref gathering and fatal analysis
19646 (vect_analyze_data_refs): Remove now redudnant code and simplify.
19647 * tree-vect-loop.c (vect_get_datarefs_in_loop): Factor out from
19648 vect_analyze_loop_2 and use vect_find_stmt_data_reference.
19649 * tree-vect-slp.c (vect_slp_bb): Use vect_find_stmt_data_reference.
19650 * tree-vectorizer.h (vect_find_stmt_data_reference): Declare.
19652 2018-05-25 Bin Cheng <bin.cheng@arm.com>
19654 PR tree-optimization/85720
19655 * tree-loop-distribution.c (break_alias_scc_partitions): Don't merge
19656 SCC if all partitions are builtins.
19657 (version_loop_by_alias_check): New parameter. Generate cancelable
19658 runtime alias check if all partitions are builtins.
19659 (distribute_loop): Update call to above function.
19661 2018-05-25 Bin Cheng <bin.cheng@arm.com>
19663 * tree-outof-ssa.c (tree-ssa.h, tree-dfa.h): Include header files.
19664 (create_default_def, for_all_parms): Moved from tree-ssa-coalesce.c.
19665 (parm_default_def_partition_arg): Ditto.
19666 (set_parm_default_def_partition): Ditto.
19667 (get_parm_default_def_partitions): Ditto and make it static.
19668 (get_undefined_value_partitions): Ditto and make it static.
19669 (remove_ssa_form): Refactor call to init_var_map here.
19670 * tree-ssa-coalesce.c (build_ssa_conflict_graph): Support live range
19671 computation for loop region.
19672 (coalesce_partitions, compute_optimized_partition_bases): Ditto.
19673 (register_default_def): Delete.
19674 (for_all_parms, create_default_def): Move to tree-outof-ssa.c.
19675 (parm_default_def_partition_arg): Ditto.
19676 (set_parm_default_def_partition): Ditto.
19677 (get_parm_default_def_partitions): Ditto and make it static.
19678 (get_undefined_value_partitions): Ditto and make it static.
19679 (coalesce_with_default, coalesce_with_default): Update comment.
19680 (create_coalesce_list_for_region): New func factored out from
19681 create_outofssa_var_map.
19682 (populate_coalesce_list_for_outofssa): New func factored out from
19683 create_outofssa_var_map and coalesce_ssa_name.
19684 (create_outofssa_var_map): Delete.
19685 (coalesce_ssa_name): Refactor to support live range computation.
19686 * tree-ssa-coalesce.h (coalesce_ssa_name): Change decl.
19687 (get_parm_default_def_partitions): Delete.
19688 (get_undefined_value_partitions): Ditto.
19689 * tree-ssa-live.c (init_var_map, delete_var_map): Support live range
19690 computation for loop region.
19691 (new_tree_live_info, loe_visit_block): Ditto.
19692 (live_worklist, set_var_live_on_entry): Ditto.
19693 (calculate_live_on_exit, verify_live_on_entry): Ditto.
19694 * tree-ssa-live.h (struct _var_map): New fields.
19695 (init_var_map): Change decl.
19696 (region_contains_p): New.
19698 2018-05-25 Bin Cheng <bin.cheng@arm.com>
19700 * tree-ssa-live.h (live_merge_and_clear): Delete.
19702 2018-05-25 Richard Biener <rguenther@suse.de>
19705 * tree-dump.c (dequeue_and_dump): Remove access to removed
19706 operand 2 of a SWITCH_EXPR.
19708 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19710 * doc/sourcebuild.texi (vect_double_cond_arith): Include
19711 multiplication and division.
19712 * doc/md.texi (cond_mul@var{m}, cond_div@var{m}, cond_mod@var{m})
19713 (cond_udiv@var{m}, cond_umod@var{m}): Document.
19714 * optabs.def (cond_smul_optab, cond_sdiv_optab, cond_smod_optab)
19715 (cond_udiv_optab, cond_umod_optab): New optabs.
19716 * internal-fn.def (IFN_COND_MUL, IFN_COND_DIV, IFN_COND_MOD)
19717 (IFN_COND_RDIV): New internal functions.
19718 * internal-fn.c (get_conditional_internal_fn): Handle TRUNC_DIV_EXPR,
19719 TRUNC_MOD_EXPR and RDIV_EXPR.
19720 * match.pd (UNCOND_BINARY, COND_BINARY): Handle them.
19721 * config/aarch64/iterators.md (UNSPEC_COND_MUL, UNSPEC_COND_DIV):
19723 (SVE_INT_BINARY): Include mult.
19724 (SVE_COND_FP_BINARY): Include UNSPEC_MUL and UNSPEC_DIV.
19725 (optab, sve_int_op): Handle mult.
19726 (optab, sve_fp_op, commutative): Handle UNSPEC_COND_MUL and
19728 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New pattern
19729 for SVE_INT_BINARY_SD.
19731 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19733 * config/aarch64/iterators.md (SVE_INT_BINARY_SD): New code iterator.
19734 (optab, sve_int_op): Handle div and udiv.
19735 * config/aarch64/aarch64-sve.md (<optab><mode>3): New expander
19736 for SVE_INT_BINARY_SD.
19737 (*<optab><mode>3): New insn for the same.
19739 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19741 * tree-vect-patterns.c: Include predict.h.
19742 (vect_recog_divmod_pattern): Restrict check for division support
19743 to when optimizing for size.
19745 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19747 * doc/sourcebuild.texi (vect_double_cond_arith: Document.
19748 * gimple-match.h (gimple_match_op::MAX_NUM_OPS): Bump to 4.
19749 (gimple_match_op::gimple_match_op): Add an overload for 4 operands.
19750 (gimple_match_op::set_op): Likewise.
19751 (gimple_resimplify4): Declare.
19752 * genmatch.c (get_operand_type): Handle CFN_COND_* functions.
19753 (expr::gen_transform): Likewise.
19754 (decision_tree::gen): Generate a simplification routine for 4 operands.
19755 * gimple-match-head.c (gimple_simplify): Add an overload for
19756 4 operands. In the top-level function, handle up to 4 call
19757 arguments and call gimple_resimplify4.
19758 (gimple_resimplify4): New function.
19759 (build_call_internal): Pass a fourth operand.
19760 (maybe_push_to_seq): Likewise.
19761 * match.pd (UNCOND_BINARY, COND_BINARY): New operator lists.
19762 Fold VEC_COND_EXPRs of an operation and a default value into
19763 an IFN_COND_* function if possible.
19764 * config/aarch64/iterators.md (UNSPEC_COND_MAX, UNSPEC_COND_MIN):
19766 (SVE_COND_FP_BINARY): Include them.
19767 (optab, sve_fp_op): Handle them.
19768 (SVE_INT_BINARY_REV): New code iterator.
19769 (SVE_COND_FP_BINARY_REV): New int iterator.
19770 (commutative): New int attribute.
19771 * config/aarch64/aarch64-protos.h (aarch64_sve_prepare_conditional_op):
19773 * config/aarch64/aarch64.c (aarch64_sve_prepare_conditional_op): New
19775 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Use it.
19776 (*cond_<optab><mode>): New patterns for reversed operands.
19778 2018-05-25 Richard Biener <rguenther@suse.de>
19780 * tree-vectorizer.h (STMT_VINFO_GROUP_*, GROUP_*): Remove.
19781 (DR_GROUP_*): New, assert we have non-NULL ->data_ref_info.
19782 (REDUC_GROUP_*): New, assert we have NULL ->data_ref_info.
19783 (STMT_VINFO_GROUPED_ACCESS): Adjust.
19784 * tree-vect-data-refs.c (everywhere): Adjust users.
19785 * tree-vect-loop.c (everywhere): Likewise.
19786 * tree-vect-slp.c (everywhere): Likewise.
19787 * tree-vect-stmts.c (everywhere): Likewise.
19788 * tree-vect-patterns.c (vect_reassociating_reduction_p): Likewise.
19790 2018-05-25 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
19792 * configure.ac (gcc_cv_as_section_has_e): Move to common section.
19794 (gcc_cv_as_section_exclude): ... this.
19795 Try Solaris as #exclude syntax.
19796 * configure: Regenerate.
19797 * config.in: Regenerate.
19798 * config/i386/i386.c (i386_solaris_elf_named_section): Handle
19800 * config/sparc/sparc.c (sparc_solaris_elf_asm_named_section)
19801 [HAVE_GAS_SECTION_EXCLUDE]: Handle SECTION_EXCLUDE.
19803 * varasm.c (default_elf_asm_named_section): Don't check if
19804 HAVE_GAS_SECTION_EXCLUDE is defined.
19806 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19808 * doc/md.texi: Update the documentation of the cond_* optabs
19809 to mention the new final operand. Fix GET_MODE_NUNITS call.
19810 Describe the scalar case too.
19811 * internal-fn.def (IFN_EXTRACT_LAST): Change type to fold_left.
19812 * internal-fn.c (expand_cond_unary_optab_fn): Expect 3 operands
19814 (expand_cond_binary_optab_fn): Expect 4 operands instead of 3.
19815 (get_conditional_internal_fn): Update comment.
19816 * tree-vect-loop.c (vectorizable_reduction): Pass the original
19817 accumulator value as a final argument to conditional functions.
19818 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): Turn into
19819 a define_expand and add an "else" operand. Assert for now that
19820 the else operand is equal to operand 2. Use SVE_INT_BINARY and
19821 SVE_COND_FP_BINARY instead of SVE_COND_INT_OP and SVE_COND_FP_OP.
19822 (*cond_<optab><mode>): New patterns.
19823 * config/aarch64/iterators.md (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX)
19824 (UNSPEC_COND_SMIN, UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
19825 (UNSPEC_COND_EOR): Delete.
19826 (optab): Remove associated mappings.
19827 (SVE_INT_BINARY): New code iterator.
19828 (sve_int_op): Remove int attribute and add "minus" to the code
19830 (SVE_COND_INT_OP): Delete.
19831 (SVE_COND_FP_OP): Rename to...
19832 (SVE_COND_FP_BINARY): ...this.
19834 2018-05-25 Richard Sandiford <richard.sandiford@linaro.org>
19836 * optabs.c (can_reuse_operands_p): New function.
19837 (maybe_legitimize_operands): Try to reuse the results for
19840 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
19842 * config/i386/sse.md (cvtusi2<ssescalarmodesuffix>64<round_name>):
19843 Add {q} suffix to insn mnemonic.
19845 2018-05-23 Jozef Lawrynowicz <jozef.l@mittosystems.com>
19847 * config/msp430/msp430.c (TARGET_WARN_FUNC_RETURN): Define.
19848 (msp430_warn_func_return): New.
19850 2018-05-24 Roger Sayle <roger@nextmovesoftware.com>
19852 * fold-const.c (tree_nonzero_bits): New function.
19853 * fold-const.h (tree_nonzero_bits): Likewise.
19854 * match.pd (POPCOUNT): New patterns to fold BUILTIN_POPCOUNT and
19855 friends. POPCOUNT(x&1) => x&1, POPCOUNT(x)==0 => x==0, etc.
19857 2018-05-24 H.J. Lu <hongjiu.lu@intel.com>
19861 * varasm.c (assemble_alias): Check ifunc_resolver only on
19864 2018-05-24 Uros Bizjak <ubizjak@gmail.com>
19867 * config/i386/sse.md (movdi_to_sse): Do not generate pseudo
19868 when memory input operand is handled.
19870 2018-05-24 Luis Machado <luis.machado@linaro.org>
19872 * config/aarch64/aarch64.c (qdf24xx_addrcost_table): New static
19874 (qdf24xx_tunings) <addr_costs>: Set to qdf24xx_addrcost_table.
19876 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
19878 * match.pd: Delay FMA folds until after vectorization.
19880 2018-05-24 Andre Vieira <andre.simoesdiasvieira@arm.com>
19883 * config/aarch64/predicates.md (aarch64_mem_pair_lanes_operand): Make
19884 address check not strict.
19886 2018-05-24 Richard Sandiford <richard.sandiford@linaro.org>
19888 * gimple-match.h (gimple_match_op): New class.
19889 (mprts_hook): Replace parameters with a gimple_match_op *.
19890 (maybe_build_generic_op): Likewise.
19891 (gimple_simplified_result_is_gimple_val): Replace parameters with
19892 a const gimple_match_op *.
19893 (gimple_simplify): Replace code_helper * and tree * parameters with
19894 a gimple_match_op * parameter.
19895 (gimple_resimplify1): Replace code_helper *, tree and tree *
19896 parameters with a gimple_match_op * parameter.
19897 (gimple_resimplify2): Likewise.
19898 (gimple_resimplify3): Likewise.
19899 (maybe_push_res_to_seq): Replace code_helper, tree and tree *
19900 parameters with a gimple_match_op * parameter.
19901 * gimple-match-head.c (gimple_simplify): Change prototypes of
19902 auto-generated functions to take a gimple_match_op * instead of
19903 separate code_helper * and tree * parameters. Make the same
19904 change in the top-level overload and update calls to the
19905 gimple_resimplify routines. Update calls to the auto-generated
19906 functions and to maybe_push_res_to_seq in the publicly-facing
19907 operation-specific gimple_simplify overloads.
19908 (gimple_match_op::MAX_NUM_OPS): Define.
19909 (gimple_resimplify1): Replace rcode and ops with a single res_op
19910 parameter. Update call to gimple_simplify.
19911 (gimple_resimplify2): Likewise.
19912 (gimple_resimplify3): Likewise.
19913 (mprts_hook): Replace parameters with a gimple_match_op *.
19914 (maybe_build_generic_op): Likewise.
19915 (build_call_internal): Replace type, nargs and ops with
19916 a gimple_match_op *.
19917 (maybe_push_res_to_seq): Replace res_code, type and ops parameters
19918 with a single gimple_match_op *. Update calls to mprts_hook,
19919 build_call_internal and gimple_simplified_result_is_gimple_val.
19920 Factor out code that is common to the tree_code and combined_fn cases.
19921 * genmatch.c (expr::gen_transform): Replace tem_code and
19922 tem_ops with a gimple_match_op called tem_op. Update calls
19923 to the gimple_resimplify functions and maybe_push_res_to_seq.
19924 (dt_simplify::gen_1): Manipulate res_op instead of res_code and
19925 res_ops. Update call to the gimple_resimplify functions.
19926 (dt_simplify::gen): Pass res_op instead of res_code and res_ops.
19927 (decision_tree::gen): Make the functions take a gimple_match_op *
19928 called res_op instead of separate res_code and res_ops parameters.
19929 Update call accordingly.
19930 * gimple-fold.c (replace_stmt_with_simplification): Replace rcode
19931 and ops with a single res_op parameter. Update calls to
19932 maybe_build_generic_op and maybe_push_res_to_seq.
19933 (fold_stmt_1): Update calls to gimple_simplify and
19934 replace_stmt_with_simplification.
19935 (gimple_fold_stmt_to_constant_1): Update calls to gimple_simplify
19936 and gimple_simplified_result_is_gimple_val.
19937 * tree-cfgcleanup.c (cleanup_control_expr_graph): Update call to
19939 * tree-ssa-sccvn.c (vn_lookup_simplify_result): Replace parameters
19940 with a gimple_match_op *.
19941 (vn_nary_build_or_lookup): Likewise. Update call to
19942 vn_nary_build_or_lookup_1.
19943 (vn_nary_build_or_lookup_1): Replace rcode, type and ops with a
19944 gimple_match_op *. Update calls to the gimple_resimplify routines
19945 and to gimple_simplified_result_is_gimple_val.
19946 (vn_nary_simplify): Update call to vn_nary_build_or_lookup_1.
19947 Use gimple_match_op::MAX_NUM_OPS instead of a hard-coded 3.
19948 (vn_reference_lookup_3): Update call to vn_nary_build_or_lookup.
19949 (visit_nary_op): Likewise.
19950 (visit_reference_op_load): Likewise.
19952 2018-05-23 Luis Machado <luis.machado@linaro.org>
19954 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Use correct type
19955 modifier for printing the step amount.
19957 2018-05-23 Jozef Lawrynowicz <jozef.l@somniumtech.com>
19960 * tree.c (build_common_tree_nodes): Dont set TYPE_SIZE for __intN
19963 2018-05-23 Segher Boessenkool <segher@kernel.crashing.org>
19965 * doc/sourcebuild.texi (Endianness): New subsubsection.
19967 2018-05-23 Luis Machado <luis.machado@linaro.org>
19969 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19970 <prefetch_dynamic_strides>: New const bool field.
19971 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19972 prefetch_dynamic_strides.
19973 (exynosm1_prefetch_tune): Likewise.
19974 (thunderxt88_prefetch_tune): Likewise.
19975 (thunderx_prefetch_tune): Likewise.
19976 (thunderx2t99_prefetch_tune): Likewise.
19977 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides to
19979 (aarch64_override_options_internal): Update to set
19980 PARAM_PREFETCH_DYNAMIC_STRIDES.
19981 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
19982 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
19983 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
19984 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
19985 prefetch-dynamic-strides setting.
19987 2018-05-23 Luis Machado <luis.machado@linaro.org>
19989 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
19990 <minimum_stride>: New const int field.
19991 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
19992 minimum_stride field defaulting to -1.
19993 (exynosm1_prefetch_tune): Likewise.
19994 (thunderxt88_prefetch_tune): Likewise.
19995 (thunderx_prefetch_tune): Likewise.
19996 (thunderx2t99_prefetch_tune): Likewise.
19997 (qdf24xx_prefetch_tune) <minimum_stride>: Set to 2048.
19998 <default_opt_level>: Set to 3.
19999 (aarch64_override_options_internal): Update to set
20000 PARAM_PREFETCH_MINIMUM_STRIDE.
20001 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
20002 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
20003 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
20004 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
20005 stride is constant and is below the minimum stride threshold.
20007 2018-05-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20009 * config/arm/arm-cpus.in (mode26): Delete.
20010 (armv4): Delete mode26 reference.
20011 * config/arm/arm.c (arm_configure_build_target): Delete use of
20014 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
20016 * config/i386/i386.md (*floatuns<SWI48:mode><MODEF:mode>2_avx512):
20018 (floatunssi<mode>2): Also enable for AVX512F and TARGET_SSE_MATH.
20019 Rewrite expander pattern. Emit gen_floatunssi<mode>2_i387_with_xmm
20021 (floatunsdisf2): Rewrite expander pattern. Hanlde TARGET_AVX512F.
20022 (floatunsdidf2): Ditto.
20024 2018-05-23 Uros Bizjak <ubizjak@gmail.com>
20026 * config/i386/i386.md (fixuns_trunc<mode>di2): New insn pattern.
20027 (fixuns_trunc<mode>si2_avx512f): Ditto.
20028 (*fixuns_trunc<mode>si2_avx512f_zext): Ditto.
20029 (fixuns_trunc<mode>si2): Also enable for AVX512F and TARGET_SSE_MATH.
20030 Emit fixuns_trunc<mode>si2_avx512f for AVX512F targets.
20032 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
20034 PR rtl-optimization/79985
20035 * df-scan.c (df_insn_refs_collect): Remove special case for
20036 global registers and asm statements.
20038 2018-05-23 Alexander Monakov <amonakov@ispras.ru>
20040 * extend.texi (Global Register Variables): Rewrite the bullet list.
20041 Note that the register is available for allocation. Note that access
20042 via inline asm must use constraints. Add note about async-signal
20043 handlers. Remove paragraph about automagic register selection.
20045 2018-05-23 Richard Biener <rguenther@suse.de>
20047 * tree-ssa-sccvn.c (vn_reference_lookup_3): Remove restriction
20048 of fixed offset from memset VN.
20050 2018-05-23 Bill Schmidt <wschmidt@linux.ibm.com>
20052 * gimple-ssa-strength-reduction.c (struct slsr_cand_d): Add
20053 first_interp field.
20054 (alloc_cand_and_find_basis): Initialize first_interp field.
20055 (slsr_process_mul): Modify first_interp field.
20056 (slsr_process_add): Likewise.
20057 (slsr_process_cast): Modify first_interp field for each new
20059 (slsr_process_copy): Likewise.
20060 (dump_candidate): Dump first_interp field.
20061 (replace_mult_candidate): Process all interpretations, not just
20063 (replace_rhs_if_not_dup): Likewise.
20064 (replace_one_candidate): Likewise.
20066 2018-05-23 Wilco Dijkstra <wdijkstr@arm.com>
20068 * config/aarch64/aarch64.c (aarch64_use_frame_pointer):
20070 (aarch64_needs_frame_chain): New function.
20071 (aarch64_parse_override_string): Set aarch64_use_frame_pointer.
20073 2018-05-23 Sudakshina Das <sudi.das@arm.com>
20076 * common/config/aarch64/aarch64-common.c (aarch64_handle_option):
20077 Check val before adding MASK_STRICT_ALIGN to opts->x_target_flags.
20078 * config/aarch64/aarch64.opt (mstrict-align): Remove RejectNegative.
20079 * config/aarch64/aarch64.c (aarch64_attributes): Mark allow_neg
20080 as true for strict-align.
20081 (aarch64_can_inline_p): Perform checks even when callee has no
20082 attributes to check for strict alignment.
20083 * doc/extend.texi (AArch64 Function Attributes): Document
20085 * doc/invoke.texi: (AArch64 Options): Likewise.
20087 2018-05-23 Richard Sandiford <richard.sandiford@linaro.org>
20089 PR tree-optimization/85853
20090 * tree-vect-slp.c (vect_slp_analyze_node_operations): Split out
20091 the handling of the root of the node to...
20092 (vect_slp_analyze_node_operations_1): ...this new function,
20093 and run the whole thing with the child nodes' def types
20094 set according to their SLP node's def type.
20096 2018-05-23 Richard Biener <rguenther@suse.de>
20098 PR middle-end/85874
20099 * tree-data-ref.c (create_runtime_alias_checks): Defer
20100 and ignore overflow warnings.
20102 2018-05-23 Yury Gribov <tetra2005@gmail.com>
20104 PR tree-optimization/85822
20105 * tree-vrp.c (is_masked_range_test): Fix handling of negative
20108 2018-05-23 Richard Biener <rguenther@suse.de>
20110 * tree-ssa-sccvn.c (vn_reference_lookup_3): Handle arbitrary
20111 memset constants via native_interpret_expr.
20113 2018-05-22 H.J. Lu <hongjiu.lu@intel.com>
20116 * cgraph.h (cgraph_node::create): Set ifunc_resolver for ifunc
20118 (cgraph_node::create_alias): Likewise.
20119 (cgraph_node::get_availability): Check ifunc_resolver instead
20120 of looking up ifunc attribute.
20121 * cgraphunit.c (maybe_diag_incompatible_alias): Likewise.
20122 * varasm.c (do_assemble_alias): Likewise.
20123 (assemble_alias): Likewise.
20124 (default_binds_local_p_3): Likewise.
20125 * cgraph.h (cgraph_node): Add ifunc_resolver.
20126 (cgraph_node::only_called_directly_or_aliased_p): Return false
20127 for IFUNC resolver.
20128 * lto-cgraph.c (input_node): Set ifunc_resolver for ifunc
20130 * symtab.c (symtab_node::verify_base): Verify that ifunc_resolver
20131 is equivalent to lookup_attribute ("ifunc", DECL_ATTRIBUTES (decl)).
20132 (symtab_node::binds_to_current_def_p): Check ifunc_resolver
20133 instead of looking up ifunc attribute.
20135 2018-05-22 Luis Machado <luis.machado@linaro.org>
20137 * config/aarch64/aarch64.md (*ashift<mode>_extv_bfiz): New pattern.
20139 2018-05-22 Martin Sebor <msebor@redhat.com>
20141 PR middle-end/85359
20142 * builtins.c (expand_builtin_strcpy): Call maybe_warn_nonstring_arg
20143 only when expasion succeeds.
20144 (expand_builtin_strcmp): Same.
20145 (expand_builtin_strncmp): Same.
20147 2018-05-22 Martin Sebor <msebor@redhat.com>
20149 * calls.c (maybe_warn_nonstring_arg): Fix a typo in a comment.
20151 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
20152 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20154 * config/aarch64/aarch64-ldpstp.md: Replace uses of
20155 aarch64_mem_pair_operand with memory_operand and delete operand swapping
20157 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
20158 Add check for legitimate_address.
20159 (aarch64_gen_adjusted_ldpstp): Swap operands where appropriate.
20160 (aarch64_swap_ldrstr_operands): New.
20161 * config/aarch64/aarch64-protos.h (aarch64_swap_ldrstr_operands):
20164 2018-05-22 Jackson Woodruff <jackson.woodruff@arm.com>
20165 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20167 * config/aarch64/aarch64.md: New patterns to generate stp
20169 (store_pair_sw, store_pair_dw): New patterns to generate stp for
20170 single words and double words.
20171 (load_pair_sw, load_pair_dw): Likewise.
20172 (store_pair_sf, store_pair_df, store_pair_si, store_pair_di):
20174 (load_pair_sf, load_pair_df, load_pair_si, load_pair_di):
20176 * config/aarch64/aarch64-ldpstp.md: Modify peephole
20177 for different mode ldpstp and add peephole for merged zero stores.
20178 Likewise for loads.
20179 * config/aarch64/aarch64.c (aarch64_operands_ok_for_ldpstp):
20181 (aarch64_gen_store_pair): Rename calls to match new patterns.
20182 (aarch64_gen_load_pair): Rename calls to match new patterns.
20183 * config/aarch64/aarch64-simd.md (load_pair<mode>): Rename to...
20184 (load_pair<DREG:mode><DREG2:mode>): ... This.
20185 (store_pair<mode>): Rename to...
20186 (vec_store_pair<DREG:mode><DREG2:mode>): ... This.
20187 * config/aarch64/iterators.md (DREG, DREG2, DX2, SX, SX2, DSX):
20188 New mode iterators.
20189 (V_INT_EQUIV): Handle SImode.
20190 * config/aarch64/predicates.md (aarch64_reg_zero_or_fp_zero):
20193 2018-05-22 Martin Sebor <msebor@redhat.com>
20196 * calls.c (maybe_warn_nonstring_arg): Use string length to set
20197 or ajust the presumed bound on an operation to avoid unnecessary
20200 2018-05-22 Martin Sebor <msebor@redhat.com>
20202 PR tree-optimization/85826
20203 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Avoid
20204 assuming that a DECL necesarily has a constant size.
20206 2018-05-22 Richard Sandiford <richard.sandiford@linaro.org>
20208 PR middle-end/85862
20209 * internal-fn.c (expand_direct_optab_fn): Cope with a null lhs.
20211 2018-05-22 Richard Biener <rguenther@suse.de>
20213 PR tree-optimization/85834
20214 * tree-ssa-sccvn.c (vn_reference_lookup_3): Properly handle
20215 non-constant and non-zero memset arguments.
20217 2018-05-22 Martin Liska <mliska@suse.cz>
20220 * ipa-icf.c (sem_item::add_type): Do not ICE for incomplete types.
20222 2018-05-22 Richard Biener <rguenther@suse.de>
20224 PR tree-optimization/85863
20225 * tree-vect-stmts.c (vect_is_simple_cond): Only widen invariant
20226 comparisons when vectype is specified.
20227 (vectorizable_condition): Do not specify vectype for
20228 vect_is_simple_cond when SLP vectorizing.
20230 2018-05-21 Michael Meissner <meissner@linux.ibm.com>
20233 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Do not
20234 define __ibm128 as long double.
20235 * config/rs6000/rs6000.c (rs6000_init_builtins): Create __ibm128
20236 as a distinct type when IEEE 128-bit support is enabled.
20237 (init_float128_ieee): Fix up conversions between IFmode and IEEE
20238 128-bit types to use the correct functions.
20239 (rs6000_expand_float128_convert): Use explicit FLOAT_EXTEND to
20240 convert between 128-bit floating point types that have different
20241 modes but the same representation, instead of using gen_lowpart to
20243 * config/rs6000/rs6000.md (IFKF): New iterator for IFmode and
20245 (IFKF_reg): New attributes to give the register constraints for
20247 (extend<mode>tf2_internal): New insns to mark an explicit
20248 conversion between 128-bit floating point types that have a
20249 different mode but share the same representation.
20251 2018-05-21 Richard Sandiford <richard.sandiford@linaro.org>
20253 PR tree-optimization/85814
20254 * tree-ssa-strlen.c (get_stridx_plus_constant): Cope with
20255 a null return from get_strinfo when unsharing the next
20256 strinfo in the chain.
20258 2018-05-21 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
20261 * varasm.c (weak_finish): Clean up weak_decls.
20263 2018-05-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20265 * config/aarch64/aarch64.md ("unspec"): Define UNSPEC_SABAL,
20266 UNSPEC_SABDL2, UNSPEC_SADALP, UNSPEC_UABAL, UNSPEC_UABDL2,
20267 UNSPEC_UADALP values.
20268 * config/aarch64/iterators.md (ABAL): New int iterator.
20271 (sur): Add mappings for the above.
20272 * config/aarch64/aarch64-simd.md (aarch64_<sur>abdl2<mode>_3):
20274 (aarch64_<sur>abal<mode>_4): Likewise.
20275 (aarch64_<sur>adalp<mode>_3): Likewise.
20276 (<sur>sadv16qi): New define_expand.
20278 2018-05-21 Alexander Nesterovskiy <alexander.nesterovskiy@intel.com>
20280 * config/i386/i386.md (*movsf_internal): AVX falsedep fix.
20281 (*movdf_internal): Ditto.
20282 (*rcpsf2_sse): Ditto.
20283 (*rsqrtsf2_sse): Ditto.
20284 (*sqrt<mode>2_sse): Ditto.
20286 2018-05-21 Tamar Christina <tamar.christina@arm.com>
20288 * config/aarch64/aarch64-simd.md (aarch64_eor3qv8hi): Change to
20290 (aarch64_bcaxqv8hi): Change to bcaxq<mode>4.
20291 * config/aarch64/aarch64-simd-builtins.def (veor3q_u8, veor3q_u32,
20292 veor3q_u64, veor3q_s8, veor3q_s16, veor3q_s32, veor3q_s64, vbcaxq_u8,
20293 vbcaxq_u32, vbcaxq_u64, vbcaxq_s8, vbcaxq_s16, vbcaxq_s32,
20295 * config/aarch64/arm_neon.h: Likewise.
20296 * config/aarch64/iterators.md (VQ_I): New.
20298 2018-05-21 Alexey Brodkin <abrodkin@synopsys.com>
20300 * config.gcc: Add arc/t-multilib-linux to tmake_file for
20302 * config/arc/t-multilib-linux: Specify MULTILIB_OPTIONS and
20305 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
20307 * config/nds32/constraints.md (S): New constraint.
20308 * config/nds32/nds32.md (call_internal): Use constraint S.
20309 (call_value_internal): Likewise.
20310 (sibcall_internal): Likewise.
20311 (sibcall_value_internal): Likewise.
20313 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
20314 Chung-Ju Wu <jasonwucj@gmail.com>
20316 * config/nds32/nds32.c (nds32_register_move_cost): Take garywolf cpu
20317 into consideration.
20319 2018-05-20 Kito Cheng <kito.cheng@gmail.com>
20320 Chung-Ju Wu <jasonwucj@gmail.com>
20322 * config/nds32/nds32-cost.c (rtx_cost_model_t): New structure.
20323 (insn_size_16bit, insn_size_32bit): New variables for cost evaluation.
20324 (nds32_rtx_costs_impl): Simplify.
20325 (nds32_address_cost_impl): Simplify.
20326 (nds32_init_rtx_costs): New function.
20327 (nds32_rtx_costs_speed_prefer): Likewise.
20328 (nds32_rtx_costs_size_prefer): Likewise.
20329 (nds32_address_cost_speed_prefer): Likewise.
20330 (nds32_address_cost_speed_fwprop): Likewise.
20331 (nds32_address_cost_size_prefer): Likewise.
20332 * config/nds32/nds32-protos.h (nds32_init_rtx_costs): Declare.
20333 * config/nds32/nds32.c (nds32_option_override): Use
20334 nds32_init_rtx_costs function.
20336 2018-05-20 Chung-Ju Wu <jasonwucj@gmail.com>
20338 * config/nds32/nds32.c (nds32_asm_file_start): Output pipeline model.
20339 * config/nds32/nds32.h (TARGET_PIPELINE_N7): Define.
20340 (TARGET_PIPELINE_N8): Likewise.
20341 (TARGET_PIPELINE_N10): Likewise.
20342 (TARGET_PIPELINE_N13): Likewise.
20343 (TARGET_PIPELINE_GRAYWOLF): Likewise.
20345 2018-05-19 Monk Chiang <sh.chiang04@gmail.com>
20347 * config/nds32/nds32-fpu.md: Update copyright year.
20349 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
20351 * config/nds32/nds32.h (ASM_SPEC): Adjust spec rule.
20353 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
20355 * config/nds32/nds32.c
20356 (nds32_md_asm_adjust): Consider flag_inline_asm_r15 variable.
20357 * config/nds32/nds32.opt (minline-asm-r15): New option.
20359 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
20361 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS): Add
20363 * config/nds32/nds32.md (abssi2): New pattern.
20365 2018-05-19 Uros Bizjak <ubizjak@gmail.com>
20367 * config/i386/i386.md (rex64namesuffix): New mode attribute.
20368 * config/i386/sse.md (sse_cvtsi2ss<rex64namesuffix><round_name>):
20369 Merge insn pattern from sse_cvtsi2ss<round_name> and
20370 sse_cvtsi2ssq<round_name> using SWI48 mode iterator.
20371 (sse_cvtss2si<rex64namesuffix><round_name>): Merge insn pattern
20372 from sse_cvtss2si<round_name> and sse_cvtss2siq<round_name>
20373 using SWI48 mode iterator.
20374 (sse_cvtss2si<rex64namesuffix>_2): Merge insn pattern from
20375 sse_cvtss2si_2 and sse_cvtss2siq_2 using SWI48 mode iterator.
20376 (sse_cvttss2si<rex64namesuffix><round_saeonly_name>): Merge insn
20377 pattern from sse_cvttss2si<round_saeonly_name>
20378 and sse_cvttss2siq<round_saeonly_name> using SWI48 mode iterator.
20379 (avx512f_vcvtss2usi<rex64namesuffix><round_name>): Merge insn pattern
20380 from avx512f_vcvtss2usi<round_name> and avx512f_vcvtss2usiq<round_name>
20381 using SWI48 mode iterator.
20382 (avx512f_vcvttss2usi<rex64namesuffix><round_saeonly_name>): Merge
20383 insn pattern from avx512f_vcvttss2usi<round_saeonly_name> and
20384 avx512f_vcvttss2usiq<round_saeonly_name> using SWI48 mode iterator.
20385 (avx512f_vcvtsd2usi<rex64namesuffix><round_name>): Merge insn pattern
20386 from avx512f_vcvtsd2usi<round_name> and avx512f_vcvtsd2usiq<round_name>
20387 using SWI48 mode iterator.
20388 (avx512f_vcvttsd2usi<rex64namesuffix><round_saeonly_name>): Merge
20389 insn pattern from avx512f_vcvttsd2usi<round_saeonly_name> and
20390 avx512f_vcvttsd2usiq<round_saeonly_name> using SWI48 mode iterator.
20391 (sse2_cvtsd2si<rex64namesuffix><round_name>): Merge insn pattern from
20392 sse2_cvtsd2si<round_name> and sse2_cvtsd2siq<round_name> using
20393 SWI48 mode iterator.
20394 (sse2_cvtsd2si<rex64namesuffix>_2): Merge insn pattern from
20395 sse2_cvtsd2si_2 and sse2_cvtsd2siq_2 using SWI48 mode iterator.
20396 (sse_cvttsd2si<rex64namesuffix><round_saeonly_name>): Merge insn
20397 pattern from sse_cvttsd2si<round_saeonly_name>
20398 and sse_cvttsd2siq<round_saeonly_name> using SWI48 mode iterator.
20400 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
20402 * config/nds32/nds32-md-auxiliary.c
20403 (nds32_valid_smw_lwm_base_p): Refine.
20404 (nds32_output_smw_single_word): Refine.
20405 (nds32_output_smw_double_word): New.
20406 * config/nds32/nds32-protos.h (nds32_output_smw_double_word): New.
20408 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
20410 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push): Refine.
20411 (nds32_output_stack_pop): Refine.
20412 (nds32_expand_unaligned_load): Refine.
20413 (nds32_expand_unaligned_store): Refine.
20415 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20416 Chung-Ju Wu <jasonwucj@gmail.com>
20418 * config/nds32/constants.md: Add TP_REGNUM constant.
20419 (unspec_element): Add UNSPEC_GOTINIT, UNSPEC_GOT, UNSPEC_GOTOFF,
20420 UNSPEC_PLT, UNSPEC_TLSGD, UNSPEC_TLSLD, UNSPEC_TLSIE, UNSPEC_TLSLE and
20422 * config/nds32/nds32-doubleword.md: Consider flag_pic.
20423 * config/nds32/nds32-dspext.md (mov<mode>): Expand TLS and PIC cases.
20424 * config/nds32/nds32-predicates.c (nds32_const_unspec_p): New.
20425 * config/nds32/nds32-md-auxiliary.c: Implementation that support TLS
20426 and PIC code generation.
20427 * config/nds32/nds32-protos.h: Declarations that support TLS and PIC
20429 * config/nds32/nds32-relax-opt.c: Consider TLS and PIC for relax
20431 * config/nds32/nds32.md: Support TLS and PIC.
20432 * config/nds32/nds32.c: Support TLS and PIC.
20433 * config/nds32/nds32.h (nds32_relax_insn_type): New enum type.
20434 * config/nds32/predicates.md (nds32_nonunspec_symbolic_operand): New
20437 2018-05-19 Chung-Ju Wu <jasonwucj@gmail.com>
20439 * config/nds32/nds32-predicates.c (const_vector_to_hwint): Use machine
20440 mode with E_ prefix.
20442 2018-05-19 Kuan-Lin Chen <kuanlinchentw@gmail.com>
20443 Chung-Ju Wu <jasonwucj@gmail.com>
20445 * config/nds32/constants.md (unspec_element): Add UNSPEC_ICT.
20446 * config/nds32/nds32-md-auxiliary.c
20447 (symbolic_reference_mentioned_p): New.
20448 (nds32_legitimize_ict_address): New.
20449 (nds32_expand_ict_move): New.
20450 (nds32_indirect_call_referenced_p): New.
20451 (nds32_symbol_binds_local_p): Delete.
20452 (nds32_long_call_p): Modify.
20453 * config/nds32/nds32-opts.h (nds32_ict_model_type): New enum type.
20454 * config/nds32/nds32-protos.h
20455 (symbolic_reference_mentioned_p): Declare.
20456 (nds32_legitimize_ict_address): Declare.
20457 (nds32_expand_ict_move): Declare.
20458 (nds32_indirect_call_referenced_p): Declare.
20459 * config/nds32/nds32-relax-opt.c (nds32_ict_const_p): New.
20460 (nds32_relax_group): Use nds32_ict_const_p as condition.
20461 * config/nds32/nds32.c (nds32_attribute_table): Add "indirect_call".
20462 (nds32_asm_file_start): Output ict_model directive in asm code.
20463 (nds32_legitimate_address_p): Consider indirect call.
20464 (nds32_print_operand): Consider indirect call.
20465 (nds32_print_operand_address): Consider indirect call.
20466 (nds32_insert_attributes): Handle "indirect_call" attribute.
20467 (TARGET_LEGITIMATE_ADDRESS_P): Define.
20468 (TARGET_LEGITIMATE_CONSTANT_P): Define.
20469 (TARGET_CANNOT_FORCE_CONST_MEM): Define.
20470 (TARGET_DELEGITIMIZE_ADDRESS): Define.
20471 (TARGET_ASM_OUTPUT_ADDR_CONST_EXTRA): Define.
20472 * config/nds32/nds32.h (SYMBOLIC_CONST_P): Define.
20473 (TARGET_ICT_MODEL_SMALL): Define.
20474 (TARGET_ICT_MODEL_LARGE): Define.
20475 * config/nds32/nds32.md (movsi): Consider ict model.
20476 (call, call_value): Consider ict model.
20477 (sibcall, sibcall_value): Consider ict model.
20478 * config/nds32/nds32.opt (mict-model): New option.
20479 * config/nds32/predicates.md (nds32_symbolic_operand): Consider ict
20482 2018-05-18 Kito Cheng <kito.cheng@gmail.com>
20483 Monk Chiang <sh.chiang04@gmail.com>
20484 Jim Wilson <jimw@sifive.com>
20486 * common/config/riscv/riscv-common.c (riscv_parse_arch_string):
20487 Add support to parse rv32e*. Clear MASK_RVE for rv32i and rv64i.
20488 * config.gcc (riscv*-*-*): Add support for rv32e* and ilp32e.
20489 * config/riscv/riscv-c.c (riscv_cpu_cpp_builtins): Define
20490 __riscv_32e when TARGET_RVE. Handle ABI_ILP32E as soft-float ABI.
20491 * config/riscv/riscv-opts.h (riscv_abi_type): Add ABI_ILP32E.
20492 * config/riscv/riscv.c (riscv_compute_frame_info): When TARGET_RVE,
20493 compute save_libcall_adjustment properly.
20494 (riscv_option_override): Call error if TARGET_RVE and not ABI_ILP32E.
20495 (riscv_conditional_register_usage): Handle TARGET_RVE and ABI_ILP32E.
20496 * config/riscv/riscv.h (UNITS_PER_FP_ARG): Handle ABI_ILP32E.
20497 (STACK_BOUNDARY, ABI_STACK_BOUNDARY): Handle TARGET_RVE.
20498 (GP_REG_LAST, MAX_ARGS_IN_REGISTERS): Likewise.
20499 (ABI_SPEC): Handle mabi=ilp32e.
20500 * config/riscv/riscv.opt (abi_type): Add ABI_ILP32E.
20501 (RVE): Add RVE mask.
20502 * doc/invoke.texi (RISC-V options) <-mabi>: Add ilp32e info.
20503 <-march>: Add rv32e as an example.
20505 2018-05-18 Marc Glisse <marc.glisse@inria.fr>
20508 * tree-ssa-structalias.c (create_variable_info_for_1): Extra argument.
20509 (intra_create_variable_infos): Handle C++ constructors.
20511 2018-05-18 Martin Liska <mliska@suse.cz>
20513 * passes.def: Remove a redundant pass.
20515 2018-05-18 Eric Botcazou <ebotcazou@adacore.com>
20518 * config/sparc/sparc.c (sparc_expand_builtin): Always initialize op[0].
20520 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20522 * config/arm/arm-cpus.in (armv3m, mode32): Delete features.
20524 (ARMv2, ARMv3, ARMv3m): Delete fgroups.
20526 (armv2, armv2a, armv3, armv3m): Delete architectures.
20527 (arm2, arm250, arm3, arm6, arm60, arm600, arm610, arm620,
20528 arm7, arm7d, arm7di, arm70, arm700, arm700i, arm710, arm720,
20529 arm710c, arm7100, arm7500, arm7500fe, arm7m, arm7dm, arm7dmi):
20531 * config/arm/arm.md (maddsidi4): Remove check for arm_arch3m.
20532 (*mulsidi3adddi): Likewise.
20533 (mulsidi3): Likewise.
20534 (*mulsidi3_nov6): Likewise.
20535 (umulsidi3): Likewise.
20536 (umulsidi3_nov6): Likewise.
20537 (umaddsidi4): Likewise.
20538 (*umulsidi3adddi): Likewise.
20539 (smulsi3_highpart): Likewise.
20540 (*smulsi3_highpart_nov6): Likewise.
20541 (umulsi3_highpart): Likewise.
20542 (*umulsi3_highpart_nov6): Likewise.
20543 * config/arm/arm.h (arm_arch3m): Delete.
20544 * config/arm/arm.c (arm_arch3m): Delete.
20545 (arm_option_override_internal): Update armv3-related comment.
20546 (arm_configure_build_target): Delete use of isa_bit_mode32.
20547 (arm_option_reconfigure_globals): Delete set of arm_ach3m.
20548 (arm_rtx_costs_internal): Delete check of arm_arch3m.
20549 * config/arm/arm-fixed.md (mulsq3): Delete check for arm_arch3m.
20550 (mulsa3): Likewise.
20551 (mulusa3): Likewise.
20552 * config/arm/arm-protos.h (arm_arch3m): Delete.
20553 * config/arm/arm-tables.opt: Regenerate.
20554 * config/arm/arm-tune.md: Likewise.
20555 * config/arm/t-arm-elf (all_early_nofp): Delete mentions of
20556 deleted architectures.
20558 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20560 * config/arm/arm-cpus.in (armv5, armv5e): Delete features.
20561 (armv5t, armv5te): New features.
20562 (ARMv5, ARMv5e): Delete fgroups.
20563 (ARMv5t, ARMv5te): Adjust for above changes.
20564 (ARMv6m): Likewise.
20565 (armv5, armv5e): Delete arches.
20566 * config/arm/arm.md (*call_reg_armv5): Use arm_arch5t instead of
20568 (*call_reg_arm): Likewise.
20569 (*call_value_reg_armv5): Likewise.
20570 (*call_value_reg_arm): Likewise.
20571 (*call_symbol): Likewise.
20572 (*call_value_symbol): Likewise.
20573 (*sibcall_insn): Likewise.
20574 (*sibcall_value_insn): Likewise.
20575 (clzsi2): Likewise.
20576 (prefetch): Likewise.
20577 (define_split and define_peephole2 dependent on arm_arch5):
20579 * config/arm/arm.h (TARGET_LDRD): Use arm_arch5te instead of
20581 (TARGET_ARM_QBIT): Likewise.
20582 (TARGET_DSP_MULTIPLY): Likewise.
20583 (enum base_architecture): Delete BASE_ARCH_5, BASE_ARCH_5E.
20584 (arm_arch5, arm_arch5e): Delete.
20585 (arm_arch5t, arm_arch5te): Declare.
20586 * config/arm/arm.c (arm_arch5, arm_arch5e): Delete.
20587 (arm_arch5t): Declare.
20588 (arm_option_reconfigure_globals): Update for the above.
20589 (arm_options_perform_arch_sanity_checks): Update comment, replace
20590 use of arm_arch5 with arm_arch5t.
20591 (use_return_insn): Likewise.
20592 (arm_emit_call_insn): Likewise.
20593 (output_return_instruction): Likewise.
20594 (arm_final_prescan_insn): Likewise.
20595 (arm_coproc_builtin_available): Likewise.
20596 * config/arm/arm-c.c (arm_cpu_builtins): Replace arm_arch5 and
20597 arm_arch5e with arm_arch5t and arm_arch5te.
20598 * config/arm/arm-protos.h (arm_arch5, arm_arch5e): Delete.
20599 (arm_arch5t, arm_arch5te): Declare.
20600 * config/arm/arm-tables.opt: Regenerate.
20601 * config/arm/t-arm-elf: Remove references to armv5, armv5e.
20602 * config/arm/t-multilib: Likewise.
20603 * config/arm/thumb1.md (*call_reg_thumb1_v5): Check arm_arch5t
20604 instead of arm_arch5.
20605 (*call_reg_thumb1): Likewise.
20606 (*call_value_reg_thumb1_v5): Likewise.
20607 (*call_value_reg_thumb1): Likewise.
20608 * config/arm/vxworks.h (TARGET_OS_CPP_BUILTINS): Remove now
20610 * doc/invoke.texi (ARM Options): Remove references to armv5, armv5e.
20612 2018-05-18 Martin Liska <mliska@suse.cz>
20614 PR gcov-profile/84846
20615 * doc/gcov.texi: Document -t option of gcov tool.
20617 2018-05-18 Martin Liska <mliska@suse.cz>
20619 PR gcov-profile/84846
20620 * gcov.c (print_usage): Add new -t option.
20621 (process_args): Handle the option.
20622 (generate_results): Use stdout as output when requested by
20625 2018-05-18 Martin Liska <mliska@suse.cz>
20627 PR gcov-profile/84846
20628 * coverage.c (coverage_init): Write PWD to .gcno file.
20629 * doc/gcov.texi: Document how working directory is printed.
20630 * gcov-dump.c (dump_gcov_file): Print PWD.
20631 * gcov.c (output_intermediate_file): Likewise.
20632 (read_graph_file): Read PWD string.
20633 (output_lines): Print PWD.
20635 2018-05-18 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
20637 PR middle-end/85817
20638 * ipa-pure-const.c (malloc_candidate_p): Remove the check integer_zerop
20639 for retval and return false if all args to phi are zero.
20641 2018-05-18 Richard Biener <rguenther@suse.de>
20643 * gimple-ssa-evrp.c (class evrp_folder): Add simplify_stmt_using_ranges
20645 (evrp_dom_walker::before_dom_children): Call it.
20647 2018-05-18 Richard Biener <rguenther@suse.de>
20649 * tree-dfa.c (get_ref_base_and_extent): Use range-info to refine
20650 results when processing array refs with variable index.
20652 2018-05-18 Toon Moene <toon@moene.org>
20654 * doc/invoke.texi: Move -floop-unroll-and-jam documentation
20655 directly after that of -floop-interchange. Indicate that both
20656 options are enabled by default when specifying -O3.
20658 2018-05-18 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20660 * config/aarch64/aarch64-simd.md (vec_set<mode>): Use VALL_F16 mode
20661 iterator. Delete separate integer-mode vec_set<mode> expander.
20662 (aarch64_simd_vec_setv2di): Delete.
20663 (vec_setv2di): Delete.
20664 (aarch64_simd_vec_set<mode>): Delete all other patterns with that name.
20665 Use VALL_F16 mode iterator. Add LD1 alternative and use vwcore for
20666 the "w, r" alternative.
20668 2018-05-18 Martin Liska <mliska@suse.cz>
20670 * passes.def: Add pass_lower_switch and pass_lower_switch_O0.
20671 * tree-pass.h (make_pass_lower_switch_O0): New function.
20672 * tree-switch-conversion.c (node_has_low_bound): Remove.
20673 (node_has_high_bound): Likewise.
20674 (node_is_bounded): Likewise.
20675 (class pass_lower_switch): Make it a template type and create
20677 (pass_lower_switch::execute): Add template argument.
20678 (make_pass_lower_switch): New function.
20679 (make_pass_lower_switch_O0): New function.
20680 (do_jump_if_equal): Remove.
20681 (emit_case_nodes): Simplify to just handle all 3 cases and leave
20682 all the hard work to tree optimization passes.
20684 2018-05-18 Martin Liska <mliska@suse.cz>
20686 * dbgcnt.c (limit_low): Renamed from limit.
20687 (limit_high): New variable.
20688 (dbg_cnt_is_enabled): Check for upper limit.
20689 (dbg_cnt): Adjust dumping.
20690 (dbg_cnt_set_limit_by_index): Add new argument for high
20692 (dbg_cnt_set_limit_by_name): Likewise.
20693 (dbg_cnt_process_single_pair): Parse new format.
20694 (dbg_cnt_process_opt): Use strtok.
20695 (dbg_cnt_list_all_counters): Remove 'value' and add
20697 * doc/invoke.texi: Document changes.
20699 2018-05-18 Richard Sandiford <richard.sandiford@linaro.org>
20701 * doc/sourcebuild.texi (scalar_all_fma): Document.
20702 * tree.def (FMA_EXPR): Delete.
20703 * internal-fn.def (FMA, FMS, FNMA, FNMS): New internal functions.
20704 * internal-fn.c (ternary_direct): New macro.
20705 (expand_ternary_optab_fn): Likewise.
20706 (direct_ternary_optab_supported_p): Likewise.
20707 * Makefile.in (build/genmatch.o): Depend on case-fn-macros.h.
20708 * builtins.c (fold_builtin_fma): Delete.
20709 (fold_builtin_3): Don't call it.
20710 * cfgexpand.c (expand_debug_expr): Remove FMA_EXPR handling.
20711 * expr.c (expand_expr_real_2): Likewise.
20712 * fold-const.c (operand_equal_p): Likewise.
20713 (fold_ternary_loc): Likewise.
20714 * gimple-pretty-print.c (dump_ternary_rhs): Likewise.
20715 * gimple.c (DEFTREECODE): Likewise.
20716 * gimplify.c (gimplify_expr): Likewise.
20717 * optabs-tree.c (optab_for_tree_code): Likewise.
20718 * tree-cfg.c (verify_gimple_assign_ternary): Likewise.
20719 * tree-eh.c (operation_could_trap_p): Likewise.
20720 (stmt_could_throw_1_p): Likewise.
20721 * tree-inline.c (estimate_operator_cost): Likewise.
20722 * tree-pretty-print.c (dump_generic_node): Likewise.
20723 (op_code_prio): Likewise.
20724 * tree-ssa-loop-im.c (stmt_cost): Likewise.
20725 * tree-ssa-operands.c (get_expr_operands): Likewise.
20726 * tree.c (commutative_ternary_tree_code, add_expr): Likewise.
20727 * fold-const-call.h (fold_fma): Delete.
20728 * fold-const-call.c (fold_const_call_ssss): Handle CFN_FMS,
20729 CFN_FNMA and CFN_FNMS.
20730 (fold_fma): Delete.
20731 * genmatch.c (combined_fn): New enum.
20732 (commutative_ternary_tree_code): Remove FMA_EXPR handling.
20733 (commutative_op): New function.
20734 (commutate): Use it. Handle more than 2 operands.
20735 (dt_operand::gen_gimple_expr): Use commutative_op.
20736 (parser::parse_expr): Allow :c to be used with non-binary
20737 operators if the commutative operand is known.
20738 * gimple-ssa-backprop.c (backprop::process_builtin_call_use): Handle
20739 CFN_FMS, CFN_FNMA and CFN_FNMS.
20740 (backprop::process_assign_use): Remove FMA_EXPR handling.
20741 * hsa-gen.c (gen_hsa_insns_for_operation_assignment): Likewise.
20742 (gen_hsa_fma): New function.
20743 (gen_hsa_insn_for_internal_fn_call): Use it for IFN_FMA, IFN_FMS,
20744 IFN_FNMA and IFN_FNMS.
20745 * match.pd: Add folds for IFN_FMS, IFN_FNMA and IFN_FNMS.
20746 * gimple-fold.h (follow_all_ssa_edges): Declare.
20747 * gimple-fold.c (follow_all_ssa_edges): New function.
20748 * tree-ssa-math-opts.c (convert_mult_to_fma_1): Use the
20749 gimple_build interface and use follow_all_ssa_edges to fold the result.
20750 (convert_mult_to_fma): Use direct_internal_fn_suppoerted_p
20751 instead of checking for optabs directly.
20752 * config/i386/i386.c (ix86_add_stmt_cost): Recognize FMAs as calls
20753 rather than FMA_EXPRs.
20754 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Create a
20755 call to IFN_FMA instead of an FMA_EXPR.
20757 2018-05-17 Jim Wilson <jimw@sifive.com>
20759 * expr.c (do_tablejump): When converting index to Pmode, if we have a
20760 sign extended promoted subreg, and the range does not have the sign bit
20761 set, then do a sign extend.
20763 * config/riscv/riscv.c (riscv_extend_comparands): In unsigned QImode
20764 test, check for sign extended subreg and/or constant operands, and
20765 do a sign extend in that case.
20767 2018-05-17 Steve Ellcey <sellcey@cavium.com>
20769 * config/aarch64/thunderx2t99.md (thunderx2t99_ls_both): Delete.
20770 (thunderx2t99_multiple): Delete psuedo-units from used cpus.
20772 (thunderx2t99_alu_shift): Remove alu_shift_reg, alus_shift_reg.
20773 Change logics_shift_reg to logics_shift_imm.
20774 (thunderx2t99_fp_loadpair_basic): Delete.
20775 (thunderx2t99_fp_storepair_basic): Delete.
20776 (thunderx2t99_asimd_int): Add neon_sub and neon_sub_q types.
20777 (thunderx2t99_asimd_polynomial): Delete.
20778 (thunderx2t99_asimd_fp_simple): Add neon_fp_mul_s_scalar_q
20779 and neon_fp_mul_d_scalar_q.
20780 (thunderx2t99_asimd_fp_conv): Add *int_to_fp* types.
20781 (thunderx2t99_asimd_misc): Delete neon_dup and neon_dup_q.
20782 (thunderx2t99_asimd_recip_step): Add missing *sqrt* types.
20783 (thunderx2t99_asimd_lut): Add missing tbl types.
20784 (thunderx2t99_asimd_ext): Delete.
20785 (thunderx2t99_asimd_load1_1_mult): Delete.
20786 (thunderx2t99_asimd_load1_2_mult): Delete.
20787 (thunderx2t99_asimd_load1_ldp): New.
20788 (thunderx2t99_asimd_load1): New.
20789 (thunderx2t99_asimd_load2): Add missing *load2* types.
20790 (thunderx2t99_asimd_load3): New.
20791 (thunderx2t99_asimd_load4): New.
20792 (thunderx2t99_asimd_store1_1_mult): Delete.
20793 (thunderx2t99_asimd_store1_2_mult): Delete.
20794 (thunderx2t99_asimd_store2_mult): Delete.
20795 (thunderx2t99_asimd_store2_onelane): Delete.
20796 (thunderx2t99_asimd_store_stp): New.
20797 (thunderx2t99_asimd_store1): New.
20798 (thunderx2t99_asimd_store2): New.
20799 (thunderx2t99_asimd_store3): New.
20800 (thunderx2t99_asimd_store4): New.
20802 2018-05-17 Jerome Lambourg <lambourg@adacore.com>
20804 * config/arm/arm_cmse.h (cmse_nsfptr_create, cmse_is_nsfptr): Remove
20805 #include <stdint.h>. Replace intptr_t with __INTPTR_TYPE__.
20807 2018-05-17 Pat Haugen <pthaugen@us.ibm.com>
20808 Segher Boessenkool <segher@kernel.crashing.org>
20811 * config/rs6000/rs6000.c (rs6000_output_move_128bit): Check dest
20814 2018-05-17 Richard Biener <rguenther@suse.de>
20816 * tree-ssa-dse.c (dse_classify_store): Fix iterator increment
20817 for pruning loop and prune defs feeding only already visited PHIs.
20819 2018-05-17 Richard Biener <rguenther@suse.de>
20821 * tree-ssa-sccvn.c (vn_reference_lookup_3): Improve memset handling.
20823 2018-05-17 Bin Cheng <bin.cheng@arm.com>
20824 Richard Biener <rguenther@suse.de>
20826 PR tree-optimization/85793
20827 * tree-vect-stmts.c (vectorizable_load): Handle 1 element-wise load
20828 for VMAT_ELEMENTWISE.
20830 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
20832 * internal-fn.h (lookup_internal_fn): Declare
20833 * internal-fn.c (lookup_internal_fn): New function.
20834 * gimple.c (gimple_build_call_from_tree): Handle calls to
20835 internal functions.
20836 * gimple-pretty-print.c (dump_gimple_call): Print "." before
20837 internal function names.
20838 * tree-pretty-print.c (dump_generic_node): Likewise.
20839 * tree-ssa-scopedtables.c (expr_hash_elt::print): Likewise.
20841 2018-05-17 Richard Sandiford <richard.sandiford@linaro.org>
20843 * gimple-fold.h (gimple_build): Make the function forms take
20844 combined_fn rather than built_in_function.
20845 (gimple_simplify): Likewise.
20846 * gimple-match-head.c (gimple_simplify): Likewise.
20847 * gimple-fold.c (gimple_build): Likewise.
20848 * tree-vect-loop.c (get_initial_def_for_reduction): Use gimple_build
20849 rather than gimple_build_call_internal.
20850 (get_initial_defs_for_reduction): Likewise.
20851 (vect_create_epilog_for_reduction): Likewise.
20852 (vectorizable_live_operation): Likewise.
20854 2018-05-17 Martin Liska <mliska@suse.cz>
20856 * gimple-ssa-sprintf.c (format_directive): Do not use
20857 space in between 'G_' and '('.
20859 2018-05-17 Jakub Jelinek <jakub@redhat.com>
20862 * config/i386/i386.c (ix86_fold_builtin): Handle masked shifts
20863 even if the mask is not all ones.
20866 * config/i386/i386.c (ix86_fold_builtin): Fold shift builtins by
20868 (ix86_gimple_fold_builtin): Likewise.
20871 * config/i386/i386.c: Include tree-vector-builder.h.
20872 (ix86_vector_shift_count): New function.
20873 (ix86_fold_builtin): Fold shift builtins by scalar count.
20874 (ix86_gimple_fold_builtin): Likewise.
20876 * config/i386/avx512fintrin.h (_mm512_set_epi16, _mm512_set_epi8,
20877 _mm512_setzero): New intrinsics.
20879 2018-05-17 James Greenhalgh <james.greenhalgh@arm.com>
20880 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
20882 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify
20883 code generation for cases where splatting a value is not useful.
20884 * simplify-rtx.c (simplify_ternary_operation): Simplify
20885 vec_merge across a vec_duplicate and a paradoxical subreg forming
20886 a vector mode to a vec_concat.
20888 2018-05-17 Olga Makhotina <olga.makhotina@intel.com>
20890 * config.gcc: Support "goldmont-plus".
20891 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
20893 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
20894 PROCESSOR_GOLDMONT_PLUS.
20895 * config/i386/i386.c (m_GOLDMONT_PLUS): Define.
20896 (processor_target_table): Add "goldmont-plus".
20897 (PTA_GOLDMONT_PLUS): Define.
20898 (ix86_lea_outperforms): Add TARGET_GOLDMONT_PLUS.
20899 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT_PLUS.
20900 (fold_builtin_cpu): Add M_INTEL_GOLDMONT_PLUS.
20901 (fold_builtin_cpu): Add "goldmont-plus".
20902 (ix86_add_stmt_cost): Add TARGET_GOLDMONT_PLUS.
20903 (ix86_option_override_internal): Add "goldmont-plus".
20904 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT_PLUS.
20905 (processor_type): Add PROCESSOR_GOLDMONT_PLUS.
20906 * config/i386/x86-tune.def: Add m_GOLDMONT_PLUS.
20907 * doc/invoke.texi: Add goldmont-plus as x86 -march=/-mtune= CPU type.
20909 2018-05-17 Richard Biener <rguenther@suse.de>
20911 PR tree-optimization/85757
20912 * tree-ssa-dse.c (dse_classify_store): Record a PHI def and
20913 remove defs that only feed that PHI from further processing.
20915 2018-05-16 Jim Wilson <jimw@sifive.com>
20917 * config/riscv/riscv.md (<optab>si3_mask, <optab>si3_mask_1): Prepend
20919 (<optab>di3_mask, <optab>di3_mask_1): Likewise.
20921 2018-05-16 Mark Wielaard <mark@klomp.org>
20923 * dwarf2out.c (count_index_strings): New function.
20924 (output_indirect_strings): Call count_index_strings and generate
20925 header for dwarf_version >= 5.
20927 2018-05-16 Mark Wielaard <mark@klomp.org>
20929 * dwarf2out.c (dwarf_FORM): New function.
20930 (set_indirect_string): Use dwarf_FORM.
20931 (reset_indirect_string): Likewise.
20932 (size_of_die): Likewise.
20933 (value_format): Likewise.
20934 (output_die): Likewise.
20935 (add_skeleton_AT_string): Likewise.
20936 (output_macinfo_op): Likewise.
20937 (index_string): Likewise.
20938 (output_index_string_offset): Likewise.
20939 (output_index_string): Likewise.
20940 (count_index_strings): Likewise.
20942 2018-05-16 Carl Love <cel@us.ibm.com>
20944 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
20945 dcbt and dcbtstt with TH=16 if operands[2] is 0 and Power 8 or newer.
20947 2018-05-16 Martin Jambor <mjambor@suse.cz>
20949 * ipa-prop.c (ipa_free_all_edge_args): Remove.
20950 * ipa-prop.h (ipa_free_all_edge_args): Likewise.
20952 2018-05-16 Wilco Dijkstra <wdijkstr@arm.com>
20954 * config/aarch64/aarch64.md (fma<mode>4): Change into expand pattern.
20955 (fnma<mode>4): Likewise.
20956 (fms<mode>4): Likewise.
20957 (fnms<mode>4): Likewise.
20958 (aarch64_fma<mode>4): Rename insn, reorder accumulator operand.
20959 (aarch64_fnma<mode>4): Likewise.
20960 (aarch64_fms<mode>4): Likewise.
20961 (aarch64_fnms<mode>4): Likewise.
20962 (aarch64_fnmadd<mode>4): Likewise.
20964 2018-05-16 Jason Merrill <jason@redhat.com>
20966 * tree.c (warn_deprecated_use): Return bool. Simplify logic.
20968 2018-05-16 Richard Biener <rguenther@suse.de>
20970 * tree-vectorizer.h (struct stmt_info_for_cost): Add where member.
20971 (dump_stmt_cost): Declare.
20972 (add_stmt_cost): Dump cost we add.
20973 (add_stmt_costs): New function.
20974 (vect_model_simple_cost, vect_model_store_cost, vect_model_load_cost):
20975 No longer exported.
20976 (vect_analyze_stmt): Adjust prototype.
20977 (vectorizable_condition): Likewise.
20978 (vectorizable_live_operation): Likewise.
20979 (vectorizable_reduction): Likewise.
20980 (vectorizable_induction): Likewise.
20981 * tree-vect-loop.c (vect_analyze_loop_operations): Create local
20982 cost vector to pass to vectorizable_ and record afterwards.
20983 (vect_model_reduction_cost): Take cost vector argument and adjust.
20984 (vect_model_induction_cost): Likewise.
20985 (vectorizable_reduction): Likewise.
20986 (vectorizable_induction): Likewise.
20987 (vectorizable_live_operation): Likewise.
20988 * tree-vect-slp.c (vect_create_new_slp_node): Initialize
20989 SLP_TREE_NUMBER_OF_VEC_STMTS.
20990 (vect_analyze_slp_cost_1): Remove.
20991 (vect_analyze_slp_cost): Likewise.
20992 (vect_slp_analyze_node_operations): Take visited args and
20993 a target cost vector. Avoid processing already visited stmt sets.
20994 (vect_slp_analyze_operations): Use a local cost vector to gather
20995 costs and register those of non-discarded instances.
20996 (vect_bb_vectorization_profitable_p): Use add_stmt_costs.
20997 (vect_schedule_slp_instance): Remove copying of
20998 SLP_TREE_NUMBER_OF_VEC_STMTS. Instead assert that it is not
21000 * tree-vect-stmts.c (record_stmt_cost): Remove path directly
21001 adding cost. Record cost entry location.
21002 (vect_prologue_cost_for_slp_op): Function to compute cost of
21003 a constant or invariant generated for SLP vect in the prologue,
21004 split out from vect_analyze_slp_cost_1.
21005 (vect_model_simple_cost): Make static. Adjust for SLP costing.
21006 (vect_model_promotion_demotion_cost): Likewise.
21007 (vect_model_store_cost): Likewise, make static.
21008 (vect_model_load_cost): Likewise.
21009 (vectorizable_bswap): Add cost vector arg and adjust.
21010 (vectorizable_call): Likewise.
21011 (vectorizable_simd_clone_call): Likewise.
21012 (vectorizable_conversion): Likewise.
21013 (vectorizable_assignment): Likewise.
21014 (vectorizable_shift): Likewise.
21015 (vectorizable_operation): Likewise.
21016 (vectorizable_store): Likewise.
21017 (vectorizable_load): Likewise.
21018 (vectorizable_condition): Likewise.
21019 (vectorizable_comparison): Likewise.
21020 (can_vectorize_live_stmts): Likewise.
21021 (vect_analyze_stmt): Likewise.
21022 (vect_transform_stmt): Adjust calls to vectorizable_*.
21023 * tree-vectorizer.c: Include gimple-pretty-print.h.
21024 (dump_stmt_cost): New function.
21026 2018-05-16 Richard Biener <rguenther@suse.de>
21028 * params.def (PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE): New param.
21029 * doc/invoke.texi (dse-max-alias-queries-per-store): Document.
21030 * tree-ssa-dse.c: Include tree-ssa-loop.h.
21031 (check_name): New callback.
21032 (dse_classify_store): Track cycles via a visited bitmap of PHI
21033 defs and simplify handling of in-loop and across loop dead stores
21034 and properly fail for loop-variant refs. Handle byte-tracking with
21035 multiple defs. Use PARAM_DSE_MAX_ALIAS_QUERIES_PER_STORE for
21038 2018-05-16 Richard Sandiford <richard.sandiford@linaro.org>
21040 * tree-vectorizer.h (vect_get_vector_types_for_stmt): Declare.
21041 (vect_get_mask_type_for_stmt): Likewise.
21042 * tree-vect-slp.c (vect_two_operations_perm_ok_p): New function,
21044 (vect_build_slp_tree_1): ...here. Use vect_get_vector_types_for_stmt
21045 to determine the statement's vector type and the vector type that
21046 should be used for calculating nunits. Deal with cases in which
21047 the type has to be deferred.
21048 (vect_slp_analyze_node_operations): Use vect_get_vector_types_for_stmt
21049 and vect_get_mask_type_for_stmt to calculate STMT_VINFO_VECTYPE.
21050 * tree-vect-loop.c (vect_determine_vf_for_stmt_1)
21051 (vect_determine_vf_for_stmt): New functions, split out from...
21052 (vect_determine_vectorization_factor): ...here.
21053 * tree-vect-stmts.c (vect_get_vector_types_for_stmt)
21054 (vect_get_mask_type_for_stmt): New functions, split out from
21055 vect_determine_vectorization_factor.
21057 2018-05-16 Richard Biener <rguenther@suse.de>
21059 * tree-cfg.c (verify_gimple_assign_ternary): Properly
21060 verify the [VEC_]COND_EXPR embedded comparison.
21062 2018-05-15 Martin Sebor <msebor@redhat.com>
21064 PR tree-optimization/85753
21065 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Handle
21066 RECORD_TYPE in addition to ARRAY_TYPE.
21068 2018-05-15 Martin Sebor <msebor@redhat.com>
21070 PR middle-end/85643
21071 * calls.c (get_attr_nonstring_decl): Handle MEM_REF.
21073 2018-05-15 Richard Biener <rguenther@suse.de>
21075 * tree-ssa-dse.c (dse_classify_store): Remove use_stmt parameter,
21076 add by_clobber_p one. Change algorithm to collect all defs
21077 representing uses we need to walk and try reducing them to
21078 a single one before failing.
21079 (dse_dom_walker::dse_optimize_stmt): Adjust.
21081 2018-05-13 Mark Wielaard <mark@klomp.org>
21083 * dwarf2out.c (dwarf_OP): Handle DW_OP_addrx and DW_OP_constx.
21084 (size_of_loc_descr): Likewise.
21085 (output_loc_operands): Likewise.
21086 (output_loc_operands_raw): Likewise.
21087 (dw_addr_op): Use dwarf_OP () for DW_OP_constx and DW_OP_addrx.
21088 (resolve_addr_in_expr): Handle DW_OP_addrx and DW_OP_constx.
21089 (hash_loc_operands): Likewise.
21090 (compare_loc_operands): Likewise.
21092 2018-05-14 Mark Wielaard <mark@klomp.org>
21094 * dwarf2out.c (count_index_addrs): New function.
21095 (dwarf2out_finish): Use count_index_addrs to calculate addrs_length.
21097 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21099 PR tree-optimization/83648
21100 * ipa-pure-const.c (malloc_candidate_p): Allow function with NULL
21101 return value as malloc candidate.
21103 2018-05-15 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
21106 * ipa-pure-const.c (warn_function_malloc): Pass value of known_finite
21107 param as true in call to suggest_attribute.
21109 2018-05-14 Segher Boessenkool <segher@kernel.crashing.org>
21111 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete @itemx for
21112 -mreadonly-in-sdata.
21114 2018-05-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21116 * config/aarch64/aarch64-simd.md (*aarch64_crypto_aese_fused):
21118 (aarch64_crypto_aesd_fused): Likewise.
21120 2018-05-14 Wilco Dijkstra <wdijkstr@arm.com>
21122 * config/aarch64/aarch64.md (mov<mode>): Remove '*' in alternatives.
21123 (movsi_aarch64): Likewise.
21124 (load_pairsi): Likewise.
21125 (load_pairdi): Likewise.
21126 (store_pairsi): Likewise.
21127 (store_pairdi): Likewise.
21128 (load_pairsf): Likewise.
21129 (load_pairdf): Likewise.
21130 (store_pairsf): Likewise.
21131 (store_pairdf): Likewise.
21132 (zero_extend): Likewise.
21133 (trunc): Swap alternatives.
21134 (fcvt_target): Add '?' to prefer w over r.
21136 2018-05-14 Jakub Jelinek <jakub@redhat.com>
21139 * config/i386/i386.md: Disallow non-commutative arithmetics in
21140 last twpeephole for mem {+,-,&,|,^}= x; mem != 0 after cmpelim
21141 optimization. Use COMMUTATIVE_ARITH_P test rather than != MINUS
21142 in the peephole2 before it.
21144 2018-05-14 Sebastian Peryt <sebastian.peryt@intel.com>
21146 * common/config/i386/i386-common.c (OPTION_MASK_ISA_CLDEMOTE_SET,
21147 OPTION_MASK_ISA_CLDEMOTE_UNSET): New defines.
21148 (ix86_handle_option): Handle -mcldemote.
21149 * config.gcc: New header.
21150 * config/i386/cldemoteintrin.h: New file.
21151 * config/i386/cpuid.h (bit_CLDEMOTE): New bit.
21152 * config/i386/driver-i386.c (host_detect_local_cpu): Detect
21154 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21155 OPTION_MASK_ISA_CLDEMOTE.
21156 * config/i386/i386.c (ix86_target_string): Add -mcldemote.
21157 (ix86_valid_target_attribute_inner_p): Ditto.
21158 (enum ix86_builtins): Add IX86_BUILTIN_CLDEMOTE.
21159 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_cldemote.
21160 (ix86_expand_builtin): Expand IX86_BUILTIN_CLDEMOTE.
21161 * config/i386/i386.h (TARGET_CLDEMOTE, TARGET_CLDEMOTE_P): New.
21162 * config/i386/i386.md (UNSPECV_CLDEMOTE): New.
21164 * config/i386/i386.opt: Add -mcldemote.
21165 * config/i386/x86intrin.h: New header.
21166 * doc/invoke.texi: Add -mcldemote.
21168 2018-05-14 Richard Biener <rguenther@suse.de>
21170 * doc/match-and-simplify.texi: Adjust :s documentation.
21172 2018-05-14 Alexander Monakov <amonakov@ispras.ru>
21174 * sort.cc (REORDER_23): Pass the type for the temporaries instead of
21175 intended memcpy size.
21176 (REORDER_45): Likewise.
21178 2018-05-13 Alexander Monakov <amonakov@ispras.ru>
21180 * sort.cc: New file.
21181 * system.h [!CHECKING_P] (qsort): Redirect to gcc_qsort.
21182 * vec.c (qsort_chk): Use gcc_qsort.
21183 * Makefile.in (OBJS-libcommon): Add sort.o.
21184 (build/sort.o): New target. Use it...
21185 (BUILD_RTL): ... here, and...
21186 (build/gencfn-macros): ... here, and...
21187 (build/genmatch): ... here.
21189 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
21190 Chung-Ju Wu <jasonwucj@gmail.com>
21192 * config.gcc (nds32*-*-*): Check that n15 is valid to --with-cpu.
21193 * config/nds32/nds32-graywolf.md: New file.
21194 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_GRAYWOLF.
21195 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n15
21197 * config/nds32/nds32-protos.h: More declarations for n15 pipeline.
21198 * config/nds32/nds32-utils.c: More implementations for n15 pipeline.
21199 * config/nds32/nds32.md (pipeline_model): Add graywolf.
21200 * config/nds32/nds32.opt (mcpu): Support n15 pipeline cpus.
21201 * config/nds32/pipelines.md: Include n15 settings.
21203 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
21204 Chung-Ju Wu <jasonwucj@gmail.com>
21206 * config.gcc (nds32*-*-*): Check that n12/n13 are valid to --with-cpu.
21207 * config/nds32/nds32-n13.md: New file.
21208 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N12 and CPU_N13.
21209 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n12/n13
21211 * config/nds32/nds32-protos.h: More declarations for n12/n13 pipeline.
21212 * config/nds32/nds32.md (pipeline_model): Add n13.
21213 * config/nds32/nds32.opt (mcpu): Support n12/n13 pipeline cpus.
21214 * config/nds32/pipelines.md: Include n13 settings.
21216 2018-05-13 Kito Cheng <kito.cheng@gmail.com>
21217 Chung-Ju Wu <jasonwucj@gmail.com>
21219 * config.gcc (nds32*-*-*): Check that n10/d10 are valid to --with-cpu.
21220 * config/nds32/nds32-n10.md: New file.
21221 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N10.
21222 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n10
21224 * config/nds32/nds32-protos.h: More declarations for n10 pipeline.
21225 * config/nds32/nds32-utils.c: More implementations for n10 pipeline.
21226 * config/nds32/nds32.md (pipeline_model): Add n10.
21227 * config/nds32/nds32.opt (mcpu): Support n10 pipeline cpus.
21228 * config/nds32/pipelines.md: Include n10 settings.
21230 2018-05-13 Monk Chiang <sh.chiang04@gmail.com>
21231 Kito Cheng <kito.cheng@gmail.com>
21232 Chung-Ju Wu <jasonwucj@gmail.com>
21234 * config.gcc (nds32be-*-*): Handle --with-ext-dsp.
21235 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
21236 Add enum values for DSP extension instructions.
21237 * config/nds32/constraints.md (Iu06, IU06, CVp5, CVs5, CVs2, CVhi):
21239 * config/nds32/iterators.md (shifts, shiftrt, sat_plus, all_plus,
21240 sat_minus, all_minus, plus_minus, extend, sumax, sumin, sumin_max):
21241 New code iterators.
21242 (su, zs, uk, opcode, add_rsub, add_sub): New code attributes.
21243 * config/nds32/nds32-dspext.md: New file for DSP implementation.
21244 * config/nds32/nds32-intrinsic.c: Implementation of DSP extension.
21245 * config/nds32/nds32-intrinsic.md: Likewise.
21246 * config/nds32/nds32_intrinsic.h: Likewise.
21247 * config/nds32/nds32-md-auxiliary.c: Likewise.
21248 * config/nds32/nds32-memory-manipulation.c: Consider DSP extension.
21249 * config/nds32/nds32-predicates.c (const_vector_to_hwint): New.
21250 (nds32_valid_CVp5_p, nds32_valid_CVs5_p): New.
21251 (nds32_valid_CVs2_p, nds32_valid_CVhi_p): New.
21252 * config/nds32/nds32-protos.h: New declarations for DSP extension.
21253 * config/nds32/nds32-utils.c (extract_mac_non_acc_rtx): New case
21254 TYPE_DMAC in switch statement.
21255 * config/nds32/nds32.c: New checking and implementation for DSP
21256 extension instructions.
21257 * config/nds32/nds32.h: Likewise.
21258 * config/nds32/nds32.md: Likewise.
21259 * config/nds32/nds32.opt (mhw-abs, mext-dsp): New options.
21260 * config/nds32/predicates.md: Implement new predicates for DSP
21263 2018-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
21265 * config/rs6000/rs6000.md (mov<mode>_softfloat, FMOVE32):
21266 Reformat alternatives and attributes so it is easier to identify
21267 which constraints/attributes go with which instruction.
21268 (mov<mode>_hardfloat32, FMOVE64): Likewise.
21269 (mov<mode>_softfloat32, FMOVE64): Likewise.
21270 (mov<mode>_hardfloat64, FMOVE64): Likewise.
21271 (mov<mode>_softfloat64, FMOVE64): Likewise.
21273 2018-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
21275 * doc/extend.texi (PowerPC Built-in Functions): Rename this
21277 (Basic PowerPC Built-in Functions): The new name of the
21278 subsection previously known as "PowerPC Built-in Functions".
21279 (Basic PowerPC Built-in Functions Available on all Configurations):
21281 (Basic PowerPC Built-in Functions Available on ISA 2.05): Likewise.
21282 (Basic PowerPC Built-in Functions Available on ISA 2.06): Likewise.
21283 (Basic PowerPC Built-in Functions Available on ISA 2.07): Likewise.
21284 (Basic PowerPC Built-in Functions Available on ISA 3.0): Likewise.
21286 2018-05-11 Martin Jambor <mjambor@suse.cz>
21289 * ipa-cp.c (intersect_with_plats): Check that the lattice contains
21292 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
21295 * config/arm/arm-cpus.in (fgroup ARMv6m): Add be8 feature.
21297 2018-05-11 Sebastian Peryt <sebastian.peryt@intel.com>
21299 * common/config/i386/i386-common.c (OPTION_MASK_ISA_WAITPKG_SET,
21300 OPTION_MASK_ISA_WAITPKG_UNSET): New defines.
21301 (ix86_handle_option): Handle -mwaitpkg.
21302 * config.gcc: New header.
21303 * config/i386/cpuid.h (bit_WAITPKG): New bit.
21304 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mwaitpkg.
21305 * config/i386/i386-builtin-types.def ((UINT8, UNSIGNED, UINT64)): New
21307 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21308 OPTION_MASK_ISA_WAITPKG.
21309 * config/i386/i386.c (ix86_target_string): Add -mwaitpkg.
21310 (ix86_option_override_internal): Add PTA_WAITPKG.
21311 (ix86_valid_target_attribute_inner_p): Add -mwaitpkg.
21312 (enum ix86_builtins): Add IX86_BUILTIN_UMONITOR, IX86_BUILTIN_UMWAIT,
21313 IX86_BUILTIN_TPAUSE.
21314 (ix86_init_mmx_sse_builtins): Define __builtin_ia32_umonitor,
21315 __builtin_ia32_umwait and __builtin_ia32_tpause.
21316 (ix86_expand_builtin): Expand IX86_BUILTIN_UMONITOR,
21317 IX86_BUILTIN_UMWAIT, IX86_BUILTIN_TPAUSE.
21318 * config/i386/i386.h (TARGET_WAITPKG, TARGET_WAITPKG_P): New.
21319 * config/i386/i386.md (UNSPECV_UMWAIT, UNSPECV_UMONITOR,
21320 UNSPECV_TPAUSE): New.
21321 (umwait, umwait_rex64, umonitor_<mode>, tpause, tpause_rex64): New.
21322 * config/i386/i386.opt: Add -mwaitpkg.
21323 * config/i386/waitpkgintrin.h: New file.
21324 * config/i386/x86intrin.h: New header.
21325 * doc/invoke.texi: Add -mwaitpkg.
21327 2018-05-11 Richard Earnshaw <rearnsha@arm.com>
21330 * config/arm/arm-cpus.in: Add comment that ARMv6-m and ARMv6S-m are now
21332 (cortex-m0): Use armv6s-m isa.
21333 (cortex-m0plus): Likewise.
21334 (cortex-m1): Likewise.
21335 (cortex-m0.small-multiply): Likewise.
21336 (cortex-m0plus.small-multiply): Likewise.
21337 (cortex-m1.small-multiply): Likewise.
21339 2018-05-11 Allan Sandfeld Jensen <allan.jensen@qt.io>
21340 Jakub Jelinek <jakub@redhat.com>
21342 PR tree-optimization/85692
21343 * tree-ssa-forwprop.c (simplify_vector_constructor): Try two
21344 source permute as well.
21346 2018-05-11 Martin Liska <mliska@suse.cz>
21349 * doc/extend.texi: Document LLVM style format for no_sanitize
21352 2018-05-10 Michael Meissner <meissner@linux.vnet.ibm.com>
21354 * config/rs6000/rs6000.c (mode_supports_dq_form): Rename
21355 mode_supports_vsx_dform_quad to mode_supports_dq_form.
21356 (mode_supports_vsx_dform_quad): Likewise.
21357 (mode_supports_vmx_dform): Move these functions to be next to the
21358 other mode_supports functions.
21359 (mode_supports_dq_form): Likewise.
21360 (quad_address_p): Change calls of mode_supports_vsx_dform_quad to
21361 mode_supports_dq_form.
21362 (reg_offset_addressing_ok_p): Likewise.
21363 (offsettable_ok_by_alignment): Likewise.
21364 (rs6000_legitimate_offset_address_p): Likewise.
21365 (legitimate_lo_sum_address_p): Likewise.
21366 (rs6000_legitimize_address): Likewise.
21367 (rs6000_legitimize_reload_address): Likewise.
21368 (rs6000_secondary_reload_inner): Likewise.
21369 (rs6000_preferred_reload_class): Likewise.
21370 (rs6000_output_move_128bit): Likewise.
21372 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
21374 * config/i386/i386.c (ix86_expand_builtin) <case IX86_BUILTIN_RDPID>:
21375 Generate SImode target register for null target.
21376 <case IX86_BUILTIN_XGETBV>: Ditto.
21377 <case IX86_BUILTIN_XSETBV>: Optimize LSHIFTRT generation.
21378 * config/i386/xsaveintrin.h (_xgetbv): Add missing return.
21380 2018-05-10 Carl Love <cel@us.ibm.com>
21382 * config/rs6000/rs6000.md (prefetch): Generate ISA 2.06 instructions
21383 dcbtt and dcbtstt if operands[2] is 0.
21385 2018-05-10 Uros Bizjak <ubizjak@gmail.com>
21388 * config/i386/sse.md (usadv64qi): New expander.
21390 2018-05-10 Segher Boessenkool <segher@kernel.crashing.org>
21392 * config/rs6000/altivec.md (altivec_vmrghb, altivec_vmrghh,
21393 altivec_vmrghw, altivec_vmrglb, altivec_vmrglh, altivec_vmrglw): Remove
21394 -maltivec=be support.
21395 (vec_widen_umult_even_v16qi, vec_widen_smult_even_v16qi,
21396 vec_widen_umult_even_v8hi, vec_widen_smult_even_v8hi,
21397 vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
21398 vec_widen_umult_odd_v16qi, vec_widen_smult_odd_v16qi,
21399 vec_widen_umult_odd_v8hi, vec_widen_smult_odd_v8hi,
21400 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si, altivec_vpkpx,
21401 altivec_vpks<VI_char>ss, altivec_vpks<VI_char>us,
21402 altivec_vpku<VI_char>us, altivec_vpku<VI_char>um, altivec_vsum2sws,
21403 altivec_vsumsws): Adjust.
21404 (altivec_vspltb *altivec_vspltb_internal, altivec_vsplth,
21405 *altivec_vsplth_internal, altivec_vspltw, *altivec_vspltw_internal,
21406 altivec_vspltsf, *altivec_vspltsf_internal): Remove -maltivec=be
21408 (altivec_vperm_<mode>, altivec_vperm_<mode>_uns,
21409 altivec_vupkhs<VU_char>, altivec_vupkls<VU_char>, altivec_vupkhpx,
21410 altivec_vupklpx, altivec_lvsl, altivec_lvsr): Adjust.
21411 (altivec_lve<VI_char>x): Delete expand.
21412 (*altivec_lve<VI_char>x_internal): Rename to...
21413 (altivec_lve<VI_char>x): ... this.
21414 (altivec_lvxl_<mode>): Delete expand.
21415 (*altivec_lvxl_<mode>_internal): Rename to ...
21416 (altivec_lvxl_<mode>): ... this.
21417 (altivec_stvxl_<mode>): Delete expand.
21418 (*altivec_stvxl_<mode>_internal): Rename to ...
21419 (altivec_stvxl_<mode>): ... this.
21420 (altivec_stve<VI_char>x): Delete expand.
21421 (*altivec_stve<VI_char>x_internal): Rename to ...
21422 (altivec_stve<VI_char>x): ... this.
21423 (doublee<mode>2, unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2,
21424 doubleh<mode>2, unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2,
21425 reduc_plus_scal_<mode>): Adjust.
21426 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Adjust
21428 (rs6000_cpu_cpp_builtins): Adjust.
21429 (altivec_resolve_overloaded_builtin): Remove -maltivec=be support.
21430 * config/rs6000/rs6000-protos.h (altivec_expand_lvx_be,
21431 altivec_expand_stvx_be, altivec_expand_stvex_be): Delete.
21432 * config/rs6000/rs6000.c (rs6000_option_override_internal): Remove
21433 -maltivec=be support.
21434 (rs6000_split_vec_extract_var): Adjust.
21435 (rs6000_split_v4si_init): Adjust.
21436 (swap_selector_for_mode): Delete.
21437 (altivec_expand_lvx_be, altivec_expand_stvx_be,
21438 altivec_expand_stvex_be): Delete.
21439 (altivec_expand_lv_builtin, altivec_expand_stv_builtin): Remove
21440 -maltivec=be support.
21441 (rs6000_gimple_fold_builtin): Ditto.
21442 (rs6000_generate_float2_double_code, rs6000_generate_float2_code):
21444 * config/rs6000/rs6000.h (VECTOR_ELT_ORDER_BIG): Delete.
21445 (TARGET_DIRECT_MOVE_64BIT): Adjust.
21446 * config/rs6000/rs6000.md (split for extendsidi2 for vectors): Adjust.
21447 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Delete.
21448 * config/rs6000/vsx.md (floate<mode>, unsfloatev2di, floato<mode>,
21449 unsfloatov2di, vsignedo_v2df, vsignede_v2df, vunsignedo_v2df,
21450 vunsignede_v2df, vsx_extract_<mode>_p9, *vsx_extract_si,
21451 *vsx_extract_<mode>_p8, *vsx_extract_si_<uns>float_df,
21452 *vsx_extract_si_<uns>float_<mode>, vsx_set_<mode>_p9, vsx_set_v4sf_p9,
21453 *vsx_insert_extract_v4sf_p9, *vsx_insert_extract_v4sf_p9_2, and an
21454 anonymous split): Adjust.
21455 (vsx_mergel_<mode>, vsx_mergeh_<mode>): Remove -maltivec=be support.
21456 (vsx_xxspltd_<mode>, extract4b, insert4b): Adjust.
21458 2018-05-10 Eric Botcazou <ebotcazou@adacore.com>
21460 * configure.ac (gcc_gxx_include_dir_add_sysroot): Set it to 1 only
21461 when --with-gxx-include-dir is also specified.
21462 * configure: Regenerate.
21464 2018-05-09 Jim Wilson <jimw@sifive.com>
21467 * config.gcc (riscv*-*-*): Handle --with-multilib-list.
21468 * config/riscv/t-withmultilib: New.
21469 * config/riscv/withmultilib.h: New.
21470 * doc/install.texi: Document RISC-V --with-multilib-list support.
21472 2018-05-09 Richard Biener <rguenther@suse.de>
21474 * tree-vect-slp.c (vect_bb_slp_scalar_cost): Fill a cost
21476 (vect_bb_vectorization_profitable_p): Adjust. Compute
21477 actual scalar cost using the cost vector and the add_stmt_cost
21480 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
21482 PR rtl-optimization/85645
21483 * config/rs6000/rs6000.c (rs6000_emit_prologue_components): Put a SET
21484 in the REG_CFA_REGISTER note for LR, don't leave it empty.
21486 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
21488 PR rtl-optimization/85645
21489 * shrink-wrap.c (spread_components): Return a boolean saying if
21490 anything was changed.
21491 (try_shrink_wrapping_separate): Iterate spread_components until
21492 nothing changes anymore.
21494 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
21496 PR rtl-optimization/85645
21497 * regrename.c (build_def_use): Also kill the chains that include the
21498 destination of a REG_CFA_REGISTER note.
21500 2018-05-09 Segher Boessenkool <segher@kernel.crashing.org>
21502 PR rtl-optimization/85645
21503 * regcprop.c (copyprop_hardreg_forward_1): Don't propagate into an
21504 insn that has a REG_CFA_REGISTER note.
21506 2018-05-09 Richard Sandiford <richard.sandiford@linaro.org>
21508 * cfgexpand.c (expand_clobber): New function.
21509 (expand_gimple_stmt_1): Use it.
21510 * tree-vect-stmts.c (vect_clobber_variable): New function,
21512 (vectorizable_simd_clone_call): ...here.
21513 (vectorizable_store): Emit a clobber either side of an
21514 IFN_STORE_LANES sequence.
21515 (vectorizable_load): Emit a clobber after an IFN_LOAD_LANES sequence.
21517 2018-05-09 Tom de Vries <tom@codesourcery.com>
21520 * config/nvptx/nvptx.md (define_insn "trap", define_insn "trap_if_true")
21521 (define_insn "trap_if_false"): Add exit after trap.
21523 2018-05-09 Eric Botcazou <ebotcazou@adacore.com>
21525 PR rtl-optimization/85638
21526 * bb-reorder.c: Include common/common-target.h.
21527 (create_forwarder_block): New function extracted from...
21528 (fix_up_crossing_landing_pad): ...here. Rename into...
21529 (dw2_fix_up_crossing_landing_pad): ...this.
21530 (sjlj_fix_up_crossing_landing_pad): New function.
21531 (find_rarely_executed_basic_blocks_and_crossing_edges): In SJLJ mode,
21532 call sjlj_fix_up_crossing_landing_pad if there are incoming EH edges
21533 from both partitions and exit the loop after one iteration.
21535 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21538 * doc/extend.texi (PowerPC Built-in Functions): Rename this
21540 (Basic PowerPC Built-in Functions): The new name of the
21541 subsection previously known as "PowerPC Built-in Functions".
21542 (Basic PowerPC Built-in Functions Available on all Configurations):
21544 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
21546 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
21548 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
21550 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
21553 2018-05-08 Jim Wilson <jimw@sifive.com>
21555 * config/riscv/linux.h (MUSL_ABI_SUFFIX): Delete unnecessary backslash.
21556 (LD_EMUL_SUFFIX): New.
21557 (LINK_SPEC): Use it.
21559 2018-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
21561 * doc/extend.texi (PowerPC Built-in Functions): Rename this
21563 (Basic PowerPC Built-in Functions): The new name of the
21564 subsection previously known as "PowerPC Built-in Functions".
21565 (Basic PowerPC Built-in Functions Available on all Configurations):
21567 (Basic PowerPC Built-in Functions Available on ISA 2.05): New
21569 (Basic PowerPC Built-in Functions Available on ISA 2.06): New
21571 (Basic PowerPC Built-in Functions Available on ISA 2.07): New
21573 (Basic PowerPC Built-in Functions Available on ISA 3.0): New
21576 2018-05-08 Jakub Jelinek <jakub@redhat.com>
21579 * config/i386/i386.md: Add peepholes for mem {+,-,&,|,^}= x; mem != 0
21580 after cmpelim optimization.
21582 2018-05-08 Olga Makhotina <olga.makhotina@intel.com>
21584 * config.gcc: Support "goldmont".
21585 * config/i386/driver-i386.c (host_detect_local_cpu): Detect "goldmont".
21586 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
21587 PROCESSOR_GOLDMONT.
21588 * config/i386/i386.c (m_GOLDMONT): Define.
21589 (processor_target_table): Add "goldmont".
21590 (PTA_GOLDMONT): Define.
21591 (ix86_lea_outperforms): Add TARGET_GOLDMONT.
21592 (get_builtin_code_for_version): Handle PROCESSOR_GOLDMONT.
21593 (fold_builtin_cpu): Add M_INTEL_GOLDMONT.
21594 (fold_builtin_cpu): Add "goldmont".
21595 (ix86_add_stmt_cost): Add TARGET_GOLDMONT.
21596 (ix86_option_override_internal): Add "goldmont".
21597 * config/i386/i386.h (processor_costs): Define TARGET_GOLDMONT.
21598 (processor_type): Add PROCESSOR_GOLDMONT.
21599 * config/i386/i386.md: Add CPU "glm".
21600 * config/i386/glm.md: New file.
21601 * config/i386/x86-tune.def: Add m_GOLDMONT.
21602 * doc/invoke.texi: Add goldmont as x86 -march=/-mtune= CPU type.
21604 2018-05-08 Jakub Jelinek <jakub@redhat.com>
21607 * config/i386/i386.c (ix86_expand_sse2_abs): Handle E_V2DImode and
21609 * config/i386/sse.md (abs<mode>2): Use VI_AVX2 iterator instead of
21610 VI1248_AVX512VL_AVX512BW. Handle V2DImode and V4DImode if not
21611 TARGET_AVX512VL using ix86_expand_sse2_abs. Formatting fixes.
21614 * config/i386/i386.c (ix86_fold_builtin): Handle
21615 IX86_BUILTIN_{,P}MOVMSK{PS,PD,B}{,128,256}.
21618 * config/i386/sse.md (ssequaterinsnmode): New mode attribute.
21619 (*<extract_type>_vinsert<shuffletype><extract_suf>_0): New pattern.
21621 2018-05-08 Richard Earnshaw <rearnsha@arm.com>
21624 * config/arm/parsecpu.awk (check_cpu): Fix operator precedence.
21625 (check_arch): Likewise.
21626 (check_fpu): Return the result rather than printing it.
21627 (end arch): Fix operator precedence.
21628 (end cpu): Likewise.
21629 (END): Print the result from check_fpu.
21631 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
21632 Alan Hayward <alan.hayward@arm.com>
21633 David Sherwood <david.sherwood@arm.com>
21635 * config/aarch64/aarch64-sve.md (*pred_cmp<cmp_op><mode>_combine)
21636 (*pred_cmp<cmp_op><mode>, *fcm<cmp_op><mode>_and_combine)
21637 (*fcmuo<mode>_and_combine, *fcm<cmp_op><mode>_and)
21638 (*fcmuo<mode>_and): New patterns.
21640 2018-05-08 Richard Sandiford <richard.sandiford@linaro.org>
21642 * config/aarch64/iterators.md (UNSPEC_COND_LO, UNSPEC_COND_LS)
21643 (UNSPEC_COND_HI, UNSPEC_COND_HS, UNSPEC_COND_UO): Delete.
21644 (SVE_INT_CMP, SVE_FP_CMP): New code iterators.
21645 (cmp_op, sve_imm_con): New code attributes.
21646 (SVE_COND_INT_CMP, imm_con): Delete.
21647 (cmp_op): Remove above unspecs from int attribute.
21648 * config/aarch64/aarch64-sve.md (*vec_cmp<cmp_op>_<mode>): Rename
21650 (*cmp<cmp_op><mode>): ...this. Use UNSPEC_MERGE_PTRUE instead of
21651 comparison-specific unspecs.
21652 (*vec_cmp<cmp_op>_<mode>_ptest): Rename to...
21653 (*cmp<cmp_op><mode>_ptest): ...this and adjust likewise.
21654 (*vec_cmp<cmp_op>_<mode>_cc): Rename to...
21655 (*cmp<cmp_op><mode>_cc): ...this and adjust likewise.
21656 (*vec_fcm<cmp_op><mode>): Rename to...
21657 (*fcm<cmp_op><mode>): ...this and adjust likewise.
21658 (*vec_fcmuo<mode>): Rename to...
21659 (*fcmuo<mode>): ...this and adjust likewise.
21660 (*pred_fcm<cmp_op><mode>): New pattern.
21661 * config/aarch64/aarch64.c (aarch64_emit_unop, aarch64_emit_binop)
21662 (aarch64_emit_sve_ptrue_op, aarch64_emit_sve_ptrue_op_cc): New
21664 (aarch64_unspec_cond_code): Remove handling of LTU, GTU, LEU, GEU
21666 (aarch64_gen_unspec_cond, aarch64_emit_unspec_cond): Delete.
21667 (aarch64_emit_sve_predicated_cond): New function.
21668 (aarch64_expand_sve_vec_cmp_int): Use aarch64_emit_sve_ptrue_op_cc.
21669 (aarch64_emit_unspec_cond_or): Replace with...
21670 (aarch64_emit_sve_or_conds): ...this new function. Use
21671 aarch64_emit_sve_ptrue_op for the individual comparisons and
21672 aarch64_emit_binop to OR them together.
21673 (aarch64_emit_inverted_unspec_cond): Replace with...
21674 (aarch64_emit_sve_inverted_cond): ...this new function. Use
21675 aarch64_emit_sve_ptrue_op for the comparison and
21676 aarch64_emit_unop to invert the result.
21677 (aarch64_expand_sve_vec_cmp_float): Update after the above
21678 changes. Use aarch64_emit_sve_ptrue_op for native comparisons.
21680 2018-05-07 Nathan Sidwell <nathan@acm.org>
21682 * doc/invoke.texi (C++ Dialect Options): Remove -ffor-scope.
21683 * doc/extend.texi (Deprecated Features): Remove -fno-for-scope
21684 (Backwards Compatibility): Likewise.
21686 2018-05-07 Luis Machado <luis.machado@linaro.org>
21690 2018-05-07 Luis Machado <luis.machado@linaro.org>
21692 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
21693 <prefetch_dynamic_strides>: New const bool field.
21694 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
21695 prefetch_dynamic_strides.
21696 (exynosm1_prefetch_tune): Likewise.
21697 (thunderxt88_prefetch_tune): Likewise.
21698 (thunderx_prefetch_tune): Likewise.
21699 (thunderx2t99_prefetch_tune): Likewise.
21700 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
21702 (aarch64_override_options_internal): Update to set
21703 PARAM_PREFETCH_DYNAMIC_STRIDES.
21704 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
21705 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
21706 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
21707 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
21708 prefetch-dynamic-strides setting.
21710 2018-05-07 Luis Machado <luis.machado@linaro.org>
21712 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
21713 <minimum_stride>: New const int field.
21714 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
21715 minimum_stride field.
21716 (exynosm1_prefetch_tune): Likewise.
21717 (thunderxt88_prefetch_tune): Likewise.
21718 (thunderx_prefetch_tune): Likewise.
21719 (thunderx2t99_prefetch_tune): Likewise.
21720 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
21721 (aarch64_override_options_internal): Update to set
21722 PARAM_PREFETCH_MINIMUM_STRIDE.
21723 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
21724 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
21725 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
21726 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
21727 stride is constant and is below the minimum stride threshold.
21729 2018-05-07 Luis Machado <luis.machado@linaro.org>
21731 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune) <l2_cache_size>: Set
21734 2018-05-07 Luis Machado <luis.machado@linaro.org>
21736 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
21737 <prefetch_dynamic_strides>: New const bool field.
21738 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
21739 prefetch_dynamic_strides.
21740 (exynosm1_prefetch_tune): Likewise.
21741 (thunderxt88_prefetch_tune): Likewise.
21742 (thunderx_prefetch_tune): Likewise.
21743 (thunderx2t99_prefetch_tune): Likewise.
21744 (qdf24xx_prefetch_tune): Likewise. Set prefetch_dynamic_strides
21746 (aarch64_override_options_internal): Update to set
21747 PARAM_PREFETCH_DYNAMIC_STRIDES.
21748 * doc/invoke.texi (prefetch-dynamic-strides): Document new option.
21749 * params.def (PARAM_PREFETCH_DYNAMIC_STRIDES): New.
21750 * params.h (PARAM_PREFETCH_DYNAMIC_STRIDES): Define.
21751 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Account for
21752 prefetch-dynamic-strides setting.
21754 2018-05-07 Luis Machado <luis.machado@linaro.org>
21756 * config/aarch64/aarch64-protos.h (cpu_prefetch_tune)
21757 <minimum_stride>: New const int field.
21758 * config/aarch64/aarch64.c (generic_prefetch_tune): Update to include
21759 minimum_stride field.
21760 (exynosm1_prefetch_tune): Likewise.
21761 (thunderxt88_prefetch_tune): Likewise.
21762 (thunderx_prefetch_tune): Likewise.
21763 (thunderx2t99_prefetch_tune): Likewise.
21764 (qdf24xx_prefetch_tune): Likewise. Set minimum_stride to 2048.
21765 (aarch64_override_options_internal): Update to set
21766 PARAM_PREFETCH_MINIMUM_STRIDE.
21767 * doc/invoke.texi (prefetch-minimum-stride): Document new option.
21768 * params.def (PARAM_PREFETCH_MINIMUM_STRIDE): New.
21769 * params.h (PARAM_PREFETCH_MINIMUM_STRIDE): Define.
21770 * tree-ssa-loop-prefetch.c (should_issue_prefetch_p): Return false if
21771 stride is constant and is below the minimum stride threshold.
21773 2018-05-06 Jakub Jelinek <jakub@redhat.com>
21776 * cfgexpand.c (expand_asm_stmt): Don't create a temporary if
21777 the type is addressable. Don't force op into register if it has
21780 2018-05-05 Roland McGrath <mcgrathr@google.com>
21783 * varasm.c (default_section_type_flags): Set SECTION_NOTYPE for
21784 any section for which we don't know a specific type it should have,
21785 regardless of name. Previously this was done only for the exact
21786 names ".init_array", ".fini_array", and ".preinit_array".
21787 (default_elf_asm_named_section): Add comment about
21788 relationship with default_section_type_flags and SECTION_NOTYPE.
21789 (get_section): Don't consider it a type conflict if one side has
21790 SECTION_NOTYPE and the other doesn't, as long as neither has the
21791 SECTION_BSS et al used in the default_section_type_flags logic.
21793 2018-05-05 Tom de Vries <tom@codesourcery.com>
21796 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_3): Define.
21797 (workaround_barsyncs): New function.
21798 (nvptx_reorg): Use workaround_barsyncs.
21799 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
21800 (define_expand "nvptx_membar_cta"): New define_expand.
21801 (define_insn "*nvptx_membar_cta"): New insn.
21803 2018-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
21805 * brig-builtins.def: Add consts to ptrs etc. in BRIG builtin defs.
21806 To improve optimization opportunities.
21807 * builtin-types.def: The new needed builtin types for the above.
21809 2018-05-04 Richard Biener <rguenther@suse.de>
21811 * bb-reorder.c (sanitize_hot_paths): Release hot_bbs_to_check.
21812 * gimple-ssa-store-merging.c
21813 (imm_store_chain_info::output_merged_store): Remove redundant create,
21814 release split_store vector contents on failure.
21815 * tree-vect-slp.c (vect_schedule_slp_instance): Avoid leaking
21816 scalar stmt vector on cache hit.
21818 2018-05-04 Segher Boessenkool <segher@kernel.crashing.org>
21820 * common/config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
21822 * config.gcc (powerpc-xilinx-eabi*): Remove.
21823 * config/rs6000/predicates.md (easy_fp_constant): Remove Xilinx FP
21825 (fusion_addis_mem_combo_load): Ditto.
21826 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Remove Xilinx
21828 (rs6000_cpu_cpp_builtins): Ditto.
21829 * config/rs6000/rs6000-linux.c
21830 (rs6000_linux_float_exceptions_rounding_supported_p): Ditto.
21831 * config/rs6000/rs6000-opts.h (enum fpu_type_t): Delete.
21832 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove Xilinx FP
21834 (rs6000_setup_reg_addr_masks): Ditto.
21835 (rs6000_init_hard_regno_mode_ok): Ditto.
21836 (rs6000_option_override_internal): Ditto.
21837 (legitimate_lo_sum_address_p): Ditto.
21838 (rs6000_legitimize_address): Ditto.
21839 (rs6000_legitimize_reload_address): Ditto.
21840 (rs6000_legitimate_address_p): Ditto.
21841 (abi_v4_pass_in_fpr): Ditto.
21842 (setup_incoming_varargs): Ditto.
21843 (rs6000_gimplify_va_arg): Ditto.
21844 (rs6000_split_multireg_move): Ditto.
21845 (rs6000_savres_strategy): Ditto.
21846 (rs6000_emit_prologue_components): Ditto.
21847 (rs6000_emit_epilogue_components): Ditto.
21848 (rs6000_emit_prologue): Ditto.
21849 (rs6000_emit_epilogue): Ditto.
21850 (rs6000_elf_file_end): Ditto.
21851 (rs6000_function_value): Ditto.
21852 (rs6000_libcall_value): Ditto.
21853 * config/rs6000/rs6000.h: Ditto.
21854 (TARGET_MINMAX_SF, TARGET_MINMAX_DF): Delete, merge to ...
21855 (TARGET_MINMAX): ... this. New.
21856 (TARGET_SF_FPR, TARGET_DF_FPR, TARGET_SF_INSN, TARGET_DF_INSN): Delete.
21857 * config/rs6000/rs6000.md: Remove Xilinx FP support.
21858 (*movsi_internal1_single): Delete.
21859 * config/rs6000/rs6000.opt (msingle-float, mdouble-float, msimple-fpu,
21860 mfpu=, mxilinx-fpu): Delete.
21861 * config/rs6000/singlefp.h: Delete.
21862 * config/rs6000/sysv4.h: Remove Xilinx FP support.
21863 * config/rs6000/t-rs6000: Ditto.
21864 * config/rs6000/t-xilinx: Delete.
21865 * config/rs6000/titan.md: Adjust for fp_type removal.
21866 * config/rs6000/vsx.md: Remove Xilinx FP support.
21867 (VStype_simple): Delete.
21868 (VSfptype_simple, VSfptype_mul, VSfptype_div, VSfptype_sqrt): Delete.
21869 * config/rs6000/xfpu.h: Delete.
21870 * config/rs6000/xfpu.md: Delete.
21871 * config/rs6000/xilinx.h: Delete.
21872 * config/rs6000/xilinx.opt: Delete.
21873 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove
21874 -msingle-float, -mdouble-float, -msimple-fpu, -mfpu=, and -mxilinx-fpu.
21876 2018-05-04 Tom de Vries <tom@codesourcery.com>
21879 * builtins.c (expand_builtin_goacc_parlevel_id_size): Handle null target
21882 2018-05-04 Richard Biener <rguenther@suse.de>
21884 PR middle-end/85627
21885 * tree-complex.c (update_complex_assignment): We are always in SSA form.
21886 (expand_complex_div_wide): Likewise.
21887 (expand_complex_operations_1): Likewise.
21888 (expand_complex_libcall): Preserve EH info of the original stmt.
21889 (tree_lower_complex): Handle removed blocks.
21890 * tree.c (build_common_builtin_nodes): Do not set ECF_NOTRHOW
21891 on complex multiplication and division libcall builtins.
21893 2018-05-04 Richard Biener <rguenther@suse.de>
21895 PR middle-end/85574
21896 * fold-const.c (negate_expr_p): Restrict negation of operand
21897 zero of a division to when we know that can happen without
21899 (fold_negate_expr_1): Likewise.
21901 2018-05-04 Jakub Jelinek <jakub@redhat.com>
21904 * real.h (real_nextafter): Declare.
21905 * real.c (real_nextafter): New function.
21906 * fold-const-call.c (fold_const_nextafter): New function.
21907 (fold_const_call_sss): Call it for CASE_CFN_NEXTAFTER and
21908 CASE_CFN_NEXTTOWARD.
21909 (fold_const_call_1): For CASE_CFN_NEXTTOWARD call fold_const_call_sss
21910 even when arg1_mode is different from arg0_mode.
21912 2018-05-03 Nathan Sidwell <nathan@acm.org>
21914 * doc/extend.texi (Deprecated Features): Remove
21915 -ffriend-injection.
21916 (Backwards Compatibility): Likewise.
21917 * doc/invoke.texi (C++ Language Options): Likewise.
21918 (C++ Dialect Options): Likewise.
21920 2018-05-03 Jakub Jelinek <jakub@redhat.com>
21923 * config/i386/avx512fintrin.h (_mm512_mullox_epi64,
21924 _mm512_mask_mullox_epi64): New intrinsics.
21926 2018-05-03 Tom de Vries <tom@codesourcery.com>
21929 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
21930 dump files): Add offload-tree.
21932 2018-05-03 Richard Biener <rguenther@suse.de>
21934 PR tree-optimization/85615
21935 * tree-ssa-threadupdate.c (thread_block_1): Only allow exits
21936 to loops not nested in BBs loop father to avoid creating multi-entry
21939 2018-05-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
21941 PR tree-optimization/70291
21942 * tree-complex.c (expand_complex_libcall): Add type, inplace_p
21943 arguments. Change return type to tree. Emit libcall as a new
21944 statement rather than replacing existing one when inplace_p is true.
21945 (expand_complex_multiplication_components): New function.
21946 (expand_complex_multiplication): Expand floating-point complex
21947 multiplication using the above.
21948 (expand_complex_division): Rename inner_type parameter to type.
21949 Update expand_complex_libcall call-site.
21950 (expand_complex_operations_1): Update expand_complex_multiplication
21951 and expand_complex_division call-sites.
21953 2018-05-02 Jakub Jelinek <jakub@redhat.com>
21956 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
21957 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
21958 *<shift_insn><dwi>3_doubleword_mask_1): In condition require that
21959 the highest significant bit of the shift count mask is clear. In
21960 check whether and[sq]i3 is needed verify that all significant bits
21961 of the shift count other than the highest are set.
21963 2018-05-02 Tom de Vries <tom@codesourcery.com>
21966 * builtins.def (DEF_GOACC_BUILTIN_ONLY): Define.
21967 * omp-builtins.def (BUILT_IN_GOACC_PARLEVEL_ID)
21968 (BUILT_IN_GOACC_PARLEVEL_SIZE): New builtin.
21969 * builtins.c (expand_builtin_goacc_parlevel_id_size): New function.
21970 (expand_builtin): Call expand_builtin_goacc_parlevel_id_size.
21971 * doc/extend.texi (Other Builtins): Add __builtin_goacc_parlevel_id and
21972 __builtin_goacc_parlevel_size.
21974 2018-05-02 Richard Biener <rguenther@suse.de>
21976 PR tree-optimization/85597
21977 * tree-vect-stmts.c (vectorizable_operation): For ternary SLP
21978 do not use split vect_get_vec_defs call but call vect_get_slp_defs
21981 2018-05-02 Tom de Vries <tom@codesourcery.com>
21984 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
21985 dump files): Add ltrans-tree.
21987 2018-05-02 Tom de Vries <tom@codesourcery.com>
21990 * doc/sourcebuild.texi (Commands for use in dg-final, Scan optimization
21991 dump files): Add wpa-ipa.
21993 2018-05-02 Segher Boessenkool <segher@kernel.crashing.org>
21995 * config.gcc (powerpc*-*-*): Remove paired.h. Unsupport the
21996 powerpc*-*-linux*paired* target.
21997 * config/rs6000/750cl.h: Delete.
21998 * config/rs6000/paired.h: Delete.
21999 * config/rs6000/paired.md: Delete.
22000 * config/rs6000/predicates.md (easy_vector_constant): Remove paired
22002 * config/rs6000/rs6000-builtin.def: Remove paired float support.
22003 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Update
22004 comment. Remove paired float support.
22005 * config/rs6000/rs6000-modes.def: Remove V2SF and V2SI.
22006 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete
22008 * config/rs6000/rs6000-protos.h (paired_expand_vector_init,
22009 paired_emit_vector_cond_expr, paired_expand_vector_move): Delete
22011 * config/rs6000/rs6000.c: Remove paired float support.
22012 (paired_expand_vector_init, paired_expand_vector_move,
22013 paired_emit_vector_compare, paired_emit_vector_cond_expr,
22014 (paired_expand_lv_builtin, paired_expand_stv_builtin,
22015 paired_expand_builtin, paired_expand_predicate_builtin,
22016 paired_init_builtins): Delete.
22017 * config/rs6000/rs6000.h: Remove paired float support.
22018 * config/rs6000/rs6000.md: Remove paired float support.
22019 (move_from_CR_ov_bit): Delete.
22020 * config/rs6000/rs6000.opt (mpaired): Delete.
22021 * config/rs6000/t-rs6000: Remove paired.md from MD_INCLUDES.
22022 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mpaired.
22024 2018-05-02 Richard Biener <rguenther@suse.de>
22026 PR middle-end/85567
22027 * gimplify.c (gimplify_save_expr): When in SSA form allow
22028 SAVE_EXPRs to compute to SSA vars.
22030 2018-05-02 Jakub Jelinek <jakub@redhat.com>
22033 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask,
22034 *ashl<dwi>3_doubleword_mask_1, *<shift_insn><dwi>3_doubleword_mask,
22035 *<shift_insn><dwi>3_doubleword_mask_1): If and[sq]i3 is needed, don't
22036 clobber operands[2], instead use a new pseudo. Formatting fixes.
22038 2018-05-02 Richard Sandiford <richard.sandiford@linaro.org>
22040 PR tree-optimization/85586
22041 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence): Only
22042 exit early for statements in the same group if the accesses are
22045 2018-05-02 Tom de Vries <tom@codesourcery.com>
22048 * lto-wrapper.c (compile_offload_image): Add "could not find mkoffload"
22051 2018-05-01 Marc Glisse <marc.glisse@inria.fr>
22053 PR tree-optimization/85143
22054 * match.pd (A<B&A<C): Extend to BIT_IOR_EXPR.
22056 2018-05-01 Tom de Vries <tom@codesourcery.com>
22059 * config/nvptx/mkoffload.c (main): Suggest using -B in "offload compiler
22060 not found" error message.
22062 2018-05-01 Tom de Vries <tom@codesourcery.com>
22065 * vec.h (VEC_ORDERED_REMOVE_IF, VEC_ORDERED_REMOVE_IF_FROM_TO): Define.
22066 * vec.c (test_ordered_remove_if): New function.
22067 (vec_c_tests): Call test_ordered_remove_if.
22068 * dwarf2cfi.c (connect_traces): Use VEC_ORDERED_REMOVE_IF_FROM_TO.
22069 * lto-streamer-out.c (prune_offload_funcs): Use VEC_ORDERED_REMOVE_IF.
22070 * tree-vect-patterns.c (vect_pattern_recog_1): Use
22071 VEC_ORDERED_REMOVE_IF.
22073 2018-05-01 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
22075 PR tree-optimization/82665
22076 * vr-values.c (vr_values::extract_range_from_binary_expr): Handle
22077 pointer subtraction where arguments come from a memchr call.
22079 2018-05-01 Jakub Jelinek <jakub@redhat.com>
22081 * configure.ac (LD_AS_NEEDED_OPTION, LD_NO_AS_NEEDED_OPTION): Use
22082 --push-state --as-needed and --pop-state instead of --as-needed and
22083 --no-as-needed if ld supports it.
22084 * configure: Regenerated.
22087 * doc/install.texi2html: Replace _002d with - and _002a with * in
22088 generated html files using sed.
22090 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22093 * gcc-rich-location.c (blank_line_before_p): New function.
22094 (use_new_line): New function.
22095 (gcc_rich_location::add_fixit_insert_formatted): New function.
22096 * gcc-rich-location.h
22097 (gcc_rich_location::add_fixit_insert_formatted): New function.
22099 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22101 * selftest.c (assert_streq): Rename "expected" and "actual" to
22102 "val1" and "val2". Extend NULL-handling to cover both inputs
22103 symmetrically, while still requiring both to be non-NULL for a pass.
22104 * selftest.h (assert_streq): Rename "expected" and "actual" to
22106 (ASSERT_EQ): Likewise.
22107 (ASSERT_EQ_AT): Likewise.
22108 (ASSERT_KNOWN_EQ): Likewise.
22109 (ASSERT_KNOWN_EQ_AT): Likewise.
22110 (ASSERT_NE): Likewise.
22111 (ASSERT_MAYBE_NE): Likewise.
22112 (ASSERT_MAYBE_NE_AT): Likewise.
22113 (ASSERT_STREQ): Likewise. Clarify that both must be non-NULL for
22114 the assertion to pass.
22115 (ASSERT_STREQ_AT): Likewise.
22117 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
22119 * doc/invoke.texi (-fpie, -fPIE): Fix grammar and clarify
22120 interaction with -pie.
22122 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22124 * selftest.h: Fix alphabetization of per-source-file selftest
22127 2018-04-30 Jason Merrill <jason@redhat.com>
22129 PR c++/61982 - dead stores to destroyed objects.
22130 * gimplify.c (gimplify_modify_expr): Simplify complex lvalue on LHS
22133 2018-04-30 Jason Merrill <jason@redhat.com>
22135 * tree.c (build_clobber): New.
22136 * tree.h: Declare it.
22137 * gimplify.c (gimplify_bind_expr, gimplify_target_expr): Use it.
22139 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22141 * diagnostic-show-locus.c (layout::layout): Update for
22142 location_get_source_line returning a char_span.
22143 (struct char_span): Move to input.h.
22144 (struct correction): Update for fields in char_span becoming
22146 (struct source_line): Update for location_get_source_line
22147 returning a char_span.
22148 (layout::print_line): Likewise.
22149 * edit-context.c (edited_file::print_content): Likewise.
22150 (edited_file::print_diff_hunk): Likewise.
22151 (edited_file::print_run_of_changed_lines): Likewise.
22152 (edited_file::get_num_lines): Likewise.
22153 (edited_line::edited_line): Likewise.
22154 * final.c (asm_show_source): Likewise.
22155 * input.c (location_get_source_line): Convert return type
22156 from const char * to char_span, losing the final "line_len"
22158 (dump_location_info): Update for the above.
22159 (get_substring_ranges_for_loc): Likewise. Use a char_span
22160 when handling the literal within the line.
22161 (test_reading_source_line): Update for location_get_source_line
22162 returning a char_span.
22163 * input.h (class char_span): Move here from
22164 diagnostic-show-locus.c, converting from a struct to a class.
22165 Make data members private.
22166 (char_span::operator bool): New.
22167 (char_span::length): New.
22168 (char_span::get_buffer): New.
22169 (char_span::operator[]): New.
22170 (char_span::subspan): Make const.
22171 (char_span::xstrdup): New.
22172 (location_get_source_line): Convert return type from const char *
22173 to char_span, losing the final "line_size" param.
22175 2018-04-30 Jan Hubicka <jh@suse.cz>
22177 * lto-wrapper.c (ltrans_priorities): New static var.
22178 (cmp_priority): New.
22179 (run_gcc): Read priorities and if doing parallel build order
22180 the Makefile by them.
22182 2018-04-30 David Malcolm <dmalcolm@redhat.com>
22184 * input.h (builtins_location_check): Convert to a STATIC_ASSERT.
22186 2018-04-30 Richard Biener <rguenther@suse.de>
22188 * tree-cfg.c (verify_address): Remove base argument, add
22189 flag whether to check TREE_ADDRESSABLE and do that.
22190 (verify_expr): Remove.
22191 (verify_types_in_gimple_reference): Add pieces from verify_expr.
22192 (verify_gimple_assign_single): Likewise.
22193 (verify_gimple_switch): Likewise.
22194 (verify_expr_location_1): Dereference tp once. Add (disabled)
22195 piece from verify_expr.
22196 (verify_gimple_in_cfg): Do not call verify_expr on all ops.
22198 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
22200 * config/arc/linux.h (CLEAR_INSN_CACHE): Define.
22202 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
22204 * config/arc/arc-protos.h (prepare_extend_operands): Remove.
22205 (small_data_pattern): Likewise.
22206 (arc_rewrite_small_data): Likewise.
22207 * config/arc/arc.c (LEGITIMATE_SMALL_DATA_OFFSET_P): Remove.
22208 (LEGITIMATE_SMALL_DATA_ADDRESS_P): Likewise.
22209 (get_symbol_alignment): New function.
22210 (legitimate_small_data_address_p): Likewise.
22211 (legitimate_scaled_address): Update, call
22212 legitimate_small_data_address_p.
22213 (output_sdata): New static variable.
22214 (arc_print_operand): Update how we handle small data operands.
22215 (arc_print_operand_address): Likewise.
22216 (arc_legitimate_address_p): Update, use
22217 legitimate_small_data_address_p.
22218 (arc_rewrite_small_data_p): Remove.
22219 (arc_rewrite_small_data_1): Likewise.
22220 (arc_rewrite_small_data): Likewise.
22221 (small_data_pattern): Likewise.
22222 (compact_sda_memory_operand): Update to use
22223 legitimate_small_data_address_p and get_symbol_alignment.
22224 (prepare_move_operands): Don't rewite sdata pattern.
22225 (prepare_extend_operands): Remove.
22226 * config/arc/arc.md (zero_extendqihi2): Don't rewrite sdata
22228 (zero_extendqisi2): Likewise.
22229 (zero_extendhisi2): Likewise.
22230 (extendqihi2): Likewise.
22231 (extendqisi2): Likewise.
22232 (extendhisi2): Likewise.
22233 (addsi3): Likewise.
22234 (subsi3): Likewise.
22235 (andsi3): Likewise.
22236 * config/arc/constraints.md (Usd): Change it to memory constraint.
22238 2018-04-30 Claudiu Zissulescu <claziss@synopsys.com>
22240 * config/arc/arc.c (arc_split_move): Allow signed 6-bit constants
22241 as source of std instructions.
22242 * config/arc/arc.md (movsi_insn): Update pattern predicate to
22243 allow 6-bit constants as source for store instructions.
22244 (movdi_insn): Update instruction pattern to allow 6-bit constants
22245 as source for store instructions.
22247 2018-04-30 Jonathan Wakely <jwakely@redhat.com>
22249 * doc/invoke.texi (-fdebug-types-section): Fix grammar.
22251 2018-04-30 Nathan Sidwell <nathan@acm.org>
22252 Sandra Loosemore <sandra@codesourcery.com>
22254 * dumpfile.c (dump_open): Allow '-' for stdout.
22255 * doc/invoke.texi (Developer Options): Document dump filename
22256 determination early. Document stdin/stdout selection.
22258 2018-04-30 Andrew Sadek <andrew.sadek.se@gmail.com>
22260 Microblaze Target: PIC data text relative
22262 * config/microblaze/microblaze.opt: add new option -mpic-data-text-rel.
22263 * config/microblaze/microblaze-protos.h (microblaze_constant_address_p):
22265 * config/microblaze/microblaze.h (microblaze_constant_address_p):
22266 CONSTANT_ADDRESS_P definition to microblaze_constant_address_p.
22267 * config/microblaze/microblaze.c (TARGET_PIC_DATA_TEXT_REL):
22268 New addressing mode for data-text relative position indepenedent code.
22269 (microblaze_classify_unspec): add 'UNSPEC_TEXT' case ->
22270 'ADDRESS_SYMBOLIC_TXT_REL'.
22271 (microblaze_classify_address): Add handling for UNSPEC + CONST_INT.
22272 (microblaze_legitimate_pic_operand): Exclude function calls from
22273 pic operands in case of TARGET_PIC_DATA_TEXT_REL option.
22274 (microblaze_legitimize_address): Generate 'UNSPEC_TEXT' for all possible
22276 (microblaze_address_insns): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
22277 (print_operand): Add 'ADDRESS_SYMBOLIC_TXT_REL' case.
22278 (print_operand_address): Add 'ADDRESS_SYMBOLIC_TXT_REL' case + handling
22279 for 'address + offset'.
22280 (microblaze_expand_prologue): Add new function prologue call for
22282 (microblaze_asm_generate_pic_addr_dif_vec): Override new target hook
22283 'TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC' to disable address diff vector
22284 table in case of TARGET_PIC_DATA_TEXT_REL.
22285 (expand_pic_symbol_ref): Add handling for 'UNSPEC_TEXT'.
22286 * config/microblaze/microblaze.md (TARGET_PIC_DATA_TEXT_REL):
22287 Add new macros 'UNSPEC_TEXT',
22288 'UNSPEC_SET_TEXT' + add rule for setting r20 in function prologue
22289 + exclude function calls from 'UNSPEC_PLT' in case of data text
22291 * doc/tm.texi.in (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
22292 new target hook for generating address diff vector tables in case of
22294 * doc/tm.texi : Regenerate.
22295 * stmt.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Append new condition
22296 'targetm.asm_out.generate_pic_addr_diff_vec' to flag_pic in case
22297 of addr diff vector generation.
22298 * target.def (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC): Add
22299 target hook definition.
22300 * targhooks.h, targhooks.c (TARGET_ASM_GENERATE_PIC_ADDR_DIFF_VEC):
22301 Add default function for generate_pic_addr_diff_vec -> flag_pic.
22302 * doc/invoke.texi (Add new pic option): Add new microblaze pic
22303 option for data text relative.
22305 2018-04-30 Richard Biener <rguenther@suse.de>
22307 * tree-chrec.h (evolution_function_is_constant_p): Remove
22309 * tree-cfg.c (tree_node_can_be_shared): Re-order checks.
22311 2018-04-30 Richard Biener <rguenther@suse.de>
22314 * dwarf2out.c (gen_producer_string): Ignore -fchecking[=].
22316 2018-04-30 Richard Biener <rguenther@suse.de>
22318 PR tree-optimization/28364
22319 PR tree-optimization/85275
22320 * tree-ssa-loop-ch.c (ch_base::copy_headers): Stop after
22321 copying first exit test.
22323 2018-04-28 Mark Wielaard <mark@klomp.org>
22325 * dwarf2out.c (dwarf2out_finish): Add .debug_addr table header for
22326 dwarf_version >= 5.
22327 (dwarf_AT): Handle DW_AT_addr_base.
22328 (add_top_level_skeleton_die_attrs): Use dwarf_AT for DW_AT_addr_base.
22330 2018-04-28 Uros Bizjak <ubizjak@gmail.com>
22333 * config/i386/i386.md (*ashl<dwi>3_doubleword_mask): New pattern.
22334 (*ashl<dwi>3_doubleword_mask_1): Ditto.
22335 (*<shift_insn><dwi>3_doubleword_mask): Ditto.
22336 (*<shift_insn><dwi>3_doubleword_mask_1): Ditto.
22338 2018-04-28 Richard Biener <rguenther@suse.de>
22340 * tree-cfg.c (verify_gimple_phi): Take a gphi * argument.
22341 (verify_gimple_in_cfg): Rename visited_stmts to visited_throwing_stmts
22342 to reflect use. Only add interesting stmts.
22344 2018-04-27 Martin Jambor <mjambor@suse.cz>
22347 * ipa-cp.c (find_aggregate_values_for_callers_subset): Make sure
22348 the jump function allows for passing through aggregate values.
22350 2018-04-27 David Malcolm <dmalcolm@redhat.com>
22352 * input.h (in_system_header_at): Convert from macro to inline
22354 (from_macro_expansion_at): Likewise.
22355 (from_macro_definition_at): Likewise.
22357 2018-04-27 Jeff Law <law@redhat.com>
22359 * config.gcc: Mark tile* targets as deprecated/obsolete.
22361 2018-04-27 Richard Biener <rguenther@suse.de>
22363 * config/aarch64/aarch64.c: Simplify ap.__stack advance and
22366 2018-04-27 Richard Biener <rguenther@suse.de>
22368 * tree-cfg.c (verify_expr): Make dead code hit gcc_unreachable.
22370 2018-04-27 Uros Bizjak <ubizjak@gmail.com>
22372 * config/i386/i386.md (*movti_internal): Substitute Ye constraint
22373 with Yd constraint. Set "preferred_for_speed" attribute from
22374 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for alternatives
22375 with Yd constraint.
22376 (*movdi_internal): Ditto.
22377 (movti_interunit splitters): Remove
22378 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC from insn condition.
22379 (movdi_interunit splitters): Ditto.
22380 * config/i386/constraints.md (Ye): Remove.
22381 (Yd): Do not depend on TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC.
22383 2018-04-27 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22386 * config/aarch64/constraints.md (Usg): Limit to 31.
22387 (Usj): Limit to 63.
22389 2018-04-27 Jakub Jelinek <jakub@redhat.com>
22391 PR tree-optimization/85529
22392 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): Add FIRST_BB
22393 argument. Don't call get_nonzero_bits if opcode is ERROR_MARK_NODE,
22394 rhs2 def stmt's bb is dominated by first_bb and it isn't an obvious
22395 zero extension or masking of the MSB bit.
22396 (optimize_range_tests): Add FIRST_BB argument, pass it through
22397 to optimize_range_tests_var_bound.
22398 (maybe_optimize_range_tests, reassociate_bb): Adjust
22399 optimize_range_tests callers.
22401 2018-04-26 Richard Biener <rguenther@suse.de>
22402 Jakub Jelinek <jakub@redhat.com>
22404 * cgraph.h (symbol_table): Just declare debug method here.
22405 * symtab.c (symbol_table::debug): Define.
22407 2018-04-26 Eric Botcazou <ebotcazou@adacore.com>
22409 * loop-invariant.c (may_assign_reg_p): Return false for frame pointer.
22411 2018-04-26 Uros Bizjak <ubizjak@gmail.com>
22413 * config/i386/i386.md ("isa" attribute): Add x64_sse2.
22414 ("enabled" attribute): Handle x64_sse2 "isa" attribute.
22415 (*movdi_internal): Substitute Yi and Yj constraint with x
22416 and Ym and Yn constraint with y constraint. Update "isa"
22417 attribute and set "preferred_for_speed" attribute from
22418 TARGET_INTER_UNIT_MOVES_{FROM,TO}_VEC for updated alternatives.
22419 (*movsi_internal): Ditto.
22420 (*movdf_internal): Ditto.
22421 (*movsf_internal): Ditto.
22422 (*zero_extendsidi2): Ditto.
22423 * config/i386/sse.md (vec_set<mode>_0): Ditto.
22424 (sse2_loadld): Ditto.
22425 (*vec_extract<ssevecmodelower>_0): Ditto.
22426 (*vec_extractv4si_0_zext_sse4): Ditto.
22427 (vec_concatv2di): Ditto.
22428 (*vec_dup<mode>): Ditto.
22429 * config/i386/mmx.md (*mov<mode>_internal): Ditto.
22430 * config/i386/constraints.md (Yi): Remove.
22435 2018-04-26 Nathan Sidwell <nathan@acm.org>
22437 * dumpfile.c (dump_open): New.
22438 (dump_open_alternate_stream, dump_start, dump_begin): Call it.
22439 (dump_finish): Detect stdio/stderr by value not name.
22441 2018-04-26 Jonathan Wakely <jwakely@redhat.com>
22443 * doc/invoke.texi (-Wreturn-type): Document default status for C++.
22445 2018-04-26 Tom de Vries <tom@codesourcery.com>
22448 * config/nvptx/nvptx.c (verify_neutering_jumps)
22449 (verify_neutering_labels): New function
22450 (nvptx_single): Use verify_neutering_jumps and verify_neutering_labels.
22452 2018-04-26 Tom de Vries <tom@codesourcery.com>
22455 * config/nvptx/nvptx.c (needs_neutering_p): New function.
22456 (nvptx_single): Use needs_neutering_p to skip over insns that do not
22459 2018-04-26 Richard Biener <rguenther@suse.de>
22460 Tom de Vries <tom@codesourcery.com>
22463 * lto-streamer-out.c (output_function): Fixup loops if required to match
22464 discovery done in the reader.
22466 2018-04-26 Richard Biener <rguenther@suse.de>
22468 PR tree-optimization/85116
22469 * tree-ssa-loop-ch.c (do_while_loop_p): A do-while loop should
22470 have a loop exit from the single latch predecessor. Remove
22471 case of header with just condition.
22472 (ch_base::copy_headers): Exclude infinite loops from any
22474 (pass_ch::execute): Record exits.
22476 2018-04-26 Richard Biener <rguenther@suse.de>
22478 * tree-vect-data-refs.c (vect_get_data_access_cost): Get
22479 prologue cost vector and pass it to vect_get_load_cost.
22480 (vect_get_peeling_costs_all_drs): Likewise.
22481 (vect_peeling_hash_get_lowest_cost): Likewise.
22482 (vect_enhance_data_refs_alignment): Likewise.
22484 2018-04-26 Richard Biener <rguenther@suse.de>
22486 PR middle-end/85450
22487 * tree-cfg.c (verify_gimple_assign_unary): Restore proper
22488 checking of integer<->pointer conversions.
22489 * omp-expand.c (expand_omp_for_static_nochunk): Avoid
22490 sign-/zero-extending pointer types.
22491 (expand_omp_for_static_chunk): Likewise.
22493 2018-03-22 Hans-Peter Nilsson <hp@axis.com>
22494 Jean Lee <xiaoyur347@gmail.com>
22496 * config/mips/mips.c (mips_asan_shadow_offset): New function.
22497 (TARGET_ASAN_SHADOW_OFFSET): Define.
22498 * config/mips/mips.h (FRAME_GROWS_DOWNWARD): Augment to also be
22499 true for -fsanitize=address.
22501 2018-04-25 Mark Wielaard <mark@klomp.org>
22503 * dwarf2out.c (file_info_cmp): Sort longer dir prefixes before
22506 2018-04-25 Jakub Jelinek <jakub@redhat.com>
22508 * config/i386/i386.md (*x86_mov<mode>cc_0_m1): Use type "alu1" rather
22509 than "alu", remove explicit "memory" and "imm_disp" attributes.
22510 (*x86_mov<mode>cc_0_m1_se, *x86_mov<mode>cc_0_m1_neg): Likewise.
22512 PR middle-end/85414
22513 * simplify-rtx.c (simplify_unary_operation_1) <case SIGN_EXTEND,
22514 case ZERO_EXTEND>: Pass SUBREG_REG (op) rather than op to
22515 gen_lowpart_no_emit.
22517 2018-04-25 Sebastian Peryt <sebastian.peryt@intel.com>
22520 * config/i386/i386.c (ix86_expand_builtin): Change memory
22521 operand to XI, extend p0 to Pmode.
22522 * config/i386/i386.md: Change unspec volatile and operand
22523 1 mode to XI, change operand 0 mode to P.
22525 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
22527 * config/nds32/nds32-predicates.c (nds32_can_use_bclr_p): Mask with
22528 GET_MODE_MASK before any checking.
22529 (nds32_can_use_bset_p): Likewise.
22530 (nds32_can_use_btgl_p): Likewise.
22532 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
22534 * config/nds32/nds32-doubleword.md: New define_split pattern for
22535 illegal register number.
22537 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
22539 * config/nds32/nds32.c (nds32_print_operand): Set op_value ealier.
22541 2018-04-25 Chung-Ju Wu <jasonwucj@gmail.com>
22543 * config/nds32/nds32.h (ASM_APP_ON): Add missing newline character.
22545 2018-04-25 Richard Biener <rguenther@suse.de>
22547 * lto-streamer.h (LTO_major_version): Bump to 8.
22549 2018-04-25 Jakub Jelinek <jakub@redhat.com>
22551 * BASE-VER: Set to 9.0.0.
22553 2018-04-24 Segher Boessenkool <segher@kernel.crashing.org>
22555 * config/rs6000/rs6000.c (init_float128_ieee): Fix spelling mistakes
22556 in __abskf2 and __powikf2.
22558 2018-04-24 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
22561 * config/aarch64/constraints.md (Usg, Usj): New constraints.
22562 * config/aarch64/iterators.md (cmode_simd): New mode attribute.
22563 * config/aarch64/aarch64.md (*aarch64_ashr_sisd_or_int_<mode>3):
22564 Use the above on operand 2. Reindent.
22565 (*aarch64_lshr_sisd_or_int_<mode>3): Likewise.
22567 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
22570 * common/config/i386/i386-common.c (ix86_handle_option): Don't
22572 * config/i386/i386.opt (mcet): Removed.
22573 * doc/install.texi: Remove -mcet documentation.
22574 * doc/invoke.texi: Likewise.
22576 2018-04-24 H.J. Lu <hongjiu.lu@intel.com>
22579 * doc/install.texi: Remove -mcet from bootstrap-cet.
22581 2018-04-24 Jakub Jelinek <jakub@redhat.com>
22584 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Don't define
22585 __builtin_ia32_readeflags_u32 and __builtin_ia32_writeeflags_u32
22589 * config/rs6000/rs6000-p8swap.c (const_load_sequence_p): Punt if
22590 const_vector is not CONST_VECTOR or SYMBOL_REF for a constant pool
22591 containing a CONST_VECTOR.
22593 2018-04-24 Cesar Philippidis <cesar@codesourcery.com>
22595 * doc/install.texi: Update newlib dependency for nvptx.
22597 2018-04-24 Jakub Jelinek <jakub@redhat.com>
22600 * config/i386/i386.c (ix86_expand_vector_init_one_var): Use UINTVAL
22601 instead of INTVAL when shifting x left.
22603 2018-04-24 Andreas Krebbel <krebbel@linux.ibm.com>
22605 PR tree-optimization/85478
22606 * tree-vect-loop.c (vect_analyze_loop_2): Do not call
22607 vect_grouped_store_supported for single element vectors.
22609 2018-04-24 Richard Biener <rguenther@suse.de>
22612 * config/i386/i386.c (ix86_add_stmt_cost): Restrict strided
22613 load cost increase to the case of non-constant step.
22615 2018-04-24 Jakub Jelinek <jakub@redhat.com>
22618 * reg-stack.c (move_for_stack_reg): Don't fail assertion about dead
22619 destination if any_malformed_asm.
22621 2018-04-23 Eric Botcazou <ebotcazou@adacore.com>
22623 PR middle-end/85496
22624 * expr.c (store_field): In the bitfield case, if the value comes from
22625 a function call and is returned in registers by means of a PARALLEL,
22626 do not change the mode of the temporary unless BLKmode and VOIDmode.
22628 2018-04-23 Andrey Belevantsev <abel@ispras.ru>
22630 PR rtl-optimization/85423
22631 * sel-sched-ir.c (has_dependence_note_mem_dep): Only discard
22632 dependencies to debug insns when the previous insn is non-debug.
22634 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
22636 * config/arc/arc.md ("vunspec"): Delete it, unify all the unspec
22637 enums into a single definition.
22638 (fls): Fix predicates and printing.
22641 2018-04-23 Claudiu Zissulescu <claziss@synopsys.com>
22643 * config/arc/arc-protos.h (check_if_valid_sleep_operand): Remove.
22644 * config/arc/arc.c (arc_expand_builtin): Sleep accepts registers
22645 and short u6 immediate.
22646 (check_if_valid_sleep_operand): Remove.
22647 * config/arc/arc.md (Sleep): Accepts registers and u6 immediates.
22649 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22651 * config/nds32/nds32.c (nds32_compute_stack_frame): Consider
22652 flag_always_save_lp condition.
22653 * config/nds32/nds32.opt (malways-save-lp): New option.
22655 2018-04-22 Shiva Chen <shiva0217@gmail.com>
22657 * config/nds32/nds32-protos.h (nds32_use_load_post_increment): Declare.
22658 * config/nds32/nds32.c (nds32_use_load_post_increment): New.
22659 * config/nds32/nds32.h
22660 (USE_LOAD_POST_INCREMENT, USE_LOAD_POST_DECREMENT): Define.
22661 (USE_STORE_POST_INCREMENT, USE_STORE_POST_DECREMENT): Define.
22663 2018-04-22 Shiva Chen <shiva0217@gmail.com>
22665 * config/nds32/nds32-protos.h (nds32_ls_333_p): Remove.
22666 * config/nds32/nds32.c (nds32_ls_333_p): Remove.
22668 2018-04-22 Shiva Chen <shiva0217@gmail.com>
22669 Chung-Ju Wu <jasonwucj@gmail.com>
22671 * config/nds32/nds32-protos.h (nds32_case_vector_shorten_mode):
22673 * config/nds32/nds32.c (nds32_case_vector_shorten_mode): New function.
22674 * config/nds32/nds32.h (CASE_VECTOR_SHORTEN_MODE): Modify.
22676 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22678 * config/nds32/nds32.c (nds32_compute_stack_frame): Fix wrong value.
22680 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22682 * config/nds32/nds32-protos.h (nds32_data_alignment,
22683 nds32_local_alignment): Declare.
22684 * config/nds32/nds32.c (nds32_data_alignment, nds32_constant_alignment,
22685 nds32_local_alignment): New functions.
22686 (TARGET_CONSTANT_ALIGNMENT): Define.
22687 * config/nds32/nds32.h (DATA_ALIGNMENT, LOCAL_ALIGNMENT): Define.
22689 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22691 * config/nds32/nds32.c
22692 (TARGET_HARD_REGNO_MODE_OK): Move to the bottom of file.
22693 (TARGET_MODES_TIEABLE_P): Likewise.
22695 2018-04-22 Chung-Ju Wu <jasonwucj@gmail.com>
22697 * config/nds32/nds32.c (nds32_asm_file_start): Display optimization
22698 level Ofast and Og.
22700 2018-04-22 Monk Chiang <sh.chiang04@gmail.com>
22701 Chung-Ju Wu <jasonwucj@gmail.com>
22703 * config/nds32/constants.md (unspec_volatile_element): Add enum values
22704 for unaligned access.
22705 * config/nds32/nds32-intrinsic.c: Implementation of expanding
22707 * config/nds32/nds32-intrinsic.md: Likewise.
22708 * config/nds32/nds32_intrinsic.h: Likewise.
22709 * config/nds32/nds32.h (nds32_builtins): Likewise.
22710 * config/nds32/nds32.opt (munaligned-access): New option.
22711 * config/nds32/nds32.c (nds32_asm_file_start): Display
22712 flag_unaligned_access status.
22714 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
22716 * config/riscv/elf.h (LINK_SPEC): Pass --no-relax if
22717 -mno-relax is present.
22718 * config/riscv/linux.h (LINK_SPEC): Ditto.
22720 2018-04-20 Martin Sebor <msebor@redhat.com>
22723 * gimple-fold.c (gimple_fold_builtin_strcpy): Suppress -Wrestrict
22725 (gimple_fold_builtin_stxcpy_chk): Same.
22726 * gimple-ssa-warn-restrict.c (check_bounds_or_overlap): Same.
22728 2018-04-20 Michael Meissner <meissner@linux.ibm.com>
22731 * config/rs6000/rs6000.c (init_float128_ieee): Add support to call
22732 __powikf2 when long double is IEEE 128-bit.
22734 2018-04-20 Kito Cheng <kito.cheng@gmail.com>
22736 * config/riscv/riscv.c (riscv_first_stack_step): Round up min
22737 step to make sure stack always aligned.
22739 2018-04-20 Carl Love <cel@us.ibm.com>
22742 * config/rs6000/rs6000-c.c (rs6000_gimple_fold_builtin): Add
22743 size check for arg0.
22745 2018-04-20 Nathan Sidwell <nathan@codesourcery.com>
22746 Tom de Vries <tom@codesourcery.com>
22749 * config/nvptx/nvptx.c (nvptx_emit_forking, nvptx_emit_joining):
22750 Emit insns for calls too.
22751 (nvptx_find_par): Always look for worker-level predecessor insn.
22752 (nvptx_propagate): Add is_call parm, return bool. Copy frame for
22754 (nvptx_vpropagate, nvptx_wpropagate): Adjust.
22755 (nvptx_process_pars): Propagate frames for calls.
22757 2018-04-20 H.J. Lu <hongjiu.lu@intel.com>
22760 * common/config/i386/i386-common.c (OPTION_MASK_ISA_IBT_SET):
22762 (OPTION_MASK_ISA_IBT_UNSET): Likewise.
22763 (ix86_handle_option): Don't handle OPT_mibt.
22764 * config/i386/cet.h: Check __CET__ instead of __IBT__ and
22766 * config/i386/driver-i386.c (host_detect_local_cpu): Remove
22768 * config/i386/i386-c.c (ix86_target_macros_internal): Don't
22769 check OPTION_MASK_ISA_IBT nor flag_cf_protection.
22770 (ix86_target_macros): Define __CET__ with flag_cf_protection
22771 for -fcf-protection.
22772 * config/i386/i386.c (isa2_opts): Remove -mibt.
22773 * config/i386/i386.h (TARGET_IBT): Removed.
22774 (TARGET_IBT_P): Likewise.
22775 (ix86_valid_target_attribute_inner_p): Don't check OPT_mibt.
22776 * config/i386/i386.md (nop_endbr): Don't check TARGET_IBT.
22777 * config/i386/i386.opt (mcet): Update help message.
22778 (mshstk): Likewise.
22780 * doc/invoke.texi: Remove -mibt. Document __CET__. Document
22781 -mcet as an alias for -mshstk.
22783 2018-04-20 Richard Biener <rguenther@suse.de>
22785 PR middle-end/85475
22786 * match.pd ((X * CST) * Y -> (X * Y) * CST): Avoid exponential
22787 complexity by forcing a single use of the multiply operand.
22789 2018-04-20 Martin Jambor <mjambor@suse.cz>
22792 * ipa-cp.c (cgraph_edge_brings_value_p): Move check for self-feeding
22793 recursion dependency to only apply to non-clones.
22795 2018-04-20 Martin Jambor <mjambor@suse.cz>
22798 * ipa-cp.c (create_specialized_node): Check that clones of
22799 self-recursive edges exist during IPA-CP.
22801 2018-04-19 Toon Moene <toon@moene.org>
22803 * doc/invoke.texi: Add -floop-unroll-and-jam to options enabled
22806 2018-04-19 Jakub Jelinek <jakub@redhat.com>
22808 PR tree-optimization/85467
22809 * fold-const.c (fold_ternary_loc) <case BIT_FIELD_REF>: Use
22810 VECTOR_TYPE_P macro. If type is vector type, VIEW_CONVERT_EXPR the
22811 VECTOR_CST element to type.
22813 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
22816 * config/i386/i386.h (STACK_SAVEAREA_MODE): New.
22817 * config/i386/i386.md (builtin_setjmp_setup): Removed.
22818 (builtin_longjmp): Likewise.
22819 (save_stack_nonlocal): New pattern.
22820 (restore_stack_nonlocal): Likewise.
22822 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
22825 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
22826 Replace ASM_OUTPUT_LABEL with fprintf.
22828 2018-04-19 H.J. Lu <hongjiu.lu@intel.com>
22831 * config/i386/cet.c (file_end_indicate_exec_stack_and_cet):
22832 Check flag_cf_protection instead of TARGET_IBT and TARGET_SHSTK.
22833 * config/i386/i386-c.c (ix86_target_macros_internal): Also
22834 define __IBT__ and __SHSTK__ for -fcf-protection.
22835 * config/i386/i386.c (pass_insert_endbranch::gate): Don't check
22837 (ix86_trampoline_init): Likewise.
22838 (x86_output_mi_thunk): Likewise.
22839 (ix86_notrack_prefixed_insn_p): Likewise.
22840 (ix86_option_override_internal): Don't disallow -fcf-protection.
22841 * config/i386/i386.md (rdssp<mode>): Also enable for
22843 (incssp<mode>): Likewise.
22844 (nop_endbr): Likewise.
22845 * config/i386/i386.opt (mcet): Change help message to built-in
22848 (mshstk): Likewise.
22849 * doc/invoke.texi: Remove -mcet, -mibt and -mshstk condition
22850 on -fcf-protection. Change -mcet, -mibt and -mshstk to only
22851 enable CET built-in functions.
22853 2018-04-19 Sebastian Peryt <sebastian.peryt@intel.com>
22855 * common/config/i386/i386-common.c
22856 (OPTION_MASK_ISA_MOVDIRI_SET, OPTION_MASK_ISA_MOVDIR64B_SET,
22857 OPTION_MASK_ISA_MOVDIRI_UNSET,
22858 OPTION_MASK_ISA_MOVDIR64B_UNSET): New defines.
22859 (ix86_handle_option): Handle -mmovdiri and -mmovdir64b.
22860 * config.gcc (movdirintrin.h): New header.
22861 * config/i386/cpuid.h (bit_MOVDIRI,
22862 bit_MOVDIR64B): New bits.
22863 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mmovdiri
22865 * config/i386/i386-builtin-types.def ((VOID, PUNSIGNED, UNSIGNED),
22866 (VOID, PVOID, PCVOID)): New function types.
22867 * config/i386/i386-builtin.def (__builtin_ia32_directstoreu_u32,
22868 __builtin_ia32_directstoreu_u64,
22869 __builtin_ia32_movdir64b): New builtins.
22870 * config/i386/i386-c.c (__MOVDIRI__, __MOVDIR64B__): New.
22871 * config/i386/i386.c (ix86_target_string): Added -mmovdir64b
22873 (ix86_valid_target_attribute_inner_p): Ditto.
22874 (ix86_expand_special_args_builtin): Added VOID_FTYPE_PUNSIGNED_UNSIGNED
22875 and VOID_FTYPE_PUNSIGNED_UNSIGNED.
22876 (ix86_expand_builtin): Expand IX86_BUILTIN_MOVDIR64B.
22877 * config/i386/i386.h (TARGET_MOVDIRI, TARGET_MOVDIRI_P,
22878 TARGET_MOVDIR64B, TARGET_MOVDIR64B_P): New.
22879 * config/i386/i386.md (UNSPECV_MOVDIRI, UNSPECV_MOVDIR64B): New.
22880 (movdiri<mode>, movdir64b_<mode>): New.
22881 * config/i386/i386.opt: Add -mmovdiri and -mmovdir64b.
22882 * config/i386/immintrin.h: Include movdirintrin.h.
22883 * config/i386/movdirintrin.h: New file.
22884 * doc/invoke.texi: Added -mmovdiri and -mmovdir64b.
22886 2018-04-19 Richard Biener <rguenther@suse.de>
22888 PR middle-end/85455
22889 * cfg.c (clear_bb_flags): When loop state says we have
22890 marked irreducible regions also preserve BB_IRREDUCIBLE_LOOP.
22892 2018-04-19 Richard Biener <rguenther@suse.de>
22894 PR tree-optimization/84737
22895 * tree-vect-data-refs.c (vect_copy_ref_info): New function
22896 copying restrict info.
22897 (vect_setup_realignment): Use it.
22898 * tree-vectorizer.h (vect_copy_ref_info): Declare.
22899 * tree-vect-stmts.c (vectorizable_store): Copy ref info from
22900 the first DR to all generated stores.
22901 (vectorizable_load): Likewise for loads.
22903 2018-04-19 Jakub Jelinek <jakub@redhat.com>
22905 PR tree-optimization/85446
22906 * match.pd ((intptr_t) x eq/ne CST to x eq/ne (typeof x) cst): Require
22907 the integral and pointer types to have the same precision.
22909 * doc/install.texi: Document --disable-cet being the default and
22912 2018-04-18 Martin Liska <mliska@suse.cz>
22914 * ipa-devirt.c (odr_subtypes_equivalent_p): Fix GNU coding
22917 2018-04-18 Martin Liska <mliska@suse.cz>
22920 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
22923 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
22924 arguments if they are comparable.
22926 2018-04-18 Martin Liska <mliska@suse.cz>
22929 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
22932 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
22935 2018-04-18 H.J. Lu <hongjiu.lu@intel.com>
22938 * config/i386/i386.c (ix86_expand_split_stack_prologue): Insert
22939 ENDBR after calling __morestack.
22941 2018-04-18 David Malcolm <dmalcolm@redhat.com>
22944 * configure.ac (gcc-driver-name.h): Honor --with-gcc-major-version
22945 by using gcc_base_ver to generate a gcc_driver_version, and use
22946 it when generating GCC_DRIVER_NAME.
22947 * configure: Regenerate.
22949 2018-04-18 Jakub Jelinek <jakub@redhat.com>
22952 * config.gcc: Obsolete powerpc*-*-*spe*.
22954 2018-04-17 Jakub Jelinek <jakub@redhat.com>
22957 * dbxout.c (dbxout_int): Perform negation in unsigned int type.
22958 (stabstr_D): Change type of unum from unsigned int to
22959 unsigned HOST_WIDE_INT. Perform negation in unsigned HOST_WIDE_INT
22962 2018-04-17 Jim Wilson <jimw@sifive.com>
22965 * config/riscv/riscv.c (riscv_compute_frame_info): Add calls to
22966 RISCV_STACK_ALIGN when using outgoing_args_size and pretend_args_size.
22967 Set arg_pointer_offset after using pretend_args_size.
22969 2018-04-17 Jakub Jelinek <jakub@redhat.com>
22971 PR rtl-optimization/85431
22972 * dse.c (record_store): Ignore zero width stores.
22975 * asan.c (handle_builtin_stack_restore): Adjust comment. Emit
22976 __asan_allocas_unpoison call and last_alloca_addr = new_sp before
22977 __builtin_stack_restore rather than after it.
22978 * builtins.c (expand_asan_emit_allocas_unpoison): Pass
22979 arg1 + (virtual_dynamic_stack_rtx - stack_pointer_rtx) as second
22980 argument instead of virtual_dynamic_stack_rtx.
22982 2018-04-17 Kelvin Nilsen <kelvin@gcc.gnu.org>
22984 * config/rs6000/rs6000-protos.h (rs6000_builtin_is_supported_p):
22986 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
22987 Add note to error message to explain internal mapping of overloaded
22988 built-in function name to non-overloaded built-in function name.
22989 * config/rs6000/rs6000.c (rs6000_builtin_is_supported_p): New
22992 2018-04-17 Michael Meissner <meissner@linux.vnet.ibm.com>
22995 * config/rs6000/rs6000.md (pack<mode>): Do not try handle a pack
22996 where the inputs overlap with the output.
22998 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23000 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): Add
23001 (=v, v) alternative and explicit "memory" attribute.
23002 (vec_extract_lo_<mode><mask_name>): Likewise. Also add
23003 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23005 (vec_extract_lo_<mode><mask_name>): Add (=v, v) alternative and use
23006 "sselog1" type instead of "sselog".
23007 (vec_extract_hi_<mode><mask_name>): Use "sselog1" type instead of
23008 "sselog". Remove explicit "memory" attribute.
23009 (vec_extract_lo_v32hi): Add (=v, v) alternative and explicit "memory",
23010 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23012 (vec_extract_hi_v32hi): Merge all alternatives into one, use
23013 "sselog1" type instead of "sselog". Remove explicit "memory"
23015 (vec_extract_hi_v16hi): Merge each pair of alternatives into one,
23016 use "sselog1" type instead of "sselog". Remove explicit "memory"
23018 (vec_extract_lo_v64qi): Add (=v, v) alternative and explicit "memory",
23019 "type", "prefix", "prefix_extra", "length_immediate" and "mode"
23021 (vec_extract_hi_v64qi): Merge all alternatives into one, use
23022 "sselog1" type instead of "sselog". Remove explicit "memory"
23024 (vec_extract_hi_v32qi): Merge each pair of alternatives into one,
23025 use "sselog1" type instead of "sselog". Remove explicit "memory"
23029 * config/i386/i386.md (*ashlqi3_1_slp): Use alu1 type instead of alu.
23031 PR middle-end/85414
23032 * rtlhooks.c (gen_lowpart_if_possible): Don't call gen_lowpart_SUBREG
23035 2018-04-17 Martin Jambor <mjambor@suse.cz>
23038 * ipa-cp.c (create_specialized_node): Call
23039 expand_all_artificial_thunks if necessary.
23041 2018-04-17 Martin Liska <mliska@suse.cz>
23044 * ipa-devirt.c (odr_types_equivalent_p): Remove trailing
23045 in message, remote space in between '_G' and '('.
23047 2018-04-17 Jakub Jelinek <jakub@redhat.com>
23050 * config/i386/sse.md (reduces<mode><mask_scalar_name>,
23051 avx512f_vmcmp<mode>3<round_saeonly_name>,
23052 avx512f_vmcmp<mode>3_mask<round_saeonly_name>,
23053 avx512f_sgetexp<mode><mask_scalar_name><round_saeonly_scalar_name>,
23054 avx512f_rndscale<mode><round_saeonly_name>,
23055 avx512dq_ranges<mode><mask_scalar_name><round_saeonly_scalar_name>,
23056 avx512f_vgetmant<mode><mask_scalar_name><round_saeonly_scalar_name>):
23057 Use %<iptr>2 instead of %2 for -masm=intel.
23058 (avx512f_vcvtss2usi<round_name>, avx512f_vcvtss2usiq<round_name>,
23059 avx512f_vcvttss2usi<round_saeonly_name>,
23060 avx512f_vcvttss2usiq<round_saeonly_name>): Use %k1 instead of %1 for
23062 (avx512f_vcvtsd2usi<round_name>, avx512f_vcvtsd2usiq<round_name>,
23063 avx512f_vcvttsd2usi<round_saeonly_name>,
23064 avx512f_vcvttsd2usiq<round_saeonly_name>, ufloatv2siv2df2<mask_name>):
23065 Use %q1 instead of %1 for -masm=intel.
23066 (avx512f_sfixupimm<mode><sd_maskz_name><round_saeonly_name>,
23067 avx512f_sfixupimm<mode>_mask<round_saeonly_name>): Use %<iptr>3 instead
23068 of %3 for -masm=intel.
23069 (sse2_shufpd_v2df_mask): Fix a typo, change %{6%} to %{%6%} for
23071 (*avx512vl_<code>v2div2qi2_store): Use %w0 instead of %0 for
23073 (*avx512vl_<code><mode>v4qi2_store): Use %k0 instead of %0 for
23075 (avx512vl_<code><mode>v4qi2_mask_store): Use a single pattern with
23076 %k0 and %1 for -masm=intel rather than two patterns, one with %0 and
23078 (*avx512vl_<code><mode>v8qi2_store): Use %q0 instead of %0 for
23080 (avx512vl_<code><mode>v8qi2_mask_store): Use a single pattern with
23081 %q0 and %1 for -masm=intel rather than two patterns, one with %0 and
23082 %g1 and one with %0 and %1.
23083 (avx512er_vmrcp28<mode><round_saeonly_name>,
23084 avx512er_vmrsqrt28<mode><round_saeonly_name>): Use %<iptr>1 instead of
23085 %1 for -masm=intel.
23086 (avx5124fmaddps_4fmaddps_mask, avx5124fmaddps_4fmaddss_mask,
23087 avx5124fmaddps_4fnmaddps_mask, avx5124fmaddps_4fnmaddss_mask,
23088 avx5124vnniw_vp4dpwssd_mask, avx5124vnniw_vp4dpwssds_mask): Swap order
23089 of %0 and %{%4%} for -masm=intel.
23090 (avx5124fmaddps_4fmaddps_maskz, avx5124fmaddps_4fmaddss_maskz,
23091 avx5124fmaddps_4fnmaddps_maskz, avx5124fmaddps_4fnmaddss_maskz,
23092 avx5124vnniw_vp4dpwssd_maskz, avx5124vnniw_vp4dpwssds_maskz): Swap
23093 order of %0 and %{%5%}%{z%} for -masm=intel.
23095 2018-04-17 Jan Hubicka <jh@suse.cz>
23098 * ipa-devirt.c (odr_types_equivalent_p): Handle bit fields.
23100 2018-04-17 Martin Liska <mliska@suse.cz>
23103 * multiple_target.c (create_dispatcher_calls): Set apostrophes
23104 for target_clone error message. Make default implementation
23105 clone to be a local declaration.
23106 (separate_attrs): Add new argument and check for an empty
23108 (expand_target_clones): Handle it.
23109 (ipa_target_clone): Make redirection just for target_clones
23112 2018-04-16 Cesar Philippidis <cesar@codesourcery.com>
23113 Tom de Vries <tom@codesourcery.com>
23115 PR middle-end/84955
23116 * omp-expand.c (expand_oacc_for): Add dummy false branch for
23117 tiled basic blocks without omp continue statements.
23119 2018-04-16 Aaron Sawdey <acsawdey@linux.ibm.com>
23122 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Mark
23123 vec_extract expression as having side effects to make sure it gets
23126 2018-04-16 H.J. Lu <hongjiu.lu@intel.com>
23129 * config/i386/i386.c (get_builtin_code_for_version): Check
23132 2018-04-16 Olga Makhotina <olga.makhotina@intel.com>
23135 * config.gcc: Support "skylake".
23136 * config/i386/i386-c.c (ix86_target_macros_internal): Handle
23138 * config/i386/i386.c (m_SKYLAKE): Define.
23139 (processor_target_table): Add "skylake".
23140 (ix86_option_override_internal): Add "skylake".
23141 (get_builtin_code_for_version): Handle PROCESSOR_SKYLAKE,
23142 PROCESSOR_CANNONLAKE.
23143 (get_builtin_code_for_version): Fix priority for
23144 PROCESSOR_ICELAKE_CLIENT, PROCESSOR_ICELAKE_SERVER,
23145 PROCESSOR_SKYLAKE-AVX512.
23146 * config/i386/i386.h (processor_costs): Define TARGET_SKYLAKE.
23147 (processor_type): Add PROCESSOR_SKYLAKE.
23149 2018-04-16 Paolo Carlini <paolo.carlini@oracle.com>
23150 Jason Merrill <jason@redhat.com>
23153 * convert.c (convert_to_integer_1): Use direct recursion for
23154 enumeral types and types with a precision less than the number
23155 of bits in their mode.
23157 2018-04-16 Julia Koval <julia.koval@intel.com>
23160 * config/i386/x86-tune.def (X86_TUNE_SSE_UNALIGNED_LOAD_OPTIMAL,
23161 X86_TUNE_SSE_UNALIGNED_STORE_OPTIMAL): Add m_SKYLAKE_AVX512
23163 2018-04-14 Segher Boessenkool <segher@kernel.crashing.org>
23166 * config/rs6000/rs6000.opt (mdirect-move): Make deprecated.
23167 * doc/invoke.texi (RS/6000 and PowerPC Options): Remove -mdirect-move
23168 and -mno-direct-move.
23170 2018-04-13 Paul A. Clarke <pc@us.ibm.com>
23173 * config/rs6000/emmintrin.h (_mm_slli_epi{16,32,64}):
23174 Ensure that vec_splat_s32 is only called with 0 <= shift < 16.
23175 Ensure negative shifts result in {0}.
23177 2018-04-13 Vladimir Makarov <vmakarov@redhat.com>
23179 PR rtl-optimization/79916
23180 * config/rs6000/rs6000.c (rs6000_emit_move): Use assigned hard
23181 regs (if any) to define how to gnerate SD moves when LRA is in
23184 2018-04-13 Jakub Jelinek <jakub@redhat.com>
23186 PR rtl-optimization/85393
23187 * except.h (expand_dw2_landing_pad_for_region): Remove declaration.
23188 * except.c (expand_dw2_landing_pad_for_region): Make static.
23189 * bb-reorder.c (fix_up_crossing_landing_pad): In new_bb emit just
23190 a label and unconditional jump to old_bb, rather than
23191 expand_dw2_landing_pad_for_region insn(s) and jump to single_succ
23194 PR rtl-optimization/85376
23195 * simplify-rtx.c (simplify_const_unary_operation): For CLZ and CTZ and
23196 zero op0, if C?Z_DEFINED_VALUE_AT_ZERO is false, return NULL_RTX
23197 instead of a specific value.
23199 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
23200 Bin Cheng <bin.cheng@arm.com>
23202 PR tree-optimization/82965
23203 PR tree-optimization/83991
23204 * cfgloopanal.c (expected_loop_iterations_unbounded): Add
23205 by_profile_only parameter.
23206 * cfgloopmanip.c (scale_loop_profile): Further scale loop's profile
23207 information if the loop was predicted to iterate too many times.
23208 * cfgloop.h (expected_loop_iterations_unbounded): Update prototype
23210 2018-04-13 Jan Hubicka <hubicka@ucw.cz>
23213 * config/i386/i386.c (ix86_can_inline_p): Allow safe transitions for
23216 2018-04-13 Martin Liska <mliska@suse.cz>
23217 Jakub Jelinek <jakub@redhat.com>
23219 PR middle-end/81657
23220 * expr.h (enum block_op_methods): Add BLOCK_OP_NO_LIBCALL_RET.
23221 * expr.c (emit_block_move_hints): Handle BLOCK_OP_NO_LIBCALL_RET.
23222 * builtins.c (expand_builtin_memory_copy_args): Use
23223 BLOCK_OP_NO_LIBCALL_RET method for mempcpy with non-ignored target,
23224 handle dest_addr == pc_rtx.
23226 2018-04-12 Segher Boessenkool <segher@kernel.crashing.org>
23229 * config/rs6000/rs6000.md (fix_trunc<mode>si2): Use legacy code if
23230 asked to not generate direct moves.
23231 (fix_trunc<mode>si2_stfiwx): Similar.
23232 (fix_trunc<mode>si2_internal): Similar.
23234 2018-04-12 Jakub Jelinek <jakub@redhat.com>
23237 * var-tracking.c (add_stores): Handle STRICT_LOW_PART SET_DEST.
23238 * cselib.c (cselib_record_sets): For STRICT_LOW_PART dest,
23239 lookup if dest in some wider mode is known to be const0_rtx and
23240 if so, record permanent equivalence for it to be ZERO_EXTEND of
23241 the narrower mode destination.
23243 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
23245 * lto-streamer-out.c (output_function): Revert 259346.
23246 * omp-expand.c (expand_oacc_for): Likewise.
23248 2018-04-12 Alexander Monakov <amonakov@ispras.ru>
23250 PR rtl-optimization/85354
23251 * sel-sched-ir.c (sel_init_pipelining): Move cfg_cleanup call...
23252 * sel-sched.c (sel_global_init): ... here.
23254 2018-04-12 Eric Botcazou <ebotcazou@adacore.com>
23257 * lto-wrapper.c (debug_objcopy): Open the files in binary mode.
23258 * dwarf2out.c (dwarf2out_early_finish): Do not generate assembly in LTO
23259 mode for PE-COFF targets.
23260 * config/i386/i386-protos.h (i386_pe_asm_lto_start): Declare.
23261 (i386_pe_asm_lto_end): Likewise.
23262 * config/i386/cygming.h (TARGET_ASM_LTO_START): Define.
23263 (TARGET_ASM_LTO_END): Likewise.
23264 * config/i386/winnt.c (saved_debug_info_level): New static variable.
23265 (i386_pe_asm_lto_start): New function.
23266 (i386_pe_asm_lto_end): Likewise.
23268 2018-04-12 Cesar Philippidis <cesar@codesourcery.com>
23269 Richard Biener <rguenther@suse.de>
23271 PR middle-end/84955
23272 * lto-streamer-out.c (output_function): Fix CFG loop state before
23274 * omp-expand.c (expand_oacc_for): Handle calls to internal
23275 functions like regular functions.
23277 2018-04-12 Richard Biener <rguenther@suse.de>
23280 * dwarf2out.c (init_sections_and_labels): Use debug_line_section[_label]
23281 for the early LTO debug to properly generate references to it
23282 during DIE emission. Do not re-use that for the skeleton for
23284 (dwarf2out_early_finish): Likewise.
23286 2018-04-12 Jakub Jelinek <jakub@redhat.com>
23289 * config/i386/sse.md
23290 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name> split,
23291 <mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name> split,
23292 vec_extract_lo_<mode><mask_name> split, vec_extract_lo_v32hi,
23293 vec_extract_lo_v64qi): For non-AVX512VL if input is xmm16+ reg
23294 and output is a reg, avoid creating invalid lowpart subreg, but
23295 instead split into a 512-bit move. Don't split if not AVX512VL,
23296 input is xmm16+ reg and output is a mem.
23297 (vec_extract_lo_<mode><mask_name>, vec_extract_lo_v32hi,
23298 vec_extract_lo_v64qi): Don't require split if not AVX512VL, input is
23299 xmm16+ reg and output is a mem.
23301 2018-04-12 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23303 * config/s390/s390.c (s390_output_indirect_thunk_function): Check
23304 also for flag_dwarf2_cfi_asm.
23306 2018-04-12 Jakub Jelinek <jakub@redhat.com>
23308 PR rtl-optimization/85342
23309 * regcprop.c (copyprop_hardreg_forward_1): Remove replaced array, use
23310 a bool scalar var inside of the loop instead. Don't try to update
23311 recog_data.operand after failed apply_change_group.
23313 2018-04-12 Tom de Vries <tom@codesourcery.com>
23316 * config/nvptx/nvptx.c (flexible_array_member_type_p): New function.
23317 (nvptx_assemble_decl_begin): Add undefined param. Declare undefined
23318 array with flexible array member as array without given dimension.
23319 (nvptx_assemble_undefined_decl): Set nvptx_assemble_decl_begin call
23320 argument for undefined param to true.
23322 2018-04-11 Aaron Sawdey <acsawdey@linux.ibm.com>
23325 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
23326 -mcall- and -mtraceback=. Remove options -mabi=spe and -mabi=no-spe
23327 from PowerPC section.
23328 * config/rs6000/sysv4.opt (mcall-): Improve help text.
23329 * config/rs6000/rs6000.opt (mblock-compare-inline-limit=): Trim
23330 help text that is too long.
23331 * config/rs6000/rs6000.opt (mblock-compare-inline-loop-limit=): Trim
23332 help text that is too long.
23333 * config/rs6000/rs6000.opt (mstring-compare-inline-limit=): Trim
23334 help text that is too long.
23336 2018-04-11 Uros Bizjak <ubizjak@gmail.com>
23338 * config/alpha/alpha.md (stack_probe_internal): Rename
23339 from "probe_stack". Update all callers.
23341 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
23343 PR rtl-optimization/84566
23344 * sched-deps.c (sched_analyze_insn): Check deps->readonly when invoking
23345 sched_macro_fuse_insns.
23347 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
23350 * sched-rgn.c (add_branch_dependences): Move sel_sched_p check here...
23351 (compute_block_dependences): ... from here.
23353 2018-04-11 Jakub Jelinek <jakub@redhat.com>
23355 PR tree-optimization/85331
23356 * vec-perm-indices.h (vec_perm_indices::clamp): Change input type
23357 from int to HOST_WIDE_INT.
23359 2018-04-11 Martin Jambor <mjambor@suse.cz>
23362 * ipa-cp.c (propagate_vals_across_pass_through): Expand comment.
23363 (cgraph_edge_brings_value_p): New parameter dest_val, check if it is
23364 not the same as the source val.
23365 (cgraph_edge_brings_value_p): New parameter.
23366 (gather_edges_for_value): Pass destination value to
23367 cgraph_edge_brings_value_p.
23368 (perhaps_add_new_callers): Likewise.
23369 (get_info_about_necessary_edges): Likewise and exclude values brought
23370 only by self-recursive edges.
23371 (create_specialized_node): Redirect only clones of self-calling edges.
23372 (+self_recursive_pass_through_p): New function.
23373 (find_more_scalar_values_for_callers_subset): Use it.
23374 (find_aggregate_values_for_callers_subset): Likewise.
23375 (known_aggs_to_agg_replacement_list): Removed.
23376 (decide_whether_version_node): Re-calculate known constants for all
23377 remaining context clones.
23379 2018-04-11 Richard Biener <rguenther@suse.de>
23382 * dwarf2out.c (dwarf2out_finish): Remove DW_AT_stmt_list attribute
23383 from early DWARF output.
23384 (dwarf2out_early_finish): Output line info unconditionally into
23385 early DWARF and add reference to it.
23387 2018-04-11 Jakub Jelinek <jakub@redhat.com>
23390 * config/i386/sse.md (iptr): Add V16SFmode and V8DFmode cases.
23391 (<avx512>_vec_dup<mode><mask_name>): Use a single pattern for modes
23392 other than V2DFmode using iptr mode attribute.
23393 (<avx512>_vec_dup<mode><mask_name>): Use iptr mode attribute.
23395 2018-04-11 Alexander Monakov <amonakov@ispras.ru>
23397 PR rtl-optimization/84659
23398 * sel-sched-ir.c (sel_init_pipelining): Invoke cleanup_cfg.
23400 2018-04-11 Jakub Jelinek <jakub@redhat.com>
23403 * dwarf2out.c (skip_loc_list_entry): Don't call size_of_locs if
23405 (output_loc_list): Pass address of a dummy size variable even in the
23406 locview handling loop.
23407 (index_location_lists): Add comment on why skip_loc_list_entry can't
23410 2018-04-11 Thomas Preud'homme <thomas.preudhomme@arm.com>
23413 * config/arm/arm-builtins.c (arm_expand_builtin): Force input operand
23416 2018-04-10 Aaron Sawdey <acsawdey@linux.ibm.com>
23419 * doc/invoke.texi (RS/6000 and PowerPC Options): Document options
23420 -mblock-compare-inline-limit, -mblock-compare-inline-loop-limit,
23421 and -mstring-compare-inline-limit.
23423 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
23426 * config/rs6000/rs6000.md (allocate_stack): Put the residual size
23427 for stack clash protection in a register whenever we need it to be in
23430 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
23432 * common/config/rs6000/rs6000-common.c (rs6000_option_init_struct):
23433 Enable -fasynchronous-unwind-tables by default if OBJECT_FORMAT_ELF.
23435 2018-04-10 Segher Boessenkool <segher@kernel.crashing.org>
23438 * config/rs6000/rs6000.opt (mtraceback=): Show the allowed values in
23440 (mlong-double-): Ditto.
23441 * config/rs6000/sysv4.opt (msdata=): Ditto.
23442 (mtls-size=): Ditto.
23444 2018-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
23446 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
23447 erroneous entries for
23448 "vector int vec_ldl (int, long int *)", and
23449 "vector unsigned int vec_ldl (int, unsigned long int *)".
23450 Add comments and entries for
23451 "vector bool char vec_ldl (int, bool char *)",
23452 "vector bool short vec_ldl (int, bool short *)",
23453 "vector bool int vec_ldl (int, bool int *)",
23454 "vector bool long long vec_ldl (int, bool long long *)",
23455 "vector pixel vec_ldl (int, pixel *)",
23456 "vector long long vec_ldl (int, long long *)",
23457 "vector unsigned long long vec_ldl (int, unsigned long long *)".
23458 * config/rs6000/rs6000.c (rs6000_init_builtins): Initialize new
23459 type tree bool_long_long_type_node and correct definition of
23460 bool_V2DI_type_node to make reference to this new type tree.
23461 (rs6000_mangle_type): Replace erroneous reference to
23462 bool_long_type_node with bool_long_long_type_node.
23463 * config/rs6000/rs6000.h (enum rs6000_builtin_type_index): Add
23464 comments to emphasize sign distinctions for char and int types and
23465 replace RS6000_BTI_bool_long constant with
23466 RS6000_BTI_bool_long_long constant. Also add comment to restrict
23467 use of RS6000_BTI_pixel.
23468 (bool_long_type_node): Remove this macro definition.
23469 (bool_long_long_type_node): New macro definition
23471 2018-04-10 Jakub Jelinek <jakub@redhat.com>
23473 PR rtl-optimization/85300
23474 * combine.c (subst): Handle subst of CONST_SCALAR_INT_P new_rtx also
23475 into FLOAT and UNSIGNED_FLOAT like ZERO_EXTEND, return a CLOBBER if
23476 simplify_unary_operation fails.
23478 2018-04-10 Martin Liska <mliska@suse.cz>
23480 * gdbhooks.py: Add pretty-printers for varpool_node, symtab_node,
23481 cgraph_edge and ipa_ref.
23483 2018-04-10 Jakub Jelinek <jakub@redhat.com>
23487 * config/i386/sse.md
23488 (<extract_type>_vinsert<shuffletype><extract_suf>_mask): Fix
23489 computation of the VEC_MERGE selector from mask.
23490 (<extract_type>_vinsert<shuffletype><extract_suf>_1<mask_name>):
23491 Fix decoding of the VEC_MERGE selector into mask.
23493 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
23495 PR tree-optimization/85286
23496 * tree-vect-data-refs.c (vect_get_smallest_scalar_type):
23498 2018-04-10 Richard Sandiford <richard.sandiford@linaro.org>
23500 * final.c (final_1): Set insn_last_address as well as
23501 insn_current_address.
23503 2018-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
23506 * explow.c (emit_stack_probe): Call validize_mem on memory location
23507 before passing it to gen_probe_stack. Create address operand and
23508 legitimize it for the probe_stack_address case.
23510 2018-04-09 Jan Hubicka <jh@suse.cz>
23513 * ipa-devirt.c (rebuild_type_inheritance-hash): New.
23514 * ipa-utils.h (rebuild_type_inheritance-hash): Declare.
23515 * tree.c (free_lang_data_in_type): Fix handling of binfos;
23517 (free_lang_data): Rebuild type inheritance graph.
23519 2018-04-09 Martin Sebor <msebor@redhat.com>
23521 * invoke.texi (-finline-small-functions): Mention other optimization
23523 (-findirect-inlining, -fpartial-inlining): Same.
23524 (-finline-functions-called-once): Same.
23525 (-freorder-blocks-and-partition): Same.
23527 2018-04-09 Jan Hubicka <jh@suse.cz>
23530 * cfgcleanup.c (try_forward_edges): Do not give up on crossing
23531 jumps; choose last target that matches the criteria (i.e.
23532 no partition changes for non-crossing jumps).
23533 * cfgrtl.c (cfg_layout_redirect_edge_and_branch): Add basic
23534 support for redirecting crossing jumps to non-crossing.
23536 2018-04-09 Alexey Brodkin <abrodkin@synopsys.com>
23538 * config/arc/arc.c (arc_expand_prologue): Set stack usage info
23539 also for naked functions.
23541 2018-04-09 Claudiu Zissulescu <claziss@synopsys.com>
23543 * config/arc/arc.md (add_shift): New pattern.
23544 (add_shift2): Likewise.
23545 (sub_shift): Likewise.
23546 (sub_shift_cmp0_noout): Likewise.
23547 (compare_si_ashiftsi): Likewise.
23548 (xbfu_cmp0_noout): New combine pattern.
23549 (xbfu_cmp0"): Likewise.
23550 (movsi_set_cc_insn): Place the predicable variant first.
23551 (commutative_binary_cmp0_noout): Remove clobber.
23552 (commutative_binary_cmp0): New pattern.
23553 (noncommutative_binary_cmp0): Likewise.
23554 (noncommutative_binary_cmp0_noout): Likewise.
23555 (noncommutative_binary_comparison_result_used): Removed.
23556 (rsub_cmp0): New pattern.
23557 (rsub_cmp0_noout): Likewise.
23558 (extzvsi): Changed, keep only meaningful variants.
23559 (SQH, SEZ): New iterators.
23560 (SQH_postfix): New mode attribute.
23561 (SEZ_prefix): New code attribute.
23562 (<SEZ_prefix>xt<SQH_postfix>_cmp0_noout): New instruction pattern.
23563 (<SEZ_prefix>xt<SQH_postfix>_cmp0): Likewise.
23564 * config/arc/predicates.md (cc_set_register): Use CC_REG instead
23565 of numerical value.
23566 (noncommutative_operator): Check the availability of barrel
23569 2018-04-09 Richard Biener <rguenther@suse.de>
23571 PR tree-optimization/85284
23572 * tree-ssa-loop-niter.c (number_of_iterations_exit_assumptions):
23573 Only use the niter constraining form of simple_iv when the exit
23574 is always executed.
23576 2018-04-09 Tom de Vries <tom@codesourcery.com>
23579 * config/nvptx/nvptx.md (define_c_enum "unspecv"): Add UNSPECV_MEMBAR.
23580 (define_expand "*memory_barrier"): New define_expand.
23581 (define_insn "memory_barrier"): New insn.
23583 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
23585 PR rtl-optimization/80463
23586 PR rtl-optimization/83972
23587 PR rtl-optimization/83480
23589 * sel-sched-ir.c (has_dependence_note_mem_dep): Take into account the
23590 correct producer for the insn.
23591 (tidy_control_flow): Fixup seqnos in case of debug insns.
23593 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
23595 PR rtl-optimization/83913
23597 * sel-sched-ir.c (merge_expr_data): Choose the middle between two
23598 different sched-times when merging exprs.
23600 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
23602 PR rtl-optimization/83962
23604 * sel-sched-ir.c (tidy_control_flow): Correct the order in which we call
23605 tidy_fallthru_edge and tidy_control_flow.
23607 2018-04-09 Andrey Belevantsev <abel@ispras.ru>
23609 PR rtl-optimization/83530
23611 * sel-sched.c (force_next_insn): New global variable.
23612 (remove_insn_for_debug): When force_next_insn is true, also leave only
23613 next insn in the ready list.
23614 (sel_sched_region): When the region wasn't scheduled, make another pass
23615 over it with force_next_insn set to 1.
23617 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
23619 * config.gcc (nds32le-*-*, nds32be-*-*): Add nds32/nds32_intrinsic.h
23621 * config/nds32/constants.md (unspec_volatile_element): Add enum values
23622 for interrupt control.
23623 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
23624 functions for interrupt control.
23625 * config/nds32/nds32-intrinsic.md: Likewise.
23626 * config/nds32/nds32_intrinsic.h: Likewise.
23627 * config/nds32/nds32.h (nds32_builtins): Likewise.
23629 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
23631 * config/nds32/nds32.c (nds32_init_machine_status,
23632 nds32_legitimate_index_p, nds32_legitimate_address_p): Consider
23633 strict_aligned_p field.
23634 (nds32_expand_to_rtl_hook): New function.
23635 (TARGET_EXPAND_TO_RTL_HOOK): Define.
23636 * config/nds32/nds32.h (machine_function): Add strict_aligned_p field.
23638 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
23639 Chung-Ju Wu <jasonwucj@gmail.com>
23641 * config.gcc (nds32*-*-*): Check that n7 is valid to --with-cpu.
23642 * config/nds32/nds32-n7.md: New file.
23643 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N7.
23644 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n7
23646 * config/nds32/nds32-protos.h: More declarations for n7 pipeline.
23647 * config/nds32/nds32.md (pipeline_model): Add n7.
23648 * config/nds32/nds32.opt (mcpu): Support n7 pipeline cpus.
23649 * config/nds32/pipelines.md: Include n7 settings.
23651 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
23652 Chung-Ju Wu <jasonwucj@gmail.com>
23654 * config.gcc (nds32*-*-*): Check that e8 is valid to --with-cpu.
23655 * config/nds32/nds32-e8.md: New file.
23656 * config/nds32/nds32-opts.h (nds32-cpu_type): Add CPU_E8.
23657 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for e8
23659 * config/nds32/nds32-protos.h: More declarations for e8 pipeline.
23660 * config/nds32/nds32.md (pipeline_model): Add e8.
23661 * config/nds32/nds32.opt (mcpu): Support e8 pipeline cpus.
23662 * config/nds32/pipelines.md: Include e8 settings.
23664 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
23665 Chung-Ju Wu <jasonwucj@gmail.com>
23667 * config.gcc (nds32*-*-*): Check that n6/n8/s8 are valid to --with-cpu.
23668 * config/nds32/nds32-n8.md: New file.
23669 * config/nds32/nds32-opts.h (nds32_cpu_type): Add CPU_N6 and CPU_N8.
23670 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n8
23672 * config/nds32/nds32-protos.h: More declarations for n8 pipeline.
23673 * config/nds32/nds32-utils.c: More implementations for n8 pipeline.
23674 * config/nds32/nds32.md (pipeline_model): Add n8.
23675 * config/nds32/nds32.opt (mcpu): Support n8 pipeline cpus.
23676 * config/nds32/pipelines.md: Include n8 settings.
23678 2018-04-08 Kito Cheng <kito.cheng@gmail.com>
23679 Chung-Ju Wu <jasonwucj@gmail.com>
23681 * config.gcc (nds32*): Add nds32-utils.o into extra_objs.
23682 * config/nds32/nds32-n9-2r1w.md: New file.
23683 * config/nds32/nds32-n9-3r2w.md: New file.
23684 * config/nds32/nds32-opts.h (nds32_cpu_type, nds32_mul_type,
23685 nds32_register_ports): New or modify for cpu n9.
23686 * config/nds32/nds32-pipelines-auxiliary.c: Implementation for n9
23688 * config/nds32/nds32-protos.h: More declarations for n9 pipeline.
23689 * config/nds32/nds32-utils.c: New file.
23690 * config/nds32/nds32.h (TARGET_PIPELINE_N9, TARGET_PIPELINE_SIMPLE,
23691 TARGET_MUL_SLOW): Define.
23692 * config/nds32/nds32.md (pipeline_model): New attribute.
23693 * config/nds32/nds32.opt (mcpu, mconfig-mul, mconfig-register-ports):
23694 New options that support cpu n9.
23695 * config/nds32/pipelines.md: Include n9 settings.
23696 * config/nds32/t-nds32 (nds32-utils.o): Add dependency.
23698 2018-04-08 Chung-Ju Wu <jasonwucj@gmail.com>
23700 * config/nds32/nds32-md-auxiliary.c (output_cond_branch): Output align
23701 information if necessary.
23702 (output_cond_branch_compare_zero): Likewise.
23703 * config/nds32/nds32.c (nds32_adjust_insn_length): Consider align case.
23704 (nds32_target_alignment): Refine for alignment.
23705 * config/nds32/nds32.h (NDS32_ALIGN_P): Define.
23706 (FUNCTION_BOUNDARY): Modify.
23707 * config/nds32/nds32.md (call_internal, call_value_internal): Consider
23709 * config/nds32/nds32.opt (malways-align, malign-functions): New.
23711 2018-04-08 Monk Chiang <sh.chiang04@gmail.com>
23713 * config/nds32/constants.md (unspec_volatile_element): Add values for
23714 TLB operation and data prefetch.
23715 * config/nds32/nds32-intrinsic.c: Implementation of intrinsic
23716 functions for TLB operation and data prefetch.
23717 * config/nds32/nds32-intrinsic.md: Likewise.
23718 * config/nds32/nds32_intrinsic.h: Likewise.
23719 * config/nds32/nds32.c (nds32_dpref_names): Likewise.
23720 (nds32_print_operand): Likewise.
23721 * config/nds32/nds32.h (nds32_builtins): Likewise.
23723 2018-04-07 Thomas Koenig <tkoenig@gcc.gnu.org>
23724 Andrew Pinski <pinsika@gcc.gnu.org>
23726 PR middle-end/82976
23727 * match.pd: Use constant_boolean_node of correct type instead of
23728 boolean_true_node or boolean_false_node for simplifying
23729 pointer comparisons to zero.
23731 2018-04-07 Jakub Jelinek <jakub@redhat.com>
23733 PR tree-optimization/80021
23734 * tree.c (verify_type_variant): Make error call in verify_variant_match
23735 translatable and remove final full stop.
23737 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
23739 * config/nds32/constants.md (unspec_volatile_element): Add
23740 UNSPEC_VOLATILE_EH_RETURN.
23741 * config/nds32/nds32-md-auxiliary.c (nds32_output_stack_push,
23742 nds32_output_stack_pop): Support dwarf exception handling process.
23743 * config/nds32/nds32-protos.h (nds32_dynamic_chain_address): Declare.
23744 * config/nds32/nds32.c (nds32_init_machine_status): Support dwarf
23745 exception handling process.
23746 (nds32_compute_stack_frame): Likewise.
23747 (nds32_return_addr_rtx): Likewise.
23748 (nds32_initial_elimination_offset): Likewise.
23749 (nds32_expand_prologue): Likewise.
23750 (nds32_expand_epilogue): Likewise.
23751 (nds32_dynamic_chain_address): New function.
23752 * config/nds32/nds32.h (machine_function): Add fields for dwarf
23753 exception handling.
23754 (DYNAMIC_CHAIN_ADDRESS): Define.
23755 (EH_RETURN_DATA_REGNO): Define.
23756 (EH_RETURN_STACKADJ_RTX): Define.
23757 * config/nds32/nds32.md (eh_return, nds32_eh_return): Implement
23758 patterns for dwarf exception handling.
23760 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
23762 * config/nds32/nds32.h: Clean up obsolete macros.
23764 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
23766 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
23767 Add enum values for particular instructions.
23768 * config/nds32/nds32-intrinsic.c: Implementation of expanding
23769 particular intrinsic functions.
23770 * config/nds32/nds32-intrinsic.md: Likewise.
23771 * config/nds32/nds32_intrinsic.h: Likewise.
23772 * config/nds32/nds32.h (nds32_builtins): Likewise.
23773 * config/nds32/nds32.md (type): Add pbsad and pbsada.
23774 (btst, ave): New patterns for particular instructions.
23776 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
23778 * config/nds32/constants.md (unspec_element, unspec_volatile_element):
23779 Add enum values for atomic load/store and memory sync.
23780 * config/nds32/nds32-intrinsic.c: Implementation for atomic load/store
23782 * config/nds32/nds32-intrinsic.md: Likewise.
23783 * config/nds32/nds32_intrinsic.h: Likewise.
23784 * config/nds32/nds32.h (nds32_builtins): Likewise.
23786 2018-04-07 Jakub Jelinek <jakub@redhat.com>
23788 PR tree-optimization/85257
23789 * fold-const.c (native_encode_vector): If not all elts could fit
23790 and off is -1, return 0 rather than offset.
23791 * tree-ssa-sccvn.c (vn_reference_lookup_3): Pass
23792 (offseti - offset2) / BITS_PER_UNIT as 4th argument to
23793 native_encode_expr. Verify len * BITS_PER_UNIT >= maxsizei. Don't
23794 adjust buffer in native_interpret_expr call.
23796 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
23798 * config/nds32/constants.md (unspec_volatile_element): Add cache
23799 control enum values.
23800 * config/nds32/nds32-intrinsic.c: Add cache control expand functions.
23801 * config/nds32/nds32-intrinsic.md: Add cache control patterns.
23802 * config/nds32/nds32.c (nds32_cctl_names): New.
23803 (nds32_print_operand): Handle cache control register names.
23804 * config/nds32/nds32.h (nds32_builtins): New enum values.
23805 * config/nds32/nds32_intrinsic.h: Add cache control enum types and
23807 * config/nds32/nds32.md (type): Add mmu.
23808 * config/nds32/pipelines.md (simple_insn): Add mmu.
23810 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
23812 * config/nds32/nds32.md (type): Remove call.
23813 * config/nds32/pipelines.md (simple_insn): Likewise.
23815 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
23817 * config/nds32/constants.md (unspec_volatile_element): Add
23818 UNSPEC_VOLATILE_FMFCSR, UNSPEC_VOLATILE_FMTCSR and
23819 UNSPEC_VOLATILE_FMFCFG.
23820 * config/nds32/nds32-intrinsic.c (bdesc_noarg): New builtin
23821 description for fmfcfg and fmfcsr.
23822 (bdesc_1arg): Add fmtcsr.
23823 (bdesc_2arg): Add fcpynss, fcpyss, fcpynsd and fcpysd.
23824 (nds32_expand_builtin_impl): Deal with FPU intrinsic functions.
23825 * config/nds32/nds32-intrinsic.md (unspec_fcpynsd, unspec_fcpysd,
23826 unspec_fcpynss, unspec_fcpysd, unspec_fcpyss, unspec_fmfcsr,
23827 unspec_fmfcfg): New patterns.
23828 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FMFCFG,
23829 NDS32_BUILTIN_FMFCSR, NDS32_BUILTIN_FMTCSR, NDS32_BUILTIN_FCPYNSS,
23830 NDS32_BUILTIN_FCPYSS,NDS32_BUILTIN_FCPYNSD and NDS32_BUILTIN_FCPYSD.
23831 * config/nds32/nds32_intrinsic.h (__nds32__fcpynsd, __nds32__fcpynss,
23832 __nds32__fcpysd, __nds32__fcpyss, __nds32__fmfcsr, __nds32__fmtcsr,
23833 __nds32__fmfcfg): Define.
23835 2018-04-07 Monk Chiang <sh.chiang04@gmail.com>
23837 * config/nds32/nds32.c (nds32_intrinsic_register_names): Add more
23838 intrinsic register names.
23839 * config/nds32/nds32_intrinsic.h (nds32_intrinsic_registers): Add more
23840 intrinsic register enum values and macros.
23842 2018-04-07 Chung-Ju Wu <jasonwucj@gmail.com>
23844 * config/nds32/nds32.c (nds32_legitimate_index_p): Modify condition
23845 for load/store addressing form.
23846 (nds32_print_operand_address): Likewise.
23848 2018-04-06 Eric Botcazou <ebotcazou@adacore.com>
23851 * config/sparc/sparc.c (sparc_expand_move): Deal with symbolic operands
23852 based on LABEL_REF. Remove useless assertion.
23853 (pic_address_needs_scratch): Fix formatting.
23854 (sparc_legitimize_pic_address): Minor tweaks.
23855 (sparc_delegitimize_address): Adjust assertion accordingly.
23856 * config/sparc/sparc.md (movsi_pic_label_ref): Change label_ref_operand
23857 into symbolic_operand.
23858 (movsi_high_pic_label_ref): Likewise.
23859 (movsi_lo_sum_pic_label_ref): Likewise.
23860 (movdi_pic_label_ref): Likewise.
23861 (movdi_high_pic_label_ref): Likewise.
23862 (movdi_lo_sum_pic_label_ref): Likewise.
23864 2018-04-06 Amaan Cheval <amaan.cheval@gmail.com>
23866 * config.gcc (x86_64-*-rtems*): Add rtems.h to tm_file for
23867 custom LIB_SPEC setup.
23869 2018-04-06 Ruslan Bukin <br@bsdpad.com>
23870 Kito Cheng <kito.cheng@gmail.com>
23872 * config.gcc (riscv*-*-freebsd*): Add RISC-V FreeBSD support.
23873 * config/riscv/freebsd.h: New.
23875 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
23877 * config/nds32/nds32.c (nds32_adjust_insn_length): Refine.
23878 * config/nds32/nds32.h (ADJUST_INSN_LENGTH): Change the location in
23881 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
23882 Kito Cheng <kito.cheng@gmail.com>
23884 * config/nds32/nds32-md-auxiliary.c (nds32_output_return,
23885 nds32_output_call, nds32_symbol_binds_local_p): New functions.
23886 * config/nds32/nds32-protos.h (nds32_output_call,
23887 nds32_output_return): Declare.
23888 * config/nds32/nds32.md: Refine all the call and return patterns.
23890 2018-04-06 Jakub Jelinek <jakub@redhat.com>
23893 * dwarf2out.c (rtl_for_decl_init): For STRING_CST initializer only
23894 build CONST_STRING if TYPE_MAX_VALUE is non-NULL and is INTEGER_CST.
23896 PR rtl-optimization/84872
23897 * cfgloopmanip.c (create_preheader): Use make_forwarder_block even if
23898 nentry == 1 when CP_FALLTHRU_PREHEADERS and single_entry is
23899 EDGE_CROSSING edge.
23901 2018-04-06 Tamar Christina <tamar.christina@arm.com>
23903 * expr.c (copy_blkmode_to_reg): Revert 254862.
23904 * doc/sourcebuild.texi (word_mode_no_slow_unalign): Likewise.
23906 2018-04-06 Richard Biener <rguenther@suse.de>
23908 PR middle-end/85244
23909 * tree-dfa.c (get_ref_base_and_extent): Reset seen_variable_array_ref
23910 after seeing a component reference with an adjacent field. Treat
23911 refs to arrays at struct end of external decls similar to
23912 refs to unconstrained commons.
23914 2018-04-06 Jakub Jelinek <jakub@redhat.com>
23917 * fold-const.c (twoval_comparison_p): Remove SAVE_P argument and don't
23918 look through SAVE_EXPRs with non-side-effects argument. Adjust
23920 (fold_comparison): Adjust twoval_comparison_p caller, don't handle
23923 2018-04-06 Richard Biener <rguenther@suse.de>
23925 PR middle-end/85180
23926 * alias.c (find_base_term): New wrapper around find_base_term
23927 unwinding CSELIB_VAL_PTR changes.
23928 (find_base_term): Do not restore CSELIB_VAL_PTR during the
23931 2018-04-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
23933 * config/s390/s390.c (s390_z10_optimize_cmp): Expand dedicated NOP
23935 * config/s390/s390.md (UNSPECV_NOP_LR_0, UNSPECV_NOP_LR_1): New
23936 constant definitions.
23937 ("nop"): lr 0,0 -> nopr r0
23938 ("nop_lr0", "nop_lr1"): New insn definitions.
23940 2018-04-06 Chung-Ju Wu <jasonwucj@gmail.com>
23942 * config/nds32/nds32.md (*stack_push, *stack_pop): Use
23943 NDS32_V3PUSH_AVAILABLE_P macro.
23945 2018-04-06 Monk Chiang <sh.chiang04@gmail.com>
23946 Chung-Ju Wu <jasonwucj@gmail.com>
23948 * config.gcc (nds32*-*-*): Add v2j v3f v3s checking.
23949 (nds32*-*-*): Add float and fpu_config into supported_defaults.
23950 * common/config/nds32/nds32-common.c (TARGET_DEFAULT_TARGET_FLAGS):
23951 Include TARGET_DEFAULT_FPU_ISA and TARGET_DEFAULT_FPU_FMA.
23952 * config/nds32/constants.md (unspec_element): Add UNSPEC_COPYSIGN,
23953 UNSPEC_FCPYNSD, UNSPEC_FCPYNSS, UNSPEC_FCPYSD and UNSPEC_FCPYSS.
23954 * config/nds32/constraints.md: New constraints and checking for hard
23955 float configuration.
23956 * config/nds32/iterators.md: New mode iterator and attribute for hard
23957 float configuration.
23958 * config/nds32/nds32-doubleword.md: Use hard float alternatives and
23960 * config/nds32/nds32-fpu.md: New file.
23961 * config/nds32/nds32-md-auxiliary.c: New functions and checkings to
23962 deal with hard float code generation.
23963 * config/nds32/nds32-opts.h (nds32_arch_type): Add ARCH_V3F and
23965 (abi_type, float_reg_number): New enum type.
23966 * config/nds32/nds32-predicates.c: New predicates for hard float.
23967 * config/nds32/nds32-protos.h: Declare functions for hard float.
23968 * config/nds32/nds32.c: Implementation for hard float configuration.
23969 * config/nds32/nds32.h: Definitions for hard float configuration.
23970 * config/nds32/nds32.md: Include hard float machine description and
23971 modify patterns for hard float configuration.
23972 * config/nds32/nds32.opt: New options for hard float configuration.
23973 * config/nds32/predicates.md: New predicates for hard float
23976 2018-04-06 Kuan-Lin Chen <kuanlinchentw@gmail.com>
23978 * common/config/nds32/nds32-common.c
23979 (nds32_option_optimization_table): Enable -mreleax-hint by default.
23981 2018-04-05 Jakub Jelinek <jakub@redhat.com>
23983 PR middle-end/85195
23984 * match.pd (BIT_FIELD_REF CONSTRUCTOR@0 @1 @2): Use view_convert around
23985 CONSTRUCTOR_ELT (ctor, ...)->value.
23987 2018-04-05 Uros Bizjak <ubizjak@gmail.com>
23990 * config/i386/i386.md (define_attr "memory"): Handle rotate1 type.
23992 2018-04-05 Tom de Vries <tom@codesourcery.com>
23995 * config/nvptx/nvptx.c (nvptx_single): Fix neutering of bb with only
23998 2018-04-05 Shiva Chen <shiva0217@gmail.com>
23999 Kito Cheng <kito.cheng@gmail.com>
24001 * config/nds32/constraints.md (U33): Fine-tune checking condition.
24002 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format): Ditto.
24003 * config/nds32/nds32.h (nds32_16bit_address_type): Add
24004 ADDRESS_POST_MODIFY_LO_REG_IMM3U.
24006 2018-04-05 Shiva Chen <shiva0217@gmail.com>
24007 Kito Cheng <kito.cheng@gmail.com>
24009 * config/nds32/constraints.md (Ufe): New memory constraint.
24010 * config/nds32/nds32-md-auxiliary.c (nds32_mem_format,
24011 nds32_output_16bit_load): Consider r8 register for lwi45.fe format.
24012 * config/nds32/nds32.c (nds32_print_operand): Output lwi45.fe
24014 * config/nds32/nds32.h (nds32_16bit_address_type): Add ADDRESS_R8_IMM7U.
24015 * config/nds32/nds32.md (*mov<mode>): Adjust pattern.
24017 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24019 * config/nds32/nds32.md: Use optimize_size in the condition for
24020 alu-shift instructions.
24022 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24024 * config/nds32/nds32.md (divsi4, udivsi4): New patterns.
24026 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24028 * config/nds32/nds32.md (negsi2): Refine pattern.
24030 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
24031 Chung-Ju Wu <jasonwucj@gmail.com>
24033 * config/nds32/iterators.md (shift_rotate): New code iterator.
24034 (shift): New code attribute.
24035 * config/nds32/nds32-md-auxiliary.c (nds32_expand_constant): New.
24036 * config/nds32/nds32-protos.h (nds32_expand_constant): Declare.
24037 * config/nds32/nds32.c (nds32_print_operand): Deal with more cases.
24038 * config/nds32/nds32.md (addsi3, *add_srli): Refine implementation for
24039 bit-wise operations.
24040 (andsi3, *andsi3): Ditto.
24041 (iorsi3, *iorsi3, *or_slli, *or_srli): Ditto.
24042 (xorsi3, *xorsi3, *xor_slli, *xor_srli): Ditto.
24043 (<shift>si3, *ashlsi3, *ashrsi3, *lshrsi3, *rotrsi3): Ditto.
24044 * config/nds32/predicates.md (nds32_rimm5u_operand, nds32_and_operand,
24045 nds32_ior_operand, nds32_xor_operand): New predicates.
24047 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24049 * config/nds32/nds32.md (add<mode>3, sub<mode>3): Rename to ...
24050 (addsi3, subsi3): ... this.
24052 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24054 * config/nds32/nds32.md (*sub_srli, *and_slli): Fine-tune predicator.
24056 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24058 * config/nds32/nds32.md: Adjust indention.
24060 2018-04-05 Kito Cheng <kito.cheng@gmail.com>
24062 * config/nds32/nds32.md (feature): New attribute.
24064 2018-04-05 Chung-Ju Wu <jasonwucj@gmail.com>
24066 * config/nds32/nds32.md (subtype): New attribute.
24068 2018-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
24071 * config/arm/arm-builtins.c (arm_expand_builtin): Change
24072 expansion to perform a bitwise AND of the argument followed by a
24073 boolean negation of the result.
24075 2018-04-04 Peter Bergner <bergner@vnet.ibm.com>
24077 PR rtl-optimization/84878
24078 * ddg.c (add_cross_iteration_register_deps): Use DF_REF_BB to determine
24079 the basic block. Assert the use reference is not artificial and that
24080 it has an associated insn.
24082 2018-04-04 Michael Matz <matz@suse.de>
24084 * builtins.c (compute_objsize): Pass correct operand
24085 to array_at_struct_end_p.
24087 2018-04-04 Richard Biener <rguenther@suse.de>
24090 * dwarf2out.c (dwarf2out_register_external_die): Peel namespaces
24091 from contexts for DINFO_LEVEL_TERSE and below.
24093 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
24095 * config/nds32/nds32-doubleword.md (move_<mode>): Require
24096 resiter_operand condition.
24097 * config/nds32/nds32.md (*move<mode>): Ditto.
24099 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
24100 Monk Chiang <sh.chiang04@gmail.com>
24102 * config/nds32/nds32.md (movmisalign<mode>): New pattern.
24104 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24106 * config/nds32/nds32.md (movqi, movhi): Merge into mov<mode>.
24108 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24109 Kito Cheng <kito.cheng@gmail.com>
24111 * config/nds32/nds32-md-auxiliary.c (nds32_inverse_cond_code,
24112 nds32_cond_code_str, output_cond_branch,
24113 output_cond_branch_compare_zero, nds32_expand_cbranch,
24114 nds32_expand_cstore, nds32_expand_movcc,
24115 nds32_output_cbranchsi4_equality_zero,
24116 nds32_output_cbranchsi4_equality_reg,
24117 nds32_output_cbranchsi4_equality_reg_or_const_int,
24118 nds32_output_cbranchsi4_greater_less_zero: New functions.
24119 * config/nds32/nds32-protos.h (nds32_expand_cbranch,
24120 nds32_expand_cstore, nds32_expand_movcc,
24121 nds32_output_cbranchsi4_equality_zero,
24122 nds32_output_cbranchsi4_equality_reg,
24123 nds32_output_cbranchsi4_equality_reg_or_const_int,
24124 nds32_output_cbranchsi4_greater_less_zero): Declare.
24125 * config/nds32/predicates.md (nds32_movecc_comparison_operator,
24126 nds32_rimm11s_operand): New predicates.
24127 * config/nds32/nds32.h (nds32_expand_result_type): New enum type.
24128 * config/nds32/nds32.md: Rewrite all the branch and conditional move
24131 2018-04-04 Kito Cheng <kito.cheng@gmail.com>
24133 * config/nds32/nds32-doubleword.md: Refine all the instruction type.
24134 * config/nds32/nds32.md: Ditto.
24135 * config/nds32/pipelines.md: Ditto.
24137 2018-04-04 Richard Biener <rguenther@suse.de>
24139 PR tree-optimization/85168
24140 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Avoid
24141 propagating abnormals.
24143 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24145 * config/nds32/nds32.md (enabled): Use yes/no for this attribute.
24147 2018-04-04 Chung-Ju Wu <jasonwucj@gmail.com>
24148 Kito Cheng <kito.cheng@gmail.com>
24150 * config/nds32/nds32-md-auxiliary.c (nds32_long_call_p): New function.
24151 * config/nds32/nds32-protos.h (nds32_long_call_p): Declare.
24152 * config/nds32/nds32.c (nds32_function_ok_for_sibcall): New function.
24153 (TARGET_FUNCTION_OK_FOR_SIBCALL): Define.
24154 * config/nds32/nds32.md (sibcall_internal): New.
24155 (sibcall_register): Remove.
24156 (sibcall_immediate): Remove.
24157 (sibcall_value_internal): New.
24158 (sibcall_value_register): Remove.
24159 (sibcall_value_immediate): Remove.
24160 * config/nds32/predicates.md (nds32_general_register_operand): New.
24161 (nds32_call_address_operand): New.
24163 2018-04-03 Jakub Jelinek <jakub@redhat.com>
24165 PR rtl-optimization/85167
24166 * shrink-wrap.c (move_insn_for_shrink_wrap): Don't set bb_uses and
24167 bb_defs if *split_p, instead preinitialize it to NULL.
24169 PR tree-optimization/85156
24170 * builtins.c (fold_builtin_expect): Use save_expr on arg1 to avoid
24171 evaluating the argument multiple times.
24173 2018-04-03 Bill Schmidt <wschmidt@linux.ibm.com>
24175 * config/rs6000/emmintrin.h (_mm_cvtpd_epi32): Use __vector rather
24177 (_mm_cvtpd_ps): Likewise.
24178 (_mm_cvttpd_epi32): Likewise.
24179 * config/rs6000/mmintrin.h (_mm_unpacklo_pi8): Likewise.
24180 * config/rs6000/xmmintrin.h: For strict-ANSI C++ or C11, undefine
24181 vector, pixel, and bool following altivec.h include.
24183 2018-04-03 Martin Sebor <msebor@redhat.com>
24185 * doc/extend.texi (Common Function Attributes): Clarify.
24186 (const attribute): Likewise.
24187 (pure attribute): Likewise.
24189 2018-04-03 Jakub Jelinek <jakub@redhat.com>
24192 * config/i386/i386.c (ix86_expand_vector_set): Use
24193 HOST_WIDE_INT_1U << elt instead of 1 << elt. Formatting fix.
24195 2018-04-03 Uros Bizjak <ubizjak@gmail.com>
24197 * config/i386/i386.c (emit_i387_cw_initialization): Always use logic
24198 instructions when changing rounding bits to preserve precision bits
24199 in the x87 control word.
24201 2018-04-03 Martin Liska <mliska@suse.cz>
24203 PR tree-optimization/82491
24204 * rtl.h (strip_offset_and_add): Replace += suboffset with
24205 poly_uint64 () + suboffset.
24207 2018-03-29 Martin Liska <mliska@suse.cz>
24208 Martin Jambor <mjambor@suse.cz>
24211 * ipa-cp.c (propagate_bits_across_jump_function): Bail out if
24212 param_type is not an integral or pointer type.
24214 2018-04-03 Richard Biener <rguenther@suse.de>
24216 * sese.h (recompute_all_dominators): Remove.
24218 2018-04-02 Martin Sebor <msebor@redhat.com>
24220 * doc/invoke.texi (-Wrestrict): Fix typos.
24222 2018-04-02 Jim Wilson <jimw@sifive.com>
24224 * config/riscv/riscv.h (SHIFT_COUNT_TRUNCATED): Set to zero.
24225 * config/riscv/riscv.md (<optab>si3): Use QImode shift count.
24226 (<optab>di3, <optab>si3_extend): Likewise.
24227 (<optab>si3_mask, <optab>si3_mask_1): New.
24228 (<optab>di3_mask, <optab>di3_mask_1): New.
24229 (<optab>si3_extend_mask, <optab>si3_extend_mask_1): New.
24230 (lshrsi3_zero_extend_1): Use VOIDmode shift count.
24231 * config/riscv/sync.md (atomic_test_and_set): Emit QImode shift count.
24233 2018-04-02 Gerald Pfeifer <gerald@pfeifer.com>
24235 * doc/cpp.texi (Variadic Macros): Fix line continuation in an
24238 2018-04-02 Chung-Ju Wu <jasonwucj@gmail.com>
24240 * config/nds32/nds32.c (TARGET_CANONICALIZE_COMPARISON): Define.
24241 (nds32_canonicalize_comparison): New function.
24243 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
24244 Kito Cheng <kito.cheng@gmail.com>
24245 Kuan-Lin Chen <kuanlinchentw@gmail.com>
24247 * config.gcc (nds32): Add nds32-relax-opt.o into extra_objs.
24248 * config/nds32/constants.md (unspec_volatile_element): Add
24249 UNSPEC_VOLATILE_RELAX_GROUP.
24250 * config/nds32/nds32-relax-opt.c: New file.
24251 * config/nds32/nds32-predicates.c
24252 (nds32_symbol_load_store_p): New function.
24253 * config/nds32/nds32-protos.h
24254 (nds32_symbol_load_store_p): Declare function.
24255 (make_pass_nds32_relax_opt): Declare new rtl pass function.
24256 * config/nds32/nds32.c
24257 (nds32_register_pass): New function to register pass.
24258 (nds32_register_passes): New function to register passes.
24259 * config/nds32/nds32.md (relax_group): New pattern.
24260 * config/nds32/nds32.opt (mrelax-hint): New option.
24261 * config/nds32/t-nds32 (nds32-relax-opt.o): New dependency.
24263 2018-04-01 Kito Cheng <kito.cheng@gmail.com>
24265 * config/nds32/t-nds32: Modify files dependency.
24267 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
24269 * config/nds32/nds32.h (FUNCTION_PROFILER): Output newline character.
24270 (PROFILE_HOOK): Define its implementation.
24272 2018-04-01 Chung-Ju Wu <jasonwucj@gmail.com>
24274 * config/nds32/nds32.h (WCHAR_TYPE, WCHAR_TYPE_SIZE): Use unsigned int
24275 type and 32-bit size.
24277 2018-04-01 Jakub Jelinek <jakub@redhat.com>
24279 PR middle-end/85090
24280 * config/i386/sse.md (V): Add V64QI and V32HI for TARGET_AVX512F.
24281 (V_128_256): New mode iterator.
24282 (*avx512dq_vextract<shuffletype>64x2_1 splitter): New define_split.
24283 (*avx512f_vextract<shuffletype>32x4_1 splitter): Likewise.
24284 (xop_pcmov_<mode><avxsizesuffix>): Use V_128_256 mode iterator instead
24286 * config/i386/i386.c (ix86_expand_vector_set): Improve V32HImode and
24287 V64QImode expansion for !TARGET_AVX512BW && TARGET_AVX512F.
24289 2018-03-31 Segher Boessenkool <segher@kernel.crashing.org>
24292 * config/rs6000/xmmintrin.h (_mm_set_ps, _mm_max_ps): Handle (quiet)
24293 NaN inputs correctly.
24295 2018-03-30 Peter Bergner <bergner@vnet.ibm.com>
24298 * config/rs6000/vsx.md (??r): New mode attribute.
24299 (*vsx_mov<mode>_64bit): Use it.
24300 (*vsx_mov<mode>_32bit): Likewise.
24302 2018-03-30 Martin Sebor <msebor@redhat.com>
24304 PR tree-optimization/84818
24305 * builtins.c (check_access): Use warning_n.
24307 2018-03-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
24310 * config/rs6000/rs6000-string.c (expand_compare_loop): Fix redundant
24312 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): Fix redundant
24315 2018-03-30 Julia Koval <julia.koval@intel.com>
24318 * x86-tune.def (movx, partial_reg_dependency): Enable for
24321 2018-03-29 Vladimir Makarov <vmakarov@redhat.com>
24323 PR inline-asm/84985
24324 * lra-constraints.c (process_alt_operands): Move setting
24325 this_alternative_matches below.
24327 2018-03-29 Martin Liska <mliska@suse.cz>
24330 * doc/invoke.texi: Document how LTO works with debug info.
24331 Describe auto-load support of binutils. Mention 'x86-64'
24332 as valid option value of -march option.
24334 2018-03-29 Jakub Jelinek <jakub@redhat.com>
24336 * config/i386/sse.md (<avx512>_blendm<mode>): Use <sseintprefix>.
24339 * fold-const.c (operand_equal_p): Handle DEBUG_BEGIN_STMT.
24340 For STATEMENT_LIST, pass down OEP_LEXICOGRAPHIC and maybe
24341 OEP_NO_HASH_CHECK for recursive call, to avoid exponential
24344 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
24347 * config/rs6000/rs6000.h (RS6000_BTM_POWERPC64): New define.
24348 (RS6000_BTM_COMMON): Add RS6000_BTM_POWERPC64.
24349 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add support
24350 for RS6000_BTM_POWERPC64.
24351 (rs6000_invalid_builtin): Add handling for RS6000_BTM_POWERPC64
24352 (rs6000_builtin_mask_names): Add RS6000_BTM_POWERPC64.
24353 * config/rs6000/rs6000-builtin.def (BU_P7_POWERPC64_MISC_2): New macro
24356 (DIVDEU): Likewise.
24358 2018-03-28 Carl Love <cel@us.ibm.com>
24361 2017-09-27 Carl Love <cel@us.ibm.com>
24363 * config/rs6000/rs6000-builtin.def (BU_FP_1MISC_1): Add define macro.
24364 (FCTID, FCTIW): Add BU_FP_MISC_1 macro expansion for builtins.
24365 * config/rs6000/rs6000.md (lrintsfsi2): Add define_insn for the
24368 2018-03-28 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
24370 * config/rs6000/xmmintrin.h (_mm_max_pi16): Use __vector __bool
24371 instead of __vector bool.
24372 (_mm_max_pu8): Likewise.
24373 (_mm_min_pi16): Likewise.
24375 2018-03-28 Peter Bergner <bergner@vnet.ibm.com>
24378 * config/rs6000/rs6000-builtin.def (DIVWEO): Delete macro expansion.
24379 (DIVWEUO): Likewise.
24380 (DIVDEO): Likewise.
24381 (DIVDEUO): Likewise.
24382 * config/rs6000/rs6000.c (builtin_function_type): Remove support for
24383 DIVWEUO and DIVDEUO.
24384 * config/rs6000/rs6000.md (UNSPEC_DIVEO, UNSPEC_DIVEUO): Delete unspecs.
24385 (UNSPEC_DIV_EXTEND): Remove deleted unspecs.
24386 (div_extend): Likewise.
24387 * doc/extend.texi (__builtin_divweo): Remove documentation for deleted
24389 (__builtin_divweuo): Likewise.
24390 (__builtin_divdeo): Likewise.
24391 (__builtin_divdeuo): Likewise.
24393 2018-03-28 Jakub Jelinek <jakub@redhat.com>
24396 * config/i386/i386.md (*add<mode>3_carry_0, *addsi3_carry_zext_0,
24397 *sub<mode>3_carry_0, *subsi3_carry_zext_0): New patterns.
24399 PR tree-optimization/82004
24400 * gimple-match-head.c (optimize_pow_to_exp): New function.
24401 * match.pd (pow(C,x) -> exp(log(C)*x)): Wrap with #if GIMPLE.
24402 Don't fold to exp if optimize_pow_to_exp is false.
24404 2018-03-28 Martin Liska <mliska@suse.cz>
24407 * calls.c (initialize_argument_information): Fix trailing space.
24408 * common.opt: Fix typo and provide better explanation for
24409 -fsanitize-coverage option.
24410 * config/i386/i386.opt: Fix typo.
24412 2018-03-28 Jakub Jelinek <jakub@redhat.com>
24413 Martin Liska <mliska@suse.cz>
24416 * gimplify.c (asan_poison_variable): Don't do the check for
24417 gimplify_omp_ctxp here.
24418 (gimplify_decl_expr): Do it here.
24419 (gimplify_target_expr): Likewise.
24421 2018-03-28 Martin Liska <mliska@suse.cz>
24424 * config/i386/i386.c (ix86_function_arg_advance): Do not call
24425 chkp_type_bounds_count if MPX is not enabled.
24427 2018-03-27 Chung-Ju Wu <jasonwucj@gmail.com>
24429 * config/nds32/nds32.h (BRANCH_COST): Adjust cost.
24431 2018-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
24434 * config/rs6000/rs6000.c (create_complex_muldiv): New helper
24435 function to create the function decl for complex long double
24436 multiply and divide for -mabi=ieeelongdouble.
24437 (init_float128_ieee): Call it.
24439 2018-03-27 H.J. Lu <hongjiu.lu@intel.com>
24442 * config/i386/i386.c (ix86_trampoline_init): Insert ENDBR for
24443 -fcf-protection=branch -mibt.
24444 * config/i386/i386.h (TRAMPOLINE_SIZE): Increased by 4 bytes.
24446 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24449 * config/arm/arm.c (arm_valid_symbolic_address): Handle
24450 arm_word_relocations.
24452 2018-03-27 Cesar Philippidis <cesar@codesourcery.com>
24455 * config/nvptx/nvptx.c (nvptx_assemble_decl_begin): Add '[]' to
24456 extern array declarations.
24458 2018-03-27 Richard Biener <rguenther@suse.de>
24460 PR middle-end/84067
24461 * match.pd ((A * C) +- (B * C) -> (A+-B) * C): Guard with
24462 explicit single_use checks.
24464 2018-03-27 Richard Biener <rguenther@suse.de>
24466 PR tree-optimization/85082
24467 * tree-ssa-sccvn.c (vn_reference_lookup_or_insert_for_pieces):
24470 2018-03-27 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
24472 * config.gcc (aarch64*-*-linux*): New TARGET_DEFAULT_ASYNC_UNWIND_TABLES
24473 * common/config/aarch64/aarch64-common.c (aarch64_optimization_table[]):
24474 Turn on fasynchronous-unwind-tables and funwind-tables.
24476 2018-03-26 Uros Bizjak <ubizjak@gmail.com>
24479 * config/i386/i386.md (*bmi_blsr_<mode>_cmp): New insn pattern.
24480 (*bmi_blsr_<mode>_ccz): Ditto.
24482 2018-03-26 Tom de Vries <tom@codesourcery.com>
24484 PR tree-optimization/85063
24485 * omp-general.c (offloading_function_p): New function. Factor out
24487 * omp-offload.c (pass_omp_target_link::gate): ... here.
24488 * omp-general.h (offloading_function_p): Declare.
24489 * tree-switch-conversion.c (build_one_array): Mark CSWTCH.x variable
24490 with attribute omp declare target for offloading functions.
24492 2018-03-24 Richard Sandiford <richard.sandiford@linaro.org>
24494 PR tree-optimization/84005
24495 * tree-data-ref.h (get_base_for_alignment): Declare.
24496 * tree-data-ref.c (get_base_for_alignment_1): New function.
24497 (get_base_for_alignment): Likewise.
24498 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
24499 get_base_for_alignment to find a suitable base object, instead
24500 of always using drb->base_address.
24502 2018-03-23 Jakub Jelinek <jakub@redhat.com>
24504 PR inline-asm/85022
24505 * emit-rtl.c (init_emit_regs): Indicate that VOIDmode MEMs don't have
24506 known size by default.
24508 2018-03-23 Vladimir Makarov <vmakarov@redhat.com>
24510 PR inline-asm/85030
24511 * lra-constraints.c (process_alt_operands): Don't match BLKmode
24512 and non BLKmode operands.
24514 2018-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24517 * config/arm/arm.md (unaligned_loadhis): Remove first alternative.
24518 Clean up attributes.
24520 2018-03-23 Richard Biener <rguenther@suse.de>
24523 * dwarf2out.c (rtl_for_decl_location): Do not generate RTL early when
24524 we are going to emit early debug for LTO.
24526 2018-03-23 Jakub Jelinek <jakub@redhat.com>
24528 PR inline-asm/85034
24529 * function.c (match_asm_constraints_1): Don't optimize if input
24530 doesn't satisfy general_operand predicate for output's mode.
24532 PR inline-asm/85022
24533 * alias.c (write_dependence_p): Don't require for x_canonicalized
24534 non-VOIDmode if x has VOIDmode.
24537 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): If DECL_REGISTER (base),
24538 just don't try to optimize it rather than assert it never happens.
24540 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
24542 * config/rs6000/rs6000-builtin.def: Remove various BU_ALTIVEC_X
24543 macro expansions for definition of ST_INTERNAL_<mode> and
24544 LD_INTERNAL_<mode> builtins.
24545 * config/rs6000/rs6000-protos.h (rs6000_address_for_altivec):
24547 * config/rs6000/rs6000.c (altivec_expand_ld_builtin): Delete this
24549 (altivec_expand_st_builtin): Likewise.
24550 (altivec_expand_builtin): Remove calls to deleted functions.
24551 (rs6000_address_for_altivec): Delete this function.
24552 * config/rs6000/vector.md: Remove expands for
24553 vector_altivec_load_<mode> and vector_altivec_store_<mode>.
24555 2018-03-22 Sudakshina Das <sudi.das@arm.com>
24558 * config/arm/arm.h (machine_function): Add static_chain_stack_bytes.
24559 * config/arm/arm.c (arm_compute_static_chain_stack_bytes): Avoid
24560 re-computing once computed.
24561 (arm_expand_prologue): Compute machine->static_chain_stack_bytes.
24562 (arm_init_machine_status): Initialize
24563 machine->static_chain_stack_bytes.
24565 2018-03-22 Kelvin Nilsen <kelvin@gcc.gnu.org>
24568 * doc/extend.texi: Add four new prototypes for vec_ld.
24569 * config/rs6000/rs6000-builtin.def (LVX_V1TI): Reorder symbol
24570 definitions for more logical presentation.
24571 * config/rs6000/rs6000-c.c: (altivec_overloaded_builtins): Add
24572 entries for V1TI variants of __builtin_altivec_ld builtin.
24573 * config/rs6000/rs6000.c: (altivec_expand_lv_builtin): Add test and
24574 handling of V1TI variant of LVX icode pattern.
24575 (altivec_expand_builtin): Add case for ALTIVEC_BUILTIN_LVX_V1TI.
24576 (rs6000_gimple_fold_builtin): Likewise.
24577 (altivec_init_builtins): Add code to define
24578 __builtin_altivec_lvx_v1ti function.
24580 2018-03-22 Jakub Jelinek <jakub@redhat.com>
24582 PR inline-asm/84941
24583 * function.c (match_asm_constraints_1): Don't do the optimization
24584 if input isn't a REG, SUBREG, MEM or constant.
24586 2018-03-22 Tom de Vries <tom@codesourcery.com>
24588 PR tree-optimization/84956
24589 * tree-ssa-tail-merge.c (find_clusters_1): Skip bbs with
24590 bb_has_abnormal_pred.
24592 2018-03-22 Jakub Jelinek <jakub@redhat.com>
24595 * dwarf2asm.c (dw2_output_indirect_constant_1): Set
24596 DECL_INITIAL (decl) to decl at the end.
24597 * varasm.c (use_blocks_for_decl_p): Revert the 2018-03-20 change,
24598 adjust the comment.
24600 2018-03-21 Joseph Myers <joseph@codesourcery.com>
24602 * doc/extend.texi (__builtin_tgmath): Document when complex
24603 integer types are treated as _Complex _Float64.
24605 2018-03-21 Tom de Vries <tom@codesourcery.com>
24607 * doc/extend.texi (__builtin_extend_pointer): Remove pasto.
24609 2018-03-21 Jakub Jelinek <jakub@redhat.com>
24611 PR tree-optimization/84960
24612 * tree-cfg.c (remove_bb): Don't move forced labels into bb->prev_bb
24613 if it is ENTRY block, move them into single succ of ENTRY in that case.
24615 2018-03-21 Richard Sandiford <richard.sandiford@linaro.org>
24617 PR tree-optimization/84811
24618 * poly-int.h (poly_span_traits): Remove the T3 parameter and
24619 promote HOST_WIDE_INT T2 - T1 results to unsigned HOST_WIDE_INT.
24620 (maybe_in_range_p, known_in_range_p, ranges_known_overlap_p):
24621 (known_subrange_p): Update accordingly. Cast each value involved
24622 in the size comparison, rather than casting the result of the
24625 2018-03-21 Jakub Jelinek <jakub@redhat.com>
24627 PR tree-optimization/84982
24628 * gimple-ssa-store-merging.c (invert_op): Handle boolean inversion
24629 by flipping the least significant bit rather than all bits from
24630 bitpos to bitpos + bitsize - 1.
24632 2018-03-21 Nathan Sidwell <nathan@acm.org>
24634 * doc/extend.texi (Deprecated Features): Remove mention of
24635 long-deleted deprecations.
24637 2018-03-21 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
24640 * configure.ac (gcc_cv_ld_soname) <*-*-solaris2*>: Set.
24641 * configure: Regenerate.
24643 2018-03-21 Tom de Vries <tom@codesourcery.com>
24645 PR tree-optimization/83126
24646 * tree-parloops.c (num_phis): New function.
24647 (gen_parallel_loop): Detect and handle canonicalize_loop_ivs failure.
24649 2018-03-21 Nathan Sidwell <nathan@acm.org>
24651 * doc/extend.texi (Deprecated Features): Update deprecated flags,
24652 mention anon-struct/union members and trailing attributes.
24654 2018-03-21 Bin Cheng <bin.cheng@arm.com>
24656 PR tree-optimization/84969
24657 * tree-loop-distribution.c (fuse_memset_builtins): Don't reorder
24658 builtin memset partitions if they set different rhs values.
24660 2018-03-21 Jakub Jelinek <jakub@redhat.com>
24662 PR rtl-optimization/84989
24663 * simplify-rtx.c (simplify_unary_operation_1): Don't try to simplify
24664 VEC_DUPLICATE with scalar result mode.
24666 2018-03-21 Martin Liska <mliska@suse.cz>
24669 * ipa-icf.c (sem_item_optimizer::fixup_points_to_sets): Remove
24670 not intended return statement.
24672 2018-03-21 Martin Liska <mliska@suse.cz>
24675 * tree-chkp.c (CHKP_ARRAY_MAX_CHECK_STEPS): Define a new macro.
24676 (chkp_find_bound_slots_1): Limit number of iterations.
24678 2018-03-20 David H. Gutteridge <dhgutteridge@sympatico.ca>
24681 * Minor grammar fixes for x86 options.
24683 2018-03-20 Jakub Jelinek <jakub@redhat.com>
24686 * dce.c (delete_unmarked_insns): Don't remove frame related noop moves
24687 holding REG_CFA_RESTORE notes, instead turn them into a USE.
24689 2018-03-20 Peter Bergner <bergner@vnet.ibm.com>
24692 * config/rs6000/altivec.md (altivec_lvx_<mode>_2op): Delete define_insn.
24693 (altivec_lvx_<mode>_1op): Likewise.
24694 (altivec_stvx_<mode>_2op): Likewise.
24695 (altivec_stvx_<mode>_1op): Likewise.
24696 (altivec_lvx_<VM2:mode>): New define_expand.
24697 (altivec_stvx_<VM2:mode>): Likewise.
24698 (altivec_lvx_<VM2:mode>_2op_<P:mptrsize>): New define_insn.
24699 (altivec_lvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
24700 (altivec_stvx_<VM2:mode>_2op_<P:mptrsize>): Likewise.
24701 (altivec_stvx_<VM2:mode>_1op_<P:mptrsize>): Likewise.
24702 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Use new expanders.
24703 (rs6000_gen_lvx): Likewise.
24704 * config/rs6000/rs6000.c (altivec_expand_lv_builtin): Likewise.
24705 (altivec_expand_stv_builtin): Likewise.
24706 (altivec_expand_builtin): Likewise.
24707 * config/rs6000/vector.md: Likewise.
24709 2018-03-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
24712 * config/arm/arm.c (arm_array_mode_supported_p): Return false for
24715 2018-03-20 Richard Biener <rguenther@suse.de>
24718 * config/i386/i386.c (ix86_add_stmt_cost): Only cost
24719 sign-conversions as zero, fall back to standard scalar_stmt
24722 2018-03-20 Martin Liska <mliska@suse.cz>
24725 * predict.c (rebuild_frequencies): Handle case when we have
24726 PROFILE_ABSENT, but flag_guess_branch_prob is false.
24728 2018-03-20 Jakub Jelinek <jakub@redhat.com>
24731 * dwarf2asm.c (dw2_output_indirect_constant_1): Temporarily turn off
24732 flag_section_anchors.
24733 * varasm.c (use_blocks_for_decl_p): Remove hack for
24734 dw2_force_const_mem.
24737 * config/aarch64/aarch64.md (*aarch64_reg_<mode>3_neg_mask2): Rename
24739 (*aarch64_<optab>_reg_<mode>3_neg_mask2): ... this. If pseudos can't
24740 be created, use lowpart_subreg of operands[0] rather than operands[0]
24742 (*aarch64_reg_<mode>3_minus_mask): Rename to ...
24743 (*aarch64_ashl_reg_<mode>3_minus_mask): ... this.
24744 (*aarch64_<optab>_reg_di3_mask2): Use const_int_operand predicate
24745 and n constraint instead of aarch64_shift_imm_di and Usd.
24746 (*aarch64_reg_<optab>_minus<mode>3): Rename to ...
24747 (*aarch64_<optab>_reg_minus<mode>3): ... this.
24749 2018-03-20 Sudakshina Das <sudi.das@arm.com>
24752 * config/arm/neon.md (ashldi3_neon): Update ?s for constraints
24753 to favor GPR over NEON registers.
24754 (<shift>di3_neon): Likewise.
24756 2018-03-20 Tom de Vries <tom@codesourcery.com>
24759 * config/nvptx/nvptx.c (nvptx_single): Don't neuter bar.sync.
24760 (nvptx_process_pars): Emit bar.sync asap and alap.
24762 2018-03-20 Tom de Vries <tom@codesourcery.com>
24765 * config/nvptx/nvptx.c (prevent_branch_around_nothing): Also update
24766 seen_label if seen_label is already set.
24768 2018-03-20 Jakub Jelinek <jakub@redhat.com>
24771 * config/i386/i386.c (fold_builtin_cpu): For features above 31
24772 use __cpu_features2 variable instead of __cpu_model.__cpu_features[0].
24773 Use 1U instead of 1. Formatting fixes.
24776 * builtins.c (fold_builtin_strpbrk): For strpbrk(x, "") use type
24777 instead of TREE_TYPE (s1) for the return value.
24779 2018-03-19 Jakub Jelinek <jakub@redhat.com>
24781 PR tree-optimization/84946
24782 * gimple-ssa-store-merging.c (mem_valid_for_store_merging): Compute
24783 bitsize + bitsize in poly_uint64 rather than poly_int64.
24786 * dwarf2asm.c: Include fold-const.c.
24787 (dw2_output_indirect_constant_1): Set DECL_INITIAL (decl) to ADDR_EXPR
24788 of decl rather than decl itself.
24790 PR rtl-optimization/84643
24791 * memmodel.h (enum memmodel): Add MEMMODEL_MAX enumerator.
24793 2018-03-19 Maxim Ostapenko <m.ostapenko@samsung.com>
24796 * dwarf2asm.c (dw2_output_indirect_constant_1): Disable ASan before
24797 calling assemble_variable.
24799 2018-03-19 Sudakshina Das <sudi.das@arm.com>
24802 * config/aarch64/aarch64-simd.md (vec_cmp<mode><v_int_equiv>): Modify
24803 instructions for UNLT, UNLE, UNGT, UNGE, UNEQ, UNORDERED and ORDERED.
24805 2018-03-19 Jim Wilson <jimw@sifive.com>
24808 * config/riscv/riscv.c (riscv_function_arg_boundary): Use
24809 PREFERRED_STACK_BOUNDARY instead of STACK_BOUNDARY.
24810 (riscv_first_stack_step): Likewise.
24811 (riscv_option_override): Use STACK_BOUNDARY instead of
24812 MIN_STACK_BOUNDARY.
24813 * config/riscv/riscv.h (STACK_BOUNDARY): Renamed from
24814 MIN_STACK_BOUNDARY.
24815 (BIGGEST_ALIGNMENT): Set to 128.
24816 (PREFERRED_STACK_BOUNDARY): Renamed from STACK_BOUNDARY.
24817 (RISCV_STACK_ALIGN): Use PREFERRED_STACK_BOUNDARY instead of
24820 2018-03-19 Richard Biener <rguenther@suse.de>
24822 PR tree-optimization/84933
24823 * tree-vrp.c (set_and_canonicalize_value_range): Treat out-of-bound
24824 values as -INF/INF when canonicalizing an ANTI_RANGE to a RANGE.
24826 2018-03-19 Richard Biener <rguenther@suse.de>
24828 PR tree-optimization/84859
24829 * tree-ssa-phiopt.c (single_trailing_store_in_bb): New function.
24830 (cond_if_else_store_replacement): Perform sinking operation on
24831 single-store BBs regardless of MAX_STORES_TO_SINK setting.
24832 Generalize what a BB with a single eligible store is.
24834 2018-03-19 Richard Biener <rguenther@suse.de>
24836 PR tree-optimization/84929
24837 * tree-data-ref.c (analyze_siv_subscript_cst_affine): Guard
24838 chrec_is_positive against non-chrec arg.
24840 2018-03-19 Tamar Christina <tamar.christina@arm.com>
24843 * config/arm/arm.c (arm_can_change_mode_class): revert r258554.
24845 2018-03-18 Martin Liska <mliska@suse.cz>
24847 PR rtl-optimization/84635
24848 * regrename.c (build_def_use): Use matches_mode only when
24851 2018-03-18 Richard Sandiford <richard.sandiford@linaro.org>
24853 PR tree-optimization/84913
24854 * tree-vect-loop.c (vectorizable_reduction): Don't try to
24855 vectorize chains of COND_EXPRs.
24857 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
24859 * config/nds32/nds32.h (MAX_REGS_PER_ADDRESS): Fix the value.
24861 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
24863 * config/nds32/nds32.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Define.
24865 2018-03-18 Chung-Ju Wu <jasonwucj@gmail.com>
24867 * config/nds32/nds32.h (CLZ_DEFINED_VALUE_AT_ZERO): Define.
24869 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
24870 Kito Cheng <kito.cheng@gmail.com>
24872 * config/nds32/nds32-protos.h (nds32_adjust_reg_alloc_order): Declare.
24873 * config/nds32/nds32.c (nds32_reg_alloc_order_for_speed): New array.
24874 (nds32_adjust_reg_alloc_order): New function.
24875 * config/nds32/nds32.h (ADJUST_REG_ALLOC_ORDER): Define.
24877 2018-03-17 Kito Cheng <kito.cheng@gmail.com>
24879 * config/nds32/nds32.c (nds32_asm_output_mi_thunk,
24880 nds32_print_operand, nds32_print_operand_address): Use
24881 HOST_WIDE_INT_PRINT_DEC instead.
24883 2018-03-17 Chung-Ju Wu <jasonwucj@gmail.com>
24885 * config/nds32/nds32.c (nds32_register_priority): Modify cost.
24887 2018-03-17 Jakub Jelinek <jakub@redhat.com>
24890 * config/i386/i386.c (initial_ix86_tune_features,
24891 initial_ix86_arch_features): Use unsigned HOST_WIDE_INT rather than
24892 unsigned long long.
24893 (set_ix86_tune_features): Change ix86_tune_mask from unsigned int
24894 to unsigned HOST_WIDE_INT, initialize to HOST_WIDE_INT_1U << ix86_tune
24895 rather than 1u << ix86_tune. Formatting fix.
24896 (ix86_option_override_internal): Change ix86_arch_mask from
24897 unsigned int to unsigned HOST_WIDE_INT, initialize to
24898 HOST_WIDE_INT_1U << ix86_arch rather than 1u << ix86_arch.
24899 (ix86_function_specific_restore): Likewise.
24901 2018-03-16 Jakub Jelinek <jakub@redhat.com>
24904 * postreload.c (reload_combine_recognize_pattern): Perform
24905 INTVAL addition in unsigned HOST_WIDE_INT type to avoid UB and
24906 truncate_int_for_mode the result for the destination's mode.
24909 * hsa-gen.c (mem_type_for_type): Fix comment typo.
24910 * tree-vect-loop-manip.c (vect_create_cond_for_niters_checks):
24912 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
24915 2018-03-16 Vladimir Makarov <vmakarov@redhat.com>
24918 * lra-assigns.c (lra_split_hard_reg_for): Don't use
24919 regno_allocno_class_array and sorted_pseudos.
24920 * lra-constraints.c (spill_hard_reg_in_range): Ignore hard regs in
24921 insns where regno is used.
24923 2018-03-16 Martin Liska <mliska@suse.cz>
24926 * multiple_target.c (create_dispatcher_calls): Redirect
24927 reference in the symbol table.
24929 2018-03-16 Martin Liska <mliska@suse.cz>
24932 * multiple_target.c (create_dispatcher_calls): Redirect also
24935 2018-03-16 Jakub Jelinek <jakub@redhat.com>
24939 * tree.h (TARGET_EXPR_NO_ELIDE): Define.
24940 * gimplify.c (gimplify_modify_expr_rhs): Don't elide TARGET_EXPRs with
24941 TARGET_EXPR_NO_ELIDE flag set unless *expr_p is INIT_EXPR.
24943 2018-03-16 Julia Koval <julia.koval@intel.com>
24945 * doc/invoke.texi (Skylake Server): Add CLWB.
24946 Cannonlake): Remove CLWB.
24948 2018-03-16 Jakub Jelinek <jakub@redhat.com>
24950 PR tree-optimization/84841
24951 * tree-ssa-reassoc.c (INTEGER_CONST_TYPE): Change to 1 << 4 from
24953 (FLOAT_ONE_CONST_TYPE): Define.
24954 (constant_type): Return FLOAT_ONE_CONST_TYPE for -1.0 and 1.0.
24955 (sort_by_operand_rank): Put entries with higher constant_type last
24956 rather than first to match comments.
24958 2018-03-15 Sandra Loosemore <sandra@codesourcery.com>
24960 * config/nios2/nios2.md (movsi_internal): Fix thinko in
24963 2018-03-15 Jakub Jelinek <jakub@redhat.com>
24966 * calls.c (expand_call): For TREE_ADDRESSABLE rettype ignore alignment
24967 check and use address of target always.
24969 2018-03-15 H.J. Lu <hongjiu.lu@intel.com>
24972 * config/i386/i386.c (indirect_thunk_needed): Update comments.
24973 (indirect_thunk_bnd_needed): Likewise.
24974 (indirect_thunks_used): Likewise.
24975 (indirect_thunks_bnd_used): Likewise.
24976 (indirect_return_needed): New.
24977 (indirect_return_bnd_needed): Likewise.
24978 (output_indirect_thunk_function): Add a bool argument for
24980 (output_indirect_thunk_function): Don't generate alias for
24981 function return thunk.
24982 (ix86_code_end): Call output_indirect_thunk_function to generate
24983 function return thunks.
24984 (ix86_output_function_return): Set indirect_return_bnd_needed
24985 and indirect_return_needed instead of indirect_thunk_bnd_needed
24986 and indirect_thunk_needed.
24988 2018-03-15 Olga Makhotina <olga.makhotina@intel.com>
24990 * config/i386/sgxintrin.h (_enclv_u32): New intrinsic.
24991 (__enclv_bc, __enclv_cd, __enclv_generic): New definitions.
24992 (ERDINFO, ETRACKC, ELDBC, ELDUC): New leaves.
24994 2018-03-15 David Malcolm <dmalcolm@redhat.com>
24995 Paul Hua <paul.hua.gm@gmail.com>
24998 * gcc.dg/fixits-pr84852-1.c: Fix filename in dg-regexp.
25000 2018-03-15 Segher Boessenkool <segher@kernel.crashing.org>
25002 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add back the
25003 TARGET_DOUBLE_FLOAT and TARGET_SINGLE_FLOAT conditions on the DFmode
25004 resp. SFmode cases.
25006 2018-03-15 Tamar Christina <tamar.christina@arm.com>
25009 * config/arm/arm.c (arm_can_change_mode_class): Use GET_MODE_UNIT_SIZE
25010 instead of GET_MODE_SIZE when comparing Units.
25012 2018-03-15 Vladimir Mezentsev <vladimir.mezentsev@oracle.com>
25015 * varasm.c (hash_section): Return an unchangeble hash value
25016 * config/aarch64/aarch64.c (aarch64_use_blocks_for_constant_p):
25017 Return !aarch64_can_use_per_function_literal_pools_p ().
25019 2018-03-15 Jakub Jelinek <jakub@redhat.com>
25022 * optabs.c (emit_conditional_move): Pass address of cmode's copy
25023 rather than address of cmode as last argument to prepare_cmp_insn.
25025 2018-03-15 Julia Koval <julia.koval@intel.com>
25027 * config/i386/i386.c (F_AVX512VBMI2, F_GFNI, F_VPCLMULQDQ,
25028 F_AVX512VNNI, F_AVX512BITALG): New.
25030 2018-03-14 John David Anglin <danglin@gcc.gnu.org>
25033 * config/pa/pa.c (pa_emit_move_sequence): Always emit secondary reload
25034 insn for floating-point loads and stores.
25036 2018-03-14 Carl Love <cel@us.ibm.com>
25038 * config/rs6000/rs6000-c.c: Add macro definitions for
25039 ALTIVEC_BUILTIN_VEC_PERMXOR.
25040 * config/rs6000/rs6000.h: Add #define for vec_permxor builtin.
25041 * config/rs6000/rs6000-builtin.def: Add macro expansions for VPERMXOR.
25042 * config/rs6000/altivec.md (altivec_vpermxor): New define expand.
25043 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Add case
25045 * config/doc/extend.texi: Add prototypes for vec_permxor.
25047 2018-03-14 David Malcolm <dmalcolm@redhat.com>
25050 * diagnostic-show-locus.c (class layout_point): Convert m_line
25051 from int to linenum_type.
25052 (line_span::comparator): Use linenum "compare" function when
25053 comparing line numbers.
25054 (test_line_span): New function.
25055 (layout_range::contains_point): Convert param "row" from int to
25057 (layout_range::intersects_line_p): Likewise.
25058 (layout::will_show_line_p): Likewise.
25059 (layout::print_source_line): Likewise.
25060 (layout::should_print_annotation_line_p): Likewise.
25061 (layout::print_annotation_line): Likewise.
25062 (layout::print_leading_fixits): Likewise.
25063 (layout::annotation_line_showed_range_p): Likewise.
25064 (struct line_corrections): Likewise for field m_row.
25065 (line_corrections::line_corrections): Likewise for param "row".
25066 (layout::print_trailing_fixits): Likewise.
25067 (layout::get_state_at_point): Likewise.
25068 (layout::get_x_bound_for_row): Likewise.
25069 (layout::print_line): Likewise.
25070 (diagnostic_show_locus): Likewise for locals "last_line" and "row".
25071 (selftest::diagnostic_show_locus_c_tests): Call test_line_span.
25072 * input.c (selftest::test_linenum_comparisons): New function.
25073 (selftest::input_c_tests): Call it.
25074 * selftest.c (selftest::test_assertions): Test ASSERT_GT,
25075 ASSERT_GT_AT, ASSERT_LT, and ASSERT_LT_AT.
25076 * selftest.h (ASSERT_GT): New macro.
25077 (ASSERT_GT_AT): New macro.
25078 (ASSERT_LT): New macro.
25079 (ASSERT_LT_AT): New macro.
25081 2018-03-14 Segher Boessenkool <segher@kernel.crashing.org>
25083 PR rtl-optimization/84780
25084 * combine.c (distribute_links): Don't make a link based on pc_rtx.
25086 2018-03-14 Martin Liska <mliska@suse.cz>
25088 * tree.c (record_node_allocation_statistics): Use
25089 get_stats_node_kind.
25090 (get_stats_node_kind): New function extracted from
25091 record_node_allocation_statistics.
25092 (free_node): Use get_stats_node_kind.
25094 2018-03-14 Richard Biener <rguenther@suse.de>
25096 * tree-ssa-pre.c (compute_antic_aux): Remove code that asserts
25097 that the value-set of ANTIC_IN doesn't grow.
25100 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
25102 (BB_VISITED_WITH_VISITED_SUCCS): New define.
25103 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
25105 2018-03-14 Julia Koval <julia.koval@intel.com>
25107 * config.gcc (icelake-client, icelake-server): New.
25109 * config/i386/i386.c (initial_ix86_tune_features): Extend to 64 bit.
25110 (initial_ix86_arch_features): Ditto.
25111 (PTA_SKYLAKE): Add SGX.
25112 (PTA_ICELAKE): Remove.
25113 (PTA_ICELAKE_CLIENT): New.
25114 (PTA_ICELAKE_SERVER): New.
25115 (ix86_option_override_internal): Split up icelake on icelake client and
25117 (get_builtin_code_for_version): Ditto.
25118 (fold_builtin_cpu): Ditto.
25119 * config/i386/driver-i386.c (config/i386/driver-i386.c): Ditto.
25120 * config/i386/i386-c.c (ix86_target_macros_internal): Ditto
25121 * config/i386/i386.h (processor_type): Ditto.
25122 * doc/invoke.texi: Ditto.
25124 2018-03-14 Jakub Jelinek <jakub@redhat.com>
25127 * sanopt.c (maybe_optimize_ubsan_ptr_ifn): Handle also
25128 INTEGER_CST offset, add it together with bitpos / 8 and
25129 sign extend based on POINTER_SIZE.
25133 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
25136 * config/i386/constraints.md (Yc): New register constraint.
25137 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
25138 Use Yc constraint for alternative 2 of operand 0. Remove
25139 preferred_for_speed attribute.
25141 2018-03-14 Richard Biener <rguenther@suse.de>
25143 PR tree-optimization/84830
25144 * tree-ssa-pre.c (compute_antic_aux): Intersect the new ANTIC_IN
25145 with the old one to avoid oscillations.
25147 2018-03-13 Vladimir Makarov <vmakarov@redhat.com>
25150 * lra-assigns.c (find_all_spills_for): Ignore uninteresting
25152 (assign_by_spills): Return a flag of reload assignment failure.
25153 Do not process the reload assignment failures. Do not spill other
25154 reload pseudos if they has the same reg class. Update n if
25156 (lra_assign): Add a return arg. Set up from the result of
25157 assign_by_spills call.
25158 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
25159 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
25160 usage_insns if it is not NULL.
25161 (spill_hard_reg_in_range): New function.
25162 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
25163 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
25164 function prototypes.
25165 (lra_assign): Change prototype.
25166 * lra.c (lra): Add code to deal with fails by splitting hard reg
25169 2018-03-01 Palmer Dabbelt <palmer@sifive.com>
25171 * config/riscv/riscv.opt (mrelax): New option.
25172 * config/riscv/riscv.c (riscv_file_start): Emit ".option
25173 "norelax" when riscv_mrelax is disabled.
25174 * doc/invoke.texi (RISC-V): Document "-mrelax" and "-mno-relax".
25176 2018-03-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
25179 * config/rs6000/rs6000.c (rs6000_reassociation_width): Disable parallel
25180 reassociation for int modes.
25182 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
25184 * tree-vect-loop-manip.c (vect_maybe_permute_loop_masks):
25185 Reverse the choice between VEC_UNPACK_LO_EXPR and VEC_UNPACK_HI_EXPR
25187 * config/aarch64/iterators.md (hi_lanes_optab): New int attribute.
25188 * config/aarch64/aarch64-sve.md
25189 (*aarch64_sve_<perm_insn><perm_hilo><mode>): Rename to...
25190 (aarch64_sve_<perm_insn><perm_hilo><mode>): ...this.
25191 (*extend<mode><Vwide>2): Rename to...
25192 (aarch64_sve_extend<mode><Vwide>2): ...this.
25193 (vec_unpack<su>_<perm_hilo>_<mode>): Turn into a define_expand,
25194 renaming the old pattern to...
25195 (aarch64_sve_punpk<perm_hilo>_<mode>): ...this. Only define
25197 (vec_unpack<su>_<perm_hilo>_<SVE_BHSI:mode>): Turn into a
25198 define_expand, renaming the old pattern to...
25199 (aarch64_sve_<su>unpk<perm_hilo>_<SVE_BHSI:mode>): ...this.
25200 (*vec_unpacku_<perm_hilo>_<mode>_no_convert): Delete.
25201 (vec_unpacks_<perm_hilo>_<mode>): Take BYTES_BIG_ENDIAN into
25202 account when deciding which SVE instruction the optab should use.
25203 (vec_unpack<su_optab>_float_<perm_hilo>_vnx4si): Likewise.
25205 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
25207 * config/aarch64/aarch64.md (V4_REGNUM, V8_REGNUM, V12_REGNUM)
25208 (V20_REGNUM, V24_REGNUM, V28_REGNUM, P1_REGNUM, P2_REGNUM, P3_REGNUM)
25209 (P4_REGNUM, P5_REGNUM, P6_REGNUM, P8_REGNUM, P9_REGNUM, P10_REGNUM)
25210 (P11_REGNUM, P12_REGNUM, P13_REGNUM, P14_REGNUM): New define_constants.
25211 (tlsdesc_small_<mode>): Turn a define_expand and use
25212 tlsdesc_small_sve_<mode> for SVE. Rename original define_insn to...
25213 (tlsdesc_small_advsimd_<mode>): ...this.
25214 (tlsdesc_small_sve_<mode>): New pattern.
25216 2018-03-13 Richard Sandiford <richard.sandiford@linaro.org>
25218 * config/aarch64/iterators.md (UNSPEC_SMUL_HIGHPART)
25219 (UNSPEC_UMUL_HIGHPART): New constants.
25220 (MUL_HIGHPART): New int iteraor.
25221 (su): Handle UNSPEC_SMUL_HIGHPART and UNSPEC_UMUL_HIGHPART.
25222 * config/aarch64/aarch64-sve.md (<su>mul<mode>3_highpart): New
25224 (*<su>mul<mode>3_highpart): New define_insn.
25226 2018-03-13 Eric Botcazou <ebotcazou@adacore.com>
25229 * ipa-devirt.c (odr_subtypes_equivalent_p): Do not get the ODR type of
25232 2018-03-13 Martin Liska <mliska@suse.cz>
25235 * (sem_item_optimizer::sem_item_optimizer): Initialize new
25237 (sem_item_optimizer::~sem_item_optimizer): Release it.
25238 (sem_item_optimizer::merge_classes): Register variable aliases.
25239 (sem_item_optimizer::fixup_pt_set): New function.
25240 (sem_item_optimizer::fixup_points_to_sets): Likewise.
25241 * ipa-icf.h: Declare new variables and functions.
25243 2018-03-13 Jakub Jelinek <jakub@redhat.com>
25245 PR middle-end/84834
25246 * match.pd ((A & C) != 0 ? D : 0): Use INTEGER_CST@2 instead of
25247 integer_pow2p@2 and test integer_pow2p in condition.
25248 (A < 0 ? C : 0): Similarly for @1.
25250 PR middle-end/84831
25251 * stmt.c (parse_output_constraint): If the CONSTRAINT_LEN (*p, p)
25252 characters starting at p contain '\0' character, don't look beyond
25256 * config/i386/i386.md (round<mode>2): For 387 fancy math, disable
25257 pattern if -ftrapping-math -fno-fp-int-builtin-inexact.
25260 * reg-stack.c (change_stack): Change update_end var from int to
25261 rtx_insn *, if non-NULL don't update just BB_END (current_block), but
25262 also call set_block_for_insn on the newly added insns and rescan.
25265 * config/i386/sse.md (sse2_loadhpd): Use Yv constraint rather than v
25266 on the last operand.
25269 * tree.c (stabilize_reference_1): Return save_expr (e) for
25270 STATEMENT_LIST even if it doesn't have side-effects.
25272 2018-03-12 Jonathan Wakely <jwakely@redhat.com>
25274 * doc/invoke.texi (-mclflushopt): Fix spelling of option.
25276 2018-03-12 Renlin Li <renlin.li@arm.com>
25278 * config/aarch64/aarch64.md (movhf_aarch64): Fix mode argument to
25279 aarch64_output_scalar_simd_mov_immediate.
25281 2018-03-12 Martin Sebor <msebor@redhat.com>
25283 PR tree-optimization/83456
25284 * gimple-fold.c (gimple_fold_builtin_memory_op): Avoid warning
25285 for perfectly overlapping calls to memcpy.
25286 (gimple_fold_builtin_memory_chk): Same.
25287 (gimple_fold_builtin_strcpy): Handle no-warning.
25288 (gimple_fold_builtin_stxcpy_chk): Same.
25289 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Handle no-warning.
25291 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
25293 * config/rs6000/rs6000.c (abi_v4_pass_in_fpr): Add bool "named"
25294 parameter. Use it for SFmode.
25295 (rs6000_function_arg_advance_1): Adjust.
25296 (rs6000_function_arg): Adjust.
25297 (rs6000_gimplify_va_arg): Pass false for that new parameter.
25299 2018-03-12 Segher Boessenkool <segher@kernel.crashing.org>
25301 PR rtl-optimization/84169
25302 PR rtl-optimization/84780
25303 * combine.c (can_combine_p): Check for a 2-insn combination whether
25304 the destination register is used between the two insns, too.
25306 2018-03-12 Richard Biener <rguenther@suse.de>
25308 PR tree-optimization/84803
25309 * tree-if-conv.c (ifcvt_memrefs_wont_trap): Don't do anything
25310 for refs DR analysis didn't process.
25312 2018-03-12 Richard Biener <rguenther@suse.de>
25314 PR tree-optimization/84777
25315 * tree-ssa-loop-ch.c (should_duplicate_loop_header_p): For
25316 force-vectorize loops ignore whether we are optimizing for size.
25318 2018-03-12 Chung-Ju Wu <jasonwucj@gmail.com>
25320 * config/nds32/nds32.c (nds32_md_asm_adjust): New function.
25321 (TARGET_MD_ASM_ADJUST): Define.
25323 2018-03-12 Monk Chiang <sh.chiang04@gmail.com>
25324 Kito Cheng <kito.cheng@gmail.com>
25325 Chung-Ju Wu <jasonwucj@gmail.com>
25327 * config/nds32/nds32.c (nds32_compute_stack_frame,
25328 nds32_emit_stack_push_multiple, nds32_emit_stack_pop_multiple,
25329 nds32_emit_stack_v3push, nds32_emit_stack_v3pop,
25330 nds32_emit_adjust_frame, nds32_expand_prologue, nds32_expand_epilogue,
25331 nds32_expand_prologue_v3push, nds32_expand_epilogue_v3pop): Refine.
25332 * config/nds32/nds32.h (NDS32_FIRST_CALLEE_SAVE_GPR_REGNUM,
25333 NDS32_LAST_CALLEE_SAVE_GPR_REGNUM, NDS32_V3PUSH_AVAILABLE_P): New.
25334 * config/nds32/nds32.md (prologue, epilogue): Use macro
25335 NDS32_V3PUSH_AVAILABLE_P to do checking.
25337 2018-03-11 Jakub Jelinek <jakub@redhat.com>
25340 * dwarf2out.c (gen_enumeration_type_die): Don't guard adding
25341 DW_AT_declaration for ENUM_IS_OPAQUE on -gdwarf-4 or -gno-strict-dwarf,
25342 but on TYPE_SIZE. Don't do anything for ENUM_IS_OPAQUE if not creating
25343 a new die. Don't set TREE_ASM_WRITTEN if ENUM_IS_OPAQUE. Guard
25344 addition of most attributes on !orig_type_die or the attribute not
25345 being present already. Assert TYPE_VALUES is NULL for ENUM_IS_OPAQUE.
25347 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
25348 Chung-Ju Wu <jasonwucj@gmail.com>
25350 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): Modify to define
25351 __NDS32_VH__ macro.
25352 * config/nds32/nds32.opt (mvh): New option.
25354 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
25355 Chung-Ju Wu <jasonwucj@gmail.com>
25357 * config/nds32/nds32-protos.h (nds32_cpu_cpp_builtins): Declare
25359 * config/nds32/nds32.c (nds32_cpu_cpp_builtins): New function.
25360 * config/nds32/nds32.h (TARGET_CPU_CPP_BUILTINS): Modify its
25363 2018-03-11 Kito Cheng <kito.cheng@gmail.com>
25364 Chung-Ju Wu <jasonwucj@gmail.com>
25366 * config/nds32/nds32-memory-manipulation.c (nds32_expand_strlen): New
25368 * config/nds32/nds32-multiple.md (strlensi): New pattern.
25369 * config/nds32/nds32-protos.h (nds32_expand_strlen): Declare function.
25371 2018-03-11 Monk Chiang <sh.chiang04@gmail.com>
25372 Kito Cheng <kito.cheng@gmail.com>
25373 Chung-Ju Wu <jasonwucj@gmail.com>
25375 * config/nds32/constants.md (unspec_element): Add UNSPEC_FFB,
25376 UNSPEC_FFMISM and UNSPEC_FLMISM.
25377 * config/nds32/nds32-intrinsic.c (bdesc_2arg): Add builtin description
25378 for ffb, ffmism and flmism.
25379 * config/nds32/nds32-intrinsic.md (unspec_ffb): Define new pattern.
25380 (unspec_ffmism): Ditto.
25381 (unspec_flmism): Ditto.
25382 (nds32_expand_builtin_impl): Check if string extension is available.
25383 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_FFB,
25384 NDS32_BUILTIN_FFMISM and NDS32_BUILTIN_FLMISM.
25386 2018-03-10 Vladimir Makarov <vmakarov@redhat.com>
25389 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
25392 * lra-assigns.c (assign_by_spills): Return a flag of reload
25393 assignment failure. Do not process the reload assignment
25394 failures. Do not spill other reload pseudos if they has the same
25396 (lra_assign): Add a return arg. Set up from the result of
25397 assign_by_spills call.
25398 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
25399 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
25400 usage_insns if it is not NULL.
25401 (spill_hard_reg_in_range): New function.
25402 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
25403 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
25404 function prototypes.
25405 (lra_assign): Change prototype.
25406 * lra.c (lra): Add code to deal with fails by splitting hard reg
25409 2018-03-10 H.J. Lu <hongjiu.lu@intel.com>
25412 * config/i386/i386.opt: Replace Enforcment with Enforcement.
25414 2018-03-10 Alexandre Oliva <aoliva@redhat.com>
25417 * dwarf2out.h (dw_val_class): Add dw_val_class_symview.
25418 (dw_val_node): Add val_symbolic_view.
25419 * dwarf2out.c (dw_line_info_table): Add symviews_since_reset.
25420 (symview_upper_bound): New.
25421 (new_line_info_table): Initialize symviews_since_reset.
25422 (dwarf2out_source_line): Count symviews_since_reset and set
25423 symview_upper_bound.
25424 (dw_val_equal_p): Handle symview.
25425 (add_AT_symview): New.
25426 (print_dw_val): Handle symview.
25427 (attr_checksum, attr_checksum_ordered): Likewise.
25428 (same_dw_val_p, size_of_die): Likewise.
25429 (value_format, output_die): Likewise.
25430 (add_high_low_attributes): Use add_AT_symview for entry_view.
25431 (dwarf2out_finish): Reset symview_upper_bound, clear
25434 2018-03-09 Peter Bergner <bergner@vnet.ibm.com>
25437 * config/rs6000/rs6000.c (rs6000_offsettable_memref_p): New prototype.
25438 Add strict argument and use it.
25439 (rs6000_split_multireg_move): Update for new strict argument.
25440 (mem_operand_gpr): Disallow all non-offsettable addresses.
25441 * config/rs6000/rs6000.md (*movdi_internal64): Use YZ constraint.
25443 2018-03-09 Jakub Jelinek <jakub@redhat.com>
25446 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Mark va_arg_tmp
25447 temporary TREE_ADDRESSABLE before gimplification of BUILT_IN_MEMCPY.
25448 * config/powerpcspe/powerpcspe.c (rs6000_gimplify_va_arg): Likewise.
25451 * tree-inline.c (copy_tree_body_r): For INDIRECT_REF of a remapped
25452 decl, use remap_type if we want to use the type.
25454 2018-03-09 Martin Sebor <msebor@redhat.com>
25456 PR tree-optimization/84526
25457 * gimple-ssa-warn-restrict.c (builtin_memref::set_base_and_offset):
25459 (builtin_access::generic_overlap): Be prepared to handle non-array
25462 2018-03-09 Alexandre Oliva <aoliva@redhat.com>
25464 PR rtl-optimization/84682
25465 * lra-constraints.c (process_address_1): Check is_address flag
25466 for address constraints.
25467 (process_alt_operands): Likewise.
25468 * lra.c (lra_set_insn_recog_data): Pass asm operand locs to
25469 preprocess_constraints.
25470 * recog.h (preprocess_constraints): Add oploc parameter.
25472 * recog.c (preprocess_constraints): Test address_operand for
25473 CT_ADDRESS constraints.
25475 2018-03-09 Vladimir Makarov <vmakarov@redhat.com>
25478 * lra-assigns.c (assign_by_spills): Return a flag of reload
25479 assignment failure. Do not process the reload assignment
25480 failures. Do not spill other reload pseudos if they has the same
25482 (lra_assign): Add a return arg. Set up from the result of
25483 assign_by_spills call.
25484 (find_reload_regno_insns, lra_split_hard_reg_for): New functions.
25485 * lra-constraints.c (split_reg): Add a new arg. Use it instead of
25486 usage_insns if it is not NULL.
25487 (spill_hard_reg_in_range): New function.
25488 (split_if_necessary, inherit_in_ebb): Pass a new arg to split_reg.
25489 * lra-int.h (spill_hard_reg_in_range, lra_split_hard_reg_for): New
25490 function prototypes.
25491 (lra_assign): Change prototype.
25492 * lra.c (lra): Add code to deal with fails by splitting hard reg
25495 2018-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25498 * common/config/arm/arm-common.c (arm_parse_arch_option_name):
25499 Accept complain bool parameter. Only emit errors if it is true.
25500 (arm_parse_cpu_option_name): Likewise.
25501 (arm_target_thumb_only): Adjust callers of the above.
25502 * config/arm/arm-protos.h (arm_parse_cpu_option_name): Adjust
25503 prototype to take a default true bool parameter.
25504 (arm_parse_arch_option_name): Likewise.
25506 2018-03-09 David Malcolm <dmalcolm@redhat.com>
25507 Francois-Xavier Coudert <fxcoudert@gcc.gnu.org>
25511 * Makefile.in (LD_VERSION_SCRIPT_OPTION, LD_SONAME_OPTION): New.
25512 * configure: Regenerate.
25513 * configure.ac ("linker --version-script option"): New.
25514 ("linker soname option"): New.
25516 2018-03-09 Richard Biener <rguenther@suse.de>
25518 PR tree-optimization/84775
25519 * tree-if-conv.c (add_bb_predicate_gimplified_stmts): Delink
25520 immediate uses of predicate stmts and mark them modified.
25523 PR tree-optimization/84178
25524 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
25526 (version_loop_for_if_conversion): Delay update_ssa call.
25527 (tree_if_conversion): Delay update_ssa until after predicate
25530 2018-03-09 Eric Botcazou <ebotcazou@adacore.com>
25533 * config/i386/winnt.c (i386_pe_seh_cold_init): Use small pre-allocation
25534 when the function accesses prior frames.
25536 2018-03-08 Jakub Jelinek <jakub@redhat.com>
25539 * dwarf2out.c (dw_loc_list): If list && loc_list->first->next, call
25540 gen_llsym, otherwise call maybe_gen_llsym.
25542 PR inline-asm/84742
25543 * recog.c (asm_operand_ok): Return 0 if multi-character constraint
25544 has ',' character inside of it.
25546 2018-03-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
25549 * config/aarch64/aarch64.md (*compare_cstore<mode>_insn): Mark pattern
25550 as clobbering CC_REGNUM.
25552 2018-03-08 Richard Biener <rguenther@suse.de>
25554 PR middle-end/84552
25555 * tree-scalar-evolution.c: Include tree-into-ssa.h.
25556 (follow_copies_to_constant): Do not follow SSA names registered
25559 2018-03-08 Richard Biener <rguenther@suse.de>
25561 PR tree-optimization/84178
25562 * tree-if-conv.c (combine_blocks): Move insert_gimplified_predicates
25564 (version_loop_for_if_conversion): Delay update_ssa call.
25565 (tree_if_conversion): Delay update_ssa until after predicate
25568 2018-03-08 David Malcolm <dmalcolm@redhat.com>
25570 PR tree-optimization/84178
25571 * tree-if-conv.c (release_bb_predicate): Remove the
25572 the assertion that the stmts have NULL use_ops.
25573 Discard the statements, asserting that they haven't
25574 yet been added to a BB.
25576 2018-03-08 Richard Biener <rguenther@suse.de>
25578 PR tree-optimization/84746
25579 * tree-ssa-pre.c (find_leader_in_sets): Deal with SET1 being NULL.
25580 (phi_translate): Pass in destination ANTIC_OUT set.
25581 (phi_translate_1): Likewise. For a simplified result lookup
25582 a leader in ANTIC_OUT and AVAIL_OUT, not the ANTIC_IN sets.
25583 (phi_translate_set): Adjust.
25584 (do_pre_regular_insertion): Likewise.
25585 (do_pre_partial_partial_insertion): Likewise.
25587 2018-03-08 Martin Liska <mliska@suse.cz>
25589 PR gcov-profile/84735
25590 * doc/gcov.texi: Document usage of profile files.
25591 * gcov-io.h: Document changes in the format.
25593 2018-03-08 Alexandre Oliva <aoliva@redhat.com>
25597 * dwarf2out.c (struct dw_line_info_table): Update comments for
25599 (FORCE_RESET_NEXT_VIEW): New.
25600 (FORCE_RESETTING_VIEW_P): New.
25601 (RESETTING_VIEW_P): Check for -1 too.
25602 (ZERO_VIEW_P): Likewise.
25603 (new_line_info_table): Force-reset next view.
25604 (dwarf2out_begin_function): Likewise.
25605 (dwarf2out_source_line): Simplify zero_view_p initialization.
25606 Test FORCE_RESETTING_VIEW_P and RESETTING_VIEW_P instead of
25607 view directly. Omit view when omitting .loc at line 0.
25609 2018-03-08 Jakub Jelinek <jakub@redhat.com>
25611 PR tree-optimization/84740
25612 * tree-switch-conversion.c (process_switch): Call build_constructors
25613 only if info.phi_count is non-zero.
25615 PR tree-optimization/84739
25616 * tree-tailcall.c (find_tail_calls): Check call arguments against
25617 DECL_ARGUMENTS (current_function_decl) rather than
25618 DECL_ARGUMENTS (func) when checking for tail recursion.
25620 2018-03-07 Jakub Jelinek <jakub@redhat.com>
25622 * doc/contrib.texi: Add entries for Martin Liska, David Malcolm,
25623 Marek Polacek, extend Vladimir Makarov's, Jonathan Wakely's and
25624 Volker Reichelt's entry and add entries for people that perform
25625 GCC fuzzy testing and report numerous bugs.
25627 2018-03-07 Segher Boessenkool <segher@kernel.crashing.org>
25630 * config/rs6000/rs6000.c (rs6000_elf_in_small_data_p): Don't put
25631 readonly data in sdata, if that is disabled.
25632 * config/rs6000/sysv4.opt (mreadonly-in-sdata): New option.
25633 * doc/invoke.texi (RS/6000 and PowerPC Options): Document
25634 -mreadonly-in-sdata option.
25636 2018-03-07 Martin Sebor <msebor@redhat.com>
25638 PR tree-optimization/84468
25639 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Consider successor
25640 basic block when looking for nul assignment.
25642 2018-03-07 Eric Botcazou <ebotcazou@adacore.com>
25645 * except.h (output_function_exception_table): Adjust prototype.
25646 * except.c (output_function_exception_table): Remove FNNAME parameter
25647 and add SECTION parameter. Ouput one part of the table at a time.
25648 * final.c (final_scan_insn_1) <NOTE_INSN_SWITCH_TEXT_SECTIONS>: Output
25649 the first part of the exception table and emit unwind directives.
25650 * config/i386/i386-protos.h (i386_pe_end_cold_function): Declare.
25651 (i386_pe_seh_cold_init): Likewise.
25652 * config/i386/cygming.h (ASM_DECLARE_COLD_FUNCTION_NAME): New macro.
25653 (ASM_DECLARE_COLD_FUNCTION_SIZE): Likewise.
25654 * config/i386/i386.c (x86_expand_epilogue): Fix wording in comment.
25655 (ix86_output_call_insn): Emit a nop in one more case for SEH.
25656 * config/i386/winnt.c: Include except.h.
25657 (struct seh_frame_state): Add reg_offset, after_prologue and
25658 in_cold_section fields.
25659 (i386_pe_seh_end_prologue): Set seh->after_prologue.
25660 (i386_pe_seh_cold_init): New function.
25661 (i386_pe_seh_fini): Add COLD parameter and bail out if it is not equal
25662 to seh->in_cold_section.
25663 (seh_emit_push): Record the offset of the push.
25664 (seh_emit_save): Record the offet of the save.
25665 (i386_pe_seh_unwind_emit): Deal with NOTE_INSN_SWITCH_TEXT_SECTIONS.
25666 Test seh->after_prologue to disregard the epilogue.
25667 (i386_pe_end_function): Pass FALSE to i386_pe_seh_fini.
25668 (i386_pe_end_cold_function): New function.
25670 2018-03-07 Jakub Jelinek <jakub@redhat.com>
25673 * config/aarch64/predicates.md (aarch64_simd_reg_or_zero): Use
25674 aarch64_simd_or_scalar_imm_zero rather than aarch64_simd_imm_zero.
25677 * gimple-expr.c (create_tmp_var_raw): Set DECL_NAMELESS flag
25679 * tree-pretty-print.c (dump_decl_name): For TDF_COMPARE_DEBUG,
25680 don't print names of DECL_NAMELESS DECL_IGNORED_P decls.
25682 PR middle-end/84723
25683 * multiple_target.c: Include tree-inline.h and intl.h.
25684 (expand_target_clones): Diagnose and fail if node->definition and
25685 !tree_versionable_function_p (node->decl).
25687 2018-03-06 John David Anglin <danglin@gcc.gnu.org>
25689 * config/pa/pa.h (ASM_GENERATE_INTERNAL_LABEL): Revise to use
25691 (ASM_OUTPUT_ADDR_VEC_ELT): Revise for above change.
25692 (ASM_OUTPUT_ADDR_DIFF_ELT): Likewise.
25693 * config/pa/pa64-hpux.h (ASM_GENERATE_INTERNAL_LABEL): Revise as above.
25695 2018-03-06 Jakub Jelinek <jakub@redhat.com>
25698 * combine.c (try_combine): Use reg_or_subregno instead of handling
25699 just paradoxical SUBREGs and REGs.
25701 2018-03-06 Claudiu Zissulescu <claziss@synopsys.com>
25703 * config/arc/arc.c (arc_finalize_pic): Remove function.
25704 (arc_must_save_register): We use single base PIC register, remove
25705 checks to save/restore the PIC register.
25706 (arc_expand_prologue): Likewise.
25707 * config/arc/arc-protos.h (arc_set_default_type_attributes):
25709 (arc_verify_short): Likewise.
25710 (arc_attr_type): Likewise.
25711 * config/arc/arc.c (arc_set_default_type_attributes): Remove.
25712 (walk_stores): Likewise.
25713 (arc_address_cost): Make it static.
25714 (arc_verify_short): Likewise.
25715 (branch_dest): Likewise.
25716 (arc_attr_type): Likewise.
25717 * config/arc/arc.c (TARGET_ADJUST_INSN_LENGTH): Remove.
25718 (TARGET_INSN_LENGTH_PARAMETERS): Likewise.
25719 (arc_final_prescan_insn): Remove inserting the nops due to
25720 hardware hazards. It is done in reorg step.
25721 (insn_length_variant_t): Remove.
25722 (insn_length_parameters_t): Likewise.
25723 (arc_insn_length_parameters): Likewise.
25724 (arc_get_insn_variants): Likewise.
25725 * config/arc/arc.h (TARGET_UPSIZE_DBR): Remove.
25727 2018-03-06 Jakub Jelinek <jakub@redhat.com>
25729 PR inline-asm/84683
25730 * reg-stack.c (move_for_stack_reg): If any_malformed_asm, avoid
25733 PR tree-optimization/84687
25734 * omp-simd-clone.c (simd_clone_create): Clear DECL_BUILT_IN_CLASS
25736 * match.pd (pow(C,x)*expN(y) -> expN(logN(C)*x+y)): New optimization.
25738 2018-03-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
25740 * config/rs6000/rs6000-builtin.def (rs6000_speculation_barrier):
25741 Rename to ppc_speculation_barrier.
25742 * config/rs6000/rs6000.c (rs6000_init_builtins): Rename builtin to
25743 __builtin_ppc_speculation_barrier.
25745 2018-03-05 Jakub Jelinek <jakub@redhat.com>
25748 * combine.c (combine_simplify_rtx): Don't try to simplify if
25749 if_then_else_cond returned non-NULL, but either true_rtx or false_rtx
25752 2018-03-05 Segher Boessenkool <segher@kernel.crashing.org>
25754 * config/rs6000/rs6000.c (rs6000_loop_align): Don't align tiny loops
25755 to 32 bytes when compiling for POWER9.
25757 2018-03-05 Jakub Jelinek <jakub@redhat.com>
25760 * config/i386/i386.c (ix86_function_ok_for_sibcall): Check for
25761 regparm >= 3 with no arg reg available also for calls with
25762 flag_force_indirect_call. Pass decl to ix86_function_regparm.
25765 * config/i386/sse.md (*<code><mode>3): Replace <mask_prefix3> with
25767 (*<plusminus_insn><mode>3): Likewise. Remove <mask_operand3> uses.
25769 2018-03-05 Peter Bergner <bergner@vnet.ibm.com>
25772 * config/rs6000/vector.md (mov<mode>): Disallow altivec memory operands.
25774 2018-03-05 Richard Biener <rguenther@suse.de>
25776 PR tree-optimization/84486
25777 * tree-ssa-pre.c (create_expression_by_pieces): Remove dead code.
25778 When inserting a __builtin_assume_aligned call set the LHS
25779 SSA name alignment info accordingly.
25781 2018-03-05 Wilco Dijkstra <wdijkstr@arm.com>
25783 PR tree-optimization/84114
25784 * config/aarch64/aarch64.c (aarch64_reassociation_width)
25785 Avoid reassociation of FLOAT_MODE addition.
25787 2018-03-05 Olga Makhotina <olga.makhotina@intel.com>
25789 * common/config/i386/i386-common.c (OPTION_MASK_ISA_PCONFIG_SET,
25790 OPTION_MASK_ISA_PCONFIG_UNSET, OPTION_MASK_ISA_WBNOINVD_SET,
25791 OPTION_MASK_ISA_WBNOINVD_UNSET): New definitions.
25792 (ix86_handle_option): Handle -mpconfig and -mwbnoinvd.
25793 * config.gcc (pconfigintrin.h, wbnoinvdintrin.h) : Add headers.
25794 * config/i386/cpuid.h (bit_PCONFIG, bit_WBNOINVD): New.
25795 * config/i386/driver-i386.c (host_detect_local_cpu): Detect -mpconfig
25797 * config/i386/i386-builtin.def (__builtin_ia32_wbnoinvd,
25798 __builtin_ia32_wbinvd): New builtins.
25799 (SPECIAL_ARGS2): New.
25800 * config/i386/i386-c.c (__WBNOINVD__, __PCONFIG__): New.
25801 (SPECIAL_ARGS2): New.
25802 * config/i386/i386.c (ix86_target_string): Add -mpconfig and -mwbnoinvd.
25803 (ix86_valid_target_attribute_inner_p): Ditto.
25804 (ix86_init_mmx_sse_builtins): Add special_args2.
25805 * config/i386/i386.h (TARGET_PCONFIG, TARGET_PCONFIG_P, TARGET_WBNOINVD,
25806 TARGET_WBNOINVD_P): New.
25807 * config/i386/i386.md (UNSPECV_WBINVD, UNSPECV_WBNOINVD): New.
25808 (define_insn "wbinvd", define_insn "wbnoinvd"): New.
25809 * config/i386/i386.opt: Add -mpconfig and -mwbnoinvd.
25810 * config/i386/immintrin.h (_wbinvd): New intrinsic.
25811 * config/i386/pconfigintrin.h: New file.
25812 * config/i386/wbnoinvdintrin.h: Ditto.
25813 * config/i386/x86intrin.h: Add headers pconfigintrin.h and
25815 * doc/invoke.texi (-mpconfig, -mwbnoinvd): New.
25817 2018-03-05 Richard Biener <rguenther@suse.de>
25819 PR tree-optimization/84670
25820 * tree-ssa-pre.c (struct bb_bitmap_sets): Add visited_with_visited_succs
25822 (BB_VISITED_WITH_VISITED_SUCCS): New define.
25823 (compute_antic): Initialize BB_VISITED_WITH_VISITED_SUCCS.
25824 (compute_antic_aux): Only assert the number of values in ANTIC_IN
25825 doesn't grow if all successors (recursively) were visited at least
25828 2018-03-05 Richard Biener <rguenther@suse.de>
25830 PR tree-optimization/84650
25831 * tree-ssa-loop-im.c (pass_lim::execute): Reset the SCEV cache
25832 if executed in the loop pipeline.
25834 2018-03-05 Sandra Loosemore <sandra@codesourcery.com>
25836 * doc/configfiles.texi (Configuration Files): Move info about
25837 conditionalizing $target-protos.h to...
25838 * doc/sourcebuild.texi (Back End): Here. Explain how $target.h
25839 differs from $target-protos.h.
25841 2018-03-05 Kito Cheng <kito.cheng@gmail.com>
25842 Chung-Ju Wu <jasonwucj@gmail.com>
25844 * config/nds32/nds32-protos.h (nds32_expand_setmem): Declare.
25845 * config/nds32/nds32-multiple.md (setmemsi): Define.
25846 * config/nds32/nds32-memory-manipulation.c
25847 (nds32_gen_dup_4_byte_to_word_value): New.
25848 (emit_setmem_word_loop): New.
25849 (emit_setmem_byte_loop): New.
25850 (nds32_expand_setmem_loop): New.
25851 (nds32_expand_setmem_loop_v3m): New.
25852 (nds32_expand_setmem_unroll): New.
25853 (nds32_expand_setmem): New.
25855 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
25856 Chung-Ju Wu <jasonwucj@gmail.com>
25858 * config/nds32/nds32-memory-manipulation.c
25859 (nds32_emit_load_store): New.
25860 (nds32_emit_post_inc_load_store): New.
25861 (nds32_emit_mem_move): New.
25862 (nds32_emit_mem_move_block): New.
25863 (nds32_expand_movmemsi_loop_unknown_size): New.
25864 (nds32_expand_movmemsi_loop_known_size): New.
25865 (nds32_expand_movmemsi_loop): New.
25866 (nds32_expand_movmemsi_unroll): New.
25867 (nds32_expand_movmemqi): Rename ...
25868 (nds32_expand_movmemsi): ... to this.
25869 * config/nds32/nds32-multiple.md (movmemqi): Rename ...
25870 (movmemsi): ... to this.
25871 * config/nds32/nds32-protos.h (nds32_expand_movmemqi): Rename ...
25872 (nds32_expand_movmemsi): ... to this.
25874 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
25875 Monk Chiang <sh.chiang04@gmail.com>
25876 Chung-Ju Wu <jasonwucj@gmail.com>
25878 * config/nds32/nds32-protos.h
25879 (nds32_expand_load_multiple): New arguments.
25880 (nds32_expand_store_multiple): Ditto.
25881 (nds32_valid_multiple_load_store): Rename ...
25882 (nds32_valid_multiple_load_store_p): ... to this.
25883 * config/nds32/nds32-memory-manipulation.c
25884 (nds32_expand_load_multiple): Refine implementation.
25885 (nds32_expand_store_multiple): Ditto.
25886 * config/nds32/nds32-multiple.md
25887 (load_multiple): Update nds32_expand_load_multiple interface.
25888 (store_multiple): Update nds32_expand_store_multiple interface.
25889 * config/nds32/nds32-predicates.c
25890 (nds32_valid_multiple_load_store): Rename ...
25891 (nds32_valid_multiple_load_store_p): ... to this and refine
25893 * config/nds32/predicates.md
25894 (nds32_load_multiple_and_update_address_operation): New predicate.
25895 (nds32_store_multiple_and_update_address_operation): New predicate.
25897 2018-03-04 Kito Cheng <kito.cheng@gmail.com>
25898 Chung-Ju Wu <jasonwucj@gmail.com>
25900 * config/nds32/nds32.md (type): Add load_multiple and store_multiple.
25901 (combo): New attribute.
25902 * config/nds32/nds32-multiple.md: Refine patterns with new attributes.
25904 2018-03-03 Chung-Ju Wu <jasonwucj@gmail.com>
25906 * config/nds32/nds32.opt: Change -mcmodel= default value.
25908 2018-03-03 Kito Cheng <kito.cheng@gmail.com>
25909 Monk Chiang <sh.chiang04@gmail.com>
25910 Chung-Ju Wu <jasonwucj@gmail.com>
25912 * config/nds32/constants.md (unspec_element): New enum.
25913 * config/nds32/constraints.md (Umw): New constraint.
25914 * config/nds32/nds32-intrinsic.c: Add more builtin functions.
25915 * config/nds32/nds32-intrinsic.md: Likewise.
25916 * config/nds32/nds32-md-auxiliary.c (nds32_regno_to_enable4): New.
25917 (nds32_valid_smw_lwm_base_p): New.
25918 (nds32_output_smw_single_word): New.
25919 (nds32_output_lmw_single_word): New.
25920 (nds32_expand_unaligned_load): New.
25921 (nds32_expand_unaligned_store): New.
25922 * config/nds32/nds32-protos.h (nds32_valid_smw_lwm_base_p): Declare.
25923 (nds32_output_smw_single_word): Declare.
25924 (nds32_output_lmw_single_word): Declare.
25925 (nds32_expand_unaligned_load): Declare.
25926 (nds32_expand_unaligned_store): Declare.
25927 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_UALOAD_HW,
25928 NDS32_BUILTIN_UALOAD_W, NDS32_BUILTIN_UALOAD_DW,
25929 NDS32_BUILTIN_UASTORE_HW, NDS32_BUILTIN_UASTORE_W,
25930 NDS32_BUILTIN_UASTORE_DW.
25931 * config/nds32/predicates.md (nds32_lmw_smw_base_operand): New
25934 2018-03-03 Monk Chiang <sh.chiang04@gmail.com>
25935 Kito Cheng <kito.cheng@gmail.com>
25936 Chung-Ju Wu <jasonwucj@gmail.com>
25938 * config/nds32/nds32-intrinsic.c
25939 (nds32_expand_builtin_null_ftype_reg): Delete.
25940 (nds32_expand_builtin_reg_ftype_imm): Ditto.
25941 (nds32_expand_builtin_null_ftype_reg_imm): Ditto.
25942 (nds32_read_argument): New.
25943 (nds32_legitimize_target): Ditto.
25944 (nds32_legitimize_argument): Ditto.
25945 (nds32_check_constant_argument): Ditto.
25946 (nds32_expand_unop_builtin): Ditto.
25947 (nds32_expand_unopimm_builtin): Ditto.
25948 (nds32_expand_binop_builtin): Ditto.
25949 (nds32_builtin_decl_impl): Ditto.
25950 (builtin_description): Ditto.
25951 (nds32_expand_builtin_impl): Rewrite with new infrastructure.
25952 (nds32_init_builtins_impl): Ditto.
25953 * config/nds32/nds32.c (TARGET_BUILTIN_DECL): Define.
25954 (nds32_builtin_decl): New.
25955 * config/nds32/nds32.h (nds32_builtins): Add NDS32_BUILTIN_COUNT.
25956 * config/nds32/nds32-protos.h (nds32_builtin_decl_impl): Declare.
25958 2018-03-02 Jeff Law <law@redhat.com>
25960 * reorg.c (stop_search_p): Handle DEBUG_INSN.
25961 (redundant_insn, fill_simple_delay_slots): Likewise.
25962 (fill_slots_from_thread): Likewise.
25963 * resource.c (mark_referenced_resources): Likewise.
25964 (mark_set_resources, find_dead_or_set_registers): Likewise.
25966 2018-03-02 Jakub Jelinek <jakub@redhat.com>
25968 * substring-locations.h (format_warning_va): Formatting fix for
25969 ATTRIBUTE_GCC_DIAG.
25970 (format_warning_at_substring): Fix up ATTRIBUTE_GCC_DIAG second
25972 (format_warning_n_va, format_warning_at_substring_n): New prototypes.
25973 * substring-locations.c: Include intl.h.
25974 (format_warning_va): Turned into small wrapper around
25975 format_warning_n_va, renamed to ...
25976 (format_warning_n_va): ... this, add N and PLURAL_GMSGID arguments,
25977 rename GMSGID to SINGULAR_GMSGID, if SINGULAR_GMSGID != PLURAL_GMSGID,
25979 (format_warning_at_substring_n): New function.
25980 * gimple-ssa-sprintf.c: Remove GCC diagnostic ignored pragma.
25981 (fmtwarn): Add ATTRIBUTE_GCC_DIAG. Turn into a copy of
25982 format_warning_at_substring with just a shorter name instead of
25983 const function pointer.
25984 (fmtwarn_n): New function.
25985 (maybe_warn, format_directive, parse_directive): Use fmtwarn_n where
25986 appropriate, get rid of all the fmtstr temporaries, move conditionals
25987 with G_() wrapped string literals directly into fmtwarn arguments,
25988 cast dir.len to (int), formatting fixes.
25990 2018-03-02 Thomas Schwinge <thomas@codesourcery.com>
25992 * doc/invoke.texi: Remove "Cilk Plus" references.
25994 2018-03-02 Jakub Jelinek <jakub@redhat.com>
25995 Richard Biener <rguenther@suse.de>
25998 * expr.c (expand_expr_real_1) <case CALL_EXPR>: Don't emit diagnostics
25999 for error or warning attributes if CALL_FROM_THUNK_P is set.
26002 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26005 * config/pa/pa.h (TARGET_CPU_CPP_BUILTINS): Predefine
26006 __SIZEOF_128__ macro if HPUX_LONG_DOUBLE_LIBRARY.
26009 * config/ia64/ia64.h (TARGET_CPU_CPP_BUILTINS): Predefine
26010 __SIZEOF_{FPREG,FLOAT{80,128}}__ macros.
26012 * predict.c (test_prediction_value_range): Use PROB_UNINITIALIZED
26013 instead of -1U in last predictors element's probability member.
26015 2018-03-02 Eric Botcazou <ebotcazou@adacore.com>
26018 * ipa-devirt.c (odr_subtypes_equivalent_p): Get the ODR type of both
26019 arguments if they are comparable.
26021 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
26023 PR tree-optimization/84634
26024 * tree-vect-stmts.c (vectorizable_store, vectorizable_load): Replace
26025 masks and masked_loop_p with a single loop_masks, making sure it's
26026 null for bb vectorization.
26028 2018-03-02 Richard Sandiford <richard.sandiford@linaro.org>
26030 * tree-vect-data-refs.c (vect_analyze_data_ref_dependence)
26031 (vect_analyze_data_ref_access): Use loop->safe_len rather than
26032 loop->force_vectorize to check whether there is no alias.
26034 2018-03-02 Jakub Jelinek <jakub@redhat.com>
26037 * rtl.h (prev_real_nondebug_insn, next_real_nondebug_insn): New
26039 * emit-rtl.c (next_real_insn, prev_real_insn): Fix up function
26041 (next_real_nondebug_insn, prev_real_nondebug_insn): New functions.
26042 * cfgcleanup.c (try_head_merge_bb): Use prev_real_nondebug_insn
26043 instead of a loop around prev_real_insn.
26044 * combine.c (move_deaths): Use prev_real_nondebug_insn instead of
26047 PR inline-asm/84625
26048 * config/i386/i386.c (ix86_print_operand): Use conditional
26049 output_operand_lossage instead of gcc_assert if CONST_VECTOR is not
26052 2018-03-02 Richard Biener <rguenther@suse.de>
26054 PR tree-optimization/84427
26055 * tree-ssa-pre.c (bitmap_remove_expr_from_set): Remove.
26056 (bitmap_set_subtract_values): Rewrite to handle multiple
26059 (prune_clobbered_mems): Likewise.
26060 (phi_translate): Take edge instead of pred/phiblock.
26061 (phi_translate_1): Likewise.
26062 (phi_translate_set): Likewise. Insert all translated
26063 exprs for a value into the set, keeping possibly multiple
26064 expressions per value.
26065 (compute_antic_aux): Adjust for phi_translate changes.
26066 When intersecting union the expressions and prune those
26067 not in the final value set, keeping possibly multiple
26068 expressions per value. Do not use value-insertion
26069 for unioning ANTIC_OUT U EXP_GEN - TMP_GEN but merge
26070 all expressions. Add verification that the value-sets
26071 only shrink during iteration.
26072 (compute_partial_antic_aux): Adjust for the phi_translate changes.
26073 (do_pre_regular_insertion): Likewise.
26074 (do_pre_partial_partial_insertion): Likewise.
26076 2018-03-02 Richard Biener <rguenther@suse.de>
26079 * config/darwin.c (saved_debug_info_level): New static global.
26080 (darwin_asm_lto_start): Disable debug info generation for LTO out.
26081 (darwin_asm_lto_end): Restore debug info generation settings.
26083 2018-03-01 Martin Liska <mliska@suse.cz>
26086 * sanopt.c (sanitize_rewrite_addressable_params): Do not handle
26087 volatile arguments.
26089 2018-03-01 Richard Biener <rguenther@suse.de>
26092 * dwarf2out.c (gen_variable_die): Properly handle late VLA
26093 type annotation with LTO when debug was disabled at compile-time.
26095 2018-03-01 Matthew Fortune <mfortune@gmail.com>
26097 * config/mips/mips.c (mips_final_prescan_insn): Fix incorrect
26099 (mips_final_postscan_insn): Likewise.
26101 2018-03-01 Richard Sandiford <richard.sandiford@linaro.org>
26103 PR rtl-optimization/84528
26104 * alias.c (init_alias_target): Add commentary.
26105 (init_alias_analysis): Only give HARD_FRAME_POINTER_REGNUM
26106 a unique base value if the frame pointer is not eliminated
26107 to the stack pointer.
26109 2018-03-01 Tom de Vries <tom@codesourcery.com>
26111 PR rtl-optimization/83327
26112 * lra-int.h (hard_regs_spilled_into): Declare.
26113 * lra.c (hard_regs_spilled_into): Define.
26114 (init_reg_info): Init hard_regs_spilled_into.
26115 * lra-spills.c (assign_spill_hard_regs): Update hard_regs_spilled_into.
26116 * lra-lives.c (make_hard_regno_born, make_hard_regno_dead)
26117 (process_bb_lives): Handle hard_regs_spilled_into.
26118 (lra_create_live_ranges_1): Before doing liveness propagation, clear
26119 regs in all_hard_regs_bitmap if set in hard_regs_spilled_into.
26121 2018-02-28 David Edelsohn <dje.gcc@gmail.com>
26123 * config.gcc (powerpc-ibm-aix7.1.*): New stanza.
26124 (powerpc-ibm-aix[789]*): Default to AIX 7.2.
26125 * config/rs6000/aix71.h (TARGET_DEFAULT): Revert to Power4 ISA.
26126 * config/rs6000/aix72.h: New file.
26128 2018-02-28 Jakub Jelinek <jakub@redhat.com>
26130 * gimple-ssa-warn-restrict.c (maybe_diag_overlap): Use warning_n
26131 instead of warning_at with conditional singular and plural messages
26135 * stor-layout.c (update_alignment_for_field): For
26136 targetm.ms_bitfield_layout_p (rli->t), if !is_bitfield
26137 && !DECL_PACKED (field), do the alignment update, just use
26138 only desired_align instead of MAX (type_align, desired_align)
26140 (place_field): Don't do known_align < desired_align handling
26141 early if targetm.ms_bitfield_layout_p (rli->t) and rli->prev_field
26142 is non-NULL, instead do it after rli->prev_field handling and
26143 only if not within a bitfield word. For DECL_PACKED (field)
26144 use type_align of BITS_PER_UNIT.
26146 2018-02-28 Eric Botcazou <ebotcazou@adacore.com>
26148 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Remove
26149 superfluous parentheses and trailing spaces.
26151 2018-02-28 Richard Biener <rguenther@suse.de>
26153 PR tree-optimization/84584
26154 * graphite-scop-detection.c (scop_detection::add_scop): Discard
26155 SCoPs with fake exit edge.
26157 2018-02-28 Martin Liska <mliska@suse.cz>
26160 * timevar.c (timer::print): Fix format to properly print 100%
26163 2018-02-28 Richard Biener <rguenther@suse.de>
26165 PR middle-end/84607
26166 * genmatch.c (capture_info::walk_match): Do not mark
26167 captured expressions without operands as expr_p given
26168 they act more like predicates and should be subject to
26169 "lost tail" side-effect preserving.
26171 2018-02-28 Alexandre Oliva <aoliva@redhat.com>
26173 PR rtl-optimization/81611
26174 * auto-inc-dec.c (attempt_change): Move dead note from
26175 mem_insn if it's the next use of regno
26176 (find_address): Take address use of reg holding
26177 non-incremented value. Add parm to limit search to the named
26179 (merge_in_block): Attempt to use a mem insn that is the next
26180 use of the original regno.
26182 2018-02-27 Martin Sebor <msebor@redhat.com>
26185 * doc/invoke.texi (-Wmissing-attributes): New option.
26186 * print-tree.c (print_node): Handle DECL_UNINLINABLE.
26188 2018-02-27 Martin Sebor <msebor@redhat.com>
26190 PR translation/84207
26191 * diagnostic-core.h (warning_n, error_n, inform_n): Change
26192 n argument to unsigned HOST_WIDE_INT.
26193 * diagnostic.c (warning_n, error_n, inform_n): Ditto.
26194 (diagnostic_n_impl): Ditto. Handle arguments in excess of LONG_MAX.
26195 * gimple-ssa-sprintf.c (format_directive): Simplify inform_n call.
26196 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Use warning_n.
26198 2018-02-27 Richard Biener <rguenther@suse.de>
26200 PR tree-optimization/84512
26201 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost):
26202 Do not use the estimate returned from record_stmt_cost for
26203 the scalar iteration cost but sum properly using add_stmt_cost.
26205 2018-02-27 Richard Biener <rguenther@suse.de>
26207 PR tree-optimization/84466
26208 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
26209 Adjust last change to less strictly validate use operands.
26211 2018-02-27 Martin Liska <mliska@suse.cz>
26213 PR gcov-profile/84548
26214 * gcov.c (process_file): Allow partial overlap and consider it
26215 also as group functions.
26216 (output_lines): Properly calculate range of lines for a group.
26218 2018-02-27 Martin Liska <mliska@suse.cz>
26220 * timevar.c (timer::print_row): Remove 'usr', 'sys', 'wall' and
26221 'ggc' suffixes. Change first column width.
26222 (timer::print): Fix formatting of the column.
26224 2018-02-27 Alexandre Oliva <aoliva@redhat.com>
26226 * tree-ssa-live.c (remove_unused_scope_block_p): Do not
26227 preserve inline entry blocks for the sake of debug inline
26228 entry point markers alone.
26229 (remove_unused_locals): Suggest in comments a better place to
26230 force the preservation of inline entry blocks that are
26231 otherwise unused, but do not preserve them.
26233 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
26235 * config/i386/i386.c (ix86_output_indirect_jmp): Update comments.
26237 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
26240 * config/i386/constraints.md (Bs): Replace
26241 ix86_indirect_branch_register with
26242 TARGET_INDIRECT_BRANCH_REGISTER.
26244 * config/i386/i386.md (indirect_jump): Likewise.
26245 (tablejump): Likewise.
26246 (*sibcall_memory): Likewise.
26247 (*sibcall_value_memory): Likewise.
26248 Peepholes of indirect call and jump via memory: Likewise.
26249 (*sibcall_GOT_32): Disallowed for TARGET_INDIRECT_BRANCH_REGISTER.
26250 (*sibcall_value_GOT_32): Likewise.
26251 * config/i386/predicates.md (indirect_branch_operand): Likewise.
26252 (GOT_memory_operand): Likewise.
26253 (call_insn_operand): Likewise.
26254 (sibcall_insn_operand): Likewise.
26255 (GOT32_symbol_operand): Likewise.
26256 * config/i386/i386.h (TARGET_INDIRECT_BRANCH_REGISTER): New.
26258 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
26260 PR rtl-optimization/83496
26261 * reorg.c (steal_delay_list_from_target): Change REDUNDANT array from
26262 booleans to RTXes. Call fix_reg_dead_note on every non-null element.
26263 (steal_delay_list_from_fallthrough): Call fix_reg_dead_note on a
26264 redundant insn, if any.
26265 (relax_delay_slots): Likewise.
26266 (update_reg_unused_notes): Rename REDUNDANT_INSN to OTHER_INSN.
26268 2018-02-26 Richard Sandiford <richard.sandiford@linaro.org>
26270 PR tree-optimization/83965
26271 * tree-vect-patterns.c (vect_reassociating_reduction_p): Assume
26272 that grouped statements are part of a reduction chain. Return
26273 true if the statement is not marked as a reduction itself but
26274 is part of a group.
26275 (vect_recog_dot_prod_pattern): Don't check whether the statement
26276 is part of a group here.
26277 (vect_recog_sad_pattern): Likewise.
26278 (vect_recog_widen_sum_pattern): Likewise.
26280 2018-02-26 Eric Botcazou <ebotcazou@adacore.com>
26283 * final.c (rest_of_clean_state): Also look for calls inside sequences.
26285 2018-02-26 H.J. Lu <hongjiu.lu@intel.com>
26288 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Remove
26290 (ix86_output_indirect_function_return): New prototype.
26291 (ix86_split_simple_return_pop_internal): Likewise.
26292 * config/i386/i386.c (indirect_return_via_cx): New.
26293 (indirect_return_via_cx_bnd): Likewise.
26294 (indirect_thunk_name): Handle return va CX_REG.
26295 (output_indirect_thunk_function): Create alias for
26296 __x86_return_thunk_[re]cx and __x86_return_thunk_[re]cx_bnd.
26297 (ix86_output_indirect_jmp): Remove the bool argument.
26298 (ix86_output_indirect_function_return): New function.
26299 (ix86_split_simple_return_pop_internal): Likewise.
26300 * config/i386/i386.md (*indirect_jump): Don't pass false
26301 to ix86_output_indirect_jmp.
26302 (*tablejump_1): Likewise.
26303 (simple_return_pop_internal): Change it to define_insn_and_split.
26304 Call ix86_split_simple_return_pop_internal to split it for
26305 -mfunction-return=.
26306 (simple_return_indirect_internal): Call
26307 ix86_output_indirect_function_return instead of
26308 ix86_output_indirect_jmp.
26310 2018-02-26 Jakub Jelinek <jakub@redhat.com>
26313 * vec.h (vec_default_construct): For BROKEN_VALUE_INITIALIZATION use
26314 memset and value initialization afterwards.
26316 2018-02-26 Christophe Lyon <christophe.lyon@linaro.org>
26318 * Makefile.in (lto-wrapper): Use ALL_LINKERFLAGS.
26320 2018-02-26 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
26323 * common/config/aarch64/aarch64-common.c
26324 (aarch_option_optimization_table[]): Switch
26325 off fomit-frame-pointer
26327 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
26328 Chung-Ju Wu <jasonwucj@gmail.com>
26330 * config/nds32/nds32-multiple.md (load_multiple): Disallow
26332 (store_multiple): Ditto.
26334 2018-02-26 Kito Cheng <kito.cheng@gmail.com>
26336 * config.gcc: Add --with-cpu support for nds32 target.
26337 * config/nds32/nds32-opts.h (nds32_cpu_type): New.
26338 * config/nds32/nds32.opt: Add -mcpu= option.
26340 2018-02-25 Segher Boessenkool <segher@kernel.crashing.org>
26342 * config/rs6000/rs6000.opt (mvrsave=no, mvrsave=yes, isel=no,
26343 isel=yes): Warn for these deprecated options.
26345 2018-02-23 David Edelsohn <dje.gcc@gmail.com>
26347 * config/rs6000/aix71.h (TARGET_DEFAULT): Change to
26348 ISA_2_5_MASKS_EMBEDDED.
26350 2018-02-23 Jakub Jelinek <jakub@redhat.com>
26352 * ipa-prop.c (ipa_vr_ggc_hash_traits::hash): Hash p->min and
26353 p->max as pointers rather than using iterative_hash_expr.
26355 2018-02-23 Carl Love <cel@us.ibm.com>
26357 * config/rs6000/rs6000-builtin.def: Change VSIGNED2 and VUNSIGNED2
26358 macro expansions from BU_VSX_2 to BU_P8V_VSX_2 and BU_VSX_OVERLOAD_2 to
26360 * config/rs6000/rs6000-c.c: Change VSX_BUILTIN_VEC_VSIGNED2 to
26361 P8V_BUILTIN_VEC_VSIGNED2. Change VSX_BUILTIN_VEC_VUNSIGNED2 to
26362 P8V_BUILTIN_VEC_VUNSIGNED2.
26364 2018-02-22 Vladimir Makarov <vmakarov@redhat.com>
26367 * lra-int.h (LRA_UNKNOWN_ALT, LRA_NON_CLOBBERED_ALT): New macros.
26368 * lra.c (lra_set_insn_recog_data, lra_update_insn_recog_data): Use
26370 * lra-constraints.c (curr_insn_transform): Set up
26371 LRA_NON_CLOBBERED_ALT for moves processed on the fast path. Use
26373 (remove_inheritance_pseudos): Use LRA_UNKNOWN_ALT.
26374 * lra-eliminations.c (spill_pseudos): Ditto.
26375 (process_insn_for_elimination): Ditto.
26376 * lra-lives.c (reg_early_clobber_p): Use the new macros.
26377 * lra-spills.c (spill_pseudos): Use LRA_UNKNOWN_ALT and
26378 LRA_NON_CLOBBERED_ALT.
26380 2018-02-22 Martin Sebor <msebor@redhat.com>
26382 PR tree-optimization/84480
26383 * gimple-fold.c (gimple_fold_builtin_strcpy): Move warnings
26384 to maybe_diag_stxncpy_trunc. Call it.
26385 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Integrate warnings
26386 from gimple_fold_builtin_strcpy. Print inlining stack.
26387 (handle_builtin_stxncpy): Print inlining stack.
26388 * tree-ssa-strlen.h (maybe_diag_stxncpy_trunc): Declare.
26390 2018-02-22 H.J. Lu <hongjiu.lu@intel.com>
26393 * config/i386/i386.c (ix86_set_indirect_branch_type): Issue an
26394 error when -mindirect-branch=thunk-extern, -fcf-protection=branch
26395 and -fcheck-pointer-bounds are used together.
26396 (indirect_thunk_prefix): New enum.
26397 (indirect_thunk_need_prefix): New function.
26398 (indirect_thunk_name): Replace need_bnd_p with need_prefix. Use
26399 "_nt" instead of "_bnd" for NOTRACK prefix.
26400 (output_indirect_thunk): Replace need_bnd_p with need_prefix.
26401 (output_indirect_thunk_function): Likewise.
26403 (ix86_code_end): Update output_indirect_thunk_function calls.
26404 (ix86_output_indirect_branch_via_reg): Replace
26405 ix86_bnd_prefixed_insn_p with indirect_thunk_need_prefix.
26406 (ix86_output_indirect_branch_via_push): Likewise.
26407 (ix86_output_function_return): Likewise.
26408 * doc/invoke.texi: Document -mindirect-branch=thunk-extern is
26409 incompatible with -fcf-protection=branch and
26410 -fcheck-pointer-bounds.
26412 2018-02-22 Steve Ellcey <sellcey@cavium.com>
26415 * config/aarch64/aarch64.c (aarch64_print_address_internal):
26416 Change gcc_assert call to output_operand_lossage.
26418 2018-02-22 Steve Ellcey <sellcey@cavium.com>
26420 * doc/extend.texi (__builtin_extend_pointer): Document builtin.
26422 2018-02-22 DJ Delorie <dj@redhat.com>
26423 Sebastian Perta <sebastian.perta@renesas.com>
26424 Oleg Endo <olegendo@gcc.gnu.org>
26426 * config/rx/rx.c (rx_rtx_costs): New function.
26427 (TARGET_RTX_COSTS): Override to use rx_rtx_costs.
26429 2018-02-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
26431 * config/arm/t-multilib: Map Armv8-R to Armv7 multilibs.
26433 2018-02-22 Martin Liska <mliska@suse.cz>
26436 * common/config/arm/arm-common.c (arm_print_hint_for_cpu_option):
26437 Add "native" as a possible value.
26439 2018-02-22 Martin Liska <mliska@suse.cz>
26442 * config/i386/i386.c (ix86_option_override_internal):
26443 Add "native" as a possible value for -march and -mtune.
26445 2018-02-22 Jakub Jelinek <jakub@redhat.com>
26448 * stor-layout.c (finalize_type_size): Propagate TYPE_EMPTY_P flag
26449 to all type variants.
26451 PR tree-optimization/84503
26452 * gimple-ssa-store-merging.c (merged_store_group::merge_into): Compute
26453 width as info->bitpos + info->bitsize - start.
26454 (merged_store_group::merge_overlapping): Simplify width computation.
26455 (check_no_overlap): New function.
26456 (imm_store_chain_info::try_coalesce_bswap): Compute expected
26457 start + width and last_order of the group, fail if check_no_overlap
26459 (imm_store_chain_info::coalesce_immediate_stores): Don't merge info
26460 to group if check_no_overlap fails.
26462 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
26464 * config/rs6000/altivec.md: Delete contraint arguments to
26465 define_expand, define_split, and define_peephole2, and in
26466 define_insn_and_split if always unused.
26467 * config/rs6000/darwin.md: Ditto.
26468 * config/rs6000/dfp.md: Ditto.
26469 * config/rs6000/rs6000.md: Ditto.
26470 * config/rs6000/sync.md: Ditto.
26471 * config/rs6000/vector.md: Ditto.
26472 * config/rs6000/vsx.md: Ditto.
26474 2018-02-21 Segher Boessenkool <segher@kernel.crashing.org>
26476 * config/rs6000/altivec.md: Write output control strings as braced
26477 blocks instead of double-quoted strings.
26478 * config/rs6000/darwin.md: Ditto.
26479 * config/rs6000/rs6000.md: Ditto.
26480 * config/rs6000/vector.md: Ditto.
26481 * config/rs6000/vsx.md: Ditto.
26483 2018-02-21 Jason Merrill <jason@redhat.com>
26485 PR c++/84314 - ICE with templates and fastcall attribute.
26486 * attribs.c (build_type_attribute_qual_variant): Remove assert.
26488 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
26490 * ipa-cp.c (determine_versionability): Fix comment typos.
26492 2018-02-21 Jan Hubicka <hubicka@ucw.cz>
26495 * ipa-cp.c (determine_versionability): Do not version functions caling
26498 2018-02-21 Martin Liska <mliska@suse.cz>
26501 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch):
26502 Add "native" as a possible value.
26503 * config/aarch64/aarch64.h (HAVE_LOCAL_CPU_DETECT): Define
26504 the macro when native cpu detection is available.
26506 2018-02-21 Martin Liska <mliska@suse.cz>
26509 * common/config/arm/arm-common.c (arm_print_hint_for_arch_option):
26510 Add "native" as a possible value.
26511 * config/arm/arm.h (HAVE_LOCAL_CPU_DETECT): Define the macro
26512 when native cpu detection is available.
26514 2018-02-21 Jakub Jelinek <jakub@redhat.com>
26515 Martin Sebor <msebor@redhat.com>
26517 PR tree-optimization/84478
26518 * gimple-fold.h (get_range_strlen): Add a bool argument defaulted to
26520 * gimple-fold.c (get_range_strlen): Make minlen const and assume it
26521 can't be NULL. Change FUZZY from bool to int, for 1 add PHI/COND_EXPR
26522 support which is conservatively correct, for 2 only stay conservative
26523 for maxlen. Formatting and comment capitalization fixes. Add STRICT
26524 argument to the 2 argument get_range_strlen, adjust 6 arg
26525 get_range_strlen caller and clear minmaxlen[0] and [1] if it returned
26527 (get_maxval_strlen): Adjust 6 arg get_range_strlen caller.
26528 (gimple_fold_builtin_strlen): Pass true as last argument to
26531 2018-02-20 Martin Sebor <msebor@redhat.com>
26533 PR middle-end/84095
26534 * gimple-ssa-warn-restrict.c (builtin_memref::extend_offset_range): New.
26535 (builtin_memref::set_base_and_offset): Same. Handle inner references.
26536 (builtin_memref::builtin_memref): Factor out parts into
26537 set_base_and_offset and call it.
26539 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
26541 PR middle-end/84406
26542 * optabs-query.c (find_widening_optab_handler_and_mode): If from_mode
26543 is a scalar_int_mode, assert that to_mode is a scalar_int_mode with
26544 greater precision. If to_mode is a MODE_PARTIAL_INT, stop the
26545 search at the associated MODE_INT.
26547 2018-02-20 Jeff Law <law@redhat.com>
26549 PR middle-end/82123
26550 PR tree-optimization/81592
26551 PR middle-end/79257
26552 * gimple-ssa-sprintf.c (format_integer): Query EVRP range analyzer
26553 for range data rather than using global data.
26554 * gimple-ssa-sprintf.c (get_int_range): Query EVRP range analyzer for
26555 range data rather than using global data.
26556 * gimple-ssa-sprintf.c (get_int_range): Accept vr_values parameter
26557 pass it to children as needed.
26558 (struct directive::fmtresult): Similarly.
26559 (struct directive::set_width): Similarly.
26560 (struct directive::set_precision): Similarly.
26561 (format_integer, format_directive, parse_directive): Similarly.
26562 (format_none): Accept unnamed vr_values parameter.
26563 (format_percent, format_floating, format_character): Similarly.
26564 (format_string, format_plain): Similarly.
26565 * gimple-ssa-sprintf.c (sprintf_dom_walker::handle_gimple_call): Query
26566 the EVRP range analyzer for range data rather than using global data.
26567 * gimple-ssa-sprintf.c: Include alloc-pool.h, vr-values.h and
26568 gimple-ssa-evrp-analyze.h
26569 (class sprintf_dom_walker): Add after_dom_children member function.
26570 Add evrp_range_analyzer member.
26571 (sprintf_dom_walker::before_dom_children): Call into the EVRP
26572 range analyzer as needed.
26573 (sprintf_dom_walker::after_dom_children): New member function.
26574 * gimple-ssa-evrp-analyze.c (evrp_range_analyzer::enter): Do nothing
26576 (evrp_range_analyzer::record_ranges_from_stmt): Likewise.
26577 (evrp_range_analyzer::pop_to_marker): Likewise.
26579 2018-02-20 Richard Sandiford <richard.sandiford@linaro.org>
26581 PR tree-optimization/84419
26582 * internal-fn.c (expand_call_mem_ref): Create a TARGET_MEM_REF
26583 with the required type if its current type is compatible but
26586 2018-02-20 Jakub Jelinek <jakub@redhat.com>
26588 PR middle-end/82004
26589 * match.pd (pow(C,x) -> exp(log(C)*x)): Delay all folding until
26590 after vectorization.
26592 2018-02-20 Martin Liska <mliska@suse.cz>
26595 * config/aarch64/aarch64.c (aarch64_print_hint_for_core_or_arch): Print
26596 possible values if we don't have a hint.
26598 2018-02-20 Martin Liska <mliska@suse.cz>
26602 * final.c (shorten_branches): Build align_tab array with one
26604 * opts.c (finish_options): Add alignment option limit check.
26605 (MAX_CODE_ALIGN): Likewise.
26606 (MAX_CODE_ALIGN_VALUE): Likewise.
26607 * doc/invoke.texi: Document maximum allowed option value for
26608 all -falign-* options.
26610 2018-02-19 Jakub Jelinek <jakub@redhat.com>
26613 * reg-notes.def (REG_CALL_ARG_LOCATION): New reg note.
26614 * insn-notes.def (NOTE_INSN_CALL_ARG_LOCATION): Remove.
26615 * var-tracking.c (emit_note_insn_var_location): Remove all references
26616 to NOTE_INSN_CALL_ARG_LOCATION.
26617 (emit_notes_in_bb): Emit arguments as REG_CALL_ARG_LOCATION note on
26618 the CALL_INSN rather than separate NOTE_INSN_CALL_ARG_LOCATION note.
26619 Use copy_rtx_if_shared.
26620 * dwarf2out.c (gen_subprogram_die): Use XEXP with 0 instead of
26621 NOTE_VAR_LOCATION on ca_loc->call_arg_loc_note.
26622 (dwarf2out_var_location): Remove handling of
26623 NOTE_INSN_CALL_ARG_LOCATION, instead handle REG_CALL_ARG_LOCATION note
26625 * final.c (final_scan_insn): Remove all references to
26626 NOTE_INSN_CALL_ARG_LOCATION.
26627 (rest_of_clean_state): Likewise. Remove REG_CALL_ARG_LOCATION notes
26628 before dumping final insns.
26629 * except.c (emit_note_eh_region_end): Remove all references to
26630 NOTE_INSN_CALL_ARG_LOCATION.
26631 * config/alpha/alpha.c (alpha_pad_function_end): Likewise.
26632 * config/c6x/c6x.c (c6x_gen_bundles): Likewise.
26633 * config/arc/arc.c (hwloop_optimize): Likewise.
26634 * config/arm/arm.c (create_fix_barrier): Likewise.
26635 * config/s390/s390.c (s390_chunkify_start): Likewise.
26636 * config/sh/sh.c (find_barrier): Likewise.
26637 * config/i386/i386.c (rest_of_insert_endbranch,
26638 ix86_seh_fixup_eh_fallthru): Likewise.
26639 * config/xtensa/xtensa.c (hwloop_optimize): Likewise.
26640 * config/iq2000/iq2000.c (final_prescan_insn): Likewise.
26641 * config/frv/frv.c (frv_function_prologue): Likewise.
26642 * emit-rtl.c (try_split): Likewise. Copy over REG_CALL_ARG_LOCATION
26644 (note_outside_basic_block_p): Remove all references to
26645 NOTE_INSN_CALL_ARG_LOCATION.
26646 * gengtype.c (adjust_field_rtx_def): Likewise.
26647 * print-rtl.c (rtx_writer::print_rtx_operand_code_0, print_insn):
26649 * jump.c (cleanup_barriers, delete_related_insns): Likewise.
26650 * cfgrtl.c (force_nonfallthru_and_redirect): Likewise.
26653 * builtins.c (builtin_mathfn_code): Don't check if CALL_EXPR_FN (t)
26656 PR tree-optimization/84452
26657 * tree-vect-patterns.c (vect_recog_pow_pattern): Don't call
26658 expand_simd_clones if targetm.simd_clone.compute_vecsize_and_simdlen
26661 2018-02-19 Martin Liska <mliska@suse.cz>
26664 * passes.def: Put pass_sancov_O0 before pass_lower_switch with -O0.
26666 2018-02-19 Martin Liska <mliska@suse.cz>
26667 Richard Sandiford <richard.sandiford@linaro.org>
26669 PR tree-optimization/82491
26670 * gimple-fold.c (get_base_constructor): Make earlier bail out
26673 2018-02-19 Carl Love <cel@us.ibm.com>
26675 * config/rs6000/rs6000-builtin.def: Change NEG macro expansions from
26676 BU_ALTIVEC_A to BU_P8V_AV_1 and BU_ALTIVEC_OVERLOAD_1 to
26678 * config/rs6000/rs6000-c.c: Change ALTIVEC_BUILTIN_VEC_NEG to
26679 P8V_BUILTIN_VEC_NEG.
26681 2018-02-19 Sebastian Perta <sebastian.perta@renesas.com>
26683 * config/rl78/rl78.md (movdf): New define expand.
26685 2018-02-19 Martin Liska <mliska@suse.cz>
26688 * doc/invoke.texi: Fix typo.
26689 * params.def (PARAM_MAX_LOOP_HEADER_INSNS): Likewise.
26691 2018-02-18 Segher Boessenkool <segher@kernel.crashing.org>
26693 * config/rs6000/rs6000.c (rs6000_option_override_internal): Don't
26694 handle rs6000_single_float and rs6000_double_float specially for
26697 2018-02-16 Jeff Law <law@redhat.com>
26699 * config/rx/rx.c (add_pop_cfi_notes): New function.;
26700 (pop_regs): Use it.
26702 2018-02-16 Jakub Jelinek <jakub@redhat.com>
26705 * ipa-inline.c (inline_small_functions): Fix a typo.
26707 2018-02-16 Nathan Sidwell <nathan@acm.org>
26709 * doc/extend.texi (Backwards Compatibility): Americanize 'behaviour'.
26711 2018-02-16 Carl Love <cel@us.ibm.com>
26713 * config/rs6000/rs6000-builtin.def: Add BU_P8V_VSX_2 macro definition.
26714 Change FLOAT2 expansions from BU_VSX_2 to BU_P8V_VSX_2 and
26715 from BU_VSX_OVERLOAD_2 to BU_P8V_OVERLOAD_2.
26716 * config/rs6000/rs6000-c.c: Changed macro VSX_BUILTIN_VEC_FLOAT2
26717 expansion to P8V_BUILTIN_VEC_FLOAT2.
26719 2018-02-16 Vladimir Makarov <vmakarov@redhat.com>
26721 PR rtl-optimization/70023
26722 * lra-constraints.c (inherit_in_ebb): Take hard reg mode of
26723 src_regno into account.
26725 2018-02-16 Carl Love <cel@us.ibm.com>
26727 * config/rs6000/altivec.h: Remove vec_vextract4b and vec_vinsert4b.
26728 * config/rs6000/rs6000-builtin.def: Remove macro expansion for
26729 VEXTRACT4B, VINSERT4B, VINSERT4B_DI and VEXTRACT4B.
26730 * config/rs6000/rs6000.c: Remove case statements for
26731 P9V_BUILTIN_VEXTRACT4B, P9V_BUILTIN_VEC_VEXTRACT4B,
26732 P9V_BUILTIN_VINSERT4B, P9V_BUILTIN_VINSERT4B_DI,
26733 and P9V_BUILTIN_VEC_VINSERT4B.
26734 * config/rs6000/rs6000-c.c (altivec_expand_builtin): Remove entries for
26735 P9V_BUILTIN_VEC_VEXTRACT4B and P9V_BUILTIN_VEC_VINSERT4B.
26736 * config/rs6000/vsx.md:
26737 * doc/extend.texi: Remove vec_vextract4b, non ABI definitions for
26740 2018-02-16 Carl Love <cel@us.ibm.com>
26742 * config/rs6000/altivec.h: Add builtin names vec_extract4b
26744 * config/rs6000/rs6000-builtin.def: Add INSERT4B and EXTRACT4B
26746 * config/rs6000/rs6000-c.c: Add the definitions for
26747 P9V_BUILTIN_VEC_EXTRACT4B and P9V_BUILTIN_VEC_INSERT4B.
26748 * config/rs6000/rs6000.c (altivec_expand_builtin): Add
26749 P9V_BUILTIN_EXTRACT4B and P9V_BUILTIN_INSERT4B case statements.
26750 * config/rs6000/vsx.md: Add define_insn extract4b. Add define_expand
26751 definition for insert4b and define insn *insert3b_internal.
26752 * doc/extend.texi: Add documentation for vec_extract4b.
26754 2018-02-16 Nathan Sidwell <nathan@acm.org>
26756 * doc/extend.texi (Backwards Compatibility): Mention friend
26757 injection. Note for-scope is deprecated.
26758 * doc/invoke.texi (-ffriend-injection): Deprecate.
26760 2018-02-16 Segher Boessenkool <segher@kernel.crashing.org>
26762 * combine.c (try_combine): When adjusting LOG_LINKS for the destination
26763 that moved to I2, also allow destinations that are a paradoxical
26764 subreg (instead of a normal reg).
26766 2018-02-16 Oleg Endo <olegendo@gcc.gnu.org>
26769 * config/rx/rx.c (rx_fuse_in_memory_bitop): Convert shift operand
26772 2018-02-16 Richard Biener <rguenther@suse.de>
26774 PR tree-optimization/84037
26775 PR tree-optimization/84016
26777 * config/i386/i386.c (ix86_builtin_vectorization_cost):
26778 Adjust vec_construct for the fact we need additional higher latency
26779 128bit inserts for AVX256 and AVX512 vector builds.
26780 (ix86_add_stmt_cost): Scale vector construction cost for
26783 2018-02-16 Richard Biener <rguenther@suse.de>
26785 PR tree-optimization/84417
26786 * tree-ssa.c (non_rewritable_mem_ref_base): Properly constrain
26787 the MEM_REF offset when conversion to BIT_FIELD_REF is desired.
26788 (non_rewritable_lvalue_p): Likewise, use poly-ints.
26790 2018-02-16 Martin Liska <mliska@suse.cz>
26793 * internal-fn.def (ASAN_CHECK): Set proper flags.
26794 (ASAN_MARK): Likewise.
26796 2018-02-16 Julia Koval <julia.koval@intel.com>
26798 * config/i386/i386.c (ix86_option_override_internal): Remove PTA_CLWB
26799 from PTA_CANNONLAKE.
26801 2018-02-16 Jakub Jelinek <jakub@redhat.com>
26804 * config/aarch64/cortex-a57-fma-steering.c (fma_forest::merge_forest):
26805 Use ++iter rather than iter++ for std::list iterators.
26806 (func_fma_steering::dfs): Likewise. Don't delete nodes right away,
26807 defer deleting them until all nodes in the forest are processed. Do
26808 free even leaf nodes. Change to_process into auto_vec.
26811 * system.h (BROKEN_VALUE_INITIALIZATION): Define for GCC < 4.3.
26812 * vec.h (vec_default_construct): Use memset instead of placement new
26813 if BROKEN_VALUE_INITIALIZATION is defined.
26814 * hash-table.h (hash_table<Descriptor, Allocator>::empty_slow): Use
26815 memset instead of value initialization if BROKEN_VALUE_INITIALIZATION
26818 PR rtl-optimization/83723
26819 * lra-int.h (lra_substitute_pseudo): Add DEBUG_P argument.
26820 * lra.c (lra_substitute_pseudo): Likewise. If true, use
26821 gen_rtx_raw_SUBREG instead of gen_rtx_SUBREG. Pass DEBUG_P to
26823 (lra_substitute_pseudo_within_insn): Adjust lra_substitute_pseudo
26825 * lra-constraints.c (inherit_reload_reg, split_reg): Likewise.
26827 2018-02-16 Eric Botcazou <ebotcazou@adacore.com>
26829 PR rtl-optimization/81443
26830 * rtlanal.c (num_sign_bit_copies1) <SUBREG>: Do not propagate results
26831 from inner REGs to paradoxical SUBREGs.
26833 2018-02-16 Richard Biener <rguenther@suse.de>
26835 PR tree-optimization/84399
26836 * graphite-scop-detection.c (scop_detection::stmt_simple_for_scop_p):
26837 For operands we can analyze at their definition make sure we can
26838 analyze them at each use as well.
26840 2018-02-16 Richard Biener <rguenther@suse.de>
26842 PR tree-optimization/84190
26843 * tree-ssa.c (non_rewritable_mem_ref_base): Do not touch
26844 volatile accesses if the decl isn't volatile.
26846 2018-02-15 Jason Merrill <jason@redhat.com>
26848 PR c++/84314 - ICE with templates and fastcall attribute.
26849 * attribs.c (build_type_attribute_qual_variant): Don't clobber
26850 TYPE_CANONICAL on an existing type.
26852 2018-02-15 Jakub Jelinek <jakub@redhat.com>
26854 PR tree-optimization/84383
26855 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Don't look at
26856 dstoff nor call operand_equal_p if dstbase is NULL.
26858 PR tree-optimization/84334
26859 * match.pd ((A +- CST1) +- CST2 -> A + CST3): If A is
26860 also a CONSTANT_CLASS_P, punt.
26862 2018-02-14 Jim Wilson <jimw@sifive.com>
26864 * config/riscv/riscv.c (riscv_first_stack_step): Move locals after
26865 first SMALL_OPERAND check. New local min_second_step. Move assert
26866 to where locals are set. Add TARGET_RVC support.
26867 * config/riscv/riscv.h (C_SxSP_BITS, SWSP_REACH, SDSP_REACH): New.
26869 2018-02-14 Indu Bhagat <indu.bhagat@oracle.com>
26871 * doc/invoke.texi: Correct -Wformat-overflow code sample.
26873 2018-02-14 Martin Sebor <msebor@redhat.com>
26875 PR tree-optimization/83698
26876 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For
26877 arrays constrain the offset range to their bounds.
26878 (builtin_access::strcat_overlap): Adjust the bounds of overlap offset.
26879 (builtin_access::overlap): Avoid setting the size of overlap if it's
26881 (maybe_diag_overlap): Also consider arrays when deciding what values
26882 of offsets to include in diagnostics.
26884 2018-02-14 Martin Sebor <msebor@redhat.com>
26887 * attribs.c (diag_attr_exclusions): Consider the exclusion(s)
26888 that correspond to the kind of a declaration.
26890 2018-02-14 John David Anglin <danglin@gcc.gnu.org>
26893 * config/pa/pa.md: Load address of PIC label using the linkage table
26894 if the label is nonlocal.
26896 2018-02-14 Kelvin Nilsen <kelvin@gcc.gnu.org>
26898 * config/rs6000/rs6000.c (rs6000_option_override_internal): Issue
26899 warning message if user requests -maltivec=be.
26900 * doc/invoke.texi: Document deprecation of -maltivec=be.
26902 2018-02-14 Will Schmidt <will_schmidt@vnet.ibm.com>
26905 * config/rs6000/rs6000-c.c: Update definitions for
26906 ALTIVEC_BUILTIN_VEC_SLD, ALTIVEC_BUILTIN_VEC_SLDW,
26907 VEC_XXSLDWI and ALTIVEC_BUILTIN_VEC_XXPERMDI builtins.
26909 2018-02-14 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
26912 * config/i386/cetintrin.h: Remove _rdssp[d|q] and
26913 add _get_ssp intrinsics. Remove argument from
26914 __builtin_ia32_rdssp[d|q].
26915 * config/i386/i386-builtin-types.def: Add UINT_FTYPE_VOID.
26916 * config/i386/i386-builtin.def: Remove argument from
26917 __builtin_ia32_rdssp[d|q].
26918 * config/i386/i386.c: Use UINT_FTYPE_VOID. Use
26919 ix86_expand_special_args_builtin for _rdssp[d|q].
26920 * config/i386/i386.md: Remove argument from rdssp[si|di] insn.
26921 Clear register before usage.
26922 * doc/extend.texi: Remove argument from __builtin_ia32_rdssp[d|q].
26923 Add documentation for new _get_ssp and _inc_ssp intrinsics.
26925 2018-02-14 Richard Sandiford <richard.sandiford@linaro.org>
26927 PR tree-optimization/84357
26928 * tree-data-ref.c (object_address_invariant_in_loop_p): Check
26929 operand 1 of an ARRAY_REF too.
26931 2018-02-14 Oleg Endo <olegendo@gcc.gnu.org>
26934 * config/rx/rx-protos.h (rx_reg_dead_or_unused_after_insn,
26935 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
26937 (set_of_reg): New struct.
26938 (rx_find_set_of_reg, rx_find_use_of_reg): New functions.
26939 * config/rx/rx.c (rx_reg_dead_or_unused_after_insn,
26940 rx_copy_reg_dead_or_unused_notes, rx_fuse_in_memory_bitop): New
26942 * config/rx/rx.md (andsi3, iorsi3, xorsi3): Convert to insn_and_split.
26943 Split into bitclr, bitset, bitinvert patterns if appropriate.
26944 (*bitset, *bitinvert, *bitclr): Convert to named insn_and_split and
26945 use rx_fuse_in_memory_bitop.
26946 (*bitset_in_memory, *bitinvert_in_memory, *bitclr_in_memory): Convert
26947 to named insn, correct maximum insn length.
26949 2018-02-14 Jozef Lawrynowicz <jozefl.gcc@gmail.com>
26952 * machmode.def: Define a complex mode for PARTIAL_INT.
26953 * genmodes.c (complex_class): Return MODE_COMPLEX_INT for
26955 * doc/rtl.texi: Document CSPImode.
26956 * config/msp430/msp430.c (msp430_hard_regno_nregs): Add CPSImode
26958 (msp430_hard_regno_nregs_with_padding): Likewise.
26960 2018-02-13 Peter Bergner <bergner@vnet.ibm.com>
26963 * config/rs6000/rs6000.c (mem_operand_gpr): Disallow altivec addresses.
26965 2018-02-13 Segher Boessenkool <segher@kernel.crashing.org>
26967 PR rtl-optimization/84169
26968 * combine.c (try_combine): New variable split_i2i3. Set it to true if
26969 we generated a parallel as new i3 and we split that to new i2 and i3
26970 instructions. Handle split_i2i3 similar to swap_i2i3: scan the
26971 LOG_LINKs of i3 to see which of those need to link to i2 now. Link
26972 those to i2, not i1. Partially rewrite this scan code.
26974 2018-02-13 Jakub Jelinek <jakub@redhat.com>
26977 * stor-layout.c (place_field): For variable length fields, adjust
26978 offset_align afterwards not just based on the field's alignment,
26979 but also on the size.
26981 PR middle-end/84309
26982 * match.pd (pow(C,x) -> exp(log(C)*x)): Use exp2s and log2s instead
26983 of exps and logs in the use_exp2 case.
26985 2018-02-13 Jeff Law <law@redhat.com>
26987 * config/rl/rl78.c (rl78_attribute_table): Fix terminator and
26988 entry for "vector".
26990 * config/rl78/rl78.c (rl78_handle_func_attribute): Mark
26993 2018-02-13 Alexandre Oliva <aoliva@redhat.com>
26997 * common.opt (gas-loc-support, gas-locview-support): New.
26998 (ginline-points, ginternal-reset-location-views): New.
26999 * doc/invoke.texi: Document them. Use @itemx where intended.
27000 (gvariable-location-views): Adjust.
27001 * target.def (reset_location_view): New.
27002 * doc/tm.texi.in (DWARF2_ASM_VIEW_DEBUG_INFO): New.
27003 (TARGET_RESET_LOCATION_VIEW): New.
27004 * doc/tm.texi: Rebuilt.
27005 * dwarf2out.c (dwarf2out_default_as_loc_support): New.
27006 (dwarf2out_default_as_locview_support): New.
27007 (output_asm_line_debug_info): Use option variables.
27008 (dwarf2out_maybe_output_loclist_view_pair): Likewise.
27009 (output_loc_list): Likewise.
27010 (add_high_low_attributes): Check option variables.
27011 Don't output entry view attribute in strict mode.
27012 (gen_inlined_subroutine_die): Check option variables.
27013 (dwarf2out_inline_entry): Likewise.
27014 (init_sections_and_labels): Likewise.
27015 (dwarf2out_early_finish): Likewise.
27016 (maybe_reset_location_view): New, from...
27017 (dwarf2out_var_location): ... here. Call it.
27018 * debug.h (dwarf2out_default_as_loc_support): Declare.
27019 (dwarf2out_default_as_locview_support): Declare.
27020 * hooks.c (hook_int_rtx_insn_0): New.
27021 * hooks.h (hook_int_rtx_insn_0): Declare.
27022 * toplev.c (process_options): Take -gas-loc-support and
27023 -gas-locview-support from dwarf2out. Enable
27024 -gvariable-location-views by default only with locview
27025 assembler support. Enable -ginternal-reset-location-views by
27026 default only if the target defines the corresponding hook.
27027 Enable -ginline-points by default if location views are
27028 enabled; force it disabled if statement frontiers are
27030 * tree-inline.c (expand_call_inline): Check option variables.
27031 * tree-ssa-live.c (remove_unused_scope_block_p): Likewise.
27033 2018-02-13 Richard Sandiford <richard.sandiford@linaro.org>
27035 PR tree-optimization/84321
27036 * tree-vrp.c (intersect_range_with_nonzero_bits): Fix VR_ANTI_RANGE
27037 handling. Also check whether the anti-range contains any values
27038 that satisfy the mask; switch to a VR_RANGE if not.
27040 2018-02-13 Paolo Bonzini <bonzini@gnu.org>
27043 * internal-fn.def (ASAN_CHECK, ASAN_MARK): Revert changes to fnspec.
27045 2018-02-13 Martin Jambor <mjambor@suse.cz>
27048 * ipa-param-manipulation.c (ipa_modify_call_arguments): Use location
27049 of call statements, also set location of a load to a temporary.
27051 2018-02-13 Sebastian Perta <sebastian.perta@renesas.com>
27053 * config/rl78/rl78.c (add_vector_labels): New function.
27054 * config/rl78/rl78.c (rl78_handle_vector_attribute): New function.
27055 * config/rl78/rl78.c (rl78_start_function): Call add_vector_labels.
27056 * config/rl78/rl78.c (rl78_handle_func_attribute): Removed the assert
27057 which checks that no arguments are passed.
27058 * config/rl78/rl78.c (rl78_attribute_table): Add "vector" attribute.
27059 * doc/extend.texi: Documentation for the new attribute.
27061 2018-02-13 Andreas Schwab <schwab@suse.de>
27063 * config/riscv/linux.h (CPP_SPEC): Define.
27065 2018-02-13 Jakub Jelinek <jakub@redhat.com>
27068 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Pass
27069 OPTION_MASK_ISA_AES | OPTION_MASK_ISA_SSE2 instead of
27070 OPTION_MASK_ISA_AES as first argument to def_builtin_const
27071 for AES builtins. Pass OPTION_MASK_ISA_PCLMUL | OPTION_MASK_ISA_SSE2
27072 instead of OPTION_MASK_ISA_PCLMUL as first argument to
27073 def_builtin_const for __builtin_ia32_pclmulqdq128 builtin.
27074 * config/i386/wmmintrin.h: If __SSE2__ is not defined, enable it
27075 temporarily for AES and PCLMUL builtins.
27077 PR tree-optimization/84339
27078 * gimple-fold.c (get_range_strlen): Set *FLEXP to true when handling
27079 ARRAY_REF where first operand is array_at_struct_end_p COMPONENT_REF.
27082 PR middle-end/84309
27083 * match.pd (pow(C,x) -> exp(log(C)*x)): Optimize instead into
27084 exp2(log2(C)*x) if C is a power of 2 and c99 runtime is available.
27085 * generic-match-head.c (canonicalize_math_after_vectorization_p): New
27087 * gimple-match-head.c (canonicalize_math_after_vectorization_p): New
27089 * omp-simd-clone.h: New file.
27090 * omp-simd-clone.c: Include omp-simd-clone.h.
27091 (expand_simd_clones): No longer static.
27092 * tree-vect-patterns.c: Include fold-const-call.h, attribs.h,
27093 cgraph.h and omp-simd-clone.h.
27094 (vect_recog_pow_pattern): Optimize pow(C,x) to exp(log(C)*x).
27095 (vect_recog_widen_shift_pattern): Formatting fix.
27096 (vect_pattern_recog_1): Don't check optab for calls.
27099 * config/i386/sse.md (<avx512>_vpermi2var<mode>3_mask): Force
27100 operands[2] into a REG before using gen_lowpart on it.
27102 2018-02-12 Jeff Law <law@redhat.com>
27105 * config/sh/sh.c (find_barrier): Consider a sibling call
27108 * cse.c (try_back_substitute_reg): Move any REG_ARGS_SIZE note when
27109 successfully back substituting a reg.
27111 2018-02-12 Richard Biener <rguenther@suse.de>
27113 PR tree-optimization/84037
27114 * tree-vect-slp.c (vect_analyze_slp_cost): Add visited
27115 parameter, move visited init to caller.
27116 (vect_slp_analyze_operations): Separate cost from validity
27117 check, initialize visited once for all instances.
27118 (vect_schedule_slp): Analyze map to CSE vectorized nodes once
27120 * tree-vect-stmts.c (vect_model_simple_cost): Make early
27122 (vect_model_promotion_demotion_cost): Likewise.
27123 (vectorizable_bswap): Guard cost modeling with !slp_node
27124 instead of !PURE_SLP_STMT to avoid double-counting on hybrid
27126 (vectorizable_call): Likewise.
27127 (vectorizable_conversion): Likewise.
27128 (vectorizable_assignment): Likewise.
27129 (vectorizable_shift): Likewise.
27130 (vectorizable_operation): Likewise.
27131 (vectorizable_store): Likewise.
27132 (vectorizable_load): Likewise.
27133 (vectorizable_condition): Likewise.
27134 (vectorizable_comparison): Likewise.
27136 2018-02-12 Paolo Bonzini <bonzini@gnu.org>
27139 * internal-fn.def (ASAN_CHECK): Fix fnspec to account for return value.
27140 (ASAN_MARK): Fix fnspec to account for return value, change pointer
27141 argument from 'R' to 'W' so that the pointed-to datum is clobbered.
27143 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
27145 PR middle-end/83665
27146 * params.def (inline-min-speedup): Increase from 8 to 15.
27147 (max-inline-insns-auto): Decrease from 40 to 30.
27148 * ipa-split.c (consider_split): Add some buffer for function to
27149 be considered inlining candidate.
27150 * invoke.texi (max-inline-insns-auto, inline-min-speedup): UPdate
27153 2018-02-12 Richard Biener <rguenther@suse.de>
27155 PR tree-optimization/84037
27156 * tree-vect-slp.c (vect_build_slp_tree_2): Try swapping the
27157 matched stmts if we cannot swap the non-matched ones.
27159 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
27161 * config/i386/avx512fintrin.h (_mm_mask_scalef_round_sd,
27162 _mm_maskz_scalef_round_sd, _mm_mask_scalef_round_ss,
27163 _mm_maskz_scalef_round_ss): New intrinsics.
27164 (__builtin_ia32_scalefsd_round, __builtin_ia32_scalefss_round): Fix.
27165 * config/i386/i386-builtin.def (__builtin_ia32_scalefsd_round,
27166 __builtin_ia32_scalefss_round): Remove.
27167 (__builtin_ia32_scalefsd_mask_round,
27168 __builtin_ia32_scalefss_mask_round): New intrinsics.
27169 * config/i386/sse.md (vmscalef<mode><round_name>): Renamed to ...
27170 (vmscalef<mode><mask_scalar_name><round_scalar_name>): ... this.
27171 ((match_operand:VF_128 2 "<round_nimm_predicate>"
27172 "<round_constraint>")): Changed to ...
27173 ((match_operand:VF_128 2 "<round_scalar_nimm_predicate>"
27174 "<round_scalar_constraint>")): ... this.
27175 ("vscalef<ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|
27176 %0, %1, %2<round_op3>}"): Changed to ...
27177 ("vscalef<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%2, %1,
27178 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %1,
27179 %2<round_scalar_mask_op3>}"): ... this.
27180 * config/i386/subst.md (round_scalar_nimm_predicate): New.
27182 2018-02-12 Olga Makhotina <olga.makhotina@intel.com>
27184 * config/i386/avx512fintrin.h (_mm_mask_sqrt_round_sd)
27185 (_mm_maskz_sqrt_round_sd, _mm_mask_sqrt_round_ss)
27186 (_mm_maskz_sqrt_round_ss): New intrinsics.
27187 (__builtin_ia32_sqrtsd_round, __builtin_ia32_sqrtss_round): Remove.
27188 (__builtin_ia32_sqrtsd_mask_round)
27189 (__builtin_ia32_sqrtss_mask_round): New builtins.
27190 * config/i386/i386-builtin.def (__builtin_ia32_sqrtsd_round)
27191 (__builtin_ia32_sqrtss_round): Remove.
27192 (__builtin_ia32_sqrtsd_mask_round)
27193 (__builtin_ia32_sqrtss_mask_round): New builtins.
27194 * config/i386/sse.md (vmsqrt<mode>2<round_name>): Renamed to ...
27195 (vmsqrt<mode>2<mask_scalar_name><round_scalar_name>): ... this.
27196 ((match_operand:VF_128 1 "vector_operand"
27197 "xBm,<round_constraint>")): Changed to ...
27198 ((match_operand:VF_128 1 "vector_operand"
27199 "xBm,<round_scalar_constraint>")): ... this.
27200 (vsqrt<ssescalarmodesuffix>\t{<round_op3>%1, %2, %0|
27201 %0, %2, %<iptr>1<round_op3>}): Changed to ...
27202 (vsqrt<ssescalarmodesuffix>\t{<round_scalar_mask_op3>%1, %2,
27203 %0<mask_scalar_operand3>|%0<mask_scalar_operand3>, %2,
27204 %<iptr>1<round_scalar_mask_op3>}): ... this.
27205 ((set_attr "prefix" "<round_prefix>")): Changed to ...
27206 ((set_attr "prefix" "<round_scalar_prefix>")): ... this.
27208 2018-02-11 Steven Munroe <munroesj@gcc.gnu.org>
27211 * config/rs6000/mmintrin.h (_mm_cmpeq_pi32 [_ARCH_PWR9]):
27212 Cast vec_cmpeq result to correct type.
27213 * config/rs6000/mmintrin.h (_mm_cmpgt_pi32 [_ARCH_PWR9]):
27214 Cast vec_cmpgt result to correct type.
27216 2018-02-11 Alexandre Oliva <aoliva@redhat.com>
27218 * final.c (final_scan_insn_1): Renamed from...
27219 (final_scan_insn): ... this. New wrapper, to recover
27220 seen from the outermost call in recursive ones.
27221 * config/sparc/sparc.c (output_return): Drop seen from call.
27222 (output_sibcall): Likewise.
27223 * config/visium/visium.c (output_branch): Likewise.
27225 2018-02-10 John David Anglin <danglin@gcc.gnu.org>
27227 * config/pa/pa.c (hppa_profile_hook): Mark SYMBOL_REF for _mcount as
27230 2018-02-10 Alan Modra <amodra@gmail.com>
27233 * config/rs6000/rs6000.md (split_stack_return): Remove (use ..).
27234 Specify LR as an input.
27236 2018-02-10 Jakub Jelinek <jakub@redhat.com>
27239 * omp-low.c (maybe_remove_omp_member_access_dummy_vars,
27240 remove_member_access_dummy_vars): New functions.
27241 (lower_omp_for, lower_omp_taskreg, lower_omp_target,
27242 lower_omp_1, execute_lower_omp): Use them.
27244 PR rtl-optimization/84308
27245 * shrink-wrap.c (spread_components): Release todo vector.
27247 2018-02-09 Vladimir Makarov <vmakarov@redhat.com>
27249 PR rtl-optimization/57193
27250 * ira-color.c (struct allocno_color_data): Add member
27251 conflict_allocno_hard_prefs.
27252 (update_conflict_allocno_hard_prefs): New.
27253 (bucket_allocno_compare_func): Add a preference based on
27254 conflict_allocno_hard_prefs.
27255 (push_allocno_to_stack): Update conflict_allocno_hard_prefs.
27256 (color_allocnos): Remove a dead code. Initiate
27257 conflict_allocno_hard_prefs. Call update_costs_from_prefs.
27259 2018-02-09 Jakub Jelinek <jakub@redhat.com>
27262 * config/rs6000/vsx.md (p9_xxbrq_v16qi): Change input operand
27263 constraint from =wa to wa. Avoid a subreg on the output operand,
27264 instead use a pseudo and subreg it in a move.
27265 (p9_xxbrd_<mode>): Changed to ...
27266 (p9_xxbrd_v2di): ... this insn, without VSX_D iterator.
27267 (p9_xxbrd_v2df): New expander.
27268 (p9_xxbrw_<mode>): Changed to ...
27269 (p9_xxbrw_v4si): ... this insn, without VSX_W iterator.
27270 (p9_xxbrw_v4sf): New expander.
27272 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
27274 * config/rx/rx.md (movsicc): Update expander to be matched by GCC.
27276 2018-02-09 Peter Bergner <bergner@vnet.ibm.com>
27279 * config/rs6000/vsx.md (vsx_mul_v2di): Handle generating a 64-bit
27280 multiply in 32-bit mode.
27281 (vsx_div_v2di): Handle generating a 64-bit signed divide in 32-bit mode.
27282 (vsx_udiv_v2di): Handle generating a 64-bit unsigned divide in 32-bit
27285 2018-02-09 Sebastian Perta <sebastian.perta@renesas.com>
27287 * config/rx/constraints.md (CALL_OP_SYMBOL_REF): Added new constraint
27288 to allow or block "symbol_ref" depending on the value of TARGET_JSR.
27289 * config/rx/rx.md (call_internal): Use CALL_OP_SYMBOL_REF.
27290 * config/rx/rx.md (call_value_internal): Use CALL_OP_SYMBOL_REF.
27292 2018-02-09 Pierre-Marie de Rodat <derodat@adacore.com>
27295 * dwarf2out.c (is_trivial_indirect_ref): New function.
27296 (dwarf2out_late_global_decl): Do not generate a location
27297 attribute for variables that have a non-trivial DECL_VALUE_EXPR
27298 and that are not defined in the current unit.
27300 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
27302 * optabs.c (prepare_cmp_insn): Try harder to emit a direct comparison
27303 instead of a libcall for UNORDERED.
27305 2018-02-09 Tamar Christina <tamar.christina@arm.com>
27308 * config/arm/arm-c.c (arm_cpu_builtins): Un-define __ARM_FEATURE_LDREX,
27309 __ARM_ARCH_PROFILE, __ARM_ARCH_ISA_THUMB, __ARM_FP and __ARM_NEON_FP.
27311 2018-02-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27314 * config/s390/s390.c (s390_set_current_function): Invoke
27315 s390_indirect_branch_settings also if fndecl didn't change.
27317 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
27319 * config/rs6000/rs6000.md (blockage): Set length to zero.
27321 2018-02-09 Eric Botcazou <ebotcazou@adacore.com>
27323 * expr.c (optimize_bitfield_assignment_op): Remove obsolete assertion.
27325 2018-02-09 Jakub Jelinek <jakub@redhat.com>
27328 * gcc.c (STATIC_LIBASAN_LIBS, STATIC_LIBTSAN_LIBS,
27329 STATIC_LIBLSAN_LIBS, STATIC_LIBUBSAN_LIBS): Handle -static like
27333 * var-tracking.c (vt_add_function_parameter): Punt for non-onepart
27334 PARALLEL incoming that failed vt_get_decl_and_offset check.
27336 PR middle-end/84237
27337 * output.h (bss_initializer_p): Add NAMED argument, defaulted to false.
27338 * varasm.c (bss_initializer_p): Add NAMED argument, if true, ignore
27340 (get_variable_section): For decls in named .bss* sections pass true as
27341 second argument to bss_initializer_p.
27343 2018-02-09 Marek Polacek <polacek@redhat.com>
27344 Jakub Jelinek <jakub@redhat.com>
27347 * fold-const.c (fold_indirect_ref_1): Use VECTOR_TYPE_P macro.
27348 Formatting fixes. Verify first that tree_fits_poly_int64_p (op01).
27349 Sync some changes from cxx_fold_indirect_ref.
27351 2018-02-09 Alexandre Oliva <aoliva@redhat.com>
27353 * cfgexpand.c (expand_gimple_basic_block): Handle inline entry
27355 * dwarf2out.c (dwarf2_debug_hooks): Enable inline_entry hook.
27356 (BLOCK_INLINE_ENTRY_LABEL): New.
27357 (dwarf2out_var_location): Disregard inline entry markers.
27358 (inline_entry_data): New struct.
27359 (inline_entry_data_hasher): New hashtable type.
27360 (inline_entry_data_hasher::hash): New.
27361 (inline_entry_data_hasher::equal): New.
27362 (inline_entry_data_table): New variable.
27363 (add_high_low_attributes): Add DW_AT_entry_pc and
27364 DW_AT_GNU_entry_view attributes if a pending entry is found
27365 in inline_entry_data_table. Add old entry_pc attribute only
27366 if debug nonbinding markers are disabled.
27367 (gen_inlined_subroutine_die): Set BLOCK_DIE if nonbinding
27368 markers are enabled.
27369 (block_within_block_p, dwarf2out_inline_entry): New.
27370 (dwarf2out_finish): Check that no entries remained in
27371 inline_entry_data_table.
27372 * final.c (reemit_insn_block_notes): Handle inline entry notes.
27373 (final_scan_insn, notice_source_line): Likewise.
27374 (rest_of_clean_state): Skip inline entry markers.
27375 * gimple-pretty-print.c (dump_gimple_debug): Handle inline entry
27377 * gimple.c (gimple_build_debug_inline_entry): New.
27378 * gimple.h (enum gimple_debug_subcode): Add
27379 GIMPLE_DEBUG_INLINE_ENTRY.
27380 (gimple_build_debug_inline_entry): Declare.
27381 (gimple_debug_inline_entry_p): New.
27382 (gimple_debug_nonbind_marker_p): Adjust.
27383 * insn-notes.def (INLINE_ENTRY): New.
27384 * print-rtl.c (rtx_writer::print_rtx_operand_code_0): Handle
27385 inline entry marker notes.
27386 (print_insn): Likewise.
27387 * rtl.h (NOTE_MARKER_P): Add INLINE_ENTRY support.
27388 (INSN_DEBUG_MARKER_KIND): Likewise.
27389 (GEN_RTX_DEBUG_MARKER_INLINE_ENTRY_PAT): New.
27390 * tree-inline.c (expand_call_inline): Build and insert
27391 debug_inline_entry stmt.
27392 * tree-ssa-live.c (remove_unused_scope_block_p): Preserve
27393 inline entry blocks early, if nonbind markers are enabled.
27394 (dump_scope_block): Dump fragment info.
27395 * var-tracking.c (reemit_marker_as_note): Handle inline entry note.
27396 * doc/gimple.texi (gimple_debug_inline_entry_p): New.
27397 (gimple_build_debug_inline_entry): New.
27398 * doc/invoke.texi (gstatement-frontiers, gno-statement-frontiers):
27399 Enable/disable inline entry points too.
27400 * doc/rtl.texi (NOTE_INSN_INLINE_ENTRY): New.
27401 (DEBUG_INSN): Describe inline entry markers.
27403 * common.opt (gvariable-location-views): New.
27404 (gvariable-location-views=incompat5): New.
27405 * config.in: Rebuilt.
27406 * configure: Rebuilt.
27407 * configure.ac: Test assembler for view support.
27408 * dwarf2asm.c (dw2_asm_output_symname_uleb128): New.
27409 * dwarf2asm.h (dw2_asm_output_symname_uleb128): Declare.
27410 * dwarf2out.c (var_loc_view): New typedef.
27411 (struct dw_loc_list_struct): Add vl_symbol, vbegin, vend.
27412 (dwarf2out_locviews_in_attribute): New.
27413 (dwarf2out_locviews_in_loclist): New.
27414 (dw_val_equal_p): Compare val_view_list of dw_val_class_view_lists.
27415 (enum dw_line_info_opcode): Add LI_adv_address.
27416 (struct dw_line_info_table): Add view.
27417 (RESET_NEXT_VIEW, RESETTING_VIEW_P): New macros.
27418 (DWARF2_ASM_VIEW_DEBUG_INFO): Define default.
27419 (zero_view_p): New variable.
27420 (ZERO_VIEW_P): New macro.
27421 (output_asm_line_debug_info): New.
27422 (struct var_loc_node): Add view.
27423 (add_AT_view_list, AT_loc_list): New.
27424 (add_var_loc_to_decl): Add view param. Test it against last.
27425 (new_loc_list): Add view params. Record them.
27426 (AT_loc_list_ptr): Handle loc and view lists.
27427 (view_list_to_loc_list_val_node): New.
27428 (print_dw_val): Handle dw_val_class_view_list.
27429 (size_of_die): Likewise.
27430 (value_format): Likewise.
27431 (loc_list_has_views): New.
27432 (gen_llsym): Set vl_symbol too.
27433 (maybe_gen_llsym, skip_loc_list_entry): New.
27434 (dwarf2out_maybe_output_loclist_view_pair): New.
27435 (output_loc_list): Output view list or entries too.
27436 (output_view_list_offset): New.
27437 (output_die): Handle dw_val_class_view_list.
27438 (output_dwarf_version): New.
27439 (output_compilation_unit_header): Use it.
27440 (output_skeleton_debug_sections): Likewise.
27441 (output_rnglists, output_line_info): Likewise.
27442 (output_pubnames, output_aranges): Update version comments.
27443 (output_one_line_info_table): Output view numbers in asm comments.
27444 (dw_loc_list): Determine current endview, pass it to new_loc_list.
27445 Call maybe_gen_llsym.
27446 (loc_list_from_tree_1): Adjust.
27447 (add_AT_location_description): Create view list attribute if
27448 needed, check it's absent otherwise.
27449 (convert_cfa_to_fb_loc_list): Adjust.
27450 (maybe_emit_file): Call output_asm_line_debug_info for test.
27451 (dwarf2out_var_location): Reset views as needed. Precompute
27452 add_var_loc_to_decl args. Call get_attr_min_length only if we have the
27453 attribute. Set view.
27454 (new_line_info_table): Reset next view.
27455 (set_cur_line_info_table): Call output_asm_line_debug_info for test.
27456 (dwarf2out_source_line): Likewise. Output view resets and labels to
27457 the assembler, or select appropriate line info opcodes.
27458 (prune_unused_types_walk_attribs): Handle dw_val_class_view_list.
27459 (optimize_string_length): Catch it. Adjust.
27460 (resolve_addr): Copy vl_symbol along with ll_symbol. Handle
27461 dw_val_class_view_list, and remove it if no longer needed.
27462 (hash_loc_list): Hash view numbers.
27463 (loc_list_hasher::equal): Compare them.
27464 (optimize_location_lists): Check whether a view list symbol is
27465 needed, and whether the locview attribute is present, and
27466 whether they match. Remove the locview attribute if no longer
27468 (index_location_lists): Call skip_loc_list_entry for test.
27469 (dwarf2out_finish): Call output_asm_line_debug_info for test.
27470 Use output_dwarf_version.
27471 * dwarf2out.h (enum dw_val_class): Add dw_val_class_view_list.
27472 (struct dw_val_node): Add val_view_list.
27473 * final.c (SEEN_NEXT_VIEW): New.
27474 (set_next_view_needed): New.
27475 (clear_next_view_needed): New.
27476 (maybe_output_next_view): New.
27477 (final_start_function): Rename to...
27478 (final_start_function_1): ... this. Take pointer to FIRST,
27479 add SEEN parameter. Emit param bindings in the initial view.
27480 (final_start_function): Reintroduce SEEN-less interface.
27481 (final): Rename to...
27482 (final_1): ... this. Take SEEN parameter. Output final pending
27483 next view at the end.
27484 (final): Reintroduce seen-less interface.
27485 (final_scan_insn): Output pending next view before switching
27486 sections or ending a block. Mark the next view as needed when
27487 outputting variable locations. Notify debug backend of section
27488 changes, and of location view changes.
27489 (rest_of_handle_final): Adjust.
27490 * toplev.c (process_options): Autodetect value for debug variable
27491 location views option. Warn on incompat5 without -gdwarf-5.
27492 * doc/invoke.texi (gvariable-location-views): New.
27493 (gvariable-location-views=incompat5): New.
27494 (gno-variable-location-views): New.
27496 2018-02-08 David Malcolm <dmalcolm@redhat.com>
27498 PR tree-optimization/84136
27499 * tree-cfg.c (find_taken_edge_computed_goto): Remove assertion
27500 that the result of find_edge is non-NULL.
27502 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
27505 * config/i386/x86-tune-costs.h (skylake_cost): Fix cost of
27506 storing integer register in SImode. Fix cost of 256 and 512
27507 byte aligned SSE register store.
27509 2018-02-08 Sergey Shalnov <sergey.shalnov@intel.com>
27511 * config/i386/i386.c (ix86_multiplication_cost): Fix
27512 multiplication cost for TARGET_AVX512DQ.
27514 2018-02-08 Marek Polacek <polacek@redhat.com>
27516 PR tree-optimization/84238
27517 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Verify the result of
27520 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
27522 PR tree-optimization/84265
27523 * tree-vect-stmts.c (vectorizable_store): Don't treat
27524 VMAT_CONTIGUOUS accesses as grouped.
27525 (vectorizable_load): Likewise.
27527 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
27529 PR tree-optimization/81635
27530 * wide-int.h (wi::round_down_for_mask, wi::round_up_for_mask): Declare.
27531 * wide-int.cc (wi::round_down_for_mask, wi::round_up_for_mask)
27532 (test_round_for_mask): New functions.
27533 (wide_int_cc_tests): Call test_round_for_mask.
27534 * tree-vrp.h (intersect_range_with_nonzero_bits): Declare.
27535 * tree-vrp.c (intersect_range_with_nonzero_bits): New function.
27536 * tree-data-ref.c (split_constant_offset_1): Use it to refine the
27537 range returned by get_range_info.
27539 2018-02-08 Jan Hubicka <hubicka@ucw.cz>
27542 * cgraph.h (symtab_node::output_to_lto_symbol_table_p): Declare
27543 * symtab.c: Include builtins.h
27544 (symtab_node::output_to_lto_symbol_table_p): Move here
27545 from lto-streamer-out.c:output_symbol_p.
27546 * lto-streamer-out.c (write_symbol): Turn early exit to assert.
27547 (output_symbol_p): Move all logic to symtab.c
27548 (produce_symtab): Update.
27550 2018-02-08 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
27552 * config/s390/s390-opts.h (enum indirect_branch): Define.
27553 * config/s390/s390-protos.h (s390_return_addr_from_memory)
27554 (s390_indirect_branch_via_thunk)
27555 (s390_indirect_branch_via_inline_thunk): Add function prototypes.
27556 (enum s390_indirect_branch_type): Define.
27557 * config/s390/s390.c (struct s390_frame_layout, struct
27558 machine_function): Remove.
27559 (indirect_branch_prez10thunk_mask, indirect_branch_z10thunk_mask)
27560 (indirect_branch_table_label_no, indirect_branch_table_name):
27562 (INDIRECT_BRANCH_NUM_OPTIONS): Define macro.
27563 (enum s390_indirect_branch_option): Define.
27564 (s390_return_addr_from_memory): New function.
27565 (s390_handle_string_attribute): New function.
27566 (s390_attribute_table): Add new attribute handler.
27567 (s390_execute_label): Handle UNSPEC_EXECUTE_JUMP patterns.
27568 (s390_indirect_branch_via_thunk): New function.
27569 (s390_indirect_branch_via_inline_thunk): New function.
27570 (s390_function_ok_for_sibcall): When jumping via thunk disallow
27571 sibling call optimization for non z10 compiles.
27572 (s390_emit_call): Force indirect branch target to be a single
27573 register. Add r1 clobber for non-z10 compiles.
27574 (s390_emit_epilogue): Emit return jump via return_use expander.
27575 (s390_reorg): Handle JUMP_INSNs as execute targets.
27576 (s390_option_override_internal): Perform validity checks for the
27577 new command line options.
27578 (s390_indirect_branch_attrvalue): New function.
27579 (s390_indirect_branch_settings): New function.
27580 (s390_set_current_function): Invoke s390_indirect_branch_settings.
27581 (s390_output_indirect_thunk_function): New function.
27582 (s390_code_end): Implement target hook.
27583 (s390_case_values_threshold): Implement target hook.
27584 (TARGET_ASM_CODE_END, TARGET_CASE_VALUES_THRESHOLD): Define target
27586 * config/s390/s390.h (struct s390_frame_layout)
27587 (struct machine_function): Move here from s390.c.
27588 (TARGET_INDIRECT_BRANCH_NOBP_RET)
27589 (TARGET_INDIRECT_BRANCH_NOBP_JUMP)
27590 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_THUNK)
27591 (TARGET_INDIRECT_BRANCH_NOBP_JUMP_INLINE_THUNK)
27592 (TARGET_INDIRECT_BRANCH_NOBP_CALL)
27593 (TARGET_DEFAULT_INDIRECT_BRANCH_TABLE)
27594 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EXRL)
27595 (TARGET_INDIRECT_BRANCH_THUNK_NAME_EX)
27596 (TARGET_INDIRECT_BRANCH_TABLE): Define macros.
27597 * config/s390/s390.md (UNSPEC_EXECUTE_JUMP)
27598 (INDIRECT_BRANCH_THUNK_REGNUM): Define constants.
27599 (mnemonic attribute): Add values which aren't recognized
27601 ("*cjump_long", "*icjump_long", "*basr", "*basr_r"): Disable
27602 pattern for branch conversion. Fix mnemonic attribute.
27603 ("*c<code>", "*sibcall_br", "*sibcall_value_br", "*return"): Emit
27604 indirect branch via thunk if requested.
27605 ("indirect_jump", "<code>"): Expand patterns for branch conversion.
27606 ("*indirect_jump"): Disable for branch conversion using out of
27608 ("indirect_jump_via_thunk<mode>_z10")
27609 ("indirect_jump_via_thunk<mode>")
27610 ("indirect_jump_via_inlinethunk<mode>_z10")
27611 ("indirect_jump_via_inlinethunk<mode>", "*casesi_jump")
27612 ("casesi_jump_via_thunk<mode>_z10", "casesi_jump_via_thunk<mode>")
27613 ("casesi_jump_via_inlinethunk<mode>_z10")
27614 ("casesi_jump_via_inlinethunk<mode>", "*basr_via_thunk<mode>_z10")
27615 ("*basr_via_thunk<mode>", "*basr_r_via_thunk_z10")
27616 ("*basr_r_via_thunk", "return<mode>_prez10"): New pattern.
27617 ("*indirect2_jump"): Disable for branch conversion.
27618 ("casesi_jump"): Turn into expander and expand patterns for branch
27620 ("return_use"): New expander.
27621 ("*return"): Emit return via thunk and rename it to ...
27622 ("*return<mode>"): ... this one.
27623 * config/s390/s390.opt: Add new options and and enum for the
27626 2018-02-08 Richard Sandiford <richard.sandiford@linaro.org>
27628 * lra-constraints.c (match_reload): Unconditionally use
27629 gen_lowpart_SUBREG, rather than selecting between that
27630 and equivalent gen_rtx_SUBREG code.
27632 2018-02-08 Richard Biener <rguenther@suse.de>
27634 PR tree-optimization/84233
27635 * tree-ssa-phiprop.c (propagate_with_phi): Use separate
27636 changed flag instead of boguously re-using phi_inserted.
27638 2018-02-08 Martin Jambor <mjambor@suse.cz>
27640 * hsa-gen.c (get_symbol_for_decl): Set program allocation for
27641 static local variables.
27643 2018-02-08 Richard Biener <rguenther@suse.de>
27645 PR tree-optimization/84278
27646 * tree-vect-stmts.c (vectorizable_store): When looking for
27647 smaller vector types to perform grouped strided loads/stores
27648 make sure the mode is supported by the target.
27649 (vectorizable_load): Likewise.
27651 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
27653 * config/aarch64/aarch64.c (aarch64_components_for_bb):
27654 Increase LDP/STP opportunities by adding adjacent callee-saves.
27656 2018-02-08 Wilco Dijkstra <wdijkstr@arm.com>
27658 PR rtl-optimization/84068
27659 PR rtl-optimization/83459
27660 * haifa-sched.c (rank_for_schedule): Fix SCHED_PRESSURE_MODEL sorting.
27662 2018-02-08 Aldy Hernandez <aldyh@redhat.com>
27664 PR tree-optimization/84224
27665 * gimple-ssa-warn-alloca.c (pass_walloca::execute): Remove assert.
27666 * calls.c (gimple_alloca_call_p): Only return TRUE when we have
27667 non-zero arguments.
27669 2018-02-07 Iain Sandoe <iain@codesourcery.com>
27672 * config/rs6000/altivec.md (*restore_world): Remove LR use.
27673 * config/rs6000/predicates.md (restore_world_operation): Adjust op
27674 count, remove one USE.
27676 2018-02-07 Michael Meissner <meissner@linux.vnet.ibm.com>
27678 * doc/install.texi (Configuration): Document the
27679 --with-long-double-format={ibm,ieee} PowerPC configuration
27683 * config/rs6000/rs6000.md (fix_trunc<SFDF:mode><QHI:mode>2):
27684 Convert from define_expand to be define_insn_and_split. Rework
27685 float/double/_Float128 conversions to QI/HI/SImode to work with
27686 both ISA 2.07 (power8) or ISA 3.0 (power9). Fix regression where
27687 conversions to QI/HImode types did a store and then a load to
27688 truncate the value. For conversions to VSX registers, don't split
27689 the insn, instead emit the code directly. Use the code iterator
27690 any_fix to combine signed and unsigned conversions.
27691 (fix<uns>_trunc<SFDF:mode>si2_p8): Likewise.
27692 (fixuns_trunc<SFDF:mode><QHI:mode>2): Likewise.
27693 (fix_trunc<IEEE128:mode><QHI:mode>2): Likewise.
27694 (fix<uns>_trunc<SFDF:mode><QHI:mode>2): Likewise.
27695 (fix_<mode>di2_hw): Likewise.
27696 (fixuns_<mode>di2_hw): Likewise.
27697 (fix_<mode>si2_hw): Likewise.
27698 (fixuns_<mode>si2_hw): Likewise.
27699 (fix<uns>_<IEEE128:mode><SDI:mode>2_hw): Likewise.
27700 (fix<uns>_trunc<IEEE128:mode><QHI:mode>2): Likewise.
27701 (fctiw<u>z_<mode>_smallint): Rename fctiw<u>z_<mode>_smallint to
27702 fix<uns>_trunc<SFDF:mode>si2_p8.
27703 (fix_trunc<SFDF:mode><QHI:mode>2_internal): Delete, no longer used.
27704 (fixuns_trunc<SFDF:mode><QHI:mode>2_internal): Likewise.
27705 (fix<uns>_<mode>_mem): Likewise.
27706 (fctiw<u>z_<mode>_mem): Likewise.
27707 (fix<uns>_<mode>_mem): Likewise.
27708 (fix<uns>_trunc<SFDF:mode><QHSI:mode>2_mem): On ISA 3.0, prevent
27709 the register allocator from doing a direct move to the GPRs to do
27710 a store, and instead use the ISA 3.0 store byte/half-word from
27711 vector register instruction. For IEEE 128-bit floating point,
27712 also optimize stores of 32-bit ints.
27713 (fix<uns>_trunc<IEEE128:mode><QHSI:mode>2_mem): Likewise.
27715 2018-02-07 Alan Hayward <alan.hayward@arm.com>
27717 * genextract.c (push_pathstr_operand): New function to support
27719 (walk_rtx): Call push_pathstr_operand.
27720 (print_path): Support [a-zA-Z].
27722 2018-02-07 Richard Biener <rguenther@suse.de>
27724 PR tree-optimization/84037
27725 * tree-vectorizer.h (struct _loop_vec_info): Add ivexpr_map member.
27726 (cse_and_gimplify_to_preheader): Declare.
27727 (vect_get_place_in_interleaving_chain): Likewise.
27728 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
27730 (_loop_vec_info::~_loop_vec_info): Delete it.
27731 (cse_and_gimplify_to_preheader): New function.
27732 * tree-vect-slp.c (vect_get_place_in_interleaving_chain): Export.
27733 * tree-vect-stmts.c (vectorizable_store): CSE base and steps.
27734 (vectorizable_load): Likewise. For grouped stores always base
27735 the IV on the first element.
27736 * tree-vect-loop-manip.c (vect_loop_versioning): Unshare versioning
27737 condition before gimplifying.
27739 2018-02-07 Jakub Jelinek <jakub@redhat.com>
27741 * tree-eh.c (operation_could_trap_helper_p): Ignore honor_trapv for
27742 *DIV_EXPR and *MOD_EXPR.
27744 2018-02-07 H.J. Lu <hongjiu.lu@intel.com>
27747 * config/i386/i386.c (ix86_option_override_internal): Mask out
27748 the CF_SET bit when checking -fcf-protection.
27750 2018-02-07 Tom de Vries <tom@codesourcery.com>
27753 * omp-expand.c (expand_oacc_collapse_init): Ensure diff_type is large
27756 2018-02-07 Richard Biener <rguenther@suse.de>
27758 PR tree-optimization/84204
27759 * tree-chrec.c (chrec_fold_plus_1): Remove size limiting in
27762 PR tree-optimization/84205
27763 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Also
27764 special-case isl_ast_op_zdiv_r.
27766 PR tree-optimization/84223
27767 * graphite-scop-detection.c (gather_bbs::before_dom_children):
27768 Only add conditions from within the region.
27769 (gather_bbs::after_dom_children): Adjust.
27771 2018-02-07 Georg-Johann Lay <avr@gjlay.de>
27774 * config/avr/avr.h (GENERAL_REGNO_P, GENERAL_REG_P): New macros.
27775 * config/avr/avr.md: Only post-reload split REG-REG moves if
27776 either register is GENERAL_REG_P.
27778 2018-02-07 Jakub Jelinek <jakub@redhat.com>
27780 PR tree-optimization/84235
27781 * tree-ssa-scopedtables.c
27782 (avail_exprs_stack::simplify_binary_operation): Fir MINUS_EXPR, punt
27783 if the subtraction is performed in floating point type where NaNs are
27784 honored. For *DIV_EXPR, punt for ALL_FRACT_MODE_Ps where we can't
27785 build 1. Formatting fix.
27787 2018-02-06 Jakub Jelinek <jakub@redhat.com>
27790 * config/i386/i386.c (rest_of_insert_endbranch): Only skip
27791 NOTE_INSN_CALL_ARG_LOCATION after a call, not anything else,
27792 and skip it regardless of bb boundaries. Use CALL_P macro,
27793 don't test INSN_P (insn) together with CALL_P or JUMP_P check
27794 unnecessarily, formatting fix.
27796 2018-02-06 Michael Collison <michael.collison@arm.com>
27798 * config/arm/thumb2.md:
27799 (*thumb2_mov_negscc): Split only if TARGET_THUMB2 && !arm_restrict_it.
27800 (*thumb_mov_notscc): Ditto.
27802 2018-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
27805 * config/rs6000/rs6000.md (su code attribute): Use "u" for
27806 unsigned_fix, not "s".
27808 2018-02-06 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
27810 * configure.ac (gcc_fn_eh_frame_ro): New function.
27811 (gcc_cv_as_cfi_directive): Check both 32 and 64-bit assembler for
27812 correct .eh_frame permissions.
27813 * configure: Regenerate.
27815 2018-02-06 Andrew Jenner <andrew@codeourcery.com>
27817 * doc/invoke.texi: Add section for the PowerPC SPE backend. Remove
27818 irrelevant options.
27820 2018-02-06 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
27822 * config/rs6000/rs6000.c (rs6000_option_override_internal):
27823 Display warning message for -mno-speculate-indirect-jumps.
27825 2018-02-06 Andrew Jenner <andrew@codesourcery.com>
27827 * config/powerpcspe/powerpcspe.opt: (msimple-fpu, mfpu) Add
27829 * config/powerpcspe/sysv4.opt (mbit-align): Likewise.
27831 2018-02-06 Aldy Hernandez <aldyh@redhat.com>
27833 PR tree-optimization/84225
27834 * tree-eh.c (find_trapping_overflow): Only call
27835 operation_no_trapping_overflow when ANY_INTEGRAL_TYPE_P.
27837 2018-02-06 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
27840 * config/i386/i386.c: Reimplement the check of possible options
27841 -mibt/-mshstk conbination. Change error messages.
27842 * doc/invoke.texi: Fix a typo: remove extra '='.
27844 2018-02-06 Marek Polacek <polacek@redhat.com>
27846 PR tree-optimization/84228
27847 * tree-ssa-strlen.c (maybe_diag_stxncpy_trunc): Skip debug statements.
27849 2018-02-06 Tamar Christina <tamar.christina@arm.com>
27852 * config/arm/arm.c (arm_print_asm_arch_directives): Record already
27853 emitted arch directives.
27854 * config/arm/arm-c.c (arm_cpu_builtins): Undefine __ARM_ARCH and
27855 __ARM_FEATURE_COPROC before changing architectures.
27857 2018-02-06 Richard Biener <rguenther@suse.de>
27859 * config/i386/i386.c (print_reg): Fix typo.
27860 (ix86_loop_unroll_adjust): Do not unroll beyond the original nunroll.
27862 2018-02-06 Eric Botcazou <ebotcazou@adacore.com>
27864 * configure: Regenerate.
27866 2018-02-05 Martin Sebor <msebor@redhat.com>
27868 PR tree-optimization/83369
27869 * tree-ssa-ccp.c (pass_post_ipa_warn::execute): Use %G to print
27872 2018-02-05 Martin Liska <mliska@suse.cz>
27874 * doc/invoke.texi: Cherry-pick upstream r323995.
27876 2018-02-05 Richard Sandiford <richard.sandiford@linaro.org>
27878 * ira.c (ira_init_register_move_cost): Adjust comment.
27880 2018-02-05 Martin Liska <mliska@suse.cz>
27882 PR gcov-profile/84137
27883 * doc/gcov.texi: Fix typo in documentation.
27885 2018-02-05 Martin Liska <mliska@suse.cz>
27887 PR gcov-profile/83879
27888 * doc/gcov.texi: Document necessity of --dynamic-list-data when
27889 using dlopen functionality.
27891 2018-02-05 Olga Makhotina <olga.makhotina@intel.com>
27893 * config/i386/avx512dqintrin.h (_mm_mask_range_sd, _mm_maskz_range_sd,
27894 _mm_mask_range_round_sd, _mm_maskz_range_round_sd, _mm_mask_range_ss,
27895 _mm_maskz_range_ss, _mm_mask_range_round_ss,
27896 _mm_maskz_range_round_ss): New intrinsics.
27897 (__builtin_ia32_rangesd128_round)
27898 (__builtin_ia32_rangess128_round): Remove.
27899 (__builtin_ia32_rangesd128_mask_round,
27900 __builtin_ia32_rangess128_mask_round): New builtins.
27901 * config/i386/i386-builtin.def (__builtin_ia32_rangesd128_round,
27902 __builtin_ia32_rangess128_round): Remove.
27903 (__builtin_ia32_rangesd128_mask_round,
27904 __builtin_ia32_rangess128_mask_round): New builtins.
27905 * config/i386/sse.md (ranges<mode><round_saeonly_name>): Renamed to ...
27906 (ranges<mode><mask_scalar_name><round_saeonly_scalar_name>): ... this.
27907 ((match_operand:VF_128 2 "<round_saeonly_nimm_predicate>"
27908 "<round_saeonly_constraint>")): Changed to ...
27909 ((match_operand:VF_128 2 "<round_saeonly_scalar_nimm_predicate>"
27910 "<round_saeonly_scalar_constraint>")): ... this.
27911 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
27912 %0, %1, %2<round_saeonly_op4>, %3}"): Changed to ...
27913 ("vrange<ssescalarmodesuffix>\t{%3, <round_saeonly_scalar_mask_op4>%2,
27914 %1, %0<mask_scalar_operand4>|%0<mask_scalar_operand4>, %1,
27915 %2<round_saeonly_scalar_mask_op4>, %3}"): ... this.
27917 2018-02-02 Andrew Jenner <andrew@codesourcery.com>
27919 * config/powerpcspe/powerpcspe.opt: Add Undocumented to irrelevant
27921 * config/powerpcspe/powerpcspe-tables.opt (rs6000_cpu_opt_value):
27922 Remove all values except native, 8540 and 8548.
27924 2018-02-02 H.J. Lu <hongjiu.lu@intel.com>
27926 * config/i386/i386.c (ix86_output_function_return): Pass
27927 INVALID_REGNUM, instead of -1, as invalid register number to
27928 indirect_thunk_name and output_indirect_thunk.
27930 2018-02-02 Julia Koval <julia.koval@intel.com>
27932 * config.gcc: Add -march=icelake.
27933 * config/i386/driver-i386.c (host_detect_local_cpu): Detect icelake.
27934 * config/i386/i386-c.c (ix86_target_macros_internal): Handle icelake.
27935 * config/i386/i386.c (processor_costs): Add m_ICELAKE.
27936 (PTA_ICELAKE, PTA_AVX512VNNI, PTA_GFNI, PTA_VAES, PTA_AVX512VBMI2,
27937 PTA_VPCLMULQDQ, PTA_RDPID, PTA_AVX512BITALG): New.
27938 (processor_target_table): Add icelake.
27939 (ix86_option_override_internal): Handle new PTAs.
27940 (get_builtin_code_for_version): Handle icelake.
27941 (M_INTEL_COREI7_ICELAKE): New.
27942 (fold_builtin_cpu): Handle icelake.
27943 * config/i386/i386.h (TARGET_ICELAKE, PROCESSOR_ICELAKE): New.
27944 * doc/invoke.texi: Add -march=icelake.
27946 2018-02-02 Julia Koval <julia.koval@intel.com>
27948 * config/i386/i386.c (ix86_option_override_internal): Change flags type
27949 to wide_int_bitmask.
27950 * wide-int-bitmask.h: New.
27952 2018-02-02 Igor Tsimbalist <igor.v.tsimbalist@intel.com>
27955 * config/i386/i386.md: Replace Pmode with word_mode in
27956 builtin_setjmp_setup and builtin_longjmp to support x32.
27958 2018-02-01 Peter Bergner <bergner@vnet.ibm.com>
27962 * config/rs6000/driver-rs6000.c: #include "diagnostic.h".
27964 (rs6000_supported_cpu_names): New static variable.
27965 (linux_cpu_translation_table): Likewise.
27966 (elf_platform) <cpu>: Define new static variable and use it.
27967 Translate kernel AT_PLATFORM name to canonical name if needed.
27968 Error if platform name is unknown.
27970 2018-02-01 Aldy Hernandez <aldyh@redhat.com>
27973 * config/pa/predicates.md (base14_operand): Handle E_VOIDmode.
27975 2018-02-01 Jeff Law <law@redhat.com>
27978 * config/i386/i386.c (release_scratch_register_on_entry): Add new
27979 OFFSET and RELEASE_VIA_POP arguments. Use SP+OFFSET to restore
27980 the scratch if RELEASE_VIA_POP is false.
27981 (ix86_adjust_stack_and_probe_stack_clash): Un-constify SIZE.
27982 If we have to save a temporary register, decrement SIZE appropriately.
27983 Pass new arguments to release_scratch_register_on_entry.
27984 (ix86_adjust_stack_and_probe): Likewise.
27985 (ix86_emit_probe_stack_range): Pass new arguments to
27986 release_scratch_register_on_entry.
27988 2018-02-01 Uros Bizjak <ubizjak@gmail.com>
27990 PR rtl-optimization/84157
27991 * combine.c (change_zero_ext): Use REG_P predicate in
27992 front of HARD_REGISTER_P predicate.
27994 2018-02-01 Georg-Johann Lay <avr@gjlay.de>
27996 * config/avr/avr.c (avr_option_override): Move disabling of
27997 -fdelete-null-pointer-checks to...
27998 * common/config/avr/avr-common.c (avr_option_optimization_table):
28001 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28003 PR tree-optimization/81635
28004 * tree-data-ref.c (split_constant_offset_1): For types that
28005 wrap on overflow, try to use range info to prove that wrapping
28008 2018-02-01 Renlin Li <renlin.li@arm.com>
28011 * config/aarch64/aarch64.c (aarch64_class_max_nregs): Handle
28012 TAILCALL_ADDR_REGS.
28013 (aarch64_register_move_cost): Likewise.
28014 * config/aarch64/aarch64.h (reg_class): Rename CALLER_SAVE_REGS to
28015 TAILCALL_ADDR_REGS.
28016 (REG_CLASS_NAMES): Likewise.
28017 (REG_CLASS_CONTENTS): Rename CALLER_SAVE_REGS to
28018 TAILCALL_ADDR_REGS. Remove IP registers.
28019 * config/aarch64/aarch64.md (Ucs): Update register constraint.
28021 2018-02-01 Richard Biener <rguenther@suse.de>
28023 * domwalk.h (dom_walker::dom_walker): Add additional constructor
28024 for specifying RPO order and allow NULL for that.
28025 * domwalk.c (dom_walker::dom_walker): Likewise.
28026 (dom_walker::walk): Handle NULL RPO order.
28027 * tree-into-ssa.c (rewrite_dom_walker): Do not walk dom children
28029 (rewrite_update_dom_walker): Likewise.
28030 (mark_def_dom_walker): Likewise.
28032 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28034 * config/aarch64/aarch64-protos.h (aarch64_split_sve_subreg_move)
28035 (aarch64_maybe_expand_sve_subreg_move): Declare.
28036 * config/aarch64/aarch64.md (UNSPEC_REV_SUBREG): New unspec.
28037 * config/aarch64/predicates.md (aarch64_any_register_operand): New
28039 * config/aarch64/aarch64-sve.md (mov<mode>): Optimize subreg moves
28040 that are semantically a reverse operation.
28041 (*aarch64_sve_mov<mode>_subreg_be): New pattern.
28042 * config/aarch64/aarch64.c (aarch64_maybe_expand_sve_subreg_move):
28043 (aarch64_replace_reg_mode, aarch64_split_sve_subreg_move): New
28045 (aarch64_can_change_mode_class): For big-endian, forbid changes
28046 between two SVE modes if they have different element sizes.
28048 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28050 * config/aarch64/aarch64.c (aarch64_expand_sve_const_vector): Prefer
28051 the TImode handling for big-endian targets.
28053 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28055 * config/aarch64/aarch64-sve.md (sve_ld1rq): Replace with...
28056 (*sve_ld1rq<Vesize>): ... this new pattern. Handle all element sizes,
28058 * config/aarch64/aarch64.c (aarch64_expand_sve_widened_duplicate):
28059 Remove BSWAP handing for big-endian targets and use the form of
28060 LD1RQ appropariate for the mode.
28062 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28064 * config/aarch64/aarch64.c (aarch64_simd_valid_immediate): Handle
28065 all CONST_VECTOR_DUPLICATE_P vectors, not just those with a single
28066 duplicated element.
28068 2018-02-01 Richard Sandiford <richard.sandiford@linaro.org>
28071 * config/aarch64/aarch64.c (aarch64_secondary_reload): Tighten
28072 check for operands that need to go through aarch64_sve_reload_be.
28074 2018-02-01 Jakub Jelinek <jakub@redhat.com>
28076 PR tree-optimization/81661
28077 PR tree-optimization/84117
28078 * tree-eh.h (rewrite_to_non_trapping_overflow): Declare.
28079 * tree-eh.c: Include gimplify.h.
28080 (find_trapping_overflow, replace_trapping_overflow,
28081 rewrite_to_non_trapping_overflow): New functions.
28082 * tree-vect-loop.c: Include tree-eh.h.
28083 (vect_get_loop_niters): Use rewrite_to_non_trapping_overflow.
28084 * tree-data-ref.c: Include tree-eh.h.
28085 (get_segment_min_max): Use rewrite_to_non_trapping_overflow.
28087 2018-01-31 Uros Bizjak <ubizjak@gmail.com>
28089 PR rtl-optimization/84123
28090 * combine.c (change_zero_ext): Check if hard register satisfies
28091 can_change_dest_mode before calling gen_lowpart_SUBREG.
28093 2018-01-31 Vladimir Makarov <vmakarov@redhat.com>
28096 * ira.c (ira_init_register_move_cost): Remove assert.
28098 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
28100 PR rtl-optimization/84071
28101 * doc/tm.texi.in (WORD_REGISTER_OPERATIONS): Add explicit case.
28102 * doc/tm.texi: Regenerate.
28104 2018-01-31 Richard Biener <rguenther@suse.de>
28106 PR tree-optimization/84132
28107 * tree-data-ref.c (analyze_miv_subscript): Properly
28108 check whether evolution_function_is_affine_multivariate_p
28109 before calling gcd_of_steps_may_divide_p.
28111 2018-01-31 Julia Koval <julia.koval@intel.com>
28114 * config/i386/i386.c (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
28115 * config/i386/i386.md (rdpid_rex64) New.
28116 (rdpid): Make 32bit only.
28118 2018-01-29 Aldy Hernandez <aldyh@redhat.com>
28121 * tree-pretty-print.c (dump_generic_node): Handle a TYPE_NAME with
28122 an IDENTIFIER_NODE for FUNCTION_TYPE's.
28124 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
28127 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
28129 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
28131 2018-01-31 Eric Botcazou <ebotcazou@adacore.com>
28133 PR rtl-optimization/84071
28134 * combine.c (record_dead_and_set_regs_1): Record the source unmodified
28135 for a paradoxical SUBREG on a WORD_REGISTER_OPERATIONS target.
28137 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
28139 * config/arc/arc.c (arc_handle_aux_attribute): New function.
28140 (arc_attribute_table): Add 'aux' attribute.
28141 (arc_in_small_data_p): Consider aux like variables.
28142 (arc_is_aux_reg_p): New function.
28143 (arc_asm_output_aligned_decl_local): Ignore 'aux' like variables.
28144 (arc_get_aux_arg): New function.
28145 (prepare_move_operands): Handle aux-register access.
28146 (arc_handle_aux_attribute): New function.
28147 * doc/extend.texi (ARC Variable attributes): Add subsection.
28149 2018-01-31 Claudiu Zissulescu <claziss@synopsys.com>
28151 * config/arc/arc-protos.h (arc_is_uncached_mem_p): Function proto.
28152 * config/arc/arc.c (arc_handle_uncached_attribute): New function.
28153 (arc_attribute_table): Add 'uncached' attribute.
28154 (arc_print_operand): Print '.di' flag for uncached memory
28156 (arc_in_small_data_p): Do not consider for small data the uncached
28158 (arc_is_uncached_mem_p): New function.
28159 * config/arc/predicates.md (compact_store_memory_operand): Check
28160 for uncached memory accesses.
28161 (nonvol_nonimm_operand): Likewise.
28162 * doc/extend.texi (ARC Type Attribute): New subsection.
28164 2018-01-31 Jakub Jelinek <jakub@redhat.com>
28167 * common.opt (falign-functions=, falign-jumps=, falign-labels=,
28168 falign-loops=): Add Optimization flag.
28170 2018-01-30 Jeff Law <law@redhat.com>
28173 * i386.c (ix86_adjust_stack_and_probe_stack_clash): New argument
28174 INT_REGISTERS_SAVED. Check it prior to calling
28175 get_scratch_register_on_entry.
28176 (ix86_adjust_stack_and_probe): Similarly.
28177 (ix86_emit_probe_stack_range): Similarly.
28178 (ix86_expand_prologue): Corresponding changes.
28180 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28183 * config/sol2.h (STARTFILE_ARCH_SPEC): Use -std=c*,
28184 -std=iso9899:199409 instead of -pedantic to select values-Xc.o.
28186 2018-01-30 Vladimir Makarov <vmakarov@redhat.com>
28189 * lra-constraints.c (curr_insn_transform): Process AND in the
28192 2018-01-30 Jakub Jelinek <jakub@redhat.com>
28194 PR rtl-optimization/83986
28195 * sched-deps.c (sched_analyze_insn): For frame related insns, add anti
28196 dependence against last_pending_memory_flush in addition to
28197 pending_jump_insns.
28199 2018-01-30 Alexandre Oliva <aoliva@redhat.com>
28201 PR tree-optimization/81611
28202 * tree-ssa-dom.c (simple_iv_increment_p): Skip intervening
28205 2018-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
28208 * config/rs6000/rs6000.c (rs6000_internal_arg_pointer): Only return
28211 2018-01-30 Richard Biener <rguenther@suse.de>
28212 Jakub Jelinek <jakub@redhat.com>
28214 PR tree-optimization/84111
28215 * tree-ssa-loop-ivcanon.c (tree_unroll_loops_completely_1): Skip
28216 inner loops added during recursion, as they don't have up-to-date
28219 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
28222 * ipa-inline.c (can_inline_edge_p): Break out late tests to...
28223 (can_inline_edge_by_limits_p): ... here.
28224 (can_early_inline_edge_p, check_callers,
28225 update_caller_keys, update_callee_keys, recursive_inlining,
28226 add_new_edges_to_heap, speculation_useful_p,
28227 inline_small_functions,
28228 inline_small_functions, flatten_function,
28229 inline_to_all_callers_1): Update.
28231 2018-01-30 Jan Hubicka <hubicka@ucw.cz>
28233 * profile-count.c (profile_count::combine_with_ipa_count): Handle
28236 2018-01-30 Richard Biener <rguenther@suse.de>
28238 PR tree-optimization/83008
28239 * tree-vect-slp.c (vect_analyze_slp_cost_1): Properly cost
28240 invariant and constant vector uses in stmts when they need
28241 more than one stmt.
28243 2018-01-30 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28246 * configure.ac (gcc_cv_as_shf_merge): Disable on Solaris 10/x86.
28247 * configure: Regenerate.
28249 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
28251 * config/aarch64/aarch64-sve.md (*vec_extract<mode><Vel>_0): New
28253 (*vec_extract<mode><Vel>_v128): Require a nonzero lane number.
28254 Use gen_rtx_REG rather than gen_lowpart.
28256 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
28258 * lra-constraints.c (match_reload): Use subreg_lowpart_offset
28259 rather than 0 when creating partial subregs.
28261 2018-01-30 Richard Sandiford <richard.sandiford@linaro.org>
28263 * vec-perm-indices.c (vec_perm_indices::series_p): Give examples
28266 2018-01-29 Michael Meissner <meissner@linux.vnet.ibm.com>
28269 * config/rs6000/rs6000.c (rs6000_setup_reg_addr_masks): If DFmode
28270 and SFmode can go in Altivec registers (-mcpu=power7 for DFmode,
28271 -mcpu=power8 for SFmode) don't set the PRE_INCDEC or PRE_MODIFY
28272 flags. This restores the settings used before the 2017-07-24.
28273 Turning off pre increment/decrement/modify allows IVOPTS to
28274 optimize DF/SF loops where the index is an int.
28276 2018-01-29 Richard Biener <rguenther@suse.de>
28277 Kelvin Nilsen <kelvin@gcc.gnu.org>
28280 * tree-vect-stmts.c (vectorizable_call): Don't call
28281 targetm.vectorize_builtin_md_vectorized_function if callee is
28284 2018-01-22 Carl Love <cel@us.ibm.com>
28286 * doc/extend.tex: Fix typo in second arg in
28287 __builtin_bcdadd_{lt|eq|gt|ov} and __builtin_bcdsub_{lt|eq|gt|ov}.
28289 2018-01-29 Richard Biener <rguenther@suse.de>
28291 PR tree-optimization/84086
28292 * tree-ssanames.c: Include cfgloop.h and tree-scalar-evolution.h.
28293 (flush_ssaname_freelist): When SSA names were released reset
28294 the SCEV hash table.
28296 2018-01-29 Richard Biener <rguenther@suse.de>
28298 PR tree-optimization/84057
28299 * tree-ssa-loop-ivcanon.c (unloop_loops): Deal with already
28300 removed paths when removing edges.
28302 2018-01-27 H.J. Lu <hongjiu.lu@intel.com>
28304 * doc/invoke.texi: Replace -mfunction-return==@var{choice} with
28305 -mfunction-return=@var{choice}.
28307 2018-01-27 Bernd Edlinger <bernd.edlinger@hotmail.de>
28309 PR diagnostic/84034
28310 * diagnostic-show-locus.c (get_line_width_without_trailing_whitespace):
28311 Handle CR like TAB.
28312 (layout::print_source_line): Likewise.
28313 (test_get_line_width_without_trailing_whitespace): Add test cases.
28315 2018-01-27 Jakub Jelinek <jakub@redhat.com>
28317 PR middle-end/84040
28318 * sched-deps.c (sched_macro_fuse_insns): Return immediately for
28321 2018-01-26 Jim Wilson <jimw@sifive.com>
28323 * config/riscv/riscv.h (MAX_FIXED_MODE_SIZE): New.
28325 * config/riscv/elf.h (LIB_SPEC): Don't include -lgloss when nosys.specs
28328 2018-01-26 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28330 * config/aarch64/aarch64.md: Add peepholes for CMP + SUB -> SUBS
28331 and CMP + SUB-immediate -> SUBS.
28333 2018-01-26 Martin Sebor <msebor@redhat.com>
28335 PR tree-optimization/83896
28336 * tree-ssa-strlen.c (get_string_len): Rename...
28337 (get_string_cst_length): ...to this. Return HOST_WIDE_INT.
28338 Avoid assuming length is constant.
28339 (handle_char_store): Use HOST_WIDE_INT for string length.
28341 2018-01-26 Uros Bizjak <ubizjak@gmail.com>
28344 * config/i386/i386.md (*andndi3_doubleword): Add earlyclobber
28345 to (=&r,r,rm) alternative. Add (=r,0,rm) and (=r,r,0) alternatives.
28347 2018-01-26 Richard Biener <rguenther@suse.de>
28349 PR rtl-optimization/84003
28350 * dse.c (record_store): Only record redundant stores when
28351 the earlier store aliases at least all accesses the later one does.
28353 2018-01-26 Jakub Jelinek <jakub@redhat.com>
28355 PR rtl-optimization/83985
28356 * dce.c (deletable_insn_p): Return false for separate shrink wrapping
28357 REG_CFA_RESTORE insns.
28358 (delete_unmarked_insns): Don't ignore separate shrink wrapping
28359 REG_CFA_RESTORE insns here.
28362 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Don't
28363 use SSA_NAME_VAR as base for SSA_NAMEs with non-NULL SSA_NAME_VAR.
28365 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
28367 * config/arc/arc-arch.h (arc_tune_attr): Add ARC_TUNE_CORE_3.
28368 * config/arc/arc.c (arc_sched_issue_rate): Use ARC_TUNE_... .
28369 (arc_init): Likewise.
28370 (arc_override_options): Likewise.
28371 (arc_file_start): Choose Tag_ARC_CPU_variation based on arc_tune
28373 (hwloop_fail): Use TARGET_DBNZ when we want to check for dbnz insn
28375 * config/arc/arc.h (TARGET_DBNZ): Define.
28376 * config/arc/arc.md (attr tune): Add core_3, use ARC_TUNE_... to
28377 properly set the tune attribute.
28378 (dbnz): Use TARGET_DBNZ guard.
28379 * config/arc/arc.opt (mtune): Add core3 option.
28381 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
28383 * config/arc/arc.c (arc_delegitimize_address_0): Refactored to
28384 recognize new pic like addresses.
28385 (arc_delegitimize_address): Clean up.
28387 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
28389 * config/arc/arc-arches.def: Option mrf16 valid for all
28391 * config/arc/arc-c.def (__ARC_RF16__): New predefined macro.
28392 * config/arc/arc-cpus.def (em_mini): New cpu with rf16 on.
28393 * config/arc/arc-options.def (FL_RF16): Add mrf16 option.
28394 * config/arc/arc-tables.opt: Regenerate.
28395 * config/arc/arc.c (arc_conditional_register_usage): Handle
28396 reduced register file case.
28397 (arc_file_start): Set must have build attributes.
28398 * config/arc/arc.h (MAX_ARC_PARM_REGS): Conditional define using
28399 mrf16 option value.
28400 * config/arc/arc.opt (mrf16): Add new option.
28401 * config/arc/elf.h (ATTRIBUTE_PCS): Define.
28402 * config/arc/genmultilib.awk: Handle new mrf16 option.
28403 * config/arc/linux.h (ATTRIBUTE_PCS): Define.
28404 * config/arc/t-multilib: Regenerate.
28405 * doc/invoke.texi (ARC Options): Document mrf16 option.
28407 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
28409 * config/arc/arc-protos.h: Add arc_is_secure_call_p proto.
28410 * config/arc/arc.c (arc_handle_secure_attribute): New function.
28411 (arc_attribute_table): Add 'secure_call' attribute.
28412 (arc_print_operand): Print secure call operand.
28413 (arc_function_ok_for_sibcall): Don't optimize tail calls when
28415 (arc_is_secure_call_p): New function. * config/arc/arc.md
28416 (call_i): Add support for sjli instruction.
28417 (call_value_i): Likewise.
28418 * config/arc/constraints.md (Csc): New constraint.
28420 2018-01-26 Claudiu Zissulescu <claziss@synopsys.com>
28421 John Eric Martin <John.Martin@emmicro-us.com>
28423 * config/arc/arc-protos.h: Add arc_is_jli_call_p proto.
28424 * config/arc/arc.c (_arc_jli_section): New struct.
28425 (arc_jli_section): New type.
28426 (rc_jli_sections): New static variable.
28427 (arc_handle_jli_attribute): New function.
28428 (arc_attribute_table): Add jli_always and jli_fixed attribute.
28429 (arc_file_end): New function.
28430 (TARGET_ASM_FILE_END): Define.
28431 (arc_print_operand): Reuse 'S' letter for JLI output instruction.
28432 (arc_add_jli_section): New function.
28433 (jli_call_scan): Likewise.
28434 (arc_reorg): Call jli_call_scan.
28435 (arc_output_addsi): Remove 'S' from printing asm operand.
28436 (arc_is_jli_call_p): New function.
28437 * config/arc/arc.md (movqi_insn): Remove 'S' from printing asm
28439 (movhi_insn): Likewise.
28440 (movsi_insn): Likewise.
28441 (movsi_set_cc_insn): Likewise.
28442 (loadqi_update): Likewise.
28443 (load_zeroextendqisi_update): Likewise.
28444 (load_signextendqisi_update): Likewise.
28445 (loadhi_update): Likewise.
28446 (load_zeroextendhisi_update): Likewise.
28447 (load_signextendhisi_update): Likewise.
28448 (loadsi_update): Likewise.
28449 (loadsf_update): Likewise.
28450 (movsicc_insn): Likewise.
28451 (bset_insn): Likewise.
28452 (bxor_insn): Likewise.
28453 (bclr_insn): Likewise.
28454 (bmsk_insn): Likewise.
28455 (bicsi3_insn): Likewise.
28456 (cmpsi_cc_c_insn): Likewise.
28457 (movsi_ne): Likewise.
28458 (movsi_cond_exec): Likewise.
28459 (clrsbsi2): Likewise.
28460 (norm_f): Likewise.
28470 (call_i): Remove 'S' asm letter, add jli instruction.
28471 (call_value_i): Likewise.
28472 * config/arc/arc.op (mjli-always): New option.
28473 * config/arc/constraints.md (Cji): New constraint.
28474 * config/arc/fpx.md (addsf3_fpx): Remove 'S' from printing asm
28476 (subsf3_fpx): Likewise.
28477 (mulsf3_fpx): Likewise.
28478 * config/arc/simdext.md (vendrec_insn): Remove 'S' from printing
28480 * doc/extend.texi (ARC): Document 'jli-always' and 'jli-fixed'
28481 function attrbutes.
28482 * doc/invoke.texi (ARC): Document mjli-always option.
28484 2018-01-26 Sebastian Perta <sebastian.perta@renesas.com>
28486 * config/rl78/rl78.c (rl78_addsi3_internal): If operand 2 is const
28487 avoid addition with 0 and use incw and decw where possible.
28489 2018-01-26 Richard Biener <rguenther@suse.de>
28491 PR tree-optimization/81082
28492 * fold-const.c (fold_plusminus_mult_expr): Do not perform the
28493 association if it requires casting to unsigned.
28494 * match.pd ((A * C) +- (B * C) -> (A+-B)): New patterns derived
28495 from fold_plusminus_mult_expr to catch important cases late when
28496 range info is available.
28498 2018-01-26 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
28500 * config/i386/sol2.h (USE_HIDDEN_LINKONCE): Remove.
28501 * configure.ac (hidden_linkonce): New test.
28502 * configure: Regenerate.
28503 * config.in: Regenerate.
28505 2018-01-26 Julia Koval <julia.koval@intel.com>
28507 * config/i386/avx512bitalgintrin.h (_mm512_bitshuffle_epi64_mask,
28508 _mm512_mask_bitshuffle_epi64_mask, _mm256_bitshuffle_epi64_mask,
28509 _mm256_mask_bitshuffle_epi64_mask, _mm_bitshuffle_epi64_mask,
28510 _mm_mask_bitshuffle_epi64_mask): Fix type.
28511 * config/i386/i386-builtin-types.def (UHI_FTYPE_V2DI_V2DI_UHI,
28512 USI_FTYPE_V4DI_V4DI_USI): Remove.
28513 * config/i386/i386-builtin.def (__builtin_ia32_vpshufbitqmb512_mask,
28514 __builtin_ia32_vpshufbitqmb256_mask,
28515 __builtin_ia32_vpshufbitqmb128_mask): Fix types.
28516 * config/i386/i386.c (ix86_expand_args_builtin): Remove old types.
28517 * config/i386/sse.md (VI1_AVX512VLBW): Change types.
28519 2018-01-26 Alan Modra <amodra@gmail.com>
28522 * config/rs6000/rs6000-p8swap.c (rtx_is_swappable_p): Exclude
28523 UNSPEC_VBPERMQ. Sort other unspecs.
28525 2018-01-25 David Edelsohn <dje.gcc@gmail.com>
28527 * doc/invoke.texi (PowerPC Options): Document 'native' cpu type.
28529 2018-01-25 Jan Hubicka <hubicka@ucw.cz>
28531 PR middle-end/83055
28532 * predict.c (drop_profile): Do not push/pop cfun; update also
28534 (handle_missing_profiles): Fix logic looking for zero profiles.
28536 2018-01-25 Jakub Jelinek <jakub@redhat.com>
28538 PR middle-end/83977
28539 * ipa-fnsummary.c (compute_fn_summary): Clear can_change_signature
28540 on functions with #pragma omp declare simd or functions with simd
28542 * omp-simd-clone.c (expand_simd_clones): Revert 2018-01-24 change.
28543 * config/i386/i386.c (ix86_simd_clone_compute_vecsize_and_simdlen):
28544 Remove trailing \n from warning_at calls.
28546 2018-01-25 Tom de Vries <tom@codesourcery.com>
28549 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
28550 for neutered workers.
28552 2018-01-24 Joseph Myers <joseph@codesourcery.com>
28555 * config/m68k/m68k.c (m68k_promote_function_mode): New function.
28556 (TARGET_PROMOTE_FUNCTION_MODE): New macro.
28558 2018-01-24 Jeff Law <law@redhat.com>
28561 * i386.c (get_probe_interval): Move to earlier point.
28562 (ix86_compute_frame_layout): If -fstack-clash-protection and
28563 the frame is larger than the probe interval, then use pushes
28564 to save registers rather than reg->mem moves.
28565 (ix86_expand_prologue): Remove conditional for int_registers_saved
28568 2018-01-24 Vladimir Makarov <vmakarov@redhat.com>
28571 * ira-build.c (setup_min_max_allocno_live_range_point): Set up
28572 min/max for never referenced object.
28574 2018-01-24 Jakub Jelinek <jakub@redhat.com>
28576 PR middle-end/83977
28577 * tree.c (free_lang_data_in_decl): Don't clear DECL_ABSTRACT_ORIGIN
28579 * omp-low.c (create_omp_child_function): Remove "omp declare simd"
28580 attributes from DECL_ATTRIBUTES (decl) without affecting
28581 DECL_ATTRIBUTES (current_function_decl).
28582 * omp-simd-clone.c (expand_simd_clones): Ignore DECL_ARTIFICIAL
28583 functions with non-NULL DECL_ABSTRACT_ORIGIN.
28585 2018-01-24 Richard Sandiford <richard.sandiford@linaro.org>
28587 PR tree-optimization/83979
28588 * fold-const.c (fold_comparison): Use constant_boolean_node
28589 instead of boolean_{true,false}_node.
28591 2018-01-24 Jan Hubicka <hubicka@ucw.cz>
28593 * ipa-profile.c (ipa_propagate_frequency_1): Fix logic skipping calls
28596 2018-01-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28598 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
28599 Simplify the clause that sets the length attribute.
28600 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
28601 (*sibcall_nonlocal_sysv<mode>): Clean up code block; simplify the
28602 clause that sets the length attribute.
28603 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
28605 2018-01-24 Tom de Vries <tom@codesourcery.com>
28608 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG_2): Define to 1.
28609 (nvptx_pc_set, nvptx_condjump_label): New function. Copy from jump.c.
28610 Add strict parameter.
28611 (prevent_branch_around_nothing): Insert dummy insn between branch to
28612 label and label with no ptx insn inbetween.
28613 * config/nvptx/nvptx.md (define_insn "fake_nop"): New insn.
28615 2018-01-24 Tom de Vries <tom@codesourcery.com>
28618 * config/nvptx/nvptx.c (nvptx_single): Add exit insn after noreturn call
28619 for neutered threads in warp.
28620 * config/nvptx/nvptx.md (define_insn "exit"): New insn.
28622 2018-01-24 Richard Biener <rguenther@suse.de>
28624 PR tree-optimization/83176
28625 * tree-chrec.c (chrec_fold_plus_1): Handle (signed T){(T) .. }
28628 2018-01-24 Richard Biener <rguenther@suse.de>
28630 PR tree-optimization/82819
28631 * graphite-isl-ast-to-gimple.c (binary_op_to_tree): Avoid
28632 code generating pluses that are no-ops in the target precision.
28634 2018-01-24 Richard Biener <rguenther@suse.de>
28636 PR middle-end/84000
28637 * tree-cfg.c (replace_loop_annotate): Handle annot_expr_parallel_kind.
28639 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
28641 * cfgcleanup.c (try_crossjump_to_edge): Use combine_with_count
28642 to merge probabilities.
28643 * predict.c (probably_never_executed): Also mark as cold functions
28644 with global 0 profile and guessed local profile.
28645 * profile-count.c (profile_probability::combine_with_count): New
28647 * profile-count.h (profile_probability::operator*,
28648 profile_probability::operator*=, profile_probability::operator/,
28649 profile_probability::operator/=): Reduce precision to adjusted
28650 and set value to guessed on contradictory divisions.
28651 (profile_probability::combine_with_freq): Remove.
28652 (profile_probability::combine_wiht_count): Declare.
28653 (profile_count::force_nonzero):: Set to adjusted.
28654 (profile_count::probability_in):: Set quality to adjusted.
28655 * tree-ssa-tail-merge.c (replace_block_by): Use
28656 combine_with_count.
28658 2018-01-23 Andrew Waterman <andrew@sifive.com>
28659 Jim Wilson <jimw@sifive.com>
28661 * config/riscv/riscv.c (riscv_stack_boundary): New.
28662 (riscv_option_override): Set riscv_stack_boundary. Handle
28663 riscv_preferred_stack_boundary_arg.
28664 * config/riscv/riscv.h (MIN_STACK_BOUNDARY, ABI_STACK_BOUNDARY): New.
28665 (BIGGEST_ALIGNMENT): Set to STACK_BOUNDARY.
28666 (STACK_BOUNDARY): Set to riscv_stack_boundary.
28667 (RISCV_STACK_ALIGN): Use STACK_BOUNDARY.
28668 * config/riscv/riscv.opt (mpreferred-stack-boundary): New.
28669 * doc/invoke.tex (RISC-V Options): Add -mpreferred-stack-boundary.
28671 2018-01-23 H.J. Lu <hongjiu.lu@intel.com>
28674 * config/i386/i386.c (ix86_expand_prologue): Use cost reference
28675 of struct ix86_frame.
28676 (ix86_expand_epilogue): Likewise. Add a local variable for
28677 the reg_save_offset field in struct ix86_frame.
28679 2018-01-23 Bin Cheng <bin.cheng@arm.com>
28681 PR tree-optimization/82604
28682 * tree-loop-distribution.c (enum partition_kind): New enum item
28683 PKIND_PARTIAL_MEMSET.
28684 (partition_builtin_p): Support above new enum item.
28685 (generate_code_for_partition): Ditto.
28686 (compute_access_range): Differentiate cases that equality can be
28687 proven at all loops, the innermost loops or no loops.
28688 (classify_builtin_st, classify_builtin_ldst): Adjust call to above
28689 function. Set PKIND_PARTIAL_MEMSET for partition appropriately.
28690 (finalize_partitions, distribute_loop): Don't fuse partition of
28691 PKIND_PARTIAL_MEMSET kind when distributing 3-level loop nest.
28692 (prepare_perfect_loop_nest): Distribute 3-level loop nest only if
28693 parloop is enabled.
28695 2018-01-23 Martin Liska <mliska@suse.cz>
28697 * predict.def (PRED_INDIR_CALL): Set probability to PROB_EVEN in
28698 order to ignore the predictor.
28699 (PRED_POLYMORPHIC_CALL): Likewise.
28700 (PRED_RECURSIVE_CALL): Likewise.
28702 2018-01-23 Martin Liska <mliska@suse.cz>
28704 * tree-profile.c (tree_profiling): Print function header to
28705 aware reader which function we are working on.
28706 * value-prof.c (gimple_find_values_to_profile): Do not print
28707 not interesting value histograms.
28709 2018-01-23 Martin Liska <mliska@suse.cz>
28711 * profile-count.h (enum profile_quality): Add
28712 profile_uninitialized as the first value. Do not number values
28713 as they are zero based.
28714 (profile_count::verify): Update sanity check.
28715 (profile_probability::verify): Likewise.
28717 2018-01-23 Nathan Sidwell <nathan@acm.org>
28719 * doc/invoke.texi (ffor-scope): Deprecate.
28721 2018-01-23 David Malcolm <dmalcolm@redhat.com>
28723 PR tree-optimization/83510
28724 * domwalk.c (set_all_edges_as_executable): New function.
28725 (dom_walker::dom_walker): Convert bool param
28726 "skip_unreachable_blocks" to enum reachability. Move setup of
28727 edge flags to set_all_edges_as_executable and only do it when
28728 reachability is REACHABLE_BLOCKS.
28729 * domwalk.h (enum dom_walker::reachability): New enum.
28730 (dom_walker::dom_walker): Convert bool param
28731 "skip_unreachable_blocks" to enum reachability.
28732 (set_all_edges_as_executable): New decl.
28733 * graphite-scop-detection.c (gather_bbs::gather_bbs): Convert
28734 from false for "skip_unreachable_blocks" to ALL_BLOCKS for
28736 * tree-ssa-dom.c (dom_opt_dom_walker::dom_opt_dom_walker): Likewise,
28737 but converting true to REACHABLE_BLOCKS.
28738 * tree-ssa-sccvn.c (sccvn_dom_walker::sccvn_dom_walker): Likewise.
28740 (check_array_bounds_dom_walker::check_array_bounds_dom_walker):
28741 Likewise, but converting it to REACHABLE_BLOCKS_PRESERVING_FLAGS.
28742 (vrp_dom_walker::vrp_dom_walker): Likewise, but converting it to
28744 (vrp_prop::vrp_finalize): Call set_all_edges_as_executable
28745 if check_all_array_refs will be called.
28747 2018-01-23 David Malcolm <dmalcolm@redhat.com>
28749 * tree.c (selftest::test_location_wrappers): Add more test
28752 2018-01-23 David Malcolm <dmalcolm@redhat.com>
28754 * sbitmap.c (selftest::test_set_range): Fix memory leaks.
28755 (selftest::test_bit_in_range): Likewise.
28757 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
28760 * doc/sourcebuild.texi (vect_float): Say that the selector
28761 only describes the situation when -funsafe-math-optimizations is on.
28762 (vect_float_strict): Document.
28764 2018-01-23 Richard Sandiford <richard.sandiford@linaro.org>
28766 PR tree-optimization/83965
28767 * tree-vect-patterns.c (vect_reassociating_reduction_p): New function.
28768 (vect_recog_dot_prod_pattern, vect_recog_sad_pattern): Use it
28769 instead of checking only for a reduction.
28770 (vect_recog_widen_sum_pattern): Likewise.
28772 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
28774 * predict.c (probably_never_executed): Only use precise profile info.
28775 (compute_function_frequency): Skip after inlining hack since we now
28776 have quality checking.
28778 2018-01-23 Jan Hubicka <hubicka@ucw.cz>
28780 * profile-count.h (profile_probability::very_unlikely,
28781 profile_probability::unlikely, profile_probability::even): Set
28782 precision to guessed.
28784 2018-01-23 Richard Biener <rguenther@suse.de>
28786 PR tree-optimization/83963
28787 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
28788 Properly terminate dominator walk when crossing the exit edge not
28789 when visiting its source block.
28791 2018-01-23 Jakub Jelinek <jakub@redhat.com>
28794 * tree.c (maybe_wrap_with_location): Use NON_LVALUE_EXPR rather than
28795 VIEW_CONVERT_EXPR to wrap CONST_DECLs.
28797 2018-01-22 Jakub Jelinek <jakub@redhat.com>
28799 PR tree-optimization/83957
28800 * omp-expand.c (expand_omp_for_generic): Ignore virtual PHIs. Remove
28801 semicolon after for body surrounded by braces.
28803 PR tree-optimization/83081
28804 * profile-count.h (profile_probability::split): New method.
28805 * dojump.c (do_jump_1) <case TRUTH_ANDIF_EXPR, case TRUTH_ORIF_EXPR>:
28806 Use profile_probability::split.
28807 (do_compare_rtx_and_jump): Fix adjustment of probabilities
28808 when splitting a single conditional jump into 2.
28810 2018-01-22 David Malcolm <dmalcolm@redhat.com>
28812 PR tree-optimization/69452
28813 * tree-ssa-loop-im.c (class move_computations_dom_walker): Remove
28816 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28818 * config/rl78/rl78-expand.md (bswaphi2): New define_expand.
28819 * config/rl78/rl78-virt.md (*bswaphi2_virt): New define_insn.
28820 * config/rl78/rl78-real.md (*bswaphi2_real): New define_insn.
28822 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28824 * config/rl78/rl78-protos.h (rl78_split_movdi): New function
28826 * config/rl78/rl78.md (movdi): New define_expand.
28827 * config/rl78/rl78.c (rl78_split_movdi): New function.
28829 2018-01-22 Michael Meissner <meissner@linux.vnet.ibm.com>
28832 * config/rs6000/rs6000-protos.h (rs6000_split_signbit): Delete,
28834 * config/rs6000/rs6000.c (rs6000_split_signbit): Likewise.
28835 * config/rs6000/rs6000.md (signbit<mode>2): Change code for IEEE
28836 128-bit to produce an UNSPEC move to get the double word with the
28837 signbit and then a shift directly to do signbit.
28838 (signbit<mode>2_dm): Replace old IEEE 128-bit signbit
28839 implementation with a new version that just does either a direct
28840 move or a regular move. Move memory interface to separate insns.
28841 Move insns so they are next to the expander.
28842 (signbit<mode>2_dm_mem_be): New combiner insns to combine load
28843 with signbit move. Split big and little endian case.
28844 (signbit<mode>2_dm_mem_le): Likewise.
28845 (signbit<mode>2_dm_<su>ext): Delete, no longer used.
28846 (signbit<mode>2_dm2): Likewise.
28848 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28850 * config/rl78/rl78.md (anddi3): New define_expand.
28852 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28854 * config/rl78/rl78.md (umindi3): New define_expand.
28856 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28858 * config/rl78/rl78.md (smindi3): New define_expand.
28860 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28862 * config/rl78/rl78.md (smaxdi3): New define_expand.
28864 2018-01-22 Carl Love <cel@us.ibm.com>
28866 * config/rs6000/rs6000-builtin.def (ST_ELEMREV_V1TI, LD_ELEMREV_V1TI,
28867 LVX_V1TI): Add macro expansion.
28868 * config/rs6000/rs6000-c.c (altivec_builtin_types): Add argument
28869 definitions for VSX_BUILTIN_VEC_XST_BE, VSX_BUILTIN_VEC_ST,
28870 VSX_BUILTIN_VEC_XL, LD_ELEMREV_V1TI builtins.
28871 * config/rs6000/rs6000-p8swap.c (insn_is_swappable_p);
28872 Change check to determine if the instruction is a byte reversing
28873 entry. Fix typo in comment.
28874 * config/rs6000/rs6000.c (altivec_expand_builtin): Add case entry
28875 for VSX_BUILTIN_ST_ELEMREV_V1TI and VSX_BUILTIN_LD_ELEMREV_V1TI.
28876 Add def_builtin calls for new builtins.
28877 * config/rs6000/vsx.md (vsx_st_elemrev_v1ti, vsx_ld_elemrev_v1ti):
28878 Add define_insn expansion.
28880 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28882 * config/rl78/rl78.md (umaxdi3): New define_expand.
28884 2018-01-22 Sebastian Perta <sebastian.perta@renesas.com>
28886 * config/rl78/rl78.c (rl78_note_reg_set): Fixed dead reg check
28887 for non-QImode registers.
28889 2018-01-22 Richard Biener <rguenther@suse.de>
28891 PR tree-optimization/83963
28892 * graphite-scop-detection.c (scop_detection::get_sese): Delay
28893 including the loop exit block.
28894 (scop_detection::merge_sese): Likewise.
28895 (scop_detection::add_scop): Do it here instead.
28897 2018-01-22 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
28899 * doc/sourcebuild.texi (arm_softfloat): Document.
28901 2018-01-21 John David Anglin <danglin@gcc.gnu.org>
28904 * config/pa/pa.c (pa_function_ok_for_sibcall): Use
28905 targetm.binds_local_p instead of TREE_PUBLIC to check local binding.
28906 Move TARGET_PORTABLE_RUNTIME check after TARGET_64BIT check.
28908 2018-01-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
28909 David Edelsohn <dje.gcc@gmail.com>
28912 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
28913 Change "crset eq" to "crset 2".
28914 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
28915 (*call_indirect_aix<mode>_nospec): Likewise.
28916 (*call_value_indirect_aix<mode>_nospec): Likewise.
28917 (*call_indirect_elfv2<mode>_nospec): Likewise.
28918 (*call_value_indirect_elfv2<mode>_nospec): Likewise.
28919 (*sibcall_nonlocal_sysv<mode>): Change "crset eq" to "crset 2";
28920 change assembly output from . to $.
28921 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
28922 (indirect_jump<mode>_nospec): Change assembly output from . to $.
28923 (*tablejump<mode>_internal1_nospec): Likewise.
28925 2018-01-21 Oleg Endo <olegendo@gcc.gnu.org>
28928 * config/sh/sh_optimize_sett_clrt.cc:
28929 Use INCLUDE_ALGORITHM and INCLUDE_VECTOR instead of direct includes.
28931 2018-01-20 Richard Sandiford <richard.sandiford@linaro.org>
28933 PR tree-optimization/83940
28934 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): Set
28935 offset_dt to vect_constant_def rather than vect_unknown_def_type.
28936 (vect_check_load_store_mask): Add a mask_dt_out parameter and
28937 use it to pass back the definition type.
28938 (vect_check_store_rhs): Likewise rhs_dt_out.
28939 (vect_build_gather_load_calls): Add a mask_dt argument and use
28940 it instead of a call to vect_is_simple_use.
28941 (vectorizable_store): Update calls to vect_check_load_store_mask
28942 and vect_check_store_rhs. Use the dt returned by the latter instead
28943 of scatter_src_dt. Use the cached mask_dt and gs_info.offset_dt
28944 instead of calls to vect_is_simple_use. Pass the scalar rather
28945 than the vector operand to vect_is_simple_use when handling
28946 second and subsequent copies of an rhs value.
28947 (vectorizable_load): Update calls to vect_check_load_store_mask
28948 and vect_build_gather_load_calls. Use the cached mask_dt and
28949 gs_info.offset_dt instead of calls to vect_is_simple_use.
28951 2018-01-20 Jakub Jelinek <jakub@redhat.com>
28953 PR middle-end/83945
28954 * tree-emutls.c: Include gimplify.h.
28955 (lower_emutls_2): New function.
28956 (lower_emutls_1): If ADDR_EXPR is a gimple invariant and walk_tree
28957 with lower_emutls_2 callback finds some TLS decl in it, unshare_expr
28958 it before further processing.
28961 * simplify-rtx.c (simplify_binary_operation_1) <case UMOD>: Use
28962 UINTVAL (trueop1) instead of INTVAL (op1).
28964 2018-01-19 Jakub Jelinek <jakub@redhat.com>
28968 * dwarf2cfi.c (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define to
28969 INCOMING_FRAME_SP_OFFSET if not defined.
28970 (scan_trace): Add ENTRY argument. If true and
28971 DEFAULT_INCOMING_FRAME_SP_OFFSET != INCOMING_FRAME_SP_OFFSET,
28972 emit a note to adjust the CFA offset.
28973 (create_cfi_notes): Adjust scan_trace callers.
28974 (create_cie_data): Use DEFAULT_INCOMING_FRAME_SP_OFFSET rather than
28975 INCOMING_FRAME_SP_OFFSET in the CIE.
28976 * config/i386/i386.h (DEFAULT_INCOMING_FRAME_SP_OFFSET): Define.
28977 * config/stormy16/stormy16.h (DEFAULT_INCOMING_FRAME_SP_OFFSET):
28979 * doc/tm.texi.in (DEFAULT_INCOMING_FRAME_SP_OFFSET): Document.
28980 * doc/tm.texi: Regenerated.
28982 2018-01-19 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
28984 PR rtl-optimization/83147
28985 * lra-constraints.c (remove_inheritance_pseudos): Use
28986 lra_substitute_pseudo_within_insn.
28988 2018-01-19 Tom de Vries <tom@codesourcery.com>
28989 Cesar Philippidis <cesar@codesourcery.com>
28992 * config/nvptx/nvptx.c (nvptx_single): Fix jit workaround.
28994 2018-01-19 Cesar Philippidis <cesar@codesourcery.com>
28997 * config/nvptx/nvptx.c (output_init_frag): Don't use generic address
28998 spaces for function labels.
29000 2018-01-19 Martin Liska <mliska@suse.cz>
29002 * predict.def (PRED_LOOP_EXIT): Change from 85 to 89.
29003 (PRED_LOOP_EXIT_WITH_RECURSION): Change from 72 to 78.
29004 (PRED_LOOP_EXTRA_EXIT): Change from 83 to 67.
29005 (PRED_OPCODE_POSITIVE): Change from 64 to 59.
29006 (PRED_TREE_OPCODE_POSITIVE): Change from 64 to 59.
29007 (PRED_CONST_RETURN): Change from 69 to 65.
29008 (PRED_NULL_RETURN): Change from 91 to 71.
29009 (PRED_LOOP_IV_COMPARE_GUESS): Change from 98 to 64.
29010 (PRED_LOOP_GUARD): Change from 66 to 73.
29012 2018-01-19 Martin Liska <mliska@suse.cz>
29014 * predict.c (predict_insn_def): Add new assert.
29015 (struct branch_predictor): Change type to signed integer.
29016 (test_prediction_value_range): Amend test to cover
29017 PROB_UNINITIALIZED.
29018 * predict.def (PRED_LOOP_ITERATIONS): Use the new constant.
29019 (PRED_LOOP_ITERATIONS_GUESSED): Likewise.
29020 (PRED_LOOP_ITERATIONS_MAX): Likewise.
29021 (PRED_LOOP_IV_COMPARE): Likewise.
29022 * predict.h (PROB_UNINITIALIZED): Define new constant.
29024 2018-01-19 Martin Liska <mliska@suse.cz>
29026 * predict.c (dump_prediction): Add new format for
29027 analyze_brprob.py script which is enabled with -details
29029 * profile-count.h (precise_p): New function.
29031 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
29033 PR tree-optimization/83922
29034 * tree-vect-loop.c (vect_verify_full_masking): Return false if
29035 there are no statements that need masking.
29036 (vect_active_double_reduction_p): New function.
29037 (vect_analyze_loop_operations): Use it when handling phis that
29038 are not in the loop header.
29040 2018-01-19 Richard Sandiford <richard.sandiford@linaro.org>
29042 PR tree-optimization/83914
29043 * tree-vect-loop.c (vectorizable_induction): Don't convert
29044 init_expr or apply the peeling adjustment for inductions
29045 that are nested within the vectorized loop.
29047 2018-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29049 * config/arm/thumb2.md (*thumb2_negsi2_short): Use RSB mnemonic
29052 2018-01-18 Jakub Jelinek <jakub@redhat.com>
29056 * function.h (gimplify_parameters): Add gimple_seq * argument.
29057 * function.c: Include gimple.h and options.h.
29058 (gimplify_parameters): Add cleanup argument, add CLOBBER stmts
29059 for the added local temporaries if needed.
29060 * gimplify.c (gimplify_body): Adjust gimplify_parameters caller,
29061 if there are any parameter cleanups, wrap whole body into a
29062 try/finally with the cleanups.
29064 2018-01-18 Wilco Dijkstra <wdijkstr@arm.com>
29067 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
29068 Use GET_MODE_CLASS for scalar floating point.
29070 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
29074 * cgraphclones.c (cgraph_node::create_version_clone_with_body):
29075 Fix call of call_cgraph_insertion_hooks.
29077 2018-01-18 Martin Sebor <msebor@redhat.com>
29079 * doc/invoke.texi (-Wclass-memaccess): Tweak text.
29081 2018-01-18 Jan Hubicka <hubicka@ucw.cz>
29084 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Update edge
29087 2018-01-18 Boris Kolpackov <boris@codesynthesis.com>
29090 * common.opt: (-ffile-prefix-map): New option.
29091 * opts.c (common_handle_option): Defer it.
29092 * opts-global.c (handle_common_deferred_options): Handle it.
29093 * debug.h (remap_debug_filename, add_debug_prefix_map): Move to...
29094 * file-prefix-map.h: New file.
29095 (remap_debug_filename, add_debug_prefix_map): ...here.
29096 (add_macro_prefix_map, add_file_prefix_map, remap_macro_filename): New.
29097 * final.c (debug_prefix_map, add_debug_prefix_map
29098 remap_debug_filename): Move to...
29099 * file-prefix-map.c: New file.
29100 (file_prefix_map, add_prefix_map, remap_filename) ...here and rename,
29101 generalize, get rid of alloca(), use strrchr() instead of strchr().
29102 (add_macro_prefix_map, add_debug_prefix_map, add_file_prefix_map):
29103 Implement in terms of add_prefix_map().
29104 (remap_macro_filename, remap_debug_filename): Implement in term of
29106 * Makefile.in (OBJS, PLUGIN_HEADERS): Add new files.
29107 * builtins.c (fold_builtin_FILE): Call remap_macro_filename().
29108 * dbxout.c: Include file-prefix-map.h.
29109 * varasm.c: Likewise.
29110 * vmsdbgout.c: Likewise.
29111 * xcoffout.c: Likewise.
29112 * dwarf2out.c: Likewise plus omit new options from DW_AT_producer.
29113 * doc/cppopts.texi (-fmacro-prefix-map): Document.
29114 * doc/invoke.texi (-ffile-prefix-map): Document.
29115 (-fdebug-prefix-map): Update description.
29117 2018-01-18 Martin Liska <mliska@suse.cz>
29119 * config/i386/i386.c (indirect_thunk_name): Document that also
29121 (output_indirect_thunk): Document why both instructions
29122 (pause and lfence) are generated.
29124 2018-01-18 Richard Biener <rguenther@suse.de>
29126 PR tree-optimization/83887
29127 * graphite-scop-detection.c
29128 (scop_detection::get_nearest_dom_with_single_entry): Remove.
29129 (scop_detection::get_nearest_pdom_with_single_exit): Likewise.
29130 (scop_detection::merge_sese): Re-implement with a flood-fill
29131 algorithm that properly finds a SESE region if it exists.
29133 2018-01-18 Jakub Jelinek <jakub@redhat.com>
29136 * match.pd ((P + A) - P, P - (P + A), (P + A) - (P + B)): For
29137 pointer_diff optimizations use view_convert instead of convert.
29139 2018-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29141 * config/rs6000/rs6000.md (*call_indirect_nonlocal_sysv<mode>):
29142 Generate different code for -mno-speculate-indirect-jumps.
29143 (*call_value_indirect_nonlocal_sysv<mode>): Likewise.
29144 (*call_indirect_aix<mode>): Disable for
29145 -mno-speculate-indirect-jumps.
29146 (*call_indirect_aix<mode>_nospec): New define_insn.
29147 (*call_value_indirect_aix<mode>): Disable for
29148 -mno-speculate-indirect-jumps.
29149 (*call_value_indirect_aix<mode>_nospec): New define_insn.
29150 (*sibcall_nonlocal_sysv<mode>): Generate different code for
29151 -mno-speculate-indirect-jumps.
29152 (*sibcall_value_nonlocal_sysv<mode>): Likewise.
29154 2018-01-17 Michael Meissner <meissner@linux.vnet.ibm.com>
29156 * config/rs6000/rs6000.c (rs6000_emit_move): If we load or store a
29157 long double type, set the flags for noting the default long double
29158 type, even if we don't pass or return a long double type.
29160 2018-01-17 Jan Hubicka <hubicka@ucw.cz>
29163 * ipa-inline.c (flatten_function): Do not overwrite final inlining
29166 2018-01-17 Will Schmidt <will_schmidt@vnet.ibm.com>
29168 * config/rs6000/rs6000.c (rs6000_gimple_builtin): Add gimple folding
29169 support for merge[hl].
29170 (fold_mergehl_helper): New helper function.
29171 (tree-vector-builder.h): New #include for tree_vector_builder usage.
29172 * config/rs6000/altivec.md (altivec_vmrghw_direct): Add xxmrghw insn.
29173 (altivec_vmrglw_direct): Add xxmrglw insn.
29175 2018-01-17 Andrew Waterman <andrew@sifive.com>
29177 * config/riscv/riscv.c (riscv_conditional_register_usage): If
29178 UNITS_PER_FP_ARG is 0, set call_used_regs to 1 for all FP regs.
29180 2018-01-17 David Malcolm <dmalcolm@redhat.com>
29183 * ipa-devirt.c (add_type_duplicate): When comparing memory layout,
29184 call the lto_location_cache before reading the
29185 DECL_SOURCE_LOCATION of the types.
29187 2018-01-17 Wilco Dijkstra <wdijkstr@arm.com>
29188 Richard Sandiford <richard.sandiford@linaro.org>
29190 * config/aarch64/aarch64.md (movti_aarch64): Use Uti constraint.
29191 * config/aarch64/aarch64.c (aarch64_mov128_immediate): New function.
29192 (aarch64_legitimate_constant_p): Just support CONST_DOUBLE
29193 SF/DF/TF mode to avoid creating illegal CONST_WIDE_INT immediates.
29194 * config/aarch64/aarch64-protos.h (aarch64_mov128_immediate):
29196 * config/aarch64/constraints.md (aarch64_movti_operand):
29198 * config/aarch64/predicates.md (Uti): Add new constraint.
29200 2018-01-17 Carl Love <cel@us.ibm.com>
29202 * config/rs6000/vsx.md (define_expand xl_len_r,
29203 define_expand stxvl, define_expand *stxvl): Add match_dup argument.
29204 (define_insn): Add, match_dup 1 argument to define_insn stxvll and
29206 (define_expand, define_insn): Move the shift left from the
29207 define_insn to the define_expand for lxvl and stxvl instructions.
29208 * config/rs6000/rs6000-builtin.def (BU_P9V_64BIT_VSX_2): Change LXVL
29209 and XL_LEN_R definitions to PURE.
29211 2018-01-17 Uros Bizjak <ubizjak@gmail.com>
29213 * config/i386/i386.c (indirect_thunk_name): Declare regno
29214 as unsigned int. Compare regno with INVALID_REGNUM.
29215 (output_indirect_thunk): Ditto.
29216 (output_indirect_thunk_function): Ditto.
29217 (ix86_code_end): Declare regno as unsigned int. Use INVALID_REGNUM
29218 in the call to output_indirect_thunk_function.
29220 2018-01-17 Richard Sandiford <richard.sandiford@linaro.org>
29222 PR middle-end/83884
29223 * expr.c (expand_expr_real_1): Use the size of GET_MODE (op0)
29224 rather than the size of inner_type to determine the stack slot size
29225 when handling VIEW_CONVERT_EXPRs on strict-alignment targets.
29227 2018-01-16 Sebastian Peryt <sebastian.peryt@intel.com>
29230 * config/i386/i386.c (ix86_option_override_internal): Add PTA_RDRND
29233 2018-01-16 Michael Meissner <meissner@linux.vnet.ibm.com>
29235 * config.gcc (powerpc*-linux*-*): Add support for 64-bit little
29236 endian Linux systems to optionally enable multilibs for selecting
29237 the long double type if the user configured an explicit type.
29238 * config/rs6000/rs6000.h (TARGET_IEEEQUAD_MULTILIB): Indicate we
29239 have no long double multilibs if not defined.
29240 * config/rs6000/rs6000.c (rs6000_option_override_internal): Do not
29241 warn if the user used -mabi={ieee,ibm}longdouble and we built
29242 multilibs for long double.
29243 * config/rs6000/linux64.h (MULTILIB_DEFAULTS_IEEE): Define as the
29244 appropriate multilib option.
29245 (MULTILIB_DEFAULTS): Add MULTILIB_DEFAULTS_IEEE to the default
29247 * config/rs6000/t-ldouble-linux64le-ibm: New configuration files
29248 for building long double multilibs.
29249 * config/rs6000/t-ldouble-linux64le-ieee: Likewise.
29251 2018-01-16 John David Anglin <danglin@gcc.gnu.org>
29253 * config.gcc (hppa*-*-linux*): Change callee copies ABI to caller
29256 * config/pa.h (MALLOC_ABI_ALIGNMENT): Set 32-bit alignment default to
29258 * config/pa/pa32-linux.h (MALLOC_ABI_ALIGNMENT): Set alignment to
29261 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Cleanup type and mode
29264 * config/pa/pa.c (pa_function_arg_size): Apply CEIL to GET_MODE_SIZE
29267 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
29269 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): For an
29270 ADDR_EXPR, do not count the offset of a COMPONENT_REF twice.
29272 2018-01-16 Kelvin Nilsen <kelvin@gcc.gnu.org>
29274 * config/rs6000/rs6000-p8swap.c (rs6000_gen_stvx): Generate
29275 different rtl trees depending on TARGET_64BIT.
29276 (rs6000_gen_lvx): Likewise.
29278 2018-01-16 Eric Botcazou <ebotcazou@adacore.com>
29280 * config/visium/visium.md (nop): Tweak comment.
29281 (hazard_nop): Likewise.
29283 2018-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
29285 * config/rs6000/rs6000.c (rs6000_opt_vars): Add entry for
29286 -mspeculate-indirect-jumps.
29287 * config/rs6000/rs6000.md (*call_indirect_elfv2<mode>): Disable
29288 for -mno-speculate-indirect-jumps.
29289 (*call_indirect_elfv2<mode>_nospec): New define_insn.
29290 (*call_value_indirect_elfv2<mode>): Disable for
29291 -mno-speculate-indirect-jumps.
29292 (*call_value_indirect_elfv2<mode>_nospec): New define_insn.
29293 (indirect_jump): Emit different RTL for
29294 -mno-speculate-indirect-jumps.
29295 (*indirect_jump<mode>): Disable for
29296 -mno-speculate-indirect-jumps.
29297 (*indirect_jump<mode>_nospec): New define_insn.
29298 (tablejump): Emit different RTL for
29299 -mno-speculate-indirect-jumps.
29300 (tablejumpsi): Disable for -mno-speculate-indirect-jumps.
29301 (tablejumpsi_nospec): New define_expand.
29302 (tablejumpdi): Disable for -mno-speculate-indirect-jumps.
29303 (tablejumpdi_nospec): New define_expand.
29304 (*tablejump<mode>_internal1): Disable for
29305 -mno-speculate-indirect-jumps.
29306 (*tablejump<mode>_internal1_nospec): New define_insn.
29307 * config/rs6000/rs6000.opt (mspeculate-indirect-jumps): New
29310 2018-01-16 Artyom Skrobov tyomitch@gmail.com
29312 * caller-save.c (insert_save): Drop unnecessary parameter. All
29315 2018-01-16 Jakub Jelinek <jakub@redhat.com>
29316 Richard Biener <rguenth@suse.de>
29319 * gimplify.c (gimplify_one_sizepos): For is_gimple_constant (expr)
29320 return early, inline manually is_gimple_sizepos. Make sure if we
29321 call gimplify_expr we don't end up with a gimple constant.
29322 * tree.c (variably_modified_type_p): Don't return true for
29323 is_gimple_constant (_t). Inline manually is_gimple_sizepos.
29324 * gimplify.h (is_gimple_sizepos): Remove.
29326 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
29328 PR tree-optimization/83857
29329 * tree-vect-loop.c (vect_analyze_loop_operations): Don't call
29330 vectorizable_live_operation for pure SLP statements.
29331 (vectorizable_live_operation): Handle PHIs.
29333 2018-01-16 Richard Biener <rguenther@suse.de>
29335 PR tree-optimization/83867
29336 * tree-vect-stmts.c (vect_transform_stmt): Precompute
29337 nested_in_vect_loop_p since the scalar stmt may get invalidated.
29339 2018-01-16 Jakub Jelinek <jakub@redhat.com>
29342 * stor-layout.c (handle_warn_if_not_align): Use byte_position and
29343 multiple_of_p instead of unchecked tree_to_uhwi and UHWI check.
29344 If off is not INTEGER_CST, issue a may not be aligned warning
29345 rather than isn't aligned. Use isn%'t rather than isn't.
29346 * fold-const.c (multiple_of_p) <case BIT_AND_EXPR>: Don't fall through
29348 <case MULT_EXPR>: Improve the case when bottom and one of the
29349 MULT_EXPR operands are INTEGER_CSTs and bottom is multiple of that
29350 operand, in that case check if the other operand is multiple of
29351 bottom divided by the INTEGER_CST operand.
29353 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
29356 * config/pa/pa.h (FUNCTION_ARG_SIZE): Delete.
29357 * config/pa/pa-protos.h (pa_function_arg_size): Declare.
29358 * config/pa/som.h (ASM_DECLARE_FUNCTION_NAME): Use
29359 pa_function_arg_size instead of FUNCTION_ARG_SIZE.
29360 * config/pa/pa.c (pa_function_arg_advance): Likewise.
29361 (pa_function_arg, pa_arg_partial_bytes): Likewise.
29362 (pa_function_arg_size): New function.
29364 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
29366 * fold-const.c (fold_ternary_loc): Construct the vec_perm_indices
29367 in a separate statement.
29369 2018-01-16 Richard Sandiford <richard.sandiford@linaro.org>
29371 PR tree-optimization/83847
29372 * tree-vect-data-refs.c (vect_analyze_data_ref_accesses): Don't
29373 group gathers and scatters.
29375 2018-01-16 Jakub Jelinek <jakub@redhat.com>
29377 PR rtl-optimization/86620
29378 * params.def (max-sched-ready-insns): Bump minimum value to 1.
29380 PR rtl-optimization/83213
29381 * recog.c (peep2_attempt): Copy over CROSSING_JUMP_P from peepinsn
29382 to last if both are JUMP_INSNs.
29384 PR tree-optimization/83843
29385 * gimple-ssa-store-merging.c
29386 (imm_store_chain_info::output_merged_store): Handle bit_not_p on
29387 store_immediate_info for bswap/nop orig_stores.
29389 2018-01-15 Andrew Waterman <andrew@sifive.com>
29391 * config/riscv/riscv.c (riscv_rtx_costs) <MULT>: Increase cost if
29393 <UDIV>: Increase cost if !TARGET_DIV.
29395 2018-01-15 Segher Boessenkool <segher@kernel.crashing.org>
29397 * config/rs6000/rs6000.md (define_attr "type"): Remove delayed_cr.
29398 (define_attr "cr_logical_3op"): New.
29399 (cceq_ior_compare): Adjust.
29400 (cceq_ior_compare_complement): Adjust.
29401 (*cceq_rev_compare): Adjust.
29402 * config/rs6000/rs6000.c (rs6000_adjust_cost): Adjust.
29403 (is_cracked_insn): Adjust.
29404 (insn_must_be_first_in_group): Adjust.
29405 * config/rs6000/40x.md: Adjust.
29406 * config/rs6000/440.md: Adjust.
29407 * config/rs6000/476.md: Adjust.
29408 * config/rs6000/601.md: Adjust.
29409 * config/rs6000/603.md: Adjust.
29410 * config/rs6000/6xx.md: Adjust.
29411 * config/rs6000/7450.md: Adjust.
29412 * config/rs6000/7xx.md: Adjust.
29413 * config/rs6000/8540.md: Adjust.
29414 * config/rs6000/cell.md: Adjust.
29415 * config/rs6000/e300c2c3.md: Adjust.
29416 * config/rs6000/e500mc.md: Adjust.
29417 * config/rs6000/e500mc64.md: Adjust.
29418 * config/rs6000/e5500.md: Adjust.
29419 * config/rs6000/e6500.md: Adjust.
29420 * config/rs6000/mpc.md: Adjust.
29421 * config/rs6000/power4.md: Adjust.
29422 * config/rs6000/power5.md: Adjust.
29423 * config/rs6000/power6.md: Adjust.
29424 * config/rs6000/power7.md: Adjust.
29425 * config/rs6000/power8.md: Adjust.
29426 * config/rs6000/power9.md: Adjust.
29427 * config/rs6000/rs64.md: Adjust.
29428 * config/rs6000/titan.md: Adjust.
29430 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
29432 * config/i386/predicates.md (indirect_branch_operand): Rewrite
29433 ix86_indirect_branch_register logic.
29435 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
29437 * config/i386/constraints.md (Bs): Update
29438 ix86_indirect_branch_register check. Don't check
29439 ix86_indirect_branch_register with GOT_memory_operand.
29441 * config/i386/predicates.md (GOT_memory_operand): Don't check
29442 ix86_indirect_branch_register here.
29443 (GOT32_symbol_operand): Likewise.
29445 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
29447 * config/i386/predicates.md (constant_call_address_operand):
29448 Rewrite ix86_indirect_branch_register logic.
29449 (sibcall_insn_operand): Likewise.
29451 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
29453 * config/i386/constraints.md (Bs): Replace
29454 ix86_indirect_branch_thunk_register with
29455 ix86_indirect_branch_register.
29457 * config/i386/i386.md (indirect_jump): Likewise.
29458 (tablejump): Likewise.
29459 (*sibcall_memory): Likewise.
29460 (*sibcall_value_memory): Likewise.
29461 Peepholes of indirect call and jump via memory: Likewise.
29462 * config/i386/i386.opt: Likewise.
29463 * config/i386/predicates.md (indirect_branch_operand): Likewise.
29464 (GOT_memory_operand): Likewise.
29465 (call_insn_operand): Likewise.
29466 (sibcall_insn_operand): Likewise.
29467 (GOT32_symbol_operand): Likewise.
29469 2018-01-15 Jakub Jelinek <jakub@redhat.com>
29471 PR middle-end/83837
29472 * omp-expand.c (expand_omp_atomic_pipeline): Use loaded_val
29473 type rather than type addr's type points to.
29474 (expand_omp_atomic_mutex): Likewise.
29475 (expand_omp_atomic): Likewise.
29477 2018-01-15 H.J. Lu <hongjiu.lu@intel.com>
29480 * config/i386/i386.c (output_indirect_thunk_function): Use
29481 ASM_OUTPUT_LABEL, instead of ASM_OUTPUT_DEF, for TARGET_MACHO
29482 for __x86_return_thunk.
29484 2018-01-15 Richard Biener <rguenther@suse.de>
29486 PR middle-end/83850
29487 * expmed.c (extract_bit_field_1): Fix typo.
29489 2018-01-15 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
29492 * config/arm/iterators.md (VF): New mode iterator.
29493 * config/arm/neon.md (neon_vabd<mode>_2): Use the above.
29494 Remove integer-related logic from pattern.
29495 (neon_vabd<mode>_3): Likewise.
29497 2018-01-15 Jakub Jelinek <jakub@redhat.com>
29499 PR middle-end/82694
29500 * common.opt (fstrict-overflow): No longer an alias.
29501 (fwrapv-pointer): New option.
29502 * tree.h (TYPE_OVERFLOW_WRAPS, TYPE_OVERFLOW_UNDEFINED): Define
29503 also for pointer types based on flag_wrapv_pointer.
29504 * opts.c (common_handle_option) <case OPT_fstrict_overflow>: Set
29505 opts->x_flag_wrap[pv] to !value, clear opts->x_flag_trapv if
29506 opts->x_flag_wrapv got set.
29507 * fold-const.c (fold_comparison, fold_binary_loc): Revert 2017-08-01
29508 changes, just use TYPE_OVERFLOW_UNDEFINED on pointer type instead of
29509 POINTER_TYPE_OVERFLOW_UNDEFINED.
29510 * match.pd: Likewise in address comparison pattern.
29511 * doc/invoke.texi: Document -fwrapv and -fstrict-overflow.
29513 2018-01-15 Richard Biener <rguenther@suse.de>
29516 * tree.c (free_lang_data_in_type): Always unlink TYPE_DECLs
29517 from TYPE_FIELDS. Free TYPE_BINFO if not used by devirtualization.
29518 Reset type names to their identifier if their TYPE_DECL doesn't
29519 have linkage (and thus is used for ODR and devirt).
29520 (save_debug_info_for_decl): Remove.
29521 (save_debug_info_for_type): Likewise.
29522 (add_tree_to_fld_list): Adjust.
29523 * tree-pretty-print.c (dump_generic_node): Make dumping of
29524 type names more robust.
29526 2018-01-15 Richard Biener <rguenther@suse.de>
29528 * BASE-VER: Bump to 8.0.1.
29530 2018-01-14 Martin Sebor <msebor@redhat.com>
29533 * builtins.c (check_access): Avoid warning when the no-warning bit
29536 2018-01-14 Cory Fields <cory-nospam-@coryfields.com>
29538 * tree-ssa-loop-im.c (sort_bbs_in_loop_postorder_cmp): Stabilize sort.
29539 * ira-color (allocno_hard_regs_compare): Likewise.
29541 2018-01-14 Nathan Rossi <nathan@nathanrossi.com>
29544 * config/microblaze/microblaze.c (microblaze_asm_output_ident):
29545 Use .pushsection/.popsection.
29547 2018-01-14 Martin Sebor <msebor@redhat.com>
29550 * doc/invoke.texi (-Wlass-memaccess): Document suppression by casting.
29552 2018-01-14 Jakub Jelinek <jakub@redhat.com>
29554 * config.gcc (i[34567]86-*-*): Remove one duplicate gfniintrin.h
29555 entry from extra_headers.
29556 (x86_64-*-*): Remove two duplicate gfniintrin.h entries from
29557 extra_headers, make the list bitwise identical to the i?86-*-* one.
29559 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29561 * config/i386/i386.c (ix86_set_indirect_branch_type): Disallow
29562 -mcmodel=large with -mindirect-branch=thunk,
29563 -mindirect-branch=thunk-extern, -mfunction-return=thunk and
29564 -mfunction-return=thunk-extern.
29565 * doc/invoke.texi: Document -mcmodel=large is incompatible with
29566 -mindirect-branch=thunk, -mindirect-branch=thunk-extern,
29567 -mfunction-return=thunk and -mfunction-return=thunk-extern.
29569 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29571 * config/i386/i386.c (print_reg): Print the name of the full
29572 integer register without '%'.
29573 (ix86_print_operand): Handle 'V'.
29574 * doc/extend.texi: Document 'V' modifier.
29576 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29578 * config/i386/constraints.md (Bs): Disallow memory operand for
29579 -mindirect-branch-register.
29581 * config/i386/predicates.md (indirect_branch_operand): Likewise.
29582 (GOT_memory_operand): Likewise.
29583 (call_insn_operand): Likewise.
29584 (sibcall_insn_operand): Likewise.
29585 (GOT32_symbol_operand): Likewise.
29586 * config/i386/i386.md (indirect_jump): Call convert_memory_address
29587 for -mindirect-branch-register.
29588 (tablejump): Likewise.
29589 (*sibcall_memory): Likewise.
29590 (*sibcall_value_memory): Likewise.
29591 Disallow peepholes of indirect call and jump via memory for
29592 -mindirect-branch-register.
29593 (*call_pop): Replace m with Bw.
29594 (*call_value_pop): Likewise.
29595 (*sibcall_pop_memory): Replace m with Bs.
29596 * config/i386/i386.opt (mindirect-branch-register): New option.
29597 * doc/invoke.texi: Document -mindirect-branch-register option.
29599 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29601 * config/i386/i386-protos.h (ix86_output_function_return): New.
29602 * config/i386/i386.c (ix86_set_indirect_branch_type): Also
29603 set function_return_type.
29604 (indirect_thunk_name): Add ret_p to indicate thunk for function
29606 (output_indirect_thunk_function): Pass false to
29607 indirect_thunk_name.
29608 (ix86_output_indirect_branch_via_reg): Likewise.
29609 (ix86_output_indirect_branch_via_push): Likewise.
29610 (output_indirect_thunk_function): Create alias for function
29611 return thunk if regno < 0.
29612 (ix86_output_function_return): New function.
29613 (ix86_handle_fndecl_attribute): Handle function_return.
29614 (ix86_attribute_table): Add function_return.
29615 * config/i386/i386.h (machine_function): Add
29616 function_return_type.
29617 * config/i386/i386.md (simple_return_internal): Use
29618 ix86_output_function_return.
29619 (simple_return_internal_long): Likewise.
29620 * config/i386/i386.opt (mfunction-return=): New option.
29621 (indirect_branch): Mention -mfunction-return=.
29622 * doc/extend.texi: Document function_return function attribute.
29623 * doc/invoke.texi: Document -mfunction-return= option.
29625 2018-01-14 H.J. Lu <hongjiu.lu@intel.com>
29627 * config/i386/i386-opts.h (indirect_branch): New.
29628 * config/i386/i386-protos.h (ix86_output_indirect_jmp): Likewise.
29629 * config/i386/i386.c (ix86_using_red_zone): Disallow red-zone
29630 with local indirect jump when converting indirect call and jump.
29631 (ix86_set_indirect_branch_type): New.
29632 (ix86_set_current_function): Call ix86_set_indirect_branch_type.
29633 (indirectlabelno): New.
29634 (indirect_thunk_needed): Likewise.
29635 (indirect_thunk_bnd_needed): Likewise.
29636 (indirect_thunks_used): Likewise.
29637 (indirect_thunks_bnd_used): Likewise.
29638 (INDIRECT_LABEL): Likewise.
29639 (indirect_thunk_name): Likewise.
29640 (output_indirect_thunk): Likewise.
29641 (output_indirect_thunk_function): Likewise.
29642 (ix86_output_indirect_branch_via_reg): Likewise.
29643 (ix86_output_indirect_branch_via_push): Likewise.
29644 (ix86_output_indirect_branch): Likewise.
29645 (ix86_output_indirect_jmp): Likewise.
29646 (ix86_code_end): Call output_indirect_thunk_function if needed.
29647 (ix86_output_call_insn): Call ix86_output_indirect_branch if
29649 (ix86_handle_fndecl_attribute): Handle indirect_branch.
29650 (ix86_attribute_table): Add indirect_branch.
29651 * config/i386/i386.h (machine_function): Add indirect_branch_type
29652 and has_local_indirect_jump.
29653 * config/i386/i386.md (indirect_jump): Set has_local_indirect_jump
29655 (tablejump): Likewise.
29656 (*indirect_jump): Use ix86_output_indirect_jmp.
29657 (*tablejump_1): Likewise.
29658 (simple_return_indirect_internal): Likewise.
29659 * config/i386/i386.opt (mindirect-branch=): New option.
29660 (indirect_branch): New.
29663 (thunk-inline): Likewise.
29664 (thunk-extern): Likewise.
29665 * doc/extend.texi: Document indirect_branch function attribute.
29666 * doc/invoke.texi: Document -mindirect-branch= option.
29668 2018-01-14 Jan Hubicka <hubicka@ucw.cz>
29671 * ipa-inline.c (edge_badness): Tolerate roundoff errors.
29673 2018-01-14 Richard Sandiford <richard.sandiford@linaro.org>
29675 * ipa-inline.c (want_inline_small_function_p): Return false if
29676 inlining has already failed with CIF_FINAL_ERROR.
29677 (update_caller_keys): Call want_inline_small_function_p before
29679 (update_callee_keys): Likewise.
29681 2018-01-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
29683 * config/rs6000/rs6000-p8swap.c (rs6000_sum_of_two_registers_p):
29685 (rs6000_quadword_masked_address_p): Likewise.
29686 (quad_aligned_load_p): Likewise.
29687 (quad_aligned_store_p): Likewise.
29688 (const_load_sequence_p): Add comment to describe the outer-most loop.
29689 (mimic_memory_attributes_and_flags): New function.
29690 (rs6000_gen_stvx): Likewise.
29691 (replace_swapped_aligned_store): Likewise.
29692 (rs6000_gen_lvx): Likewise.
29693 (replace_swapped_aligned_load): Likewise.
29694 (replace_swapped_load_constant): Capitalize argument name in
29695 comment describing this function.
29696 (rs6000_analyze_swaps): Add a third pass to search for vector loads
29697 and stores that access quad-word aligned addresses and replace
29698 with stvx or lvx instructions when appropriate.
29699 * config/rs6000/rs6000-protos.h (rs6000_sum_of_two_registers_p):
29700 New function prototype.
29701 (rs6000_quadword_masked_address_p): Likewise.
29702 (rs6000_gen_lvx): Likewise.
29703 (rs6000_gen_stvx): Likewise.
29704 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode>): For modes
29705 VSX_D (V2DF, V2DI), modify this split to select lvx instruction
29706 when memory address is aligned.
29707 (*vsx_le_perm_load_<mode>): For modes VSX_W (V4SF, V4SI), modify
29708 this split to select lvx instruction when memory address is aligned.
29709 (*vsx_le_perm_load_v8hi): Modify this split to select lvx
29710 instruction when memory address is aligned.
29711 (*vsx_le_perm_load_v16qi): Likewise.
29712 (four unnamed splitters): Modify to select the stvx instruction
29713 when memory is aligned.
29715 2018-01-13 Jan Hubicka <hubicka@ucw.cz>
29717 * predict.c (determine_unlikely_bbs): Handle correctly BBs
29718 which appears in the queue multiple times.
29720 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29721 Alan Hayward <alan.hayward@arm.com>
29722 David Sherwood <david.sherwood@arm.com>
29724 * tree-vectorizer.h (vec_lower_bound): New structure.
29725 (_loop_vec_info): Add check_nonzero and lower_bounds.
29726 (LOOP_VINFO_CHECK_NONZERO): New macro.
29727 (LOOP_VINFO_LOWER_BOUNDS): Likewise.
29728 (LOOP_REQUIRES_VERSIONING_FOR_ALIAS): Check lower_bounds too.
29729 * tree-data-ref.h (dr_with_seg_len): Add access_size and align
29730 fields. Make seg_len the distance travelled, not including the
29732 (dr_direction_indicator): Declare.
29733 (dr_zero_step_indicator): Likewise.
29734 (dr_known_forward_stride_p): Likewise.
29735 * tree-data-ref.c: Include stringpool.h, tree-vrp.h and
29737 (runtime_alias_check_p): Allow runtime alias checks with
29739 (operator ==): Compare access_size and align.
29740 (prune_runtime_alias_test_list): Rework for new distinction between
29741 the access_size and seg_len.
29742 (create_intersect_range_checks_index): Likewise. Cope with polynomial
29744 (get_segment_min_max): New function.
29745 (create_intersect_range_checks): Use it.
29746 (dr_step_indicator): New function.
29747 (dr_direction_indicator): Likewise.
29748 (dr_zero_step_indicator): Likewise.
29749 (dr_known_forward_stride_p): Likewise.
29750 * tree-loop-distribution.c (data_ref_segment_size): Return
29751 DR_STEP * (niters - 1).
29752 (compute_alias_check_pairs): Update call to the dr_with_seg_len
29754 * tree-vect-data-refs.c (vect_check_nonzero_value): New function.
29755 (vect_preserves_scalar_order_p): New function, split out from...
29756 (vect_analyze_data_ref_dependence): ...here. Check for zero steps.
29757 (vect_vfa_segment_size): Return DR_STEP * (length_factor - 1).
29758 (vect_vfa_access_size): New function.
29759 (vect_vfa_align): Likewise.
29760 (vect_compile_time_alias): Take access_size_a and access_b arguments.
29761 (dump_lower_bound): New function.
29762 (vect_check_lower_bound): Likewise.
29763 (vect_small_gap_p): Likewise.
29764 (vectorizable_with_step_bound_p): Likewise.
29765 (vect_prune_runtime_alias_test_list): Ignore cross-iteration
29766 depencies if the vectorization factor is 1. Convert the checks
29767 for nonzero steps into checks on the bounds of DR_STEP. Try using
29768 a bunds check for variable steps if the minimum required step is
29769 relatively small. Update calls to the dr_with_seg_len
29770 constructor and to vect_compile_time_alias.
29771 * tree-vect-loop-manip.c (vect_create_cond_for_lower_bounds): New
29773 (vect_loop_versioning): Call it.
29774 * tree-vect-loop.c (vect_analyze_loop_2): Clear LOOP_VINFO_LOWER_BOUNDS
29776 (vect_estimate_min_profitable_iters): Account for any bounds checks.
29778 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29779 Alan Hayward <alan.hayward@arm.com>
29780 David Sherwood <david.sherwood@arm.com>
29782 * doc/sourcebuild.texi (vect_scatter_store): Document.
29783 * optabs.def (scatter_store_optab, mask_scatter_store_optab): New
29785 * doc/md.texi (scatter_store@var{m}, mask_scatter_store@var{m}):
29787 * genopinit.c (main): Add supports_vec_scatter_store and
29788 supports_vec_scatter_store_cached to target_optabs.
29789 * gimple.h (gimple_expr_type): Handle IFN_SCATTER_STORE and
29790 IFN_MASK_SCATTER_STORE.
29791 * internal-fn.def (SCATTER_STORE, MASK_SCATTER_STORE): New internal
29793 * internal-fn.h (internal_store_fn_p): Declare.
29794 (internal_fn_stored_value_index): Likewise.
29795 * internal-fn.c (scatter_store_direct): New macro.
29796 (expand_scatter_store_optab_fn): New function.
29797 (direct_scatter_store_optab_supported_p): New macro.
29798 (internal_store_fn_p): New function.
29799 (internal_gather_scatter_fn_p): Handle IFN_SCATTER_STORE and
29800 IFN_MASK_SCATTER_STORE.
29801 (internal_fn_mask_index): Likewise.
29802 (internal_fn_stored_value_index): New function.
29803 (internal_gather_scatter_fn_supported_p): Adjust operand numbers
29804 for scatter stores.
29805 * optabs-query.h (supports_vec_scatter_store_p): Declare.
29806 * optabs-query.c (supports_vec_scatter_store_p): New function.
29807 * tree-vectorizer.h (vect_get_store_rhs): Declare.
29808 * tree-vect-data-refs.c (vect_analyze_data_ref_access): Return
29809 true for scatter stores.
29810 (vect_gather_scatter_fn_p): Handle scatter stores too.
29811 (vect_check_gather_scatter): Consider using scatter stores if
29812 supports_vec_scatter_store_p.
29813 * tree-vect-patterns.c (vect_try_gather_scatter_pattern): Handle
29814 scatter stores too.
29815 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
29816 internal_fn_stored_value_index.
29817 (check_load_store_masking): Handle scatter stores too.
29818 (vect_get_store_rhs): Make public.
29819 (vectorizable_call): Use internal_store_fn_p.
29820 (vectorizable_store): Handle scatter store internal functions.
29821 (vect_transform_stmt): Compare GROUP_STORE_COUNT with GROUP_SIZE
29822 when deciding whether the end of the group has been reached.
29823 * config/aarch64/aarch64.md (UNSPEC_ST1_SCATTER): New unspec.
29824 * config/aarch64/aarch64-sve.md (scatter_store<mode>): New expander.
29825 (mask_scatter_store<mode>): New insns.
29827 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29828 Alan Hayward <alan.hayward@arm.com>
29829 David Sherwood <david.sherwood@arm.com>
29831 * tree-vectorizer.h (vect_gather_scatter_fn_p): Declare.
29832 * tree-vect-data-refs.c (vect_gather_scatter_fn_p): Make public.
29833 * tree-vect-stmts.c (vect_truncate_gather_scatter_offset): New
29835 (vect_use_strided_gather_scatters_p): Take a masked_p argument.
29836 Use vect_truncate_gather_scatter_offset if we can't treat the
29837 operation as a normal gather load or scatter store.
29838 (get_group_load_store_type): Take the gather_scatter_info
29839 as argument. Try using a gather load or scatter store for
29840 single-element groups.
29841 (get_load_store_type): Update calls to get_group_load_store_type
29842 and vect_use_strided_gather_scatters_p.
29844 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29845 Alan Hayward <alan.hayward@arm.com>
29846 David Sherwood <david.sherwood@arm.com>
29848 * tree-vectorizer.h (vect_create_data_ref_ptr): Take an extra
29849 optional tree argument.
29850 * tree-vect-data-refs.c (vect_check_gather_scatter): Check for
29852 (vect_create_data_ref_ptr): Take the iv_step as an optional argument,
29853 but continue to use the current value as a fallback.
29854 (bump_vector_ptr): Use operand_equal_p rather than tree_int_cst_compare
29855 to compare the updates.
29856 * tree-vect-stmts.c (vect_use_strided_gather_scatters_p): New function.
29857 (get_load_store_type): Use it when handling a strided access.
29858 (vect_get_strided_load_store_ops): New function.
29859 (vect_get_data_ptr_increment): Likewise.
29860 (vectorizable_load): Handle strided gather loads. Always pass
29861 a step to vect_create_data_ref_ptr and bump_vector_ptr.
29863 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29864 Alan Hayward <alan.hayward@arm.com>
29865 David Sherwood <david.sherwood@arm.com>
29867 * doc/md.texi (gather_load@var{m}): Document.
29868 (mask_gather_load@var{m}): Likewise.
29869 * genopinit.c (main): Add supports_vec_gather_load and
29870 supports_vec_gather_load_cached to target_optabs.
29871 * optabs-tree.c (init_tree_optimization_optabs): Use
29872 ggc_cleared_alloc to allocate target_optabs.
29873 * optabs.def (gather_load_optab, mask_gather_laod_optab): New optabs.
29874 * internal-fn.def (GATHER_LOAD, MASK_GATHER_LOAD): New internal
29876 * internal-fn.h (internal_load_fn_p): Declare.
29877 (internal_gather_scatter_fn_p): Likewise.
29878 (internal_fn_mask_index): Likewise.
29879 (internal_gather_scatter_fn_supported_p): Likewise.
29880 * internal-fn.c (gather_load_direct): New macro.
29881 (expand_gather_load_optab_fn): New function.
29882 (direct_gather_load_optab_supported_p): New macro.
29883 (direct_internal_fn_optab): New function.
29884 (internal_load_fn_p): Likewise.
29885 (internal_gather_scatter_fn_p): Likewise.
29886 (internal_fn_mask_index): Likewise.
29887 (internal_gather_scatter_fn_supported_p): Likewise.
29888 * optabs-query.c (supports_at_least_one_mode_p): New function.
29889 (supports_vec_gather_load_p): Likewise.
29890 * optabs-query.h (supports_vec_gather_load_p): Declare.
29891 * tree-vectorizer.h (gather_scatter_info): Add ifn, element_type
29892 and memory_type field.
29893 (NUM_PATTERNS): Bump to 15.
29894 * tree-vect-data-refs.c: Include internal-fn.h.
29895 (vect_gather_scatter_fn_p): New function.
29896 (vect_describe_gather_scatter_call): Likewise.
29897 (vect_check_gather_scatter): Try using internal functions for
29898 gather loads. Recognize existing calls to a gather load function.
29899 (vect_analyze_data_refs): Consider using gather loads if
29900 supports_vec_gather_load_p.
29901 * tree-vect-patterns.c (vect_get_load_store_mask): New function.
29902 (vect_get_gather_scatter_offset_type): Likewise.
29903 (vect_convert_mask_for_vectype): Likewise.
29904 (vect_add_conversion_to_patterm): Likewise.
29905 (vect_try_gather_scatter_pattern): Likewise.
29906 (vect_recog_gather_scatter_pattern): New pattern recognizer.
29907 (vect_vect_recog_func_ptrs): Add it.
29908 * tree-vect-stmts.c (exist_non_indexing_operands_for_use_p): Use
29909 internal_fn_mask_index and internal_gather_scatter_fn_p.
29910 (check_load_store_masking): Take the gather_scatter_info as an
29911 argument and handle gather loads.
29912 (vect_get_gather_scatter_ops): New function.
29913 (vectorizable_call): Check internal_load_fn_p.
29914 (vectorizable_load): Likewise. Handle gather load internal
29916 (vectorizable_store): Update call to check_load_store_masking.
29917 * config/aarch64/aarch64.md (UNSPEC_LD1_GATHER): New unspec.
29918 * config/aarch64/iterators.md (SVE_S, SVE_D): New mode iterators.
29919 * config/aarch64/predicates.md (aarch64_gather_scale_operand_w)
29920 (aarch64_gather_scale_operand_d): New predicates.
29921 * config/aarch64/aarch64-sve.md (gather_load<mode>): New expander.
29922 (mask_gather_load<mode>): New insns.
29924 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29925 Alan Hayward <alan.hayward@arm.com>
29926 David Sherwood <david.sherwood@arm.com>
29928 * optabs.def (fold_left_plus_optab): New optab.
29929 * doc/md.texi (fold_left_plus_@var{m}): Document.
29930 * internal-fn.def (IFN_FOLD_LEFT_PLUS): New internal function.
29931 * internal-fn.c (fold_left_direct): Define.
29932 (expand_fold_left_optab_fn): Likewise.
29933 (direct_fold_left_optab_supported_p): Likewise.
29934 * fold-const-call.c (fold_const_fold_left): New function.
29935 (fold_const_call): Use it to fold CFN_FOLD_LEFT_PLUS.
29936 * tree-parloops.c (valid_reduction_p): New function.
29937 (gather_scalar_reductions): Use it.
29938 * tree-vectorizer.h (FOLD_LEFT_REDUCTION): New vect_reduction_type.
29939 (vect_finish_replace_stmt): Declare.
29940 * tree-vect-loop.c (fold_left_reduction_fn): New function.
29941 (needs_fold_left_reduction_p): New function, split out from...
29942 (vect_is_simple_reduction): ...here. Accept reductions that
29943 forbid reassociation, but give them type FOLD_LEFT_REDUCTION.
29944 (vect_force_simple_reduction): Also store the reduction type in
29945 the assignment's STMT_VINFO_REDUC_TYPE.
29946 (vect_model_reduction_cost): Handle FOLD_LEFT_REDUCTION.
29947 (merge_with_identity): New function.
29948 (vect_expand_fold_left): Likewise.
29949 (vectorize_fold_left_reduction): Likewise.
29950 (vectorizable_reduction): Handle FOLD_LEFT_REDUCTION. Leave the
29951 scalar phi in place for it. Check for target support and reject
29952 cases that would reassociate the operation. Defer the transform
29953 phase to vectorize_fold_left_reduction.
29954 * config/aarch64/aarch64.md (UNSPEC_FADDA): New unspec.
29955 * config/aarch64/aarch64-sve.md (fold_left_plus_<mode>): New expander.
29956 (*fold_left_plus_<mode>, *pred_fold_left_plus_<mode>): New insns.
29958 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29960 * tree-if-conv.c (predicate_mem_writes): Remove redundant
29961 call to ifc_temp_var.
29963 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29964 Alan Hayward <alan.hayward@arm.com>
29965 David Sherwood <david.sherwood@arm.com>
29967 * target.def (legitimize_address_displacement): Take the original
29968 offset as a poly_int.
29969 * targhooks.h (default_legitimize_address_displacement): Update
29971 * targhooks.c (default_legitimize_address_displacement): Likewise.
29972 * doc/tm.texi: Regenerate.
29973 * lra-constraints.c (base_plus_disp_to_reg): Take the displacement
29974 as an argument, moving assert of ad->disp == ad->disp_term to...
29975 (process_address_1): ...here. Update calls to base_plus_disp_to_reg.
29976 Try calling targetm.legitimize_address_displacement before expanding
29977 the address rather than afterwards, and adjust for the new interface.
29978 * config/aarch64/aarch64.c (aarch64_legitimize_address_displacement):
29979 Match the new hook interface. Handle SVE addresses.
29980 * config/sh/sh.c (sh_legitimize_address_displacement): Make the
29981 new hook interface.
29983 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
29985 * Makefile.in (OBJS): Add early-remat.o.
29986 * target.def (select_early_remat_modes): New hook.
29987 * doc/tm.texi.in (TARGET_SELECT_EARLY_REMAT_MODES): New hook.
29988 * doc/tm.texi: Regenerate.
29989 * targhooks.h (default_select_early_remat_modes): Declare.
29990 * targhooks.c (default_select_early_remat_modes): New function.
29991 * timevar.def (TV_EARLY_REMAT): New timevar.
29992 * passes.def (pass_early_remat): New pass.
29993 * tree-pass.h (make_pass_early_remat): Declare.
29994 * early-remat.c: New file.
29995 * config/aarch64/aarch64.c (aarch64_select_early_remat_modes): New
29997 (TARGET_SELECT_EARLY_REMAT_MODES): Define.
29999 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30000 Alan Hayward <alan.hayward@arm.com>
30001 David Sherwood <david.sherwood@arm.com>
30003 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Replace
30004 vfm1 with a bound_epilog parameter.
30005 (vect_do_peeling): Update calls accordingly, and move the prologue
30006 call earlier in the function. Treat the base bound_epilog as 0 for
30007 fully-masked loops and retain vf - 1 for other loops. Add 1 to
30008 this base when peeling for gaps.
30009 * tree-vect-loop.c (vect_analyze_loop_2): Allow peeling for gaps
30010 with fully-masked loops.
30011 (vect_estimate_min_profitable_iters): Handle the single peeled
30012 iteration in that case.
30014 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30015 Alan Hayward <alan.hayward@arm.com>
30016 David Sherwood <david.sherwood@arm.com>
30018 * tree-vect-data-refs.c (vect_analyze_group_access_1): Allow
30019 single-element interleaving even if the size is not a power of 2.
30020 * tree-vect-stmts.c (get_load_store_type): Disallow elementwise
30021 accesses for single-element interleaving if the group size is
30024 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30025 Alan Hayward <alan.hayward@arm.com>
30026 David Sherwood <david.sherwood@arm.com>
30028 * doc/md.texi (fold_extract_last_@var{m}): Document.
30029 * doc/sourcebuild.texi (vect_fold_extract_last): Likewise.
30030 * optabs.def (fold_extract_last_optab): New optab.
30031 * internal-fn.def (FOLD_EXTRACT_LAST): New internal function.
30032 * internal-fn.c (fold_extract_direct): New macro.
30033 (expand_fold_extract_optab_fn): Likewise.
30034 (direct_fold_extract_optab_supported_p): Likewise.
30035 * tree-vectorizer.h (EXTRACT_LAST_REDUCTION): New vect_reduction_type.
30036 * tree-vect-loop.c (vect_model_reduction_cost): Handle
30037 EXTRACT_LAST_REDUCTION.
30038 (get_initial_def_for_reduction): Do not create an initial vector
30039 for EXTRACT_LAST_REDUCTION reductions.
30040 (vectorizable_reduction): Leave the scalar phi in place for
30041 EXTRACT_LAST_REDUCTIONs. Try using EXTRACT_LAST_REDUCTION
30042 ahead of INTEGER_INDUC_COND_REDUCTION. Do not check for an
30043 epilogue code for EXTRACT_LAST_REDUCTION and defer the
30044 transform phase to vectorizable_condition.
30045 * tree-vect-stmts.c (vect_finish_stmt_generation_1): New function,
30047 (vect_finish_stmt_generation): ...here.
30048 (vect_finish_replace_stmt): New function.
30049 (vectorizable_condition): Handle EXTRACT_LAST_REDUCTION.
30050 * config/aarch64/aarch64-sve.md (fold_extract_last_<mode>): New
30052 * config/aarch64/aarch64.md (UNSPEC_CLASTB): New unspec.
30054 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30055 Alan Hayward <alan.hayward@arm.com>
30056 David Sherwood <david.sherwood@arm.com>
30058 * doc/md.texi (extract_last_@var{m}): Document.
30059 * optabs.def (extract_last_optab): New optab.
30060 * internal-fn.def (EXTRACT_LAST): New internal function.
30061 * internal-fn.c (cond_unary_direct): New macro.
30062 (expand_cond_unary_optab_fn): Likewise.
30063 (direct_cond_unary_optab_supported_p): Likewise.
30064 * tree-vect-loop.c (vectorizable_live_operation): Allow fully-masked
30065 loops using EXTRACT_LAST.
30066 * config/aarch64/aarch64-sve.md (aarch64_sve_lastb<mode>): Rename to...
30067 (extract_last_<mode>): ...this optab.
30068 (vec_extract<mode><Vel>): Update accordingly.
30070 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30071 Alan Hayward <alan.hayward@arm.com>
30072 David Sherwood <david.sherwood@arm.com>
30074 * target.def (empty_mask_is_expensive): New hook.
30075 * doc/tm.texi.in (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): New hook.
30076 * doc/tm.texi: Regenerate.
30077 * targhooks.h (default_empty_mask_is_expensive): Declare.
30078 * targhooks.c (default_empty_mask_is_expensive): New function.
30079 * tree-vectorizer.c (vectorize_loops): Only call optimize_mask_stores
30080 if the target says that empty masks are expensive.
30081 * config/aarch64/aarch64.c (aarch64_empty_mask_is_expensive):
30083 (TARGET_VECTORIZE_EMPTY_MASK_IS_EXPENSIVE): Redefine.
30085 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30086 Alan Hayward <alan.hayward@arm.com>
30087 David Sherwood <david.sherwood@arm.com>
30089 * tree-vectorizer.h (_loop_vec_info::mask_skip_niters): New field.
30090 (LOOP_VINFO_MASK_SKIP_NITERS): New macro.
30091 (vect_use_loop_mask_for_alignment_p): New function.
30092 (vect_prepare_for_masked_peels, vect_gen_while_not): Declare.
30093 * tree-vect-loop-manip.c (vect_set_loop_masks_directly): Add an
30094 niters_skip argument. Make sure that the first niters_skip elements
30095 of the first iteration are inactive.
30096 (vect_set_loop_condition_masked): Handle LOOP_VINFO_MASK_SKIP_NITERS.
30097 Update call to vect_set_loop_masks_directly.
30098 (get_misalign_in_elems): New function, split out from...
30099 (vect_gen_prolog_loop_niters): ...here.
30100 (vect_update_init_of_dr): Take a code argument that specifies whether
30101 the adjustment should be added or subtracted.
30102 (vect_update_init_of_drs): Likewise.
30103 (vect_prepare_for_masked_peels): New function.
30104 (vect_do_peeling): Skip prologue peeling if we're using a mask
30105 instead. Update call to vect_update_inits_of_drs.
30106 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
30108 (vect_analyze_loop_2): Allow fully-masked loops with peeling for
30109 alignment. Do not include the number of peeled iterations in
30110 the minimum threshold in that case.
30111 (vectorizable_induction): Adjust the start value down by
30112 LOOP_VINFO_MASK_SKIP_NITERS iterations.
30113 (vect_transform_loop): Call vect_prepare_for_masked_peels.
30114 Take the number of skipped iterations into account when calculating
30116 * tree-vect-stmts.c (vect_gen_while_not): New function.
30118 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30119 Alan Hayward <alan.hayward@arm.com>
30120 David Sherwood <david.sherwood@arm.com>
30122 * doc/sourcebuild.texi (vect_fully_masked): Document.
30123 * params.def (PARAM_MIN_VECT_LOOP_BOUND): Change minimum and
30124 default value to 0.
30125 * tree-vect-loop.c (vect_analyze_loop_costing): New function,
30127 (vect_analyze_loop_2): ...here. Don't check the vectorization
30128 factor against the number of loop iterations if the loop is
30131 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30132 Alan Hayward <alan.hayward@arm.com>
30133 David Sherwood <david.sherwood@arm.com>
30135 * tree-ssa-loop-ivopts.c (USE_ADDRESS): Split into...
30136 (USE_REF_ADDRESS, USE_PTR_ADDRESS): ...these new use types.
30137 (dump_groups): Update accordingly.
30138 (iv_use::mem_type): New member variable.
30139 (address_p): New function.
30140 (record_use): Add a mem_type argument and initialize the new
30142 (record_group_use): Add a mem_type argument. Use address_p.
30143 Remove obsolete null checks of base_object. Update call to record_use.
30144 (find_interesting_uses_op): Update call to record_group_use.
30145 (find_interesting_uses_cond): Likewise.
30146 (find_interesting_uses_address): Likewise.
30147 (get_mem_type_for_internal_fn): New function.
30148 (find_address_like_use): Likewise.
30149 (find_interesting_uses_stmt): Try find_address_like_use before
30150 calling find_interesting_uses_op.
30151 (addr_offset_valid_p): Use the iv mem_type field as the type
30152 of the addressed memory.
30153 (add_autoinc_candidates): Likewise.
30154 (get_address_cost): Likewise.
30155 (split_small_address_groups_p): Use address_p.
30156 (split_address_groups): Likewise.
30157 (add_iv_candidate_for_use): Likewise.
30158 (autoinc_possible_for_pair): Likewise.
30159 (rewrite_groups): Likewise.
30160 (get_use_type): Check for USE_REF_ADDRESS instead of USE_ADDRESS.
30161 (determine_group_iv_cost): Update after split of USE_ADDRESS.
30162 (get_alias_ptr_type_for_ptr_address): New function.
30163 (rewrite_use_address): Rewrite address uses in calls that were
30164 identified by find_address_like_use.
30166 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30167 Alan Hayward <alan.hayward@arm.com>
30168 David Sherwood <david.sherwood@arm.com>
30170 * expr.c (expand_expr_addr_expr_1): Handle ADDR_EXPRs of
30172 * gimple-expr.h (is_gimple_addressable: Likewise.
30173 * gimple-expr.c (is_gimple_address): Likewise.
30174 * internal-fn.c (expand_call_mem_ref): New function.
30175 (expand_mask_load_optab_fn): Use it.
30176 (expand_mask_store_optab_fn): Likewise.
30178 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30179 Alan Hayward <alan.hayward@arm.com>
30180 David Sherwood <david.sherwood@arm.com>
30182 * doc/md.texi (cond_add@var{mode}, cond_sub@var{mode})
30183 (cond_and@var{mode}, cond_ior@var{mode}, cond_xor@var{mode})
30184 (cond_smin@var{mode}, cond_smax@var{mode}, cond_umin@var{mode})
30185 (cond_umax@var{mode}): Document.
30186 * optabs.def (cond_add_optab, cond_sub_optab, cond_and_optab)
30187 (cond_ior_optab, cond_xor_optab, cond_smin_optab, cond_smax_optab)
30188 (cond_umin_optab, cond_umax_optab): New optabs.
30189 * internal-fn.def (COND_ADD, COND_SUB, COND_MIN, COND_MAX, COND_AND)
30190 (COND_IOR, COND_XOR): New internal functions.
30191 * internal-fn.h (get_conditional_internal_fn): Declare.
30192 * internal-fn.c (cond_binary_direct): New macro.
30193 (expand_cond_binary_optab_fn): Likewise.
30194 (direct_cond_binary_optab_supported_p): Likewise.
30195 (get_conditional_internal_fn): New function.
30196 * tree-vect-loop.c (vectorizable_reduction): Handle fully-masked loops.
30197 Cope with reduction statements that are vectorized as calls rather
30199 * config/aarch64/aarch64-sve.md (cond_<optab><mode>): New insns.
30200 * config/aarch64/iterators.md (UNSPEC_COND_ADD, UNSPEC_COND_SUB)
30201 (UNSPEC_COND_SMAX, UNSPEC_COND_UMAX, UNSPEC_COND_SMIN)
30202 (UNSPEC_COND_UMIN, UNSPEC_COND_AND, UNSPEC_COND_ORR)
30203 (UNSPEC_COND_EOR): New unspecs.
30204 (optab): Add mappings for them.
30205 (SVE_COND_INT_OP, SVE_COND_FP_OP): New int iterators.
30206 (sve_int_op, sve_fp_op): New int attributes.
30208 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30209 Alan Hayward <alan.hayward@arm.com>
30210 David Sherwood <david.sherwood@arm.com>
30212 * optabs.def (while_ult_optab): New optab.
30213 * doc/md.texi (while_ult@var{m}@var{n}): Document.
30214 * internal-fn.def (WHILE_ULT): New internal function.
30215 * internal-fn.h (direct_internal_fn_supported_p): New override
30216 that takes two types as argument.
30217 * internal-fn.c (while_direct): New macro.
30218 (expand_while_optab_fn): New function.
30219 (convert_optab_supported_p): Likewise.
30220 (direct_while_optab_supported_p): New macro.
30221 * wide-int.h (wi::udiv_ceil): New function.
30222 * tree-vectorizer.h (rgroup_masks): New structure.
30223 (vec_loop_masks): New typedef.
30224 (_loop_vec_info): Add masks, mask_compare_type, can_fully_mask_p
30225 and fully_masked_p.
30226 (LOOP_VINFO_CAN_FULLY_MASK_P, LOOP_VINFO_FULLY_MASKED_P)
30227 (LOOP_VINFO_MASKS, LOOP_VINFO_MASK_COMPARE_TYPE): New macros.
30228 (vect_max_vf): New function.
30229 (slpeel_make_loop_iterate_ntimes): Delete.
30230 (vect_set_loop_condition, vect_get_loop_mask_type, vect_gen_while)
30231 (vect_halve_mask_nunits, vect_double_mask_nunits): Declare.
30232 (vect_record_loop_mask, vect_get_loop_mask): Likewise.
30233 * tree-vect-loop-manip.c: Include tree-ssa-loop-niter.h,
30234 internal-fn.h, stor-layout.h and optabs-query.h.
30235 (vect_set_loop_mask): New function.
30236 (add_preheader_seq): Likewise.
30237 (add_header_seq): Likewise.
30238 (interleave_supported_p): Likewise.
30239 (vect_maybe_permute_loop_masks): Likewise.
30240 (vect_set_loop_masks_directly): Likewise.
30241 (vect_set_loop_condition_masked): Likewise.
30242 (vect_set_loop_condition_unmasked): New function, split out from
30243 slpeel_make_loop_iterate_ntimes.
30244 (slpeel_make_loop_iterate_ntimes): Rename to..
30245 (vect_set_loop_condition): ...this. Use vect_set_loop_condition_masked
30246 for fully-masked loops and vect_set_loop_condition_unmasked otherwise.
30247 (vect_do_peeling): Update call accordingly.
30248 (vect_gen_vector_loop_niters): Use VF as the step for fully-masked
30250 * tree-vect-loop.c (_loop_vec_info::_loop_vec_info): Initialize
30251 mask_compare_type, can_fully_mask_p and fully_masked_p.
30252 (release_vec_loop_masks): New function.
30253 (_loop_vec_info): Use it to free the loop masks.
30254 (can_produce_all_loop_masks_p): New function.
30255 (vect_get_max_nscalars_per_iter): Likewise.
30256 (vect_verify_full_masking): Likewise.
30257 (vect_analyze_loop_2): Save LOOP_VINFO_CAN_FULLY_MASK_P around
30258 retries, and free the mask rgroups before retrying. Check loop-wide
30259 reasons for disallowing fully-masked loops. Make the final decision
30260 about whether use a fully-masked loop or not.
30261 (vect_estimate_min_profitable_iters): Do not assume that peeling
30262 for the number of iterations will be needed for fully-masked loops.
30263 (vectorizable_reduction): Disable fully-masked loops.
30264 (vectorizable_live_operation): Likewise.
30265 (vect_halve_mask_nunits): New function.
30266 (vect_double_mask_nunits): Likewise.
30267 (vect_record_loop_mask): Likewise.
30268 (vect_get_loop_mask): Likewise.
30269 (vect_transform_loop): Handle the case in which the final loop
30270 iteration might handle a partial vector. Call vect_set_loop_condition
30271 instead of slpeel_make_loop_iterate_ntimes.
30272 * tree-vect-stmts.c: Include tree-ssa-loop-niter.h and gimple-fold.h.
30273 (check_load_store_masking): New function.
30274 (prepare_load_store_mask): Likewise.
30275 (vectorizable_store): Handle fully-masked loops.
30276 (vectorizable_load): Likewise.
30277 (supportable_widening_operation): Use vect_halve_mask_nunits for
30279 (supportable_narrowing_operation): Likewise vect_double_mask_nunits.
30280 (vect_gen_while): New function.
30281 * config/aarch64/aarch64.md (umax<mode>3): New expander.
30282 (aarch64_uqdec<mode>): New insn.
30284 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30285 Alan Hayward <alan.hayward@arm.com>
30286 David Sherwood <david.sherwood@arm.com>
30288 * optabs.def (reduc_and_scal_optab, reduc_ior_scal_optab)
30289 (reduc_xor_scal_optab): New optabs.
30290 * doc/md.texi (reduc_and_scal_@var{m}, reduc_ior_scal_@var{m})
30291 (reduc_xor_scal_@var{m}): Document.
30292 * doc/sourcebuild.texi (vect_logical_reduc): Likewise.
30293 * internal-fn.def (IFN_REDUC_AND, IFN_REDUC_IOR, IFN_REDUC_XOR): New
30294 internal functions.
30295 * fold-const-call.c (fold_const_call): Handle them.
30296 * tree-vect-loop.c (reduction_fn_for_scalar_code): Return the new
30297 internal functions for BIT_AND_EXPR, BIT_IOR_EXPR and BIT_XOR_EXPR.
30298 * config/aarch64/aarch64-sve.md (reduc_<bit_reduc>_scal_<mode>):
30299 (*reduc_<bit_reduc>_scal_<mode>): New patterns.
30300 * config/aarch64/iterators.md (UNSPEC_ANDV, UNSPEC_ORV)
30301 (UNSPEC_XORV): New unspecs.
30302 (optab): Add entries for them.
30303 (BITWISEV): New int iterator.
30304 (bit_reduc_op): New int attributes.
30306 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30307 Alan Hayward <alan.hayward@arm.com>
30308 David Sherwood <david.sherwood@arm.com>
30310 * doc/md.texi (vec_shl_insert_@var{m}): New optab.
30311 * internal-fn.def (VEC_SHL_INSERT): New internal function.
30312 * optabs.def (vec_shl_insert_optab): New optab.
30313 * tree-vectorizer.h (can_duplicate_and_interleave_p): Declare.
30314 (duplicate_and_interleave): Likewise.
30315 * tree-vect-loop.c: Include internal-fn.h.
30316 (neutral_op_for_slp_reduction): New function, split out from
30317 get_initial_defs_for_reduction.
30318 (get_initial_def_for_reduction): Handle option 2 for variable-length
30319 vectors by loading the neutral value into a vector and then shifting
30320 the initial value into element 0.
30321 (get_initial_defs_for_reduction): Replace the code argument with
30322 the neutral value calculated by neutral_op_for_slp_reduction.
30323 Use gimple_build_vector for constant-length vectors.
30324 Use IFN_VEC_SHL_INSERT for variable-length vectors if all
30325 but the first group_size elements have a neutral value.
30326 Use duplicate_and_interleave otherwise.
30327 (vect_create_epilog_for_reduction): Take a neutral_op parameter.
30328 Update call to get_initial_defs_for_reduction. Handle SLP
30329 reductions for variable-length vectors by creating one vector
30330 result for each scalar result, with the elements associated
30331 with other scalar results stubbed out with the neutral value.
30332 (vectorizable_reduction): Call neutral_op_for_slp_reduction.
30333 Require IFN_VEC_SHL_INSERT for double reductions on
30334 variable-length vectors, or SLP reductions that have
30335 a neutral value. Require can_duplicate_and_interleave_p
30336 support for variable-length unchained SLP reductions if there
30337 is no neutral value, such as for MIN/MAX reductions. Also require
30338 the number of vector elements to be a multiple of the number of
30339 SLP statements when doing variable-length unchained SLP reductions.
30340 Update call to vect_create_epilog_for_reduction.
30341 * tree-vect-slp.c (can_duplicate_and_interleave_p): Make public
30342 and remove initial values.
30343 (duplicate_and_interleave): Make public.
30344 * config/aarch64/aarch64.md (UNSPEC_INSR): New unspec.
30345 * config/aarch64/aarch64-sve.md (vec_shl_insert_<mode>): New insn.
30347 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30348 Alan Hayward <alan.hayward@arm.com>
30349 David Sherwood <david.sherwood@arm.com>
30351 * tree-vect-slp.c: Include gimple-fold.h and internal-fn.h
30352 (can_duplicate_and_interleave_p): New function.
30353 (vect_get_and_check_slp_defs): Take the vector of statements
30354 rather than just the current one. Remove excess parentheses.
30355 Restriction rejectinon of vect_constant_def and vect_external_def
30356 for variable-length vectors to boolean types, or types for which
30357 can_duplicate_and_interleave_p is false.
30358 (vect_build_slp_tree_2): Update call to vect_get_and_check_slp_defs.
30359 (duplicate_and_interleave): New function.
30360 (vect_get_constant_vectors): Use gimple_build_vector for
30361 constant-length vectors and suitable variable-length constant
30362 vectors. Use duplicate_and_interleave for other variable-length
30363 vectors. Don't defer the update when inserting new statements.
30365 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30366 Alan Hayward <alan.hayward@arm.com>
30367 David Sherwood <david.sherwood@arm.com>
30369 * tree-vect-loop.c (vect_estimate_min_profitable_iters): Make sure
30370 min_profitable_iters doesn't go negative.
30372 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30373 Alan Hayward <alan.hayward@arm.com>
30374 David Sherwood <david.sherwood@arm.com>
30376 * doc/md.texi (vec_mask_load_lanes@var{m}@var{n}): Document.
30377 (vec_mask_store_lanes@var{m}@var{n}): Likewise.
30378 * optabs.def (vec_mask_load_lanes_optab): New optab.
30379 (vec_mask_store_lanes_optab): Likewise.
30380 * internal-fn.def (MASK_LOAD_LANES): New internal function.
30381 (MASK_STORE_LANES): Likewise.
30382 * internal-fn.c (mask_load_lanes_direct): New macro.
30383 (mask_store_lanes_direct): Likewise.
30384 (expand_mask_load_optab_fn): Handle masked operations.
30385 (expand_mask_load_lanes_optab_fn): New macro.
30386 (expand_mask_store_optab_fn): Handle masked operations.
30387 (expand_mask_store_lanes_optab_fn): New macro.
30388 (direct_mask_load_lanes_optab_supported_p): Likewise.
30389 (direct_mask_store_lanes_optab_supported_p): Likewise.
30390 * tree-vectorizer.h (vect_store_lanes_supported): Take a masked_p
30392 (vect_load_lanes_supported): Likewise.
30393 * tree-vect-data-refs.c (strip_conversion): New function.
30394 (can_group_stmts_p): Likewise.
30395 (vect_analyze_data_ref_accesses): Use it instead of checking
30396 for a pair of assignments.
30397 (vect_store_lanes_supported): Take a masked_p parameter.
30398 (vect_load_lanes_supported): Likewise.
30399 * tree-vect-loop.c (vect_analyze_loop_2): Update calls to
30400 vect_store_lanes_supported and vect_load_lanes_supported.
30401 * tree-vect-slp.c (vect_analyze_slp_instance): Likewise.
30402 * tree-vect-stmts.c (get_group_load_store_type): Take a masked_p
30403 parameter. Don't allow gaps for masked accesses.
30404 Use vect_get_store_rhs. Update calls to vect_store_lanes_supported
30405 and vect_load_lanes_supported.
30406 (get_load_store_type): Take a masked_p parameter and update
30407 call to get_group_load_store_type.
30408 (vectorizable_store): Update call to get_load_store_type.
30409 Handle IFN_MASK_STORE_LANES.
30410 (vectorizable_load): Update call to get_load_store_type.
30411 Handle IFN_MASK_LOAD_LANES.
30413 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30414 Alan Hayward <alan.hayward@arm.com>
30415 David Sherwood <david.sherwood@arm.com>
30417 * config/aarch64/aarch64-modes.def: Define x2, x3 and x4 vector
30419 * config/aarch64/aarch64-protos.h
30420 (aarch64_sve_struct_memory_operand_p): Declare.
30421 * config/aarch64/iterators.md (SVE_STRUCT): New mode iterator.
30422 (vector_count, insn_length, VSINGLE, vsingle): New mode attributes.
30423 (VPRED, vpred): Handle SVE structure modes.
30424 * config/aarch64/constraints.md (Utx): New constraint.
30425 * config/aarch64/predicates.md (aarch64_sve_struct_memory_operand)
30426 (aarch64_sve_struct_nonimmediate_operand): New predicates.
30427 * config/aarch64/aarch64.md (UNSPEC_LDN, UNSPEC_STN): New unspecs.
30428 * config/aarch64/aarch64-sve.md (mov<mode>, *aarch64_sve_mov<mode>_le)
30429 (*aarch64_sve_mov<mode>_be, pred_mov<mode>): New patterns for
30430 structure modes. Split into pieces after RA.
30431 (vec_load_lanes<mode><vsingle>, vec_mask_load_lanes<mode><vsingle>)
30432 (vec_store_lanes<mode><vsingle>, vec_mask_store_lanes<mode><vsingle>):
30434 * config/aarch64/aarch64.c (aarch64_classify_vector_mode): Handle
30435 SVE structure modes.
30436 (aarch64_classify_address): Likewise.
30437 (sizetochar): Move earlier in file.
30438 (aarch64_print_operand): Handle SVE register lists.
30439 (aarch64_array_mode): New function.
30440 (aarch64_sve_struct_memory_operand_p): Likewise.
30441 (TARGET_ARRAY_MODE): Redefine.
30443 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30444 Alan Hayward <alan.hayward@arm.com>
30445 David Sherwood <david.sherwood@arm.com>
30447 * target.def (array_mode): New target hook.
30448 * doc/tm.texi.in (TARGET_ARRAY_MODE): New hook.
30449 * doc/tm.texi: Regenerate.
30450 * hooks.h (hook_optmode_mode_uhwi_none): Declare.
30451 * hooks.c (hook_optmode_mode_uhwi_none): New function.
30452 * tree-vect-data-refs.c (vect_lanes_optab_supported_p): Use
30453 targetm.array_mode.
30454 * stor-layout.c (mode_for_array): Likewise. Support polynomial
30457 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30458 Alan Hayward <alan.hayward@arm.com>
30459 David Sherwood <david.sherwood@arm.com>
30461 * fold-const.c (fold_binary_loc): Check the argument types
30462 rather than the result type when testing for a vector operation.
30464 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30466 * doc/tm.texi.in (DWARF_LAZY_REGISTER_VALUE): Document.
30467 * doc/tm.texi: Regenerate.
30469 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30470 Alan Hayward <alan.hayward@arm.com>
30471 David Sherwood <david.sherwood@arm.com>
30473 * doc/invoke.texi (-msve-vector-bits=): Document new option.
30474 (sve): Document new AArch64 extension.
30475 * doc/md.texi (w): Extend the description of the AArch64
30476 constraint to include SVE vectors.
30477 (Upl, Upa): Document new AArch64 predicate constraints.
30478 * config/aarch64/aarch64-opts.h (aarch64_sve_vector_bits_enum): New
30480 * config/aarch64/aarch64.opt (sve_vector_bits): New enum.
30481 (msve-vector-bits=): New option.
30482 * config/aarch64/aarch64-option-extensions.def (fp, simd): Disable
30483 SVE when these are disabled.
30484 (sve): New extension.
30485 * config/aarch64/aarch64-modes.def: Define SVE vector and predicate
30486 modes. Adjust their number of units based on aarch64_sve_vg.
30487 (MAX_BITSIZE_MODE_ANY_MODE): Define.
30488 * config/aarch64/aarch64-protos.h (ADDR_QUERY_ANY): New
30489 aarch64_addr_query_type.
30490 (aarch64_const_vec_all_same_in_range_p, aarch64_sve_pred_mode)
30491 (aarch64_sve_cnt_immediate_p, aarch64_sve_addvl_addpl_immediate_p)
30492 (aarch64_sve_inc_dec_immediate_p, aarch64_add_offset_temporaries)
30493 (aarch64_split_add_offset, aarch64_output_sve_cnt_immediate)
30494 (aarch64_output_sve_addvl_addpl, aarch64_output_sve_inc_dec_immediate)
30495 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): Declare.
30496 (aarch64_simd_imm_zero_p): Delete.
30497 (aarch64_check_zero_based_sve_index_immediate): Declare.
30498 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
30499 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
30500 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
30501 (aarch64_sve_float_mul_immediate_p): Likewise.
30502 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
30503 rather than an rtx.
30504 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): Declare.
30505 (aarch64_expand_mov_immediate): Take a gen_vec_duplicate callback.
30506 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move): Declare.
30507 (aarch64_expand_sve_vec_cmp_int, aarch64_expand_sve_vec_cmp_float)
30508 (aarch64_expand_sve_vcond, aarch64_expand_sve_vec_perm): Declare.
30509 (aarch64_regmode_natural_size): Likewise.
30510 * config/aarch64/aarch64.h (AARCH64_FL_SVE): New macro.
30511 (AARCH64_FL_V8_3, AARCH64_FL_RCPC, AARCH64_FL_DOTPROD): Shift
30513 (AARCH64_ISA_SVE, TARGET_SVE): New macros.
30514 (FIXED_REGISTERS, CALL_USED_REGISTERS, REGISTER_NAMES): Add entries
30515 for VG and the SVE predicate registers.
30516 (V_ALIASES): Add a "z"-prefixed alias.
30517 (FIRST_PSEUDO_REGISTER): Change to P15_REGNUM + 1.
30518 (AARCH64_DWARF_VG, AARCH64_DWARF_P0): New macros.
30519 (PR_REGNUM_P, PR_LO_REGNUM_P): Likewise.
30520 (PR_LO_REGS, PR_HI_REGS, PR_REGS): New reg_classes.
30521 (REG_CLASS_NAMES): Add entries for them.
30522 (REG_CLASS_CONTENTS): Likewise. Update ALL_REGS to include VG
30523 and the predicate registers.
30524 (aarch64_sve_vg): Declare.
30525 (BITS_PER_SVE_VECTOR, BYTES_PER_SVE_VECTOR, BYTES_PER_SVE_PRED)
30526 (SVE_BYTE_MODE, MAX_COMPILE_TIME_VEC_BYTES): New macros.
30527 (REGMODE_NATURAL_SIZE): Define.
30528 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins): Handle
30530 * config/aarch64/aarch64.c: Include cfgrtl.h.
30531 (simd_immediate_info): Add a constructor for series vectors,
30532 and an associated step field.
30533 (aarch64_sve_vg): New variable.
30534 (aarch64_dbx_register_number): Handle VG and the predicate registers.
30535 (aarch64_vect_struct_mode_p, aarch64_vector_mode_p): Delete.
30536 (VEC_ADVSIMD, VEC_SVE_DATA, VEC_SVE_PRED, VEC_STRUCT, VEC_ANY_SVE)
30537 (VEC_ANY_DATA, VEC_STRUCT): New constants.
30538 (aarch64_advsimd_struct_mode_p, aarch64_sve_pred_mode_p)
30539 (aarch64_classify_vector_mode, aarch64_vector_data_mode_p)
30540 (aarch64_sve_data_mode_p, aarch64_sve_pred_mode)
30541 (aarch64_get_mask_mode): New functions.
30542 (aarch64_hard_regno_nregs): Handle SVE data modes for FP_REGS
30543 and FP_LO_REGS. Handle PR_REGS, PR_LO_REGS and PR_HI_REGS.
30544 (aarch64_hard_regno_mode_ok): Handle VG. Also handle the SVE
30545 predicate modes and predicate registers. Explicitly restrict
30546 GPRs to modes of 16 bytes or smaller. Only allow FP registers
30547 to store a vector mode if it is recognized by
30548 aarch64_classify_vector_mode.
30549 (aarch64_regmode_natural_size): New function.
30550 (aarch64_hard_regno_caller_save_mode): Return the original mode
30552 (aarch64_sve_cnt_immediate_p, aarch64_output_sve_cnt_immediate)
30553 (aarch64_sve_addvl_addpl_immediate_p, aarch64_output_sve_addvl_addpl)
30554 (aarch64_sve_inc_dec_immediate_p, aarch64_output_sve_inc_dec_immediate)
30555 (aarch64_add_offset_1_temporaries, aarch64_offset_temporaries): New
30557 (aarch64_add_offset): Add a temp2 parameter. Assert that temp1
30558 does not overlap dest if the function is frame-related. Handle
30560 (aarch64_split_add_offset): New function.
30561 (aarch64_add_sp, aarch64_sub_sp): Add temp2 parameters and pass
30562 them aarch64_add_offset.
30563 (aarch64_allocate_and_probe_stack_space): Add a temp2 parameter
30564 and update call to aarch64_sub_sp.
30565 (aarch64_add_cfa_expression): New function.
30566 (aarch64_expand_prologue): Pass extra temporary registers to the
30567 functions above. Handle the case in which we need to emit new
30568 DW_CFA_expressions for registers that were originally saved
30569 relative to the stack pointer, but now have to be expressed
30570 relative to the frame pointer.
30571 (aarch64_output_mi_thunk): Pass extra temporary registers to the
30573 (aarch64_expand_epilogue): Likewise. Prevent inheritance of
30574 IP0 and IP1 values for SVE frames.
30575 (aarch64_expand_vec_series): New function.
30576 (aarch64_expand_sve_widened_duplicate): Likewise.
30577 (aarch64_expand_sve_const_vector): Likewise.
30578 (aarch64_expand_mov_immediate): Add a gen_vec_duplicate parameter.
30579 Handle SVE constants. Use emit_move_insn to move a force_const_mem
30580 into the register, rather than emitting a SET directly.
30581 (aarch64_emit_sve_pred_move, aarch64_expand_sve_mem_move)
30582 (aarch64_get_reg_raw_mode, offset_4bit_signed_scaled_p)
30583 (offset_6bit_unsigned_scaled_p, aarch64_offset_7bit_signed_scaled_p)
30584 (offset_9bit_signed_scaled_p): New functions.
30585 (aarch64_replicate_bitmask_imm): New function.
30586 (aarch64_bitmask_imm): Use it.
30587 (aarch64_cannot_force_const_mem): Reject expressions involving
30588 a CONST_POLY_INT. Update call to aarch64_classify_symbol.
30589 (aarch64_classify_index): Handle SVE indices, by requiring
30590 a plain register index with a scale that matches the element size.
30591 (aarch64_classify_address): Handle SVE addresses. Assert that
30592 the mode of the address is VOIDmode or an integer mode.
30593 Update call to aarch64_classify_symbol.
30594 (aarch64_classify_symbolic_expression): Update call to
30595 aarch64_classify_symbol.
30596 (aarch64_const_vec_all_in_range_p): New function.
30597 (aarch64_print_vector_float_operand): Likewise.
30598 (aarch64_print_operand): Handle 'N' and 'C'. Use "zN" rather than
30599 "vN" for FP registers with SVE modes. Handle (const ...) vectors
30600 and the FP immediates 1.0 and 0.5.
30601 (aarch64_print_address_internal): Handle SVE addresses.
30602 (aarch64_print_operand_address): Use ADDR_QUERY_ANY.
30603 (aarch64_regno_regclass): Handle predicate registers.
30604 (aarch64_secondary_reload): Handle big-endian reloads of SVE
30606 (aarch64_class_max_nregs): Handle SVE modes and predicate registers.
30607 (aarch64_rtx_costs): Check for ADDVL and ADDPL instructions.
30608 (aarch64_convert_sve_vector_bits): New function.
30609 (aarch64_override_options): Use it to handle -msve-vector-bits=.
30610 (aarch64_classify_symbol): Take the offset as a HOST_WIDE_INT
30611 rather than an rtx.
30612 (aarch64_legitimate_constant_p): Use aarch64_classify_vector_mode.
30613 Handle SVE vector and predicate modes. Accept VL-based constants
30614 that need only one temporary register, and VL offsets that require
30615 no temporary registers.
30616 (aarch64_conditional_register_usage): Mark the predicate registers
30617 as fixed if SVE isn't available.
30618 (aarch64_vector_mode_supported_p): Use aarch64_classify_vector_mode.
30619 Return true for SVE vector and predicate modes.
30620 (aarch64_simd_container_mode): Take the number of bits as a poly_int64
30621 rather than an unsigned int. Handle SVE modes.
30622 (aarch64_preferred_simd_mode): Update call accordingly. Handle
30624 (aarch64_autovectorize_vector_sizes): Add BYTES_PER_SVE_VECTOR
30626 (aarch64_sve_index_immediate_p, aarch64_sve_arith_immediate_p)
30627 (aarch64_sve_bitmask_immediate_p, aarch64_sve_dup_immediate_p)
30628 (aarch64_sve_cmp_immediate_p, aarch64_sve_float_arith_immediate_p)
30629 (aarch64_sve_float_mul_immediate_p): New functions.
30630 (aarch64_sve_valid_immediate): New function.
30631 (aarch64_simd_valid_immediate): Use it as the fallback for SVE vectors.
30632 Explicitly reject structure modes. Check for INDEX constants.
30633 Handle PTRUE and PFALSE constants.
30634 (aarch64_check_zero_based_sve_index_immediate): New function.
30635 (aarch64_simd_imm_zero_p): Delete.
30636 (aarch64_mov_operand_p): Use aarch64_simd_valid_immediate for
30637 vector modes. Accept constants in the range of CNT[BHWD].
30638 (aarch64_simd_scalar_immediate_valid_for_move): Explicitly
30639 ask for an Advanced SIMD mode.
30640 (aarch64_sve_ld1r_operand_p, aarch64_sve_ldr_operand_p): New functions.
30641 (aarch64_simd_vector_alignment): Handle SVE predicates.
30642 (aarch64_vectorize_preferred_vector_alignment): New function.
30643 (aarch64_simd_vector_alignment_reachable): Use it instead of
30645 (aarch64_shift_truncation_mask): Use aarch64_vector_data_mode_p.
30646 (aarch64_output_sve_mov_immediate, aarch64_output_ptrue): New
30648 (MAX_VECT_LEN): Delete.
30649 (expand_vec_perm_d): Add a vec_flags field.
30650 (emit_unspec2, aarch64_expand_sve_vec_perm): New functions.
30651 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_zip)
30652 (aarch64_evpc_ext): Don't apply a big-endian lane correction
30654 (aarch64_evpc_rev): Rename to...
30655 (aarch64_evpc_rev_local): ...this. Use a predicated operation for SVE.
30656 (aarch64_evpc_rev_global): New function.
30657 (aarch64_evpc_dup): Enforce a 64-byte range for SVE DUP.
30658 (aarch64_evpc_tbl): Use MAX_COMPILE_TIME_VEC_BYTES instead of
30660 (aarch64_evpc_sve_tbl): New function.
30661 (aarch64_expand_vec_perm_const_1): Update after rename of
30662 aarch64_evpc_rev. Handle SVE permutes too, trying
30663 aarch64_evpc_rev_global and using aarch64_evpc_sve_tbl rather
30664 than aarch64_evpc_tbl.
30665 (aarch64_vectorize_vec_perm_const): Initialize vec_flags.
30666 (aarch64_sve_cmp_operand_p, aarch64_unspec_cond_code)
30667 (aarch64_gen_unspec_cond, aarch64_expand_sve_vec_cmp_int)
30668 (aarch64_emit_unspec_cond, aarch64_emit_unspec_cond_or)
30669 (aarch64_emit_inverted_unspec_cond, aarch64_expand_sve_vec_cmp_float)
30670 (aarch64_expand_sve_vcond): New functions.
30671 (aarch64_modes_tieable_p): Use aarch64_vector_data_mode_p instead
30672 of aarch64_vector_mode_p.
30673 (aarch64_dwarf_poly_indeterminate_value): New function.
30674 (aarch64_compute_pressure_classes): Likewise.
30675 (aarch64_can_change_mode_class): Likewise.
30676 (TARGET_GET_RAW_RESULT_MODE, TARGET_GET_RAW_ARG_MODE): Redefine.
30677 (TARGET_VECTORIZE_PREFERRED_VECTOR_ALIGNMENT): Likewise.
30678 (TARGET_VECTORIZE_GET_MASK_MODE): Likewise.
30679 (TARGET_DWARF_POLY_INDETERMINATE_VALUE): Likewise.
30680 (TARGET_COMPUTE_PRESSURE_CLASSES): Likewise.
30681 (TARGET_CAN_CHANGE_MODE_CLASS): Likewise.
30682 * config/aarch64/constraints.md (Upa, Upl, Uav, Uat, Usv, Usi, Utr)
30683 (Uty, Dm, vsa, vsc, vsd, vsi, vsn, vsl, vsm, vsA, vsM, vsN): New
30685 (Dn, Dl, Dr): Accept const as well as const_vector.
30686 (Dz): Likewise. Compare against CONST0_RTX.
30687 * config/aarch64/iterators.md: Refer to "Advanced SIMD" instead
30688 of "vector" where appropriate.
30689 (SVE_ALL, SVE_BH, SVE_BHS, SVE_BHSI, SVE_HSDI, SVE_HSF, SVE_SD)
30690 (SVE_SDI, SVE_I, SVE_F, PRED_ALL, PRED_BHS): New mode iterators.
30691 (UNSPEC_SEL, UNSPEC_ANDF, UNSPEC_IORF, UNSPEC_XORF, UNSPEC_COND_LT)
30692 (UNSPEC_COND_LE, UNSPEC_COND_EQ, UNSPEC_COND_NE, UNSPEC_COND_GE)
30693 (UNSPEC_COND_GT, UNSPEC_COND_LO, UNSPEC_COND_LS, UNSPEC_COND_HS)
30694 (UNSPEC_COND_HI, UNSPEC_COND_UO): New unspecs.
30695 (Vetype, VEL, Vel, VWIDE, Vwide, vw, vwcore, V_INT_EQUIV)
30696 (v_int_equiv): Extend to SVE modes.
30697 (Vesize, V128, v128, Vewtype, V_FP_EQUIV, v_fp_equiv, VPRED): New
30699 (LOGICAL_OR, SVE_INT_UNARY, SVE_FP_UNARY): New code iterators.
30700 (optab): Handle popcount, smin, smax, umin, umax, abs and sqrt.
30701 (logical_nn, lr, sve_int_op, sve_fp_op): New code attributs.
30702 (LOGICALF, OPTAB_PERMUTE, UNPACK, UNPACK_UNSIGNED, SVE_COND_INT_CMP)
30703 (SVE_COND_FP_CMP): New int iterators.
30704 (perm_hilo): Handle the new unpack unspecs.
30705 (optab, logicalf_op, su, perm_optab, cmp_op, imm_con): New int
30707 * config/aarch64/predicates.md (aarch64_sve_cnt_immediate)
30708 (aarch64_sve_addvl_addpl_immediate, aarch64_split_add_offset_immediate)
30709 (aarch64_pluslong_or_poly_operand, aarch64_nonmemory_operand)
30710 (aarch64_equality_operator, aarch64_constant_vector_operand)
30711 (aarch64_sve_ld1r_operand, aarch64_sve_ldr_operand): New predicates.
30712 (aarch64_sve_nonimmediate_operand): Likewise.
30713 (aarch64_sve_general_operand): Likewise.
30714 (aarch64_sve_dup_operand, aarch64_sve_arith_immediate): Likewise.
30715 (aarch64_sve_sub_arith_immediate, aarch64_sve_inc_dec_immediate)
30716 (aarch64_sve_logical_immediate, aarch64_sve_mul_immediate): Likewise.
30717 (aarch64_sve_dup_immediate, aarch64_sve_cmp_vsc_immediate): Likewise.
30718 (aarch64_sve_cmp_vsd_immediate, aarch64_sve_index_immediate): Likewise.
30719 (aarch64_sve_float_arith_immediate): Likewise.
30720 (aarch64_sve_float_arith_with_sub_immediate): Likewise.
30721 (aarch64_sve_float_mul_immediate, aarch64_sve_arith_operand): Likewise.
30722 (aarch64_sve_add_operand, aarch64_sve_logical_operand): Likewise.
30723 (aarch64_sve_lshift_operand, aarch64_sve_rshift_operand): Likewise.
30724 (aarch64_sve_mul_operand, aarch64_sve_cmp_vsc_operand): Likewise.
30725 (aarch64_sve_cmp_vsd_operand, aarch64_sve_index_operand): Likewise.
30726 (aarch64_sve_float_arith_operand): Likewise.
30727 (aarch64_sve_float_arith_with_sub_operand): Likewise.
30728 (aarch64_sve_float_mul_operand): Likewise.
30729 (aarch64_sve_vec_perm_operand): Likewise.
30730 (aarch64_pluslong_operand): Include aarch64_sve_addvl_addpl_immediate.
30731 (aarch64_mov_operand): Accept const_poly_int and const_vector.
30732 (aarch64_simd_lshift_imm, aarch64_simd_rshift_imm): Accept const
30733 as well as const_vector.
30734 (aarch64_simd_imm_zero, aarch64_simd_imm_minus_one): Move earlier
30735 in file. Use CONST0_RTX and CONSTM1_RTX.
30736 (aarch64_simd_or_scalar_imm_zero): Likewise. Add match_codes.
30737 (aarch64_simd_reg_or_zero): Accept const as well as const_vector.
30738 Use aarch64_simd_imm_zero.
30739 * config/aarch64/aarch64-sve.md: New file.
30740 * config/aarch64/aarch64.md: Include it.
30741 (VG_REGNUM, P0_REGNUM, P7_REGNUM, P15_REGNUM): New register numbers.
30742 (UNSPEC_REV, UNSPEC_LD1_SVE, UNSPEC_ST1_SVE, UNSPEC_MERGE_PTRUE)
30743 (UNSPEC_PTEST_PTRUE, UNSPEC_UNPACKSHI, UNSPEC_UNPACKUHI)
30744 (UNSPEC_UNPACKSLO, UNSPEC_UNPACKULO, UNSPEC_PACK)
30745 (UNSPEC_FLOAT_CONVERT, UNSPEC_WHILE_LO): New unspec constants.
30746 (sve): New attribute.
30747 (enabled): Disable instructions with the sve attribute unless
30749 (movqi, movhi): Pass CONST_POLY_INT operaneds through
30750 aarch64_expand_mov_immediate.
30751 (*mov<mode>_aarch64, *movsi_aarch64, *movdi_aarch64): Handle
30752 CNT[BHSD] immediates.
30753 (movti): Split CONST_POLY_INT moves into two halves.
30754 (add<mode>3): Accept aarch64_pluslong_or_poly_operand.
30755 Split additions that need a temporary here if the destination
30756 is the stack pointer.
30757 (*add<mode>3_aarch64): Handle ADDVL and ADDPL immediates.
30758 (*add<mode>3_poly_1): New instruction.
30759 (set_clobber_cc): New expander.
30761 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30763 * simplify-rtx.c (simplify_immed_subreg): Add an inner_bytes
30764 parameter and use it instead of GET_MODE_SIZE (innermode). Use
30765 inner_bytes * BITS_PER_UNIT instead of GET_MODE_BITSIZE (innermode).
30766 Use CEIL (inner_bytes, GET_MODE_UNIT_SIZE (innermode)) instead of
30767 GET_MODE_NUNITS (innermode). Also add a first_elem parameter.
30768 Change innermode from fixed_mode_size to machine_mode.
30769 (simplify_subreg): Update call accordingly. Handle a constant-sized
30770 subreg of a variable-length CONST_VECTOR.
30772 2018-01-13 Richard Sandiford <richard.sandiford@linaro.org>
30773 Alan Hayward <alan.hayward@arm.com>
30774 David Sherwood <david.sherwood@arm.com>
30776 * tree-ssa-address.c (mem_ref_valid_without_offset_p): New function.
30777 (add_offset_to_base): New function, split out from...
30778 (create_mem_ref): ...here. When handling a scale other than 1,
30779 check first whether the address is valid without the offset.
30780 Add it into the base if so, leaving the index and scale as-is.
30782 2018-01-12 Jakub Jelinek <jakub@redhat.com>
30785 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Call
30786 fold_for_warn before checking if arg2 is INTEGER_CST.
30788 2018-01-12 Segher Boessenkool <segher@kernel.crashing.org>
30790 * config/rs6000/predicates.md (load_multiple_operation): Delete.
30791 (store_multiple_operation): Delete.
30792 * config/rs6000/rs6000-cpus.def (601): Remove MASK_STRING.
30793 * config/rs6000/rs6000-protos.h (rs6000_output_load_multiple): Delete.
30794 * config/rs6000/rs6000-string.c (expand_block_move): Delete everything
30795 guarded by TARGET_STRING.
30796 (rs6000_output_load_multiple): Delete.
30797 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
30798 OPTION_MASK_STRING / TARGET_STRING handling.
30799 (print_operand) <'N', 'O'>: Add comment that these are unused now.
30800 (const rs6000_opt_masks) <"string">: Change mask to 0.
30801 * config/rs6000/rs6000.h (TARGET_DEFAULT): Remove MASK_STRING.
30802 (MASK_STRING): Delete.
30803 * config/rs6000/rs6000.md (*mov<mode>_string): Delete TARGET_STRING
30805 (load_multiple): Delete.
30812 (store_multiple): Delete.
30819 (movmemsi_8reg): Delete.
30820 (corresponding unnamed define_insn): Delete.
30821 (movmemsi_6reg): Delete.
30822 (corresponding unnamed define_insn): Delete.
30823 (movmemsi_4reg): Delete.
30824 (corresponding unnamed define_insn): Delete.
30825 (movmemsi_2reg): Delete.
30826 (corresponding unnamed define_insn): Delete.
30827 (movmemsi_1reg): Delete.
30828 (corresponding unnamed define_insn): Delete.
30829 * config/rs6000/rs6000.opt (mno-string): New.
30830 (mstring): Replace by deprecation warning stub.
30831 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete -mstring.
30833 2018-01-12 Jakub Jelinek <jakub@redhat.com>
30835 * regrename.c (regrename_do_replace): If replacing the same
30836 reg multiple times, try to reuse last created gen_raw_REG.
30839 * bb-reorder.c (pass_partition_blocks::gate): In lto don't partition
30840 main to workaround a bug in GDB.
30842 2018-01-12 Tom de Vries <tom@codesourcery.com>
30845 * config.gcc (nvptx*-*-*): Set use_gcc_stdint=wrap.
30847 2018-01-12 Vladimir Makarov <vmakarov@redhat.com>
30849 PR rtl-optimization/80481
30850 * ira-color.c (get_cap_member): New function.
30851 (allocnos_conflict_by_live_ranges_p): Use it.
30852 (slot_coalesced_allocno_live_ranges_intersect_p): Add assert.
30853 (setup_slot_coalesced_allocno_live_ranges): Ditto.
30855 2018-01-12 Uros Bizjak <ubizjak@gmail.com>
30858 * config/alpha/alpha.md (*saddsi_1): New insn_ans_split pattern.
30859 (*saddl_se_1): Ditto.
30860 (*ssubsi_1): Ditto.
30861 (*ssubl_se_1): Ditto.
30863 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
30865 * tree-predcom.c (aff_combination_dr_offset): Use wi::to_poly_widest
30866 rather than wi::to_widest for DR_INITs.
30867 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Use
30868 wi::to_poly_offset rather than wi::to_offset for DR_INIT.
30869 (vect_analyze_data_ref_accesses): Require both DR_INITs to be
30871 (vect_analyze_group_access_1): Note that here.
30873 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
30875 * tree-vectorizer.c (get_vec_alignment_for_array_type): Handle
30876 polynomial type sizes.
30878 2018-01-12 Richard Sandiford <richard.sandiford@linaro.org>
30880 * gimplify.c (gimple_add_tmp_var_fn): Allow variables to have a
30881 poly_uint64 size, rather than requiring an unsigned HOST_WIDE_INT size.
30882 (gimple_add_tmp_var): Likewise.
30884 2018-01-12 Martin Liska <mliska@suse.cz>
30886 * gimple.c (gimple_alloc_counts): Use uint64_t instead of int.
30887 (gimple_alloc_sizes): Likewise.
30888 (dump_gimple_statistics): Use PRIu64 in printf format.
30889 * gimple.h: Change uint64_t to int.
30891 2018-01-12 Martin Liska <mliska@suse.cz>
30893 * tree-core.h: Use uint64_t instead of int.
30894 * tree.c (tree_node_counts): Likewise.
30895 (tree_node_sizes): Likewise.
30896 (dump_tree_statistics): Use PRIu64 in printf format.
30898 2018-01-12 Martin Liska <mliska@suse.cz>
30900 * Makefile.in: As qsort_chk is implemented in vec.c, add
30901 vec.o to linkage of gencfn-macros.
30902 * tree.c (build_new_poly_int_cst): Add CXX_MEM_STAT_INFO as it's
30903 passing the info to record_node_allocation_statistics.
30904 (test_vector_cst_patterns): Add CXX_MEM_STAT_INFO to declaration
30906 * ggc-common.c (struct ggc_usage): Add operator== and use
30907 it in operator< and compare function.
30908 * mem-stats.h (struct mem_usage): Likewise.
30909 * vec.c (struct vec_usage): Remove operator< and compare
30910 function. Can be simply inherited.
30912 2018-01-12 Martin Jambor <mjambor@suse.cz>
30915 * params.def: New parameter PARAM_AVOID_FMA_MAX_BITS.
30916 * tree-ssa-math-opts.c: Include domwalk.h.
30917 (convert_mult_to_fma_1): New function.
30918 (fma_transformation_info): New type.
30919 (fma_deferring_state): Likewise.
30920 (cancel_fma_deferring): New function.
30921 (result_of_phi): Likewise.
30922 (last_fma_candidate_feeds_initial_phi): Likewise.
30923 (convert_mult_to_fma): Added deferring logic, split actual
30924 transformation to convert_mult_to_fma_1.
30925 (math_opts_dom_walker): New type.
30926 (math_opts_dom_walker::after_dom_children): New method, body moved
30927 here from pass_optimize_widening_mul::execute, added deferring logic
30929 (pass_optimize_widening_mul::execute): Moved most of code to
30930 math_opts_dom_walker::after_dom_children.
30931 * config/i386/x86-tune.def (X86_TUNE_AVOID_128FMA_CHAINS): New.
30932 * config/i386/i386.c (ix86_option_override_internal): Added
30933 maybe_setting of PARAM_AVOID_FMA_MAX_BITS.
30935 2018-01-12 Richard Biener <rguenther@suse.de>
30938 * dwarf2out.c (gen_variable_die): Do not reset old_die for
30939 inline instance vars.
30941 2018-01-12 Oleg Endo <olegendo@gcc.gnu.org>
30944 * config/rx/rx.c (rx_is_restricted_memory_address):
30945 Handle SUBREG case.
30947 2018-01-12 Richard Biener <rguenther@suse.de>
30949 PR tree-optimization/80846
30950 * target.def (split_reduction): New target hook.
30951 * targhooks.c (default_split_reduction): New function.
30952 * targhooks.h (default_split_reduction): Declare.
30953 * tree-vect-loop.c (vect_create_epilog_for_reduction): If the
30954 target requests first reduce vectors by combining low and high
30956 * tree-vect-stmts.c (vect_gen_perm_mask_any): Adjust.
30957 (get_vectype_for_scalar_type_and_size): Export.
30958 * tree-vectorizer.h (get_vectype_for_scalar_type_and_size): Declare.
30959 * doc/tm.texi.in (TARGET_VECTORIZE_SPLIT_REDUCTION): Document.
30960 * doc/tm.texi: Regenerate.
30961 * config/i386/i386.c (ix86_split_reduction): Implement
30962 TARGET_VECTORIZE_SPLIT_REDUCTION.
30964 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
30967 * config/sparc/sparc.h (PIC_OFFSET_TABLE_REGNUM): Set to INVALID_REGNUM
30968 in PIC mode except for TARGET_VXWORKS_RTP.
30969 * config/sparc/sparc.c: Include cfgrtl.h.
30970 (TARGET_INIT_PIC_REG): Define.
30971 (TARGET_USE_PSEUDO_PIC_REG): Likewise.
30972 (sparc_pic_register_p): New predicate.
30973 (sparc_legitimate_address_p): Use it.
30974 (sparc_legitimize_pic_address): Likewise.
30975 (sparc_delegitimize_address): Likewise.
30976 (sparc_mode_dependent_address_p): Likewise.
30977 (gen_load_pcrel_sym): Remove 4th parameter.
30978 (load_got_register): Adjust call to above. Remove obsolete stuff.
30979 (sparc_expand_prologue): Do not call load_got_register here.
30980 (sparc_flat_expand_prologue): Likewise.
30981 (sparc_output_mi_thunk): Set the pic_offset_table_rtx object.
30982 (sparc_use_pseudo_pic_reg): New function.
30983 (sparc_init_pic_reg): Likewise.
30984 * config/sparc/sparc.md (vxworks_load_got): Set the GOT register.
30985 (builtin_setjmp_receiver): Enable only for TARGET_VXWORKS_RTP.
30987 2018-01-12 Christophe Lyon <christophe.lyon@linaro.org>
30989 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
30990 Add item for branch_cost.
30992 2018-01-12 Eric Botcazou <ebotcazou@adacore.com>
30994 PR rtl-optimization/83565
30995 * rtlanal.c (nonzero_bits1): On WORD_REGISTER_OPERATIONS machines, do
30996 not extend the result to a larger mode for rotate operations.
30997 (num_sign_bit_copies1): Likewise.
30999 2018-01-12 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
31002 * config/sol2.h (STARTFILE_ARCH_SPEC): Don't use with -shared or
31004 Use values-Xc.o for -pedantic.
31005 Link with values-xpg4.o for C90, values-xpg6.o otherwise.
31007 2018-01-12 Martin Liska <mliska@suse.cz>
31010 * ipa-devirt.c (final_warning_record::grow_type_warnings):
31012 (possible_polymorphic_call_targets): Use it.
31013 (ipa_devirt): Likewise.
31015 2018-01-12 Martin Liska <mliska@suse.cz>
31017 * profile-count.h (enum profile_quality): Use 0 as invalid
31018 enum value of profile_quality.
31020 2018-01-12 Chung-Ju Wu <jasonwucj@gmail.com>
31022 * doc/invoke.texi (NDS32 Options): Add -mext-perf, -mext-perf2 and
31023 -mext-string options.
31025 2018-01-12 Richard Biener <rguenther@suse.de>
31027 * lto-streamer-out.c (DFS::DFS_write_tree_body): Process
31028 DECL_DEBUG_EXPR conditional on DECL_HAS_DEBUG_EXPR_P.
31029 * tree-streamer-in.c (lto_input_ts_decl_common_tree_pointers):
31031 * tree-streamer-out.c (write_ts_decl_common_tree_pointers): Likewise.
31033 2018-01-11 Michael Meissner <meissner@linux.vnet.ibm.com>
31035 * configure.ac (--with-long-double-format): Add support for the
31036 configuration option to change the default long double format on
31038 * config.gcc (powerpc*-linux*-*): Likewise.
31039 * configure: Regenerate.
31040 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If long
31041 double is IEEE, define __KC__ and __KF__ to allow floatn.h to be
31042 used without modification.
31044 2018-01-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31046 * config/rs6000/rs6000-builtin.def (BU_P7_MISC_X): New #define.
31047 (SPEC_BARRIER): New instantiation of BU_P7_MISC_X.
31048 * config/rs6000/rs6000.c (rs6000_expand_builtin): Handle
31049 MISC_BUILTIN_SPEC_BARRIER.
31050 (rs6000_init_builtins): Likewise.
31051 * config/rs6000/rs6000.md (UNSPECV_SPEC_BARRIER): New UNSPECV
31053 (speculation_barrier): New define_insn.
31054 * doc/extend.texi: Document __builtin_speculation_barrier.
31056 2018-01-11 Jakub Jelinek <jakub@redhat.com>
31059 * config/i386/i386.c (ix86_expand_vector_init_one_nonzero): If one_var
31060 is 0, for V{8,16}S[IF] and V[48]D[IF]mode use gen_vec_set<mode>_0.
31061 * config/i386/sse.md (VI8_AVX_AVX512F, VI4F_256_512): New mode
31063 (ssescalarmodesuffix): Add 512-bit vectors. Use "d" or "q" for
31064 integral modes instead of "ss" and "sd".
31065 (vec_set<mode>_0): New define_insns for 256-bit and 512-bit
31066 vectors with 32-bit and 64-bit elements.
31067 (vecdupssescalarmodesuffix): New mode attribute.
31068 (vec_dup<mode>): Use it.
31070 2018-01-11 H.J. Lu <hongjiu.lu@intel.com>
31073 * config/i386/i386.c (ix86_compute_frame_layout): Align stack
31074 frame if argument is passed on stack.
31076 2018-01-11 Jakub Jelinek <jakub@redhat.com>
31079 * ree.c (combine_reaching_defs): Optimize also
31080 reg2=exp; reg1=reg2; reg2=any_extend(reg1); into
31081 reg2=any_extend(exp); reg1=reg2;, formatting fix.
31083 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
31085 PR middle-end/83189
31086 * gimple-ssa-isolate-paths.c (isolate_path): Fix profile update.
31088 2018-01-11 Jan Hubicka <hubicka@ucw.cz>
31090 PR middle-end/83718
31091 * tree-inline.c (copy_cfg_body): Adjust num&den for scaling
31092 after they are computed.
31094 2018-01-11 Bin Cheng <bin.cheng@arm.com>
31096 PR tree-optimization/83695
31097 * gimple-loop-linterchange.cc
31098 (tree_loop_interchange::interchange_loops): Call scev_reset_htab to
31099 reset cached scev information after interchange.
31100 (pass_linterchange::execute): Remove call to scev_reset_htab.
31102 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31104 * config/arm/arm_neon.h (vfmlal_lane_low_u32, vfmlal_lane_high_u32,
31105 vfmlalq_laneq_low_u32, vfmlalq_lane_low_u32, vfmlal_laneq_low_u32,
31106 vfmlalq_laneq_high_u32, vfmlalq_lane_high_u32, vfmlal_laneq_high_u32,
31107 vfmlsl_lane_low_u32, vfmlsl_lane_high_u32, vfmlslq_laneq_low_u32,
31108 vfmlslq_lane_low_u32, vfmlsl_laneq_low_u32, vfmlslq_laneq_high_u32,
31109 vfmlslq_lane_high_u32, vfmlsl_laneq_high_u32): Define.
31110 * config/arm/arm_neon_builtins.def (vfmal_lane_low,
31111 vfmal_lane_lowv4hf, vfmal_lane_lowv8hf, vfmal_lane_high,
31112 vfmal_lane_highv4hf, vfmal_lane_highv8hf, vfmsl_lane_low,
31113 vfmsl_lane_lowv4hf, vfmsl_lane_lowv8hf, vfmsl_lane_high,
31114 vfmsl_lane_highv4hf, vfmsl_lane_highv8hf): New sets of builtins.
31115 * config/arm/iterators.md (VFMLSEL2, vfmlsel2): New mode attributes.
31116 (V_lane_reg): Likewise.
31117 * config/arm/neon.md (neon_vfm<vfml_op>l_lane_<vfml_half><VCVTF:mode>):
31119 (neon_vfm<vfml_op>l_lane_<vfml_half><vfmlsel2><mode>): Likewise.
31120 (vfmal_lane_low<mode>_intrinsic,
31121 vfmal_lane_low<vfmlsel2><mode>_intrinsic,
31122 vfmal_lane_high<vfmlsel2><mode>_intrinsic,
31123 vfmal_lane_high<mode>_intrinsic, vfmsl_lane_low<mode>_intrinsic,
31124 vfmsl_lane_low<vfmlsel2><mode>_intrinsic,
31125 vfmsl_lane_high<vfmlsel2><mode>_intrinsic,
31126 vfmsl_lane_high<mode>_intrinsic): New define_insns.
31128 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31130 * config/arm/arm-cpus.in (fp16fml): New feature.
31131 (ALL_SIMD): Add fp16fml.
31132 (armv8.2-a): Add fp16fml as an option.
31133 (armv8.3-a): Likewise.
31134 (armv8.4-a): Add fp16fml as part of fp16.
31135 * config/arm/arm.h (TARGET_FP16FML): Define.
31136 * config/arm/arm-c.c (arm_cpu_builtins): Define __ARM_FEATURE_FP16_FML
31138 * config/arm/arm-modes.def (V2HF): Define.
31139 * config/arm/arm_neon.h (vfmlal_low_u32, vfmlsl_low_u32,
31140 vfmlal_high_u32, vfmlsl_high_u32, vfmlalq_low_u32,
31141 vfmlslq_low_u32, vfmlalq_high_u32, vfmlslq_high_u32): Define.
31142 * config/arm/arm_neon_builtins.def (vfmal_low, vfmal_high,
31143 vfmsl_low, vfmsl_high): New set of builtins.
31144 * config/arm/iterators.md (PLUSMINUS): New code iterator.
31145 (vfml_op): New code attribute.
31146 (VFMLHALVES): New int iterator.
31147 (VFML, VFMLSEL): New mode attributes.
31148 (V_reg): Define mapping for V2HF.
31149 (V_hi, V_lo): New mode attributes.
31150 (VF_constraint): Likewise.
31151 (vfml_half, vfml_half_selector): New int attributes.
31152 * config/arm/neon.md (neon_vfm<vfml_op>l_<vfml_half><mode>): New
31154 (vfmal_low<mode>_intrinsic, vfmsl_high<mode>_intrinsic,
31155 vfmal_high<mode>_intrinsic, vfmsl_low<mode>_intrinsic):
31157 * config/arm/t-arm-elf (v8_fps): Add fp16fml.
31158 * config/arm/t-multilib (v8_2_a_simd_variants): Add fp16fml.
31159 * config/arm/unspecs.md (UNSPEC_VFML_LO, UNSPEC_VFML_HI): New unspecs.
31160 * doc/invoke.texi (ARM Options): Document fp16fml. Update armv8.4-a
31162 * doc/sourcebuild.texi (arm_fp16fml_neon_ok, arm_fp16fml_neon):
31163 Document new effective target and option set.
31165 2018-01-11 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31167 * config/arm/arm-cpus.in (armv8_4): New feature.
31168 (ARMv8_4a): New fgroup.
31169 (armv8.4-a): New arch.
31170 * config/arm/arm-tables.opt: Regenerate.
31171 * config/arm/t-aprofile: Add matching rules for -march=armv8.4-a.
31172 * config/arm/t-arm-elf (all_v8_archs): Add armv8.4-a.
31173 * config/arm/t-multilib (v8_4_a_simd_variants): New variable.
31174 Add matching rules for -march=armv8.4-a and extensions.
31175 * doc/invoke.texi (ARM Options): Document -march=armv8.4-a.
31177 2018-01-11 Oleg Endo <olegendo@gcc.gnu.org>
31180 * config/rx/rx.md (BW): New mode attribute.
31181 (sync_lock_test_and_setsi): Add mode suffix to insn output.
31183 2018-01-11 Richard Biener <rguenther@suse.de>
31185 PR tree-optimization/83435
31186 * graphite.c (canonicalize_loop_form): Ignore fake loop exit edges.
31187 * graphite-scop-detection.c (scop_detection::get_sese): Likewise.
31188 * tree-vrp.c (add_assert_info): Drop TREE_OVERFLOW if they appear.
31190 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
31191 Alan Hayward <alan.hayward@arm.com>
31192 David Sherwood <david.sherwood@arm.com>
31194 * config/aarch64/aarch64.c (aarch64_address_info): Add a const_offset
31196 (aarch64_classify_address): Initialize it. Track polynomial offsets.
31197 (aarch64_print_address_internal): Use it to check for a zero offset.
31199 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
31200 Alan Hayward <alan.hayward@arm.com>
31201 David Sherwood <david.sherwood@arm.com>
31203 * config/aarch64/aarch64-modes.def (NUM_POLY_INT_COEFFS): Set to 2.
31204 * config/aarch64/aarch64-protos.h (aarch64_initial_elimination_offset):
31205 Return a poly_int64 rather than a HOST_WIDE_INT.
31206 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a poly_int64
31207 rather than a HOST_WIDE_INT.
31208 * config/aarch64/aarch64.h (aarch64_frame): Protect with
31209 HAVE_POLY_INT_H rather than HOST_WIDE_INT. Change locals_offset,
31210 hard_fp_offset, frame_size, initial_adjust, callee_offset and
31211 final_offset from HOST_WIDE_INT to poly_int64.
31212 * config/aarch64/aarch64-builtins.c (aarch64_simd_expand_args): Use
31213 to_constant when getting the number of units in an Advanced SIMD
31215 (aarch64_builtin_vectorized_function): Check for a constant number
31217 * config/aarch64/aarch64-simd.md (mov<mode>): Handle polynomial
31219 (aarch64_ld<VSTRUCT:nregs>_lane<VALLDIF:mode>): Use the nunits
31220 attribute instead of GET_MODE_NUNITS.
31221 * config/aarch64/aarch64.c (aarch64_hard_regno_nregs)
31222 (aarch64_class_max_nregs): Use the constant_lowest_bound of the
31223 GET_MODE_SIZE for fixed-size registers.
31224 (aarch64_const_vec_all_same_in_range_p): Use const_vec_duplicate_p.
31225 (aarch64_hard_regno_call_part_clobbered, aarch64_classify_index)
31226 (aarch64_mode_valid_for_sched_fusion_p, aarch64_classify_address)
31227 (aarch64_legitimize_address_displacement, aarch64_secondary_reload)
31228 (aarch64_print_operand, aarch64_print_address_internal)
31229 (aarch64_address_cost, aarch64_rtx_costs, aarch64_register_move_cost)
31230 (aarch64_short_vector_p, aapcs_vfp_sub_candidate)
31231 (aarch64_simd_attr_length_rglist, aarch64_operands_ok_for_ldpstp):
31232 Handle polynomial GET_MODE_SIZE.
31233 (aarch64_hard_regno_caller_save_mode): Likewise. Return modes
31234 wider than SImode without modification.
31235 (tls_symbolic_operand_type): Use strip_offset instead of split_const.
31236 (aarch64_pass_by_reference, aarch64_layout_arg, aarch64_pad_reg_upward)
31237 (aarch64_gimplify_va_arg_expr): Assert that we don't yet handle
31238 passing and returning SVE modes.
31239 (aarch64_function_value, aarch64_layout_arg): Use gen_int_mode
31240 rather than GEN_INT.
31241 (aarch64_emit_probe_stack_range): Take the size as a poly_int64
31242 rather than a HOST_WIDE_INT, but call sorry if it isn't constant.
31243 (aarch64_allocate_and_probe_stack_space): Likewise.
31244 (aarch64_layout_frame): Cope with polynomial offsets.
31245 (aarch64_save_callee_saves, aarch64_restore_callee_saves): Take the
31246 start_offset as a poly_int64 rather than a HOST_WIDE_INT. Track
31247 polynomial offsets.
31248 (offset_9bit_signed_unscaled_p, offset_12bit_unsigned_scaled_p)
31249 (aarch64_offset_7bit_signed_scaled_p): Take the offset as a
31250 poly_int64 rather than a HOST_WIDE_INT.
31251 (aarch64_get_separate_components, aarch64_process_components)
31252 (aarch64_expand_prologue, aarch64_expand_epilogue)
31253 (aarch64_use_return_insn_p): Handle polynomial frame offsets.
31254 (aarch64_anchor_offset): New function, split out from...
31255 (aarch64_legitimize_address): ...here.
31256 (aarch64_builtin_vectorization_cost): Handle polynomial
31257 TYPE_VECTOR_SUBPARTS.
31258 (aarch64_simd_check_vect_par_cnst_half): Handle polynomial
31260 (aarch64_simd_make_constant, aarch64_expand_vector_init): Get the
31261 number of elements from the PARALLEL rather than the mode.
31262 (aarch64_shift_truncation_mask): Use GET_MODE_UNIT_BITSIZE
31263 rather than GET_MODE_BITSIZE.
31264 (aarch64_evpc_trn, aarch64_evpc_uzp, aarch64_evpc_ext)
31265 (aarch64_evpc_rev, aarch64_evpc_dup, aarch64_evpc_zip)
31266 (aarch64_expand_vec_perm_const_1): Handle polynomial
31267 d->perm.length () and d->perm elements.
31268 (aarch64_evpc_tbl): Likewise. Use nelt rather than GET_MODE_NUNITS.
31269 Apply to_constant to d->perm elements.
31270 (aarch64_simd_valid_immediate, aarch64_vec_fpconst_pow_of_2): Handle
31271 polynomial CONST_VECTOR_NUNITS.
31272 (aarch64_move_pointer): Take amount as a poly_int64 rather
31274 (aarch64_progress_pointer): Avoid temporary variable.
31275 * config/aarch64/aarch64.md (aarch64_<crc_variant>): Use
31276 the mode attribute instead of GET_MODE.
31278 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
31279 Alan Hayward <alan.hayward@arm.com>
31280 David Sherwood <david.sherwood@arm.com>
31282 * config/aarch64/aarch64.c (aarch64_force_temporary): Assert that
31283 x exists before using it.
31284 (aarch64_add_constant_internal): Rename to...
31285 (aarch64_add_offset_1): ...this. Replace regnum with separate
31286 src and dest rtxes. Handle the case in which they're different,
31287 including when the offset is zero. Replace scratchreg with an rtx.
31288 Use 2 additions if there is no spare register into which we can
31289 move a 16-bit constant.
31290 (aarch64_add_constant): Delete.
31291 (aarch64_add_offset): Replace reg with separate src and dest
31292 rtxes. Take a poly_int64 offset instead of a HOST_WIDE_INT.
31293 Use aarch64_add_offset_1.
31294 (aarch64_add_sp, aarch64_sub_sp): Take the scratch register as
31295 an rtx rather than an int. Take the delta as a poly_int64
31296 rather than a HOST_WIDE_INT. Use aarch64_add_offset.
31297 (aarch64_expand_mov_immediate): Update uses of aarch64_add_offset.
31298 (aarch64_expand_prologue): Update calls to aarch64_sub_sp,
31299 aarch64_allocate_and_probe_stack_space and aarch64_add_offset.
31300 (aarch64_expand_epilogue): Update calls to aarch64_add_offset
31301 and aarch64_add_sp.
31302 (aarch64_output_mi_thunk): Use aarch64_add_offset rather than
31303 aarch64_add_constant.
31305 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
31307 * config/aarch64/aarch64.c (aarch64_reinterpret_float_as_int):
31308 Use scalar_float_mode.
31310 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
31312 * config/aarch64/aarch64-simd.md
31313 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): Avoid GET_MODE_NUNITS.
31314 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Likewise.
31315 (aarch64_fml<f16mac1>l_lane_lowv2sf): Likewise.
31316 (aarch64_fml<f16mac1>l_lane_highv2sf): Likewise.
31317 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Likewise.
31318 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Likewise.
31319 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Likewise.
31320 (aarch64_fml<f16mac1>l_laneq_highv2sf): Likewise.
31321 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Likewise.
31322 (aarch64_fml<f16mac1>lq_lane_highv4sf): Likewise.
31324 2018-01-11 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31327 * config/arm/arm.c (arm_declare_function_name): Set arch_to_print if
31328 targ_options->x_arm_arch_string is non NULL.
31330 2018-01-11 Tamar Christina <tamar.christina@arm.com>
31332 * config/aarch64/aarch64.h
31333 (AARCH64_FL_FOR_ARCH8_4): Add AARCH64_FL_DOTPROD.
31335 2018-01-11 Sudakshina Das <sudi.das@arm.com>
31338 * expmed.c (emit_store_flag_force): Swap if const op0
31339 and change VOIDmode to mode of op0.
31341 2018-01-11 Richard Sandiford <richard.sandiford@linaro.org>
31343 PR rtl-optimization/83761
31344 * caller-save.c (replace_reg_with_saved_mem): Pass bits rather
31345 than bytes to mode_for_size.
31347 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
31349 PR middle-end/83189
31350 * gfortran.fortran-torture/compile/pr83189.f90: New testcase.
31351 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Handle zero
31354 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
31356 PR middle-end/83575
31357 * cfgrtl.c (rtl_verify_edges): Only verify fixability of partition
31358 when in layout mode.
31359 (cfg_layout_finalize): Do not verify cfg before we are out of layout.
31360 * cfgcleanup.c (try_optimize_cfg): Only verify flow info when doing
31363 2018-01-10 Michael Collison <michael.collison@arm.com>
31365 * config/aarch64/aarch64-modes.def (V2HF): New VECTOR_MODE.
31366 * config/aarch64/aarch64-option-extension.def: Add
31367 AARCH64_OPT_EXTENSION of 'fp16fml'.
31368 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
31369 (__ARM_FEATURE_FP16_FML): Define if TARGET_F16FML is true.
31370 * config/aarch64/predicates.md (aarch64_lane_imm3): New predicate.
31371 * config/aarch64/constraints.md (Ui7): New constraint.
31372 * config/aarch64/iterators.md (VFMLA_W): New mode iterator.
31373 (VFMLA_SEL_W): Ditto.
31376 (VFMLA16_LOW): New int iterator.
31377 (VFMLA16_HIGH): Ditto.
31378 (UNSPEC_FMLAL): New unspec.
31379 (UNSPEC_FMLSL): Ditto.
31380 (UNSPEC_FMLAL2): Ditto.
31381 (UNSPEC_FMLSL2): Ditto.
31382 (f16mac): New code attribute.
31383 * config/aarch64/aarch64-simd-builtins.def
31384 (aarch64_fmlal_lowv2sf): Ditto.
31385 (aarch64_fmlsl_lowv2sf): Ditto.
31386 (aarch64_fmlalq_lowv4sf): Ditto.
31387 (aarch64_fmlslq_lowv4sf): Ditto.
31388 (aarch64_fmlal_highv2sf): Ditto.
31389 (aarch64_fmlsl_highv2sf): Ditto.
31390 (aarch64_fmlalq_highv4sf): Ditto.
31391 (aarch64_fmlslq_highv4sf): Ditto.
31392 (aarch64_fmlal_lane_lowv2sf): Ditto.
31393 (aarch64_fmlsl_lane_lowv2sf): Ditto.
31394 (aarch64_fmlal_laneq_lowv2sf): Ditto.
31395 (aarch64_fmlsl_laneq_lowv2sf): Ditto.
31396 (aarch64_fmlalq_lane_lowv4sf): Ditto.
31397 (aarch64_fmlsl_lane_lowv4sf): Ditto.
31398 (aarch64_fmlalq_laneq_lowv4sf): Ditto.
31399 (aarch64_fmlsl_laneq_lowv4sf): Ditto.
31400 (aarch64_fmlal_lane_highv2sf): Ditto.
31401 (aarch64_fmlsl_lane_highv2sf): Ditto.
31402 (aarch64_fmlal_laneq_highv2sf): Ditto.
31403 (aarch64_fmlsl_laneq_highv2sf): Ditto.
31404 (aarch64_fmlalq_lane_highv4sf): Ditto.
31405 (aarch64_fmlsl_lane_highv4sf): Ditto.
31406 (aarch64_fmlalq_laneq_highv4sf): Ditto.
31407 (aarch64_fmlsl_laneq_highv4sf): Ditto.
31408 * config/aarch64/aarch64-simd.md:
31409 (aarch64_fml<f16mac1>l<f16quad>_low<mode>): New pattern.
31410 (aarch64_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
31411 (aarch64_simd_fml<f16mac1>l<f16quad>_low<mode>): Ditto.
31412 (aarch64_simd_fml<f16mac1>l<f16quad>_high<mode>): Ditto.
31413 (aarch64_fml<f16mac1>l_lane_lowv2sf): Ditto.
31414 (aarch64_fml<f16mac1>l_lane_highv2sf): Ditto.
31415 (aarch64_simd_fml<f16mac>l_lane_lowv2sf): Ditto.
31416 (aarch64_simd_fml<f16mac>l_lane_highv2sf): Ditto.
31417 (aarch64_fml<f16mac1>lq_laneq_lowv4sf): Ditto.
31418 (aarch64_fml<f16mac1>lq_laneq_highv4sf): Ditto.
31419 (aarch64_simd_fml<f16mac>lq_laneq_lowv4sf): Ditto.
31420 (aarch64_simd_fml<f16mac>lq_laneq_highv4sf): Ditto.
31421 (aarch64_fml<f16mac1>l_laneq_lowv2sf): Ditto.
31422 (aarch64_fml<f16mac1>l_laneq_highv2sf): Ditto.
31423 (aarch64_simd_fml<f16mac>l_laneq_lowv2sf): Ditto.
31424 (aarch64_simd_fml<f16mac>l_laneq_highv2sf): Ditto.
31425 (aarch64_fml<f16mac1>lq_lane_lowv4sf): Ditto.
31426 (aarch64_fml<f16mac1>lq_lane_highv4sf): Ditto.
31427 (aarch64_simd_fml<f16mac>lq_lane_lowv4sf): Ditto.
31428 (aarch64_simd_fml<f16mac>lq_lane_highv4sf): Ditto.
31429 * config/aarch64/arm_neon.h (vfmlal_low_u32): New intrinsic.
31430 (vfmlsl_low_u32): Ditto.
31431 (vfmlalq_low_u32): Ditto.
31432 (vfmlslq_low_u32): Ditto.
31433 (vfmlal_high_u32): Ditto.
31434 (vfmlsl_high_u32): Ditto.
31435 (vfmlalq_high_u32): Ditto.
31436 (vfmlslq_high_u32): Ditto.
31437 (vfmlal_lane_low_u32): Ditto.
31438 (vfmlsl_lane_low_u32): Ditto.
31439 (vfmlal_laneq_low_u32): Ditto.
31440 (vfmlsl_laneq_low_u32): Ditto.
31441 (vfmlalq_lane_low_u32): Ditto.
31442 (vfmlslq_lane_low_u32): Ditto.
31443 (vfmlalq_laneq_low_u32): Ditto.
31444 (vfmlslq_laneq_low_u32): Ditto.
31445 (vfmlal_lane_high_u32): Ditto.
31446 (vfmlsl_lane_high_u32): Ditto.
31447 (vfmlal_laneq_high_u32): Ditto.
31448 (vfmlsl_laneq_high_u32): Ditto.
31449 (vfmlalq_lane_high_u32): Ditto.
31450 (vfmlslq_lane_high_u32): Ditto.
31451 (vfmlalq_laneq_high_u32): Ditto.
31452 (vfmlslq_laneq_high_u32): Ditto.
31453 * config/aarch64/aarch64.h (AARCH64_FL_F16SML): New flag.
31454 (AARCH64_FL_FOR_ARCH8_4): New.
31455 (AARCH64_ISA_F16FML): New ISA flag.
31456 (TARGET_F16FML): New feature flag for fp16fml.
31457 (doc/invoke.texi): Document new fp16fml option.
31459 2018-01-10 Michael Collison <michael.collison@arm.com>
31461 * config/aarch64/aarch64-builtins.c:
31462 (aarch64_types_ternopu_imm_qualifiers, TYPES_TERNOPUI): New.
31463 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
31464 (__ARM_FEATURE_SHA3): Define if TARGET_SHA3 is true.
31465 * config/aarch64/aarch64.h (AARCH64_FL_SHA3): New flags.
31466 (AARCH64_ISA_SHA3): New ISA flag.
31467 (TARGET_SHA3): New feature flag for sha3.
31468 * config/aarch64/iterators.md (sha512_op): New int attribute.
31469 (CRYPTO_SHA512): New int iterator.
31470 (UNSPEC_SHA512H): New unspec.
31471 (UNSPEC_SHA512H2): Ditto.
31472 (UNSPEC_SHA512SU0): Ditto.
31473 (UNSPEC_SHA512SU1): Ditto.
31474 * config/aarch64/aarch64-simd-builtins.def
31475 (aarch64_crypto_sha512hqv2di): New builtin.
31476 (aarch64_crypto_sha512h2qv2di): Ditto.
31477 (aarch64_crypto_sha512su0qv2di): Ditto.
31478 (aarch64_crypto_sha512su1qv2di): Ditto.
31479 (aarch64_eor3qv8hi): Ditto.
31480 (aarch64_rax1qv2di): Ditto.
31481 (aarch64_xarqv2di): Ditto.
31482 (aarch64_bcaxqv8hi): Ditto.
31483 * config/aarch64/aarch64-simd.md:
31484 (aarch64_crypto_sha512h<sha512_op>qv2di): New pattern.
31485 (aarch64_crypto_sha512su0qv2di): Ditto.
31486 (aarch64_crypto_sha512su1qv2di): Ditto.
31487 (aarch64_eor3qv8hi): Ditto.
31488 (aarch64_rax1qv2di): Ditto.
31489 (aarch64_xarqv2di): Ditto.
31490 (aarch64_bcaxqv8hi): Ditto.
31491 * config/aarch64/arm_neon.h (vsha512hq_u64): New intrinsic.
31492 (vsha512h2q_u64): Ditto.
31493 (vsha512su0q_u64): Ditto.
31494 (vsha512su1q_u64): Ditto.
31495 (veor3q_u16): Ditto.
31496 (vrax1q_u64): Ditto.
31497 (vxarq_u64): Ditto.
31498 (vbcaxq_u16): Ditto.
31499 * config/arm/types.md (crypto_sha512): New type attribute.
31500 (crypto_sha3): Ditto.
31501 (doc/invoke.texi): Document new sha3 option.
31503 2018-01-10 Michael Collison <michael.collison@arm.com>
31505 * config/aarch64/aarch64-builtins.c:
31506 (aarch64_types_quadopu_imm_qualifiers, TYPES_QUADOPUI): New.
31507 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
31508 (__ARM_FEATURE_SM3): Define if TARGET_SM4 is true.
31509 (__ARM_FEATURE_SM4): Define if TARGET_SM4 is true.
31510 * config/aarch64/aarch64.h (AARCH64_FL_SM4): New flags.
31511 (AARCH64_ISA_SM4): New ISA flag.
31512 (TARGET_SM4): New feature flag for sm4.
31513 * config/aarch64/aarch64-simd-builtins.def
31514 (aarch64_sm3ss1qv4si): Ditto.
31515 (aarch64_sm3tt1aq4si): Ditto.
31516 (aarch64_sm3tt1bq4si): Ditto.
31517 (aarch64_sm3tt2aq4si): Ditto.
31518 (aarch64_sm3tt2bq4si): Ditto.
31519 (aarch64_sm3partw1qv4si): Ditto.
31520 (aarch64_sm3partw2qv4si): Ditto.
31521 (aarch64_sm4eqv4si): Ditto.
31522 (aarch64_sm4ekeyqv4si): Ditto.
31523 * config/aarch64/aarch64-simd.md:
31524 (aarch64_sm3ss1qv4si): Ditto.
31525 (aarch64_sm3tt<sm3tt_op>qv4si): Ditto.
31526 (aarch64_sm3partw<sm3part_op>qv4si): Ditto.
31527 (aarch64_sm4eqv4si): Ditto.
31528 (aarch64_sm4ekeyqv4si): Ditto.
31529 * config/aarch64/iterators.md (sm3tt_op): New int iterator.
31530 (sm3part_op): Ditto.
31531 (CRYPTO_SM3TT): Ditto.
31532 (CRYPTO_SM3PART): Ditto.
31533 (UNSPEC_SM3SS1): New unspec.
31534 (UNSPEC_SM3TT1A): Ditto.
31535 (UNSPEC_SM3TT1B): Ditto.
31536 (UNSPEC_SM3TT2A): Ditto.
31537 (UNSPEC_SM3TT2B): Ditto.
31538 (UNSPEC_SM3PARTW1): Ditto.
31539 (UNSPEC_SM3PARTW2): Ditto.
31540 (UNSPEC_SM4E): Ditto.
31541 (UNSPEC_SM4EKEY): Ditto.
31542 * config/aarch64/constraints.md (Ui2): New constraint.
31543 * config/aarch64/predicates.md (aarch64_imm2): New predicate.
31544 * config/arm/types.md (crypto_sm3): New type attribute.
31545 (crypto_sm4): Ditto.
31546 * config/aarch64/arm_neon.h (vsm3ss1q_u32): New intrinsic.
31547 (vsm3tt1aq_u32): Ditto.
31548 (vsm3tt1bq_u32): Ditto.
31549 (vsm3tt2aq_u32): Ditto.
31550 (vsm3tt2bq_u32): Ditto.
31551 (vsm3partw1q_u32): Ditto.
31552 (vsm3partw2q_u32): Ditto.
31553 (vsm4eq_u32): Ditto.
31554 (vsm4ekeyq_u32): Ditto.
31555 (doc/invoke.texi): Document new sm4 option.
31557 2018-01-10 Michael Collison <michael.collison@arm.com>
31559 * config/aarch64/aarch64-arches.def (armv8.4-a): New architecture.
31560 * config/aarch64/aarch64.h (AARCH64_ISA_V8_4): New ISA flag.
31561 (AARCH64_FL_FOR_ARCH8_4): New.
31562 (AARCH64_FL_V8_4): New flag.
31563 (doc/invoke.texi): Document new armv8.4-a option.
31565 2018-01-10 Michael Collison <michael.collison@arm.com>
31567 * config/aarch64/aarch64-c.c (aarch64_update_cpp_builtins):
31568 (__ARM_FEATURE_AES): Define if TARGET_AES is true.
31569 (__ARM_FEATURE_SHA2): Define if TARGET_SHA2 is true.
31570 * config/aarch64/aarch64-option-extension.def: Add
31571 AARCH64_OPT_EXTENSION of 'sha2'.
31572 (aes): Add AARCH64_OPT_EXTENSION of 'aes'.
31573 (crypto): Disable sha2 and aes if crypto disabled.
31574 (crypto): Enable aes and sha2 if enabled.
31575 (simd): Disable sha2 and aes if simd disabled.
31576 * config/aarch64/aarch64.h (AARCH64_FL_AES, AARCH64_FL_SHA2):
31578 (AARCH64_ISA_AES, AARCH64_ISA_SHA2): New ISA flags.
31579 (TARGET_SHA2): New feature flag for sha2.
31580 (TARGET_AES): New feature flag for aes.
31581 * config/aarch64/aarch64-simd.md:
31582 (aarch64_crypto_aes<aes_op>v16qi): Make pattern
31583 conditional on TARGET_AES.
31584 (aarch64_crypto_aes<aesmc_op>v16qi): Ditto.
31585 (aarch64_crypto_sha1hsi): Make pattern conditional
31587 (aarch64_crypto_sha1hv4si): Ditto.
31588 (aarch64_be_crypto_sha1hv4si): Ditto.
31589 (aarch64_crypto_sha1su1v4si): Ditto.
31590 (aarch64_crypto_sha1<sha1_op>v4si): Ditto.
31591 (aarch64_crypto_sha1su0v4si): Ditto.
31592 (aarch64_crypto_sha256h<sha256_op>v4si): Ditto.
31593 (aarch64_crypto_sha256su0v4si): Ditto.
31594 (aarch64_crypto_sha256su1v4si): Ditto.
31595 (doc/invoke.texi): Document new aes and sha2 options.
31597 2018-01-10 Martin Sebor <msebor@redhat.com>
31599 PR tree-optimization/83781
31600 * gimple-fold.c (get_range_strlen): Avoid treating arrays of pointers
31603 2018-01-11 Martin Sebor <msebor@gmail.com>
31604 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
31606 PR tree-optimization/83501
31607 PR tree-optimization/81703
31609 * tree-ssa-strlen.c (get_string_cst): Rename...
31610 (get_string_len): ...to this. Handle global constants.
31611 (handle_char_store): Adjust.
31613 2018-01-10 Kito Cheng <kito.cheng@gmail.com>
31614 Jim Wilson <jimw@sifive.com>
31616 * config/riscv/riscv-protos.h (riscv_output_return): New.
31617 * config/riscv/riscv.c (struct machine_function): New naked_p field.
31618 (riscv_attribute_table, riscv_output_return),
31619 (riscv_handle_fndecl_attribute, riscv_naked_function_p),
31620 (riscv_allocate_stack_slots_for_args, riscv_warn_func_return): New.
31621 (riscv_compute_frame_info): Only compute frame->mask if not a naked
31623 (riscv_expand_prologue): Add early return for naked function.
31624 (riscv_expand_epilogue): Likewise.
31625 (riscv_function_ok_for_sibcall): Return false for naked function.
31626 (riscv_set_current_function): New.
31627 (TARGET_SET_CURRENT_FUNCTION, TARGET_ALLOCATE_STACK_SLOTS_FOR_ARGS),
31628 (TARGET_ATTRIBUTE_TABLE, TARGET_WARN_FUNC_RETURN): New.
31629 * config/riscv/riscv.md (simple_return): Call riscv_output_return.
31630 * doc/extend.texi (RISC-V Function Attributes): New.
31632 2018-01-10 Michael Meissner <meissner@linux.vnet.ibm.com>
31634 * config/rs6000/rs6000.c (is_complex_IBM_long_double): Explicitly
31635 check for 128-bit long double before checking TCmode.
31636 * config/rs6000/rs6000.h (FLOAT128_IEEE_P): Explicitly check for
31637 128-bit long doubles before checking TFmode or TCmode.
31638 (FLOAT128_IBM_P): Likewise.
31640 2018-01-10 Martin Sebor <msebor@redhat.com>
31642 PR tree-optimization/83671
31643 * builtins.c (c_strlen): Unconditionally return zero for the empty
31645 Use -Warray-bounds for warnings.
31646 * gimple-fold.c (get_range_strlen): Handle non-constant lengths
31647 for non-constant array indices with COMPONENT_REF, arrays of
31648 arrays, and pointers to arrays.
31649 (gimple_fold_builtin_strlen): Determine and set length range for
31650 non-constant character arrays.
31652 2018-01-10 Aldy Hernandez <aldyh@redhat.com>
31654 PR middle-end/81897
31655 * tree-ssa-uninit.c (convert_control_dep_chain_into_preds): Skip
31658 2018-01-10 Eric Botcazou <ebotcazou@adacore.com>
31660 * dwarf2out.c (dwarf2out_var_location): Do not pass NULL to fprintf.
31662 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
31665 * config/rs6000/rs6000.c (print_operand) <'y'>: Use
31666 VECTOR_MEM_ALTIVEC_OR_VSX_P.
31667 * config/rs6000/vsx.md (*vsx_le_perm_load_<mode> for VSX_D): Use
31668 indexed_or_indirect_operand predicate.
31669 (*vsx_le_perm_load_<mode> for VSX_W): Likewise.
31670 (*vsx_le_perm_load_v8hi): Likewise.
31671 (*vsx_le_perm_load_v16qi): Likewise.
31672 (*vsx_le_perm_store_<mode> for VSX_D): Likewise.
31673 (*vsx_le_perm_store_<mode> for VSX_W): Likewise.
31674 (*vsx_le_perm_store_v8hi): Likewise.
31675 (*vsx_le_perm_store_v16qi): Likewise.
31676 (eight unnamed splitters): Likewise.
31678 2018-01-10 Peter Bergner <bergner@vnet.ibm.com>
31680 * config/rs6000/x86intrin.h: Change #warning to #error. Update message.
31681 * config/rs6000/emmintrin.h: Likewise.
31682 * config/rs6000/mmintrin.h: Likewise.
31683 * config/rs6000/xmmintrin.h: Likewise.
31685 2018-01-10 David Malcolm <dmalcolm@redhat.com>
31688 * tree-core.h: Document EXPR_LOCATION_WRAPPER_P's usage of
31690 * tree.c (tree_nop_conversion): Return true for location wrapper
31692 (maybe_wrap_with_location): New function.
31693 (selftest::check_strip_nops): New function.
31694 (selftest::test_location_wrappers): New function.
31695 (selftest::tree_c_tests): Call it.
31696 * tree.h (STRIP_ANY_LOCATION_WRAPPER): New macro.
31697 (maybe_wrap_with_location): New decl.
31698 (EXPR_LOCATION_WRAPPER_P): New macro.
31699 (location_wrapper_p): New inline function.
31700 (tree_strip_any_location_wrapper): New inline function.
31702 2018-01-10 H.J. Lu <hongjiu.lu@intel.com>
31705 * config/i386/i386.c (ix86_compute_frame_layout): Always adjust
31706 stack_realign_offset for the largest alignment of stack slot
31708 (ix86_find_max_used_stack_alignment): New function.
31709 (ix86_finalize_stack_frame_flags): Use it. Set
31710 max_used_stack_alignment if we don't realign stack.
31711 * config/i386/i386.h (machine_function): Add
31712 max_used_stack_alignment.
31714 2018-01-10 Christophe Lyon <christophe.lyon@linaro.org>
31716 * config/arm/arm.opt (-mbranch-cost): New option.
31717 * config/arm/arm.h (BRANCH_COST): Take arm_branch_cost into
31720 2018-01-10 Segher Boessenkool <segher@kernel.crashing.org>
31723 * config/rs6000/rs6000.md (load_toc_v4_PIC_2, load_toc_v4_PIC_3b,
31724 load_toc_v4_PIC_3c): Wrap const term in CONST RTL.
31726 2018-01-10 Richard Biener <rguenther@suse.de>
31729 * dwarf2out.c (gen_subprogram_die): Hoist old_die && declaration
31730 early out so it also covers the case where we have a non-NULL
31733 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
31735 PR tree-optimization/83753
31736 * tree-vect-stmts.c (get_group_load_store_type): Use VMAT_CONTIGUOUS
31737 for non-strided grouped accesses if the number of elements is 1.
31739 2018-01-10 Jan Hubicka <hubicka@ucw.cz>
31742 * i386.c (ix86_vectorize_builtin_gather): Check TARGET_USE_GATHER.
31743 * i386.h (TARGET_USE_GATHER): Define.
31744 * x86-tune.def (X86_TUNE_USE_GATHER): New.
31746 2018-01-10 Martin Liska <mliska@suse.cz>
31749 * basic-block.h (CLEANUP_NO_PARTITIONING): New define.
31750 * bb-reorder.c (pass_reorder_blocks::execute): Do not clean up
31752 * cfgcleanup.c (try_optimize_cfg): Fix up partitioning if
31753 CLEANUP_NO_PARTITIONING is not set.
31755 2018-01-10 Richard Sandiford <richard.sandiford@linaro.org>
31757 * doc/rtl.texi: Remove documentation of (const ...) wrappers
31758 for vectors, as a partial revert of r254296.
31759 * rtl.h (const_vec_p): Delete.
31760 (const_vec_duplicate_p): Don't test for vector CONSTs.
31761 (unwrap_const_vec_duplicate, const_vec_series_p): Likewise.
31762 * expmed.c (make_tree): Likewise.
31765 * common.md (E, F): Use CONSTANT_P instead of checking for
31767 * emit-rtl.c (gen_lowpart_common): Use const_vec_p instead of
31768 checking for CONST_VECTOR.
31770 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
31772 PR middle-end/83575
31773 * predict.c (force_edge_cold): Handle in more sane way edges
31774 with no prediction.
31776 2018-01-09 Carl Love <cel@us.ibm.com>
31778 * config/rs6002/altivec.md (p8_vmrgow): Add support for V2DI, V2DF,
31780 (p8_vmrgew): Add support for V2DI, V2DF, V4SF types.
31781 * config/rs6000/rs6000-builtin.def: Add definitions for FLOAT2_V2DF,
31782 VMRGEW_V2DI, VMRGEW_V2DF, VMRGEW_V4SF, VMRGOW_V4SI, VMRGOW_V4SF,
31783 VMRGOW_V2DI, VMRGOW_V2DF. Remove definition for VMRGOW.
31784 * config/rs6000/rs6000-c.c (VSX_BUILTIN_VEC_FLOAT2,
31785 P8V_BUILTIN_VEC_VMRGEW, P8V_BUILTIN_VEC_VMRGOW): Add definitions.
31786 * config/rs6000/rs6000-protos.h: Add extern defition for
31787 rs6000_generate_float2_double_code.
31788 * config/rs6000/rs6000.c (rs6000_generate_float2_double_code): Add
31790 * config/rs6000/vsx.md (vsx_xvcdpsp): Add define_insn.
31791 (float2_v2df): Add define_expand.
31793 2018-01-09 Uros Bizjak <ubizjak@gmail.com>
31796 * combine.c (force_int_to_mode) <case ASHIFT>: Use mode instead of
31797 op_mode in the force_to_mode call.
31799 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
31801 * config/aarch64/aarch64.c (aarch64_evpc_trn): Use d.perm.series_p
31802 instead of checking each element individually.
31803 (aarch64_evpc_uzp): Likewise.
31804 (aarch64_evpc_zip): Likewise.
31805 (aarch64_evpc_ext): Likewise.
31806 (aarch64_evpc_rev): Likewise.
31807 (aarch64_evpc_dup): Test the encoding for a single duplicated element,
31808 instead of checking each element individually. Return true without
31810 (aarch64_vectorize_vec_perm_const): Use all_from_input_p to test
31811 whether all selected elements come from the same input, instead of
31812 checking each element individually. Remove calls to gen_rtx_REG,
31813 start_sequence and end_sequence and instead assert that no rtl is
31816 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
31818 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p): Fix
31819 order of HIGH and CONST checks.
31821 2018-01-09 Richard Sandiford <richard.sandiford@linaro.org>
31823 * tree-vect-stmts.c (permute_vec_elements): Create a fresh variable
31824 if the destination isn't an SSA_NAME.
31826 2018-01-09 Richard Biener <rguenther@suse.de>
31828 PR tree-optimization/83668
31829 * graphite.c (canonicalize_loop_closed_ssa): Add edge argument,
31831 (canonicalize_loop_form): ... here, renamed from ...
31832 (canonicalize_loop_closed_ssa_form): ... this and amended to
31833 swap successor edges for loop exit blocks to make us use
31834 the RPO order we need for initial schedule generation.
31836 2018-01-09 Joseph Myers <joseph@codesourcery.com>
31838 PR tree-optimization/64811
31839 * match.pd: When optimizing comparisons with Inf, avoid
31840 introducing or losing exceptions from comparisons with NaN.
31842 2018-01-09 Martin Liska <mliska@suse.cz>
31845 * asan.c (shadow_mem_size): Add gcc_assert.
31847 2018-01-09 Georg-Johann Lay <avr@gjlay.de>
31849 Don't save registers in main().
31852 * doc/invoke.texi (AVR Options) [-mmain-is-OS_task]: Document it.
31853 * config/avr/avr.opt (-mmain-is-OS_task): New target option.
31854 * config/avr/avr.c (avr_set_current_function): Don't error if
31855 naked, OS_task or OS_main are specified at the same time.
31856 (avr_function_ok_for_sibcall): Don't disable sibcalls for OS_task,
31858 (avr_insert_attributes) [-mmain-is-OS_task] <main>: Add OS_task
31860 * common/config/avr/avr-common.c (avr_option_optimization_table):
31861 Switch on -mmain-is-OS_task for optimizing compilations.
31863 2018-01-09 Richard Biener <rguenther@suse.de>
31865 PR tree-optimization/83572
31866 * graphite.c: Include cfganal.h.
31867 (graphite_transform_loops): Connect infinite loops to exit
31868 and remove fake edges at the end.
31870 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
31872 * ipa-inline.c (edge_badness): Revert accidental checkin.
31874 2018-01-09 Jan Hubicka <hubicka@ucw.cz>
31877 * ipa-comdats.c (set_comdat_group): Only set comdat group of real
31878 symbols; not inline clones.
31880 2018-01-09 Jakub Jelinek <jakub@redhat.com>
31883 * modulo-sched.c (schedule_reg_moves): Punt if we'd need to move
31884 hard registers. Formatting fixes.
31886 PR preprocessor/83722
31887 * gcc.c (try_generate_repro): Pass
31888 &temp_stderr_files[RETRY_ICE_ATTEMPTS - 1] rather than
31889 &temp_stdout_files[RETRY_ICE_ATTEMPTS - 1] as last argument to
31892 2018-01-08 Monk Chiang <sh.chiang04@gmail.com>
31893 Kito Cheng <kito.cheng@gmail.com>
31895 * config/riscv/riscv.c (machine_function::is_leaf): Remove field.
31896 (riscv_leaf_function_p): Delete.
31897 (riscv_function_ok_for_sibcall): Return false when TARGET_SAVE_RESTORE.
31899 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31901 * config/rs6000/rs6000-string.c (do_load_for_compare_from_addr): New
31903 (do_ifelse): New function.
31904 (do_isel): New function.
31905 (do_sub3): New function.
31906 (do_add3): New function.
31907 (do_load_mask_compare): New function.
31908 (do_overlap_load_compare): New function.
31909 (expand_compare_loop): New function.
31910 (expand_block_compare): Call expand_compare_loop() when appropriate.
31911 * config/rs6000/rs6000.opt (-mblock-compare-inline-limit): Change
31912 option description.
31913 (-mblock-compare-inline-loop-limit): New option.
31915 2018-01-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
31918 * config/rs6000/altivec.md (*altivec_vpermr_<mode>_internal):
31919 Reverse order of second and third operands in first alternative.
31920 * config/rs6000/rs6000.c (rs6000_expand_vector_set): Reverse order
31921 of first and second elements in UNSPEC_VPERMR vector.
31922 (altivec_expand_vec_perm_le): Likewise.
31924 2018-01-08 Jeff Law <law@redhat.com>
31926 PR rtl-optimizatin/81308
31927 * tree-switch-conversion.c (cfg_altered): New file scoped static.
31928 (process_switch): If group_case_labels makes a change, then set
31930 (pass_convert_switch::execute): If a switch is converted, then
31931 set cfg_altered. Return TODO_cfg_cleanup if cfg_altered is true.
31933 PR rtl-optimization/81308
31934 * recog.c (split_all_insns): Conditionally cleanup the CFG after
31937 2018-01-08 Vidya Praveen <vidyapraveen@arm.com>
31939 PR target/83663 - Revert r255946
31940 * config/aarch64/aarch64.c (aarch64_expand_vector_init): Modify code
31941 generation for cases where splatting a value is not useful.
31942 * simplify-rtx.c (simplify_ternary_operation): Simplify vec_merge
31943 across a vec_duplicate and a paradoxical subreg forming a vector
31944 mode to a vec_concat.
31946 2018-01-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
31948 * config/arm/t-aprofile (MULTILIB_MATCHES): Add mapping rules for
31949 -march=armv8.3-a variants.
31950 * config/arm/t-multilib: Likewise.
31951 * config/arm/t-arm-elf: Likewise. Handle dotprod extension.
31953 2018-01-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
31955 * config/rs6000/rs6000.md (cceq_ior_compare): Remove * so I can use it
31957 (cceq_ior_compare_complement): Give it a name so I can use it, and
31958 change boolean_or_operator predicate to boolean_operator so it can
31959 be used to generate a crand.
31960 (eqne): New code iterator.
31961 (bd/bd_neg): New code_attrs.
31962 (<bd>_<mode>): New name for ctr<mode>_internal[12] now combined into
31963 a single define_insn.
31964 (<bd>tf_<mode>): A new insn pattern for the conditional form branch
31965 decrement (bdnzt/bdnzf/bdzt/bdzf).
31966 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Updated
31967 with the new names of the branch decrement patterns, and added the
31968 names of the branch decrement conditional patterns.
31970 2018-01-08 Richard Biener <rguenther@suse.de>
31972 PR tree-optimization/83563
31973 * graphite.c (canonicalize_loop_closed_ssa_form): Reset the SCEV
31976 2018-01-08 Richard Biener <rguenther@suse.de>
31978 PR middle-end/83713
31979 * convert.c (do_narrow): Properly guard TYPE_OVERFLOW_WRAPS checks.
31981 2018-01-08 Richard Biener <rguenther@suse.de>
31983 PR tree-optimization/83685
31984 * tree-ssa-pre.c (create_expression_by_pieces): Do not insert
31985 references to abnormals.
31987 2018-01-08 Richard Biener <rguenther@suse.de>
31990 * dwarf2out.c (output_indirect_strings): Handle empty
31991 skeleton_debug_str_hash.
31992 (dwarf2out_early_finish): Index strings for -gsplit-dwarf.
31994 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
31996 * config/arc/arc.c (TARGET_TRAMPOLINE_ADJUST_ADDRESS): Delete.
31997 (emit_store_direct): Likewise.
31998 (arc_trampoline_adjust_address): Likewise.
31999 (arc_asm_trampoline_template): New function.
32000 (arc_initialize_trampoline): Use asm_trampoline_template.
32001 (TARGET_ASM_TRAMPOLINE_TEMPLATE): Define.
32002 * config/arc/arc.h (TRAMPOLINE_SIZE): Adjust to 16.
32003 * config/arc/arc.md (flush_icache): Delete pattern.
32005 2018-01-08 Claudiu Zissulescu <claziss@synopsys.com>
32007 * config/arc/arc-c.def (__ARC_UNALIGNED__): New define.
32008 * config/arc/arc.h (STRICT_ALIGNMENT): Control this macro using
32011 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
32014 * config/epiphany/epiphany.h (make_pass_mode_switch_use): Guard
32015 by not USED_FOR_TARGET.
32016 (make_pass_resolve_sw_modes): Likewise.
32018 2018-01-08 Sebastian Huber <sebastian.huber@embedded-brains.de>
32020 * config/nios2/nios2.h (nios2_section_threshold): Guard by not
32023 2018-01-08 Richard Biener <rguenther@suse.de>
32025 PR middle-end/83580
32026 * tree-data-ref.c (split_constant_offset): Remove STRIP_NOPS.
32028 2018-01-08 Richard Biener <rguenther@suse.de>
32030 PR middle-end/83517
32031 * match.pd ((t * 2) / 2) -> t): Add missing :c.
32033 2018-01-06 Aldy Hernandez <aldyh@redhat.com>
32035 PR middle-end/81897
32036 * tree-ssa-uninit.c (compute_control_dep_chain): Do not bail on
32037 basic blocks with a small number of successors.
32038 (convert_control_dep_chain_into_preds): Improve handling of
32040 (dump_predicates): Split apart into...
32041 (dump_pred_chain): ...here...
32042 (dump_pred_info): ...and here.
32043 (can_one_predicate_be_invalidated_p): Add debugging printfs.
32044 (can_chain_union_be_invalidated_p): Improve check for invalidation
32046 (uninit_uses_cannot_happen): Avoid unnecessary if
32047 convert_control_dep_chain_into_preds yielded nothing.
32049 2018-01-06 Martin Sebor <msebor@redhat.com>
32051 PR tree-optimization/83640
32052 * gimple-ssa-warn-restrict.c (builtin_access::builtin_access): Avoid
32053 subtracting negative offset from size.
32054 (builtin_access::overlap): Adjust offset bounds of the access to fall
32055 within the size of the object if possible.
32057 2018-01-06 Richard Sandiford <richard.sandiford@linaro.org>
32059 PR rtl-optimization/83699
32060 * expmed.c (extract_bit_field_1): Restrict the vector usage of
32061 extract_bit_field_as_subreg to cases in which the extracted
32062 value is also a vector.
32064 * lra-constraints.c (process_alt_operands): Test for the equivalence
32065 substitutions when detecting a possible reload cycle.
32067 2018-01-06 Jakub Jelinek <jakub@redhat.com>
32070 * toplev.c (process_options): Don't enable debug_nonbind_markers_p
32071 by default if flag_selective_schedling{,2}. Formatting fixes.
32073 PR rtl-optimization/83682
32074 * rtl.h (const_vec_duplicate_p): Only return true for VEC_DUPLICATE
32075 if it has non-VECTOR_MODE element mode.
32076 (vec_duplicate_p): Likewise.
32078 PR middle-end/83694
32079 * cfgexpand.c (expand_debug_expr): Punt if mode1 is VOIDmode
32080 and bitsize might be greater than MAX_BITSIZE_MODE_ANY_INT.
32082 2018-01-05 Jakub Jelinek <jakub@redhat.com>
32085 * config/i386/i386-builtin.def
32086 (__builtin_ia32_vgf2p8affineinvqb_v64qi,
32087 __builtin_ia32_vgf2p8affineqb_v64qi, __builtin_ia32_vgf2p8mulb_v64qi):
32088 Require also OPTION_MASK_ISA_AVX512F in addition to
32089 OPTION_MASK_ISA_GFNI.
32090 (__builtin_ia32_vgf2p8affineinvqb_v16qi_mask,
32091 __builtin_ia32_vgf2p8affineqb_v16qi_mask): Require
32092 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_SSE in addition
32093 to OPTION_MASK_ISA_GFNI.
32094 (__builtin_ia32_vgf2p8mulb_v32qi_mask): Require
32095 OPTION_MASK_ISA_AVX512VL in addition to OPTION_MASK_ISA_GFNI and
32096 OPTION_MASK_ISA_AVX512BW.
32097 (__builtin_ia32_vgf2p8mulb_v16qi_mask): Require
32098 OPTION_MASK_ISA_AVX512VL instead of OPTION_MASK_ISA_AVX512BW in
32099 addition to OPTION_MASK_ISA_GFNI.
32100 (__builtin_ia32_vgf2p8affineinvqb_v16qi,
32101 __builtin_ia32_vgf2p8affineqb_v16qi, __builtin_ia32_vgf2p8mulb_v16qi):
32102 Require OPTION_MASK_ISA_SSE2 instead of OPTION_MASK_ISA_SSE in addition
32103 to OPTION_MASK_ISA_GFNI.
32104 * config/i386/i386.c (def_builtin): Change to builtin isa/isa2 being
32105 a requirement for all ISAs rather than any of them with a few
32107 (ix86_add_new_builtins): Clear OPTION_MASK_ISA_64BIT from isa before
32109 (ix86_expand_builtin): Require all ISAs from builtin's isa and isa2
32110 bitmasks to be enabled with 3 exceptions, instead of requiring any
32111 enabled ISA with lots of exceptions.
32112 * config/i386/sse.md (vgf2p8affineinvqb_<mode><mask_name>,
32113 vgf2p8affineqb_<mode><mask_name>, vgf2p8mulb_<mode><mask_name>):
32114 Change avx512bw in isa attribute to avx512f.
32115 * config/i386/sgxintrin.h: Add license boilerplate.
32116 * config/i386/vaesintrin.h: Likewise. Fix macro spelling __AVX512F
32117 to __AVX512F__ and __AVX512VL to __AVX512VL__.
32118 (_mm256_aesdec_epi128, _mm256_aesdeclast_epi128, _mm256_aesenc_epi128,
32119 _mm256_aesenclast_epi128): Enable temporarily avx if __AVX__ is not
32121 * config/i386/gfniintrin.h (_mm_gf2p8mul_epi8,
32122 _mm_gf2p8affineinv_epi64_epi8, _mm_gf2p8affine_epi64_epi8): Enable
32123 temporarily sse2 rather than sse if not enabled already.
32126 * config/i386/sse.md (VI248_VLBW): Rename to ...
32127 (VI248_AVX512VL): ... this. Don't guard V32HI with TARGET_AVX512BW.
32128 (vpshrd_<mode><mask_name>, vpshld_<mode><mask_name>,
32129 vpshrdv_<mode>, vpshrdv_<mode>_mask, vpshrdv_<mode>_maskz,
32130 vpshrdv_<mode>_maskz_1, vpshldv_<mode>, vpshldv_<mode>_mask,
32131 vpshldv_<mode>_maskz, vpshldv_<mode>_maskz_1): Use VI248_AVX512VL
32132 mode iterator instead of VI248_VLBW.
32134 2018-01-05 Jan Hubicka <hubicka@ucw.cz>
32136 * ipa-fnsummary.c (record_modified_bb_info): Add OP.
32137 (record_modified): Skip clobbers; add debug output.
32138 (param_change_prob): Use sreal frequencies.
32140 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
32142 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
32143 punt for user-aligned variables.
32145 2018-01-05 Richard Sandiford <richard.sandiford@linaro.org>
32147 * tree-chrec.c (chrec_contains_symbols): Return true for
32150 2018-01-05 Sudakshina Das <sudi.das@arm.com>
32153 * simplify-rtx.c (simplify_relational_operation_1): Add simplifications
32154 of (x|y) == x for BICS pattern.
32156 2018-01-05 Jakub Jelinek <jakub@redhat.com>
32158 PR tree-optimization/83605
32159 * gimple-ssa-strength-reduction.c: Include tree-eh.h.
32160 (find_candidates_dom_walker::before_dom_children): Ignore stmts that
32163 2018-01-05 Sebastian Huber <sebastian.huber@embedded-brains.de>
32165 * config.gcc (epiphany-*-elf*): Add (epiphany-*-rtems*) configuration.
32166 * config/epiphany/rtems.h: New file.
32168 2018-01-04 Jakub Jelinek <jakub@redhat.com>
32169 Uros Bizjak <ubizjak@gmail.com>
32172 * config/i386/i386.md (*<rotate_insn>hi3_1 splitter): Use
32173 QIreg_operand instead of register_operand predicate.
32174 * config/i386/i386.c (ix86_rop_should_change_byte_p,
32175 set_rop_modrm_reg_bits, ix86_mitigate_rop): Use -mmitigate-rop in
32176 comments instead of -fmitigate[-_]rop.
32178 2018-01-04 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
32181 * cgraphunit.c (symbol_table::compile): Switch to text_section
32182 before calling assembly_start debug hook.
32183 * run-rtl-passes.c (run_rtl_passes): Likewise.
32186 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
32188 * tree-vrp.c (extract_range_from_binary_expr_1): Check
32189 range_int_cst_p rather than !symbolic_range_p before calling
32190 extract_range_from_multiplicative_op_1.
32192 2018-01-04 Jeff Law <law@redhat.com>
32194 * tree-ssa-math-opts.c (execute_cse_reciprocals_1): Remove
32195 redundant test in assertion.
32197 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
32199 * doc/rtl.texi: Document machine_mode wrapper classes.
32201 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
32203 * fold-const.c (fold_ternary_loc): Check tree_fits_uhwi_p before
32204 using tree_to_uhwi.
32206 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
32208 * tree-ssa-forwprop.c (is_combined_permutation_identity): Allow
32209 the VEC_PERM_EXPR fold to fail.
32211 2018-01-04 Jakub Jelinek <jakub@redhat.com>
32214 * bb-reorder.c (insert_section_boundary_note): Set has_bb_partition
32215 to switched_sections.
32217 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
32220 * config/arm/arm.c (arm_vectorize_vec_perm_const): Fix inverted
32221 test for d.testing.
32223 2018-01-04 Peter Bergner <bergner@vnet.ibm.com>
32226 * config/rs6000/rs6000.c (rs6000_discover_homogeneous_aggregate): Do not
32227 allow arguments in FP registers if TARGET_HARD_FLOAT is false.
32229 2018-01-04 Jakub Jelinek <jakub@redhat.com>
32232 * cfgexpand.c (expand_debug_expr) <case BIT_FIELD_REF>: Punt if mode
32233 is BLKmode and bitpos not zero or mode change is needed.
32235 2018-01-04 Richard Sandiford <richard.sandiford@linaro.org>
32238 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): Require
32241 2018-01-04 Uros Bizjak <ubizjak@gmail.com>
32244 * config/alpha/alpha.md (*sadd<modesuffix>): Use ASHIFT
32245 instead of MULT rtx. Update all corresponding splitters.
32246 (*saddl_se): Ditto.
32247 (*ssub<modesuffix>): Ditto.
32248 (*ssubl_se): Ditto.
32249 (*cmp_sadd_di): Update split patterns.
32250 (*cmp_sadd_si): Ditto.
32251 (*cmp_sadd_sidi): Ditto.
32252 (*cmp_ssub_di): Ditto.
32253 (*cmp_ssub_si): Ditto.
32254 (*cmp_ssub_sidi): Ditto.
32255 * config/alpha/predicates.md (const23_operand): New predicate.
32256 * config/alpha/alpha.c (alpha_rtx_costs) [PLUS, MINUS]:
32257 Look for ASHIFT, not MULT inner operand.
32258 (alpha_split_conditional_move): Update for *sadd<modesuffix> change.
32260 2018-01-04 Martin Liska <mliska@suse.cz>
32262 PR gcov-profile/83669
32263 * gcov.c (output_intermediate_file): Add version to intermediate
32265 * doc/gcov.texi: Document new field 'version' in intermediate
32266 file format. Fix location of '-k' option of gcov command.
32268 2018-01-04 Martin Liska <mliska@suse.cz>
32271 * ipa-icf.c (sem_function::merge): Do not cross comdat boundary.
32273 2018-01-04 Jakub Jelinek <jakub@redhat.com>
32275 * gimple-ssa-sprintf.c (parse_directive): Cast second dir.len to uhwi.
32277 2018-01-03 Martin Sebor <msebor@redhat.com>
32279 PR tree-optimization/83655
32280 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call): Avoid
32281 checking calls with invalid arguments.
32283 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32285 * tree-vect-stmts.c (vect_get_store_rhs): New function.
32286 (vectorizable_mask_load_store): Delete.
32287 (vectorizable_call): Return false for masked loads and stores.
32288 (vectorizable_store): Handle IFN_MASK_STORE. Use vect_get_store_rhs
32289 instead of gimple_assign_rhs1.
32290 (vectorizable_load): Handle IFN_MASK_LOAD.
32291 (vect_transform_stmt): Don't set is_store for call_vec_info_type.
32293 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32295 * tree-vect-stmts.c (vect_build_gather_load_calls): New function,
32297 (vectorizable_mask_load_store): ...here.
32298 (vectorizable_load): ...and here.
32300 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32302 * tree-vect-stmts.c (vect_build_all_ones_mask)
32303 (vect_build_zero_merge_argument): New functions, split out from...
32304 (vectorizable_load): ...here.
32306 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32308 * tree-vect-stmts.c (vect_check_store_rhs): New function,
32310 (vectorizable_mask_load_store): ...here.
32311 (vectorizable_store): ...and here.
32313 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32315 * tree-vect-stmts.c (vect_check_load_store_mask): New function,
32317 (vectorizable_mask_load_store): ...here.
32319 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32321 * tree-vectorizer.h (vec_load_store_type): Moved from tree-vec-stmts.c
32322 (vect_model_store_cost): Take a vec_load_store_type instead of a
32324 * tree-vect-stmts.c (vec_load_store_type): Move to tree-vectorizer.h.
32325 (vect_model_store_cost): Take a vec_load_store_type instead of a
32327 (vectorizable_mask_load_store): Update accordingly.
32328 (vectorizable_store): Likewise.
32329 * tree-vect-slp.c (vect_analyze_slp_cost_1): Update accordingly.
32331 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32333 * tree-vect-loop.c (vect_transform_loop): Stub out scalar
32334 IFN_MASK_LOAD calls here rather than...
32335 * tree-vect-stmts.c (vectorizable_mask_load_store): ...here.
32337 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32338 Alan Hayward <alan.hayward@arm.com>
32339 David Sherwood <david.sherwood@arm.com>
32341 * expmed.c (extract_bit_field_1): For vector extracts,
32342 fall back to extract_bit_field_as_subreg if vec_extract
32345 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32346 Alan Hayward <alan.hayward@arm.com>
32347 David Sherwood <david.sherwood@arm.com>
32349 * lra-spills.c (pseudo_reg_slot_compare): Sort slots by whether
32350 they are variable or constant sized.
32351 (assign_stack_slot_num_and_sort_pseudos): Don't reuse variable-sized
32352 slots for constant-sized data.
32354 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32355 Alan Hayward <alan.hayward@arm.com>
32356 David Sherwood <david.sherwood@arm.com>
32358 * tree-vect-patterns.c (vect_recog_mask_conversion_pattern): When
32359 handling COND_EXPRs with boolean comparisons, try to find a better
32360 basis for the mask type than the boolean itself.
32362 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32364 * doc/rtl.texi (MAX_BITSIZE_MODE_ANY_MODE): Describe how the default
32365 is calculated and how it can be overridden.
32366 * genmodes.c (max_bitsize_mode_any_mode): New variable.
32367 (create_modes): Initialize it from MAX_BITSIZE_MODE_ANY_MODE,
32369 (emit_max_int): Use it to set the output MAX_BITSIZE_MODE_ANY_MODE,
32372 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32373 Alan Hayward <alan.hayward@arm.com>
32374 David Sherwood <david.sherwood@arm.com>
32376 * config/aarch64/aarch64-protos.h (aarch64_output_simd_mov_immediate):
32377 Remove the mode argument.
32378 (aarch64_simd_valid_immediate): Remove the mode and inverse
32380 * config/aarch64/iterators.md (bitsize): New iterator.
32381 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>, and<mode>3)
32382 (ior<mode>3): Update calls to aarch64_output_simd_mov_immediate.
32383 * config/aarch64/constraints.md (Do, Db, Dn): Update calls to
32384 aarch64_simd_valid_immediate.
32385 * config/aarch64/predicates.md (aarch64_reg_or_orr_imm): Likewise.
32386 (aarch64_reg_or_bic_imm): Likewise.
32387 * config/aarch64/aarch64.c (simd_immediate_info): Replace mvn
32388 with an insn_type enum and msl with a modifier_type enum.
32389 Replace element_width with a scalar_mode. Change the shift
32390 to unsigned int. Add constructors for scalar_float_mode and
32391 scalar_int_mode elements.
32392 (aarch64_vect_float_const_representable_p): Delete.
32393 (aarch64_can_const_movi_rtx_p)
32394 (aarch64_simd_scalar_immediate_valid_for_move)
32395 (aarch64_simd_make_constant): Update call to
32396 aarch64_simd_valid_immediate.
32397 (aarch64_advsimd_valid_immediate_hs): New function.
32398 (aarch64_advsimd_valid_immediate): Likewise.
32399 (aarch64_simd_valid_immediate): Remove mode and inverse
32400 arguments. Rewrite to use the above. Use const_vec_duplicate_p
32401 to detect duplicated constants and use aarch64_float_const_zero_rtx_p
32402 and aarch64_float_const_representable_p on the result.
32403 (aarch64_output_simd_mov_immediate): Remove mode argument.
32404 Update call to aarch64_simd_valid_immediate and use of
32405 simd_immediate_info.
32406 (aarch64_output_scalar_simd_mov_immediate): Update call
32409 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32410 Alan Hayward <alan.hayward@arm.com>
32411 David Sherwood <david.sherwood@arm.com>
32413 * machmode.h (mode_precision): Prefix with CONST_MODE_PRECISION.
32414 (mode_nunits): Likewise CONST_MODE_NUNITS.
32415 * machmode.def (ADJUST_NUNITS): Document.
32416 * genmodes.c (mode_data::need_nunits_adj): New field.
32417 (blank_mode): Update accordingly.
32418 (adj_nunits): New variable.
32419 (print_maybe_const_decl): Replace CATEGORY with a NEEDS_ADJ
32421 (emit_mode_size_inline): Set need_bytesize_adj for all modes
32422 listed in adj_nunits.
32423 (emit_mode_nunits_inline): Set need_nunits_adj for all modes
32424 listed in adj_nunits. Don't emit case statements for such modes.
32425 (emit_insn_modes_h): Emit definitions of CONST_MODE_NUNITS
32426 and CONST_MODE_PRECISION. Make CONST_MODE_SIZE expand to
32427 nothing if adj_nunits is nonnull.
32428 (emit_mode_precision, emit_mode_nunits): Use print_maybe_const_decl.
32429 (emit_mode_unit_size, emit_mode_base_align, emit_mode_ibit)
32430 (emit_mode_fbit): Update use of print_maybe_const_decl.
32431 (emit_move_size): Likewise. Treat the array as non-const
32433 (emit_mode_adjustments): Handle adj_nunits.
32435 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32437 * machmode.def (VECTOR_MODES_WITH_PREFIX): Document.
32438 * genmodes.c (VECTOR_MODES_WITH_PREFIX): New macro.
32439 (VECTOR_MODES): Use it.
32440 (make_vector_modes): Take the prefix as an argument.
32442 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32443 Alan Hayward <alan.hayward@arm.com>
32444 David Sherwood <david.sherwood@arm.com>
32446 * mode-classes.def (MODE_VECTOR_BOOL): New mode class.
32447 * machmode.h (INTEGRAL_MODE_P, VECTOR_MODE_P): Return true
32448 for MODE_VECTOR_BOOL.
32449 * machmode.def (VECTOR_BOOL_MODE): Document.
32450 * genmodes.c (VECTOR_BOOL_MODE): New macro.
32451 (make_vector_bool_mode): New function.
32452 (complete_mode, emit_mode_wider, emit_mode_adjustments): Handle
32454 * lto-streamer-in.c (lto_input_mode_table): Likewise.
32455 * rtx-vector-builder.c (rtx_vector_builder::find_cached_value):
32457 * stor-layout.c (int_mode_for_mode): Likewise.
32458 * tree.c (build_vector_type_for_mode): Likewise.
32459 * varasm.c (output_constant_pool_2): Likewise.
32460 * emit-rtl.c (init_emit_once): Make sure that CONST1_RTX (BImode) and
32461 CONSTM1_RTX (BImode) are the same thing. Initialize const_tiny_rtx
32462 for MODE_VECTOR_BOOL.
32463 * expr.c (expand_expr_real_1): Use VECTOR_MODE_P instead of a list
32464 of mode class checks.
32465 * tree-vect-generic.c (expand_vector_operation): Use VECTOR_MODE_P
32466 instead of a list of mode class checks.
32467 (expand_vector_scalar_condition): Likewise.
32468 (type_for_widest_vector_mode): Handle BImode as an inner mode.
32470 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32471 Alan Hayward <alan.hayward@arm.com>
32472 David Sherwood <david.sherwood@arm.com>
32474 * machmode.h (mode_size): Change from unsigned short to
32476 (mode_to_bytes): Return a poly_uint16 rather than an unsigned short.
32477 (GET_MODE_SIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
32478 or if measurement_type is not polynomial.
32479 (fixed_size_mode::includes_p): Check for constant-sized modes.
32480 * genmodes.c (emit_mode_size_inline): Make mode_size_inline
32481 return a poly_uint16 rather than an unsigned short.
32482 (emit_mode_size): Change the type of mode_size from unsigned short
32483 to poly_uint16_pod. Use ZERO_COEFFS for the initializer.
32484 (emit_mode_adjustments): Cope with polynomial vector sizes.
32485 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
32487 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
32489 * auto-inc-dec.c (try_merge): Treat GET_MODE_SIZE as polynomial.
32490 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Likewise.
32491 * caller-save.c (setup_save_areas): Likewise.
32492 (replace_reg_with_saved_mem): Likewise.
32493 * calls.c (emit_library_call_value_1): Likewise.
32494 * combine-stack-adj.c (combine_stack_adjustments_for_block): Likewise.
32495 * combine.c (simplify_set, make_extraction, simplify_shift_const_1)
32496 (gen_lowpart_for_combine): Likewise.
32497 * convert.c (convert_to_integer_1): Likewise.
32498 * cse.c (equiv_constant, cse_insn): Likewise.
32499 * cselib.c (autoinc_split, cselib_hash_rtx): Likewise.
32500 (cselib_subst_to_values): Likewise.
32501 * dce.c (word_dce_process_block): Likewise.
32502 * df-problems.c (df_word_lr_mark_ref): Likewise.
32503 * dwarf2cfi.c (init_one_dwarf_reg_size): Likewise.
32504 * dwarf2out.c (multiple_reg_loc_descriptor, mem_loc_descriptor)
32505 (concat_loc_descriptor, concatn_loc_descriptor, loc_descriptor)
32506 (rtl_for_decl_location): Likewise.
32507 * emit-rtl.c (gen_highpart, widen_memory_access): Likewise.
32508 * expmed.c (extract_bit_field_1, extract_integral_bit_field): Likewise.
32509 * expr.c (emit_group_load_1, clear_storage_hints): Likewise.
32510 (emit_move_complex, emit_move_multi_word, emit_push_insn): Likewise.
32511 (expand_expr_real_1): Likewise.
32512 * function.c (assign_parm_setup_block_p, assign_parm_setup_block)
32513 (pad_below): Likewise.
32514 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
32515 * gimple-ssa-store-merging.c (rhs_valid_for_store_merging_p): Likewise.
32516 * ira.c (get_subreg_tracking_sizes): Likewise.
32517 * ira-build.c (ira_create_allocno_objects): Likewise.
32518 * ira-color.c (coalesced_pseudo_reg_slot_compare): Likewise.
32519 (ira_sort_regnos_for_alter_reg): Likewise.
32520 * ira-costs.c (record_operand_costs): Likewise.
32521 * lower-subreg.c (interesting_mode_p, simplify_gen_subreg_concatn)
32522 (resolve_simple_move): Likewise.
32523 * lra-constraints.c (get_reload_reg, operands_match_p): Likewise.
32524 (process_addr_reg, simplify_operand_subreg, curr_insn_transform)
32525 (lra_constraints): Likewise.
32526 (CONST_POOL_OK_P): Reject variable-sized modes.
32527 * lra-spills.c (slot, assign_mem_slot, pseudo_reg_slot_compare)
32528 (add_pseudo_to_slot, lra_spill): Likewise.
32529 * omp-low.c (omp_clause_aligned_alignment): Likewise.
32530 * optabs-query.c (get_best_extraction_insn): Likewise.
32531 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
32532 * optabs.c (expand_vec_perm_var, expand_vec_cond_expr): Likewise.
32533 (expand_mult_highpart, valid_multiword_target_p): Likewise.
32534 * recog.c (offsettable_address_addr_space_p): Likewise.
32535 * regcprop.c (maybe_mode_change): Likewise.
32536 * reginfo.c (choose_hard_reg_mode, record_subregs_of_mode): Likewise.
32537 * regrename.c (build_def_use): Likewise.
32538 * regstat.c (dump_reg_info): Likewise.
32539 * reload.c (complex_word_subreg_p, push_reload, find_dummy_reload)
32540 (find_reloads, find_reloads_subreg_address): Likewise.
32541 * reload1.c (eliminate_regs_1): Likewise.
32542 * rtlanal.c (for_each_inc_dec_find_inc_dec, rtx_cost): Likewise.
32543 * simplify-rtx.c (avoid_constant_pool_reference): Likewise.
32544 (simplify_binary_operation_1, simplify_subreg): Likewise.
32545 * targhooks.c (default_function_arg_padding): Likewise.
32546 (default_hard_regno_nregs, default_class_max_nregs): Likewise.
32547 * tree-cfg.c (verify_gimple_assign_binary): Likewise.
32548 (verify_gimple_assign_ternary): Likewise.
32549 * tree-inline.c (estimate_move_cost): Likewise.
32550 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
32551 * tree-ssa-loop-ivopts.c (add_autoinc_candidates): Likewise.
32552 (get_address_cost_ainc): Likewise.
32553 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
32554 (vect_supportable_dr_alignment): Likewise.
32555 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32556 (vectorizable_reduction): Likewise.
32557 * tree-vect-stmts.c (vectorizable_assignment, vectorizable_shift)
32558 (vectorizable_operation, vectorizable_load): Likewise.
32559 * tree.c (build_same_sized_truth_vector_type): Likewise.
32560 * valtrack.c (cleanup_auto_inc_dec): Likewise.
32561 * var-tracking.c (emit_note_insn_var_location): Likewise.
32562 * config/arc/arc.h (ASM_OUTPUT_CASE_END): Use as_a <scalar_int_mode>.
32563 (ADDR_VEC_ALIGN): Likewise.
32565 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32566 Alan Hayward <alan.hayward@arm.com>
32567 David Sherwood <david.sherwood@arm.com>
32569 * machmode.h (mode_to_bits): Return a poly_uint16 rather than an
32571 (GET_MODE_BITSIZE): Return a constant if ONLY_FIXED_SIZE_MODES,
32572 or if measurement_type is polynomial.
32573 * calls.c (shift_return_value): Treat GET_MODE_BITSIZE as polynomial.
32574 * combine.c (make_extraction): Likewise.
32575 * dse.c (find_shift_sequence): Likewise.
32576 * dwarf2out.c (mem_loc_descriptor): Likewise.
32577 * expmed.c (store_integral_bit_field, extract_bit_field_1): Likewise.
32578 (extract_bit_field, extract_low_bits): Likewise.
32579 * expr.c (convert_move, convert_modes, emit_move_insn_1): Likewise.
32580 (optimize_bitfield_assignment_op, expand_assignment): Likewise.
32581 (store_expr_with_bounds, store_field, expand_expr_real_1): Likewise.
32582 * fold-const.c (optimize_bit_field_compare, merge_ranges): Likewise.
32583 * gimple-fold.c (optimize_atomic_compare_exchange_p): Likewise.
32584 * reload.c (find_reloads): Likewise.
32585 * reload1.c (alter_reg): Likewise.
32586 * stor-layout.c (bitwise_mode_for_mode, compute_record_mode): Likewise.
32587 * targhooks.c (default_secondary_memory_needed_mode): Likewise.
32588 * tree-if-conv.c (predicate_mem_writes): Likewise.
32589 * tree-ssa-strlen.c (handle_builtin_memcmp): Likewise.
32590 * tree-vect-patterns.c (adjust_bool_pattern): Likewise.
32591 * tree-vect-stmts.c (vectorizable_simd_clone_call): Likewise.
32592 * valtrack.c (dead_debug_insert_temp): Likewise.
32593 * varasm.c (mergeable_constant_section): Likewise.
32594 * config/sh/sh.h (LOCAL_ALIGNMENT): Use as_a <fixed_size_mode>.
32596 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32597 Alan Hayward <alan.hayward@arm.com>
32598 David Sherwood <david.sherwood@arm.com>
32600 * expr.c (expand_assignment): Cope with polynomial mode sizes
32601 when assigning to a CONCAT.
32603 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32604 Alan Hayward <alan.hayward@arm.com>
32605 David Sherwood <david.sherwood@arm.com>
32607 * machmode.h (mode_precision): Change from unsigned short to
32609 (mode_to_precision): Return a poly_uint16 rather than an unsigned
32611 (GET_MODE_PRECISION): Return a constant if ONLY_FIXED_SIZE_MODES,
32612 or if measurement_type is not polynomial.
32613 (HWI_COMPUTABLE_MODE_P): Turn into a function. Optimize the case
32614 in which the mode is already known to be a scalar_int_mode.
32615 * genmodes.c (emit_mode_precision): Change the type of mode_precision
32616 from unsigned short to poly_uint16_pod. Use ZERO_COEFFS for the
32618 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
32619 for GET_MODE_PRECISION.
32620 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
32621 for GET_MODE_PRECISION.
32622 * combine.c (update_rsp_from_reg_equal): Treat GET_MODE_PRECISION
32624 (try_combine, find_split_point, combine_simplify_rtx): Likewise.
32625 (expand_field_assignment, make_extraction): Likewise.
32626 (make_compound_operation_int, record_dead_and_set_regs_1): Likewise.
32627 (get_last_value): Likewise.
32628 * convert.c (convert_to_integer_1): Likewise.
32629 * cse.c (cse_insn): Likewise.
32630 * expr.c (expand_expr_real_1): Likewise.
32631 * lra-constraints.c (simplify_operand_subreg): Likewise.
32632 * optabs-query.c (can_atomic_load_p): Likewise.
32633 * optabs.c (expand_atomic_load): Likewise.
32634 (expand_atomic_store): Likewise.
32635 * ree.c (combine_reaching_defs): Likewise.
32636 * rtl.h (partial_subreg_p, paradoxical_subreg_p): Likewise.
32637 * rtlanal.c (nonzero_bits1, lsb_bitfield_op_p): Likewise.
32638 * tree.h (type_has_mode_precision_p): Likewise.
32639 * ubsan.c (instrument_si_overflow): Likewise.
32641 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32642 Alan Hayward <alan.hayward@arm.com>
32643 David Sherwood <david.sherwood@arm.com>
32645 * tree.h (TYPE_VECTOR_SUBPARTS): Turn into a function and handle
32646 polynomial numbers of units.
32647 (SET_TYPE_VECTOR_SUBPARTS): Likewise.
32648 (valid_vector_subparts_p): New function.
32649 (build_vector_type): Remove temporary shim and take the number
32650 of units as a poly_uint64 rather than an int.
32651 (build_opaque_vector_type): Take the number of units as a
32652 poly_uint64 rather than an int.
32653 * tree.c (build_vector_from_ctor): Handle polynomial
32654 TYPE_VECTOR_SUBPARTS.
32655 (type_hash_canon_hash, type_cache_hasher::equal): Likewise.
32656 (uniform_vector_p, vector_type_mode, build_vector): Likewise.
32657 (build_vector_from_val): If the number of units is variable,
32658 use build_vec_duplicate_cst for constant operands and
32659 VEC_DUPLICATE_EXPR otherwise.
32660 (make_vector_type): Remove temporary is_constant ().
32661 (build_vector_type, build_opaque_vector_type): Take the number of
32662 units as a poly_uint64 rather than an int.
32663 (check_vector_cst): Handle polynomial TYPE_VECTOR_SUBPARTS and
32665 * cfgexpand.c (expand_debug_expr): Likewise.
32666 * expr.c (count_type_elements, categorize_ctor_elements_1): Likewise.
32667 (store_constructor, expand_expr_real_1): Likewise.
32668 (const_scalar_mask_from_tree): Likewise.
32669 * fold-const-call.c (fold_const_reduction): Likewise.
32670 * fold-const.c (const_binop, const_unop, fold_convert_const): Likewise.
32671 (operand_equal_p, fold_vec_perm, fold_ternary_loc): Likewise.
32672 (native_encode_vector, vec_cst_ctor_to_array): Likewise.
32673 (fold_relational_const): Likewise.
32674 (native_interpret_vector): Likewise. Change the size from an
32675 int to an unsigned int.
32676 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Handle polynomial
32677 TYPE_VECTOR_SUBPARTS.
32678 (gimple_fold_indirect_ref, gimple_build_vector): Likewise.
32679 (gimple_build_vector_from_val): Use VEC_DUPLICATE_EXPR when
32680 duplicating a non-constant operand into a variable-length vector.
32681 * hsa-brig.c (hsa_op_immed::emit_to_buffer): Handle polynomial
32682 TYPE_VECTOR_SUBPARTS and VECTOR_CST_NELTS.
32683 * ipa-icf.c (sem_variable::equals): Likewise.
32684 * match.pd: Likewise.
32685 * omp-simd-clone.c (simd_clone_subparts): Likewise.
32686 * print-tree.c (print_node): Likewise.
32687 * stor-layout.c (layout_type): Likewise.
32688 * targhooks.c (default_builtin_vectorization_cost): Likewise.
32689 * tree-cfg.c (verify_gimple_comparison): Likewise.
32690 (verify_gimple_assign_binary): Likewise.
32691 (verify_gimple_assign_ternary): Likewise.
32692 (verify_gimple_assign_single): Likewise.
32693 * tree-pretty-print.c (dump_generic_node): Likewise.
32694 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
32695 (simplify_bitfield_ref, is_combined_permutation_identity): Likewise.
32696 * tree-vect-data-refs.c (vect_permute_store_chain): Likewise.
32697 (vect_grouped_load_supported, vect_permute_load_chain): Likewise.
32698 (vect_shift_permute_load_chain): Likewise.
32699 * tree-vect-generic.c (nunits_for_known_piecewise_op): Likewise.
32700 (expand_vector_condition, optimize_vector_constructor): Likewise.
32701 (lower_vec_perm, get_compute_type): Likewise.
32702 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
32703 (get_initial_defs_for_reduction, vect_transform_loop): Likewise.
32704 * tree-vect-patterns.c (vect_recog_bool_pattern): Likewise.
32705 (vect_recog_mask_conversion_pattern): Likewise.
32706 * tree-vect-slp.c (vect_supported_load_permutation_p): Likewise.
32707 (vect_get_constant_vectors, vect_transform_slp_perm_load): Likewise.
32708 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
32709 (get_group_load_store_type, vectorizable_mask_load_store): Likewise.
32710 (vectorizable_bswap, simd_clone_subparts, vectorizable_assignment)
32711 (vectorizable_shift, vectorizable_operation, vectorizable_store)
32712 (vectorizable_load, vect_is_simple_cond, vectorizable_comparison)
32713 (supportable_widening_operation): Likewise.
32714 (supportable_narrowing_operation): Likewise.
32715 * tree-vector-builder.c (tree_vector_builder::binary_encoded_nelts):
32717 * varasm.c (output_constant): Likewise.
32719 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32720 Alan Hayward <alan.hayward@arm.com>
32721 David Sherwood <david.sherwood@arm.com>
32723 * tree-vect-data-refs.c (vect_permute_store_chain): Reorganize
32724 so that both the length == 3 and length != 3 cases set up their
32725 own permute vectors. Add comments explaining why we know the
32726 number of elements is constant.
32727 (vect_permute_load_chain): Likewise.
32729 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32730 Alan Hayward <alan.hayward@arm.com>
32731 David Sherwood <david.sherwood@arm.com>
32733 * machmode.h (mode_nunits): Change from unsigned char to
32735 (ONLY_FIXED_SIZE_MODES): New macro.
32736 (pod_mode::measurement_type, scalar_int_mode::measurement_type)
32737 (scalar_float_mode::measurement_type, scalar_mode::measurement_type)
32738 (complex_mode::measurement_type, fixed_size_mode::measurement_type):
32740 (mode_to_nunits): Return a poly_uint16 rather than an unsigned short.
32741 (GET_MODE_NUNITS): Return a constant if ONLY_FIXED_SIZE_MODES,
32742 or if measurement_type is not polynomial.
32743 * genmodes.c (ZERO_COEFFS): New macro.
32744 (emit_mode_nunits_inline): Make mode_nunits_inline return a
32746 (emit_mode_nunits): Change the type of mode_nunits to poly_uint16_pod.
32747 Use ZERO_COEFFS when emitting initializers.
32748 * data-streamer.h (bp_pack_poly_value): New function.
32749 (bp_unpack_poly_value): Likewise.
32750 * lto-streamer-in.c (lto_input_mode_table): Use bp_unpack_poly_value
32751 for GET_MODE_NUNITS.
32752 * lto-streamer-out.c (lto_write_mode_table): Use bp_pack_poly_value
32753 for GET_MODE_NUNITS.
32754 * tree.c (make_vector_type): Remove temporary shim and make
32755 the real function take the number of units as a poly_uint64
32756 rather than an int.
32757 (build_vector_type_for_mode): Handle polynomial nunits.
32758 * dwarf2out.c (loc_descriptor, add_const_value_attribute): Likewise.
32759 * emit-rtl.c (const_vec_series_p_1): Likewise.
32760 (gen_rtx_CONST_VECTOR): Likewise.
32761 * fold-const.c (test_vec_duplicate_folding): Likewise.
32762 * genrecog.c (validate_pattern): Likewise.
32763 * optabs-query.c (can_vec_perm_var_p, can_mult_highpart_p): Likewise.
32764 * optabs-tree.c (expand_vec_cond_expr_p): Likewise.
32765 * optabs.c (expand_vector_broadcast, expand_binop_directly): Likewise.
32766 (shift_amt_for_vec_perm_mask, expand_vec_perm_var): Likewise.
32767 (expand_vec_cond_expr, expand_mult_highpart): Likewise.
32768 * rtlanal.c (subreg_get_info): Likewise.
32769 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
32770 (vect_grouped_load_supported): Likewise.
32771 * tree-vect-generic.c (type_for_widest_vector_mode): Likewise.
32772 * tree-vect-loop.c (have_whole_vector_shift): Likewise.
32773 * simplify-rtx.c (simplify_unary_operation_1): Likewise.
32774 (simplify_const_unary_operation, simplify_binary_operation_1)
32775 (simplify_const_binary_operation, simplify_ternary_operation)
32776 (test_vector_ops_duplicate, test_vector_ops): Likewise.
32777 (simplify_immed_subreg): Use GET_MODE_NUNITS on a fixed_size_mode
32778 instead of CONST_VECTOR_NUNITS.
32779 * varasm.c (output_constant_pool_2): Likewise.
32780 * rtx-vector-builder.c (rtx_vector_builder::build): Only include the
32781 explicit-encoded elements in the XVEC for variable-length vectors.
32783 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32785 * lra-constraints.c (curr_insn_transform): Use partial_subreg_p.
32787 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32788 Alan Hayward <alan.hayward@arm.com>
32789 David Sherwood <david.sherwood@arm.com>
32791 * coretypes.h (fixed_size_mode): Declare.
32792 (fixed_size_mode_pod): New typedef.
32793 * builtins.h (target_builtins::x_apply_args_mode)
32794 (target_builtins::x_apply_result_mode): Change type to
32795 fixed_size_mode_pod.
32796 * builtins.c (apply_args_size, apply_result_size, result_vector)
32797 (expand_builtin_apply_args_1, expand_builtin_apply)
32798 (expand_builtin_return): Update accordingly.
32800 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32802 * cse.c (hash_rtx_cb): Hash only the encoded elements.
32803 * cselib.c (cselib_hash_rtx): Likewise.
32804 * expmed.c (make_tree): Build VECTOR_CSTs directly from the
32805 CONST_VECTOR encoding.
32807 2018-01-03 Jakub Jelinek <jakub@redhat.com>
32808 Jeff Law <law@redhat.com>
32811 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): For
32812 noreturn probe, use gen_pop instead of ix86_emit_restore_reg_using_pop,
32813 only set RTX_FRAME_RELATED_P on both the push and pop if cfa_reg is sp
32814 and add REG_CFA_ADJUST_CFA notes in that case to both insns.
32817 * config/i386/i386.c (ix86_adjust_stack_and_probe_stack_clash): Do not
32818 explicitly probe *sp in a noreturn function if there were any callee
32819 register saves or frame pointer is needed.
32821 2018-01-03 Jakub Jelinek <jakub@redhat.com>
32824 * cfgexpand.c (expand_debug_expr): Return NULL if mode is
32825 BLKmode for ternary, binary or unary expressions.
32828 * var-tracking.c (delete_vta_debug_insn): New inline function.
32829 (delete_vta_debug_insns): Add USE_CFG argument, if true, walk just
32830 insns from get_insns () to NULL instead of each bb separately.
32831 Use delete_vta_debug_insn. No longer static.
32832 (vt_debug_insns_local, variable_tracking_main_1): Adjust
32833 delete_vta_debug_insns callers.
32834 * rtl.h (delete_vta_debug_insns): Declare.
32835 * final.c (rest_of_handle_final): Call delete_vta_debug_insns
32836 instead of variable_tracking_main.
32838 2018-01-03 Martin Sebor <msebor@redhat.com>
32840 PR tree-optimization/83603
32841 * calls.c (maybe_warn_nonstring_arg): Avoid accessing function
32842 arguments past the endof the argument list in functions declared
32843 without a prototype.
32844 * gimple-ssa-warn-restrict.c (wrestrict_dom_walker::check_call):
32845 Avoid checking when arguments are null.
32847 2018-01-03 Martin Sebor <msebor@redhat.com>
32850 * doc/extend.texi (attribute const): Fix a typo.
32851 * ipa-pure-const.c ((warn_function_const, warn_function_pure): Avoid
32852 issuing -Wsuggest-attribute for void functions.
32854 2018-01-03 Martin Sebor <msebor@redhat.com>
32856 * gimple-ssa-warn-restrict.c (builtin_memref::builtin_memref): Use
32857 offset_int::from instead of wide_int::to_shwi.
32858 (maybe_diag_overlap): Remove assertion.
32859 Use HOST_WIDE_INT_PRINT_DEC instead of %lli.
32860 * gimple-ssa-sprintf.c (format_directive): Same.
32861 (parse_directive): Same.
32862 (sprintf_dom_walker::compute_format_length): Same.
32863 (try_substitute_return_value): Same.
32865 2018-01-03 Jeff Law <law@redhat.com>
32867 PR middle-end/83654
32868 * explow.c (anti_adjust_stack_and_probe_stack_clash): Test a
32869 non-constant residual for zero at runtime and avoid probing in
32870 that case. Reorganize code for trailing problem to mirror handling
32873 2018-01-03 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
32875 PR tree-optimization/83501
32876 * tree-ssa-strlen.c (get_string_cst): New.
32877 (handle_char_store): Call get_string_cst.
32879 2018-01-03 Martin Liska <mliska@suse.cz>
32881 PR tree-optimization/83593
32882 * tree-ssa-strlen.c: Include tree-cfg.h.
32883 (strlen_check_and_optimize_stmt): Add new argument cleanup_eh.
32884 (strlen_dom_walker): Add new member variable m_cleanup_cfg.
32885 (strlen_dom_walker::strlen_dom_walker): Initialize m_cleanup_cfg
32887 (strlen_dom_walker::before_dom_children): Call
32888 gimple_purge_dead_eh_edges. Dump tranformation with details
32890 (strlen_dom_walker::before_dom_children): Update call by adding
32891 new argument cleanup_eh.
32892 (pass_strlen::execute): Return TODO_cleanup_cfg if needed.
32894 2018-01-03 Martin Liska <mliska@suse.cz>
32897 * cif-code.def (VARIADIC_THUNK): New enum value.
32898 * ipa-fnsummary.c (compute_fn_summary): Do not inline variadic
32901 2018-01-03 Jan Beulich <jbeulich@suse.com>
32903 * sse.md (mov<mode>_internal): Tighten condition for when to use
32904 vmovdqu<ssescalarsize> for TI and OI modes.
32906 2018-01-03 Jakub Jelinek <jakub@redhat.com>
32908 Update copyright years.
32910 2018-01-03 Martin Liska <mliska@suse.cz>
32913 * ipa-visibility.c (function_and_variable_visibility): Skip
32914 functions with noipa attribure.
32916 2018-01-03 Jakub Jelinek <jakub@redhat.com>
32918 * gcc.c (process_command): Update copyright notice dates.
32919 * gcov-dump.c (print_version): Ditto.
32920 * gcov.c (print_version): Ditto.
32921 * gcov-tool.c (print_version): Ditto.
32922 * gengtype.c (create_file): Ditto.
32923 * doc/cpp.texi: Bump @copying's copyright year.
32924 * doc/cppinternals.texi: Ditto.
32925 * doc/gcc.texi: Ditto.
32926 * doc/gccint.texi: Ditto.
32927 * doc/gcov.texi: Ditto.
32928 * doc/install.texi: Ditto.
32929 * doc/invoke.texi: Ditto.
32931 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32933 * vector-builder.h (vector_builder::m_full_nelts): Change from
32934 unsigned int to poly_uint64.
32935 (vector_builder::full_nelts): Update prototype accordingly.
32936 (vector_builder::new_vector): Likewise.
32937 (vector_builder::encoded_full_vector_p): Handle polynomial full_nelts.
32938 (vector_builder::operator ==): Likewise.
32939 (vector_builder::finalize): Likewise.
32940 * int-vector-builder.h (int_vector_builder::int_vector_builder):
32941 Take the number of elements as a poly_uint64 rather than an
32943 * vec-perm-indices.h (vec_perm_indices::m_nelts_per_input): Change
32944 from unsigned int to poly_uint64.
32945 (vec_perm_indices::vec_perm_indices): Update prototype accordingly.
32946 (vec_perm_indices::new_vector): Likewise.
32947 (vec_perm_indices::length): Likewise.
32948 (vec_perm_indices::nelts_per_input): Likewise.
32949 (vec_perm_indices::input_nelts): Likewise.
32950 * vec-perm-indices.c (vec_perm_indices::new_vector): Take the
32951 number of elements per input as a poly_uint64 rather than an
32952 unsigned int. Use the original encoding for variable-length
32953 vectors, rather than clamping each individual element.
32954 For the second and subsequent elements in each pattern,
32955 clamp the step and base before clamping their sum.
32956 (vec_perm_indices::series_p): Handle polynomial element counts.
32957 (vec_perm_indices::all_in_range_p): Likewise.
32958 (vec_perm_indices_to_tree): Likewise.
32959 (vec_perm_indices_to_rtx): Likewise.
32960 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise.
32961 * tree-vector-builder.c (tree_vector_builder::new_unary_operation)
32962 (tree_vector_builder::new_binary_operation): Handle polynomial
32963 element counts. Return false if we need to know the number
32964 of elements at compile time.
32965 * fold-const.c (fold_vec_perm): Punt if the number of elements
32966 isn't known at compile time.
32968 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32970 * vec-perm-indices.h (vec_perm_builder): Change element type
32971 from HOST_WIDE_INT to poly_int64.
32972 (vec_perm_indices::element_type): Update accordingly.
32973 (vec_perm_indices::clamp): Handle polynomial element_types.
32974 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
32975 (vec_perm_indices::all_in_range_p): Likewise.
32976 (tree_to_vec_perm_builder): Check for poly_int64 trees rather
32978 * vector-builder.h (vector_builder::stepped_sequence_p): Handle
32979 polynomial vec_perm_indices element types.
32980 * int-vector-builder.h (int_vector_builder::equal_p): Likewise.
32981 * fold-const.c (fold_vec_perm): Likewise.
32982 * optabs.c (shift_amt_for_vec_perm_mask): Likewise.
32983 * tree-vect-generic.c (lower_vec_perm): Likewise.
32984 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
32985 * config/aarch64/aarch64.c (aarch64_evpc_tbl): Cast d->perm
32986 element type to HOST_WIDE_INT.
32988 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32989 Alan Hayward <alan.hayward@arm.com>
32990 David Sherwood <david.sherwood@arm.com>
32992 * alias.c (addr_side_effect_eval): Take the size as a poly_int64
32993 rather than an int. Use plus_constant.
32994 (memrefs_conflict_p): Take the sizes as poly_int64s rather than ints.
32995 Take the offset "c" as a poly_int64 rather than a HOST_WIDE_INT.
32997 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
32998 Alan Hayward <alan.hayward@arm.com>
32999 David Sherwood <david.sherwood@arm.com>
33001 * calls.c (emit_call_1, expand_call): Change struct_value_size from
33002 a HOST_WIDE_INT to a poly_int64.
33004 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33005 Alan Hayward <alan.hayward@arm.com>
33006 David Sherwood <david.sherwood@arm.com>
33008 * calls.c (load_register_parameters): Cope with polynomial
33009 mode sizes. Require a constant size for BLKmode parameters
33010 that aren't described by a PARALLEL. If BLOCK_REG_PADDING
33011 forces a parameter to be padded at the lsb end in order to
33012 fill a complete number of words, require the parameter size
33013 to be ordered wrt UNITS_PER_WORD.
33015 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33016 Alan Hayward <alan.hayward@arm.com>
33017 David Sherwood <david.sherwood@arm.com>
33019 * reload1.c (spill_stack_slot_width): Change element type
33020 from unsigned int to poly_uint64_pod.
33021 (alter_reg): Treat mode sizes as polynomial.
33023 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33024 Alan Hayward <alan.hayward@arm.com>
33025 David Sherwood <david.sherwood@arm.com>
33027 * reload.c (complex_word_subreg_p): New function.
33028 (reload_inner_reg_of_subreg, push_reload): Use it.
33030 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33031 Alan Hayward <alan.hayward@arm.com>
33032 David Sherwood <david.sherwood@arm.com>
33034 * lra-constraints.c (process_alt_operands): Reject matched
33035 operands whose sizes aren't ordered.
33036 (match_reload): Refer to this check here.
33038 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33039 Alan Hayward <alan.hayward@arm.com>
33040 David Sherwood <david.sherwood@arm.com>
33042 * builtins.c (expand_ifn_atomic_compare_exchange_into_call): Assert
33043 that the mode size is in the set {1, 2, 4, 8, 16}.
33045 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33046 Alan Hayward <alan.hayward@arm.com>
33047 David Sherwood <david.sherwood@arm.com>
33049 * var-tracking.c (adjust_mems): Treat mode sizes as polynomial.
33050 Use plus_constant instead of gen_rtx_PLUS.
33052 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33053 Alan Hayward <alan.hayward@arm.com>
33054 David Sherwood <david.sherwood@arm.com>
33056 * config/cr16/cr16-protos.h (cr16_push_rounding): Declare.
33057 * config/cr16/cr16.h (PUSH_ROUNDING): Move implementation to...
33058 * config/cr16/cr16.c (cr16_push_rounding): ...this new function.
33059 * config/h8300/h8300-protos.h (h8300_push_rounding): Declare.
33060 * config/h8300/h8300.h (PUSH_ROUNDING): Move implementation to...
33061 * config/h8300/h8300.c (h8300_push_rounding): ...this new function.
33062 * config/i386/i386-protos.h (ix86_push_rounding): Declare.
33063 * config/i386/i386.h (PUSH_ROUNDING): Move implementation to...
33064 * config/i386/i386.c (ix86_push_rounding): ...this new function.
33065 * config/m32c/m32c-protos.h (m32c_push_rounding): Take and return
33067 * config/m32c/m32c.c (m32c_push_rounding): Likewise.
33068 * config/m68k/m68k-protos.h (m68k_push_rounding): Declare.
33069 * config/m68k/m68k.h (PUSH_ROUNDING): Move implementation to...
33070 * config/m68k/m68k.c (m68k_push_rounding): ...this new function.
33071 * config/pdp11/pdp11-protos.h (pdp11_push_rounding): Declare.
33072 * config/pdp11/pdp11.h (PUSH_ROUNDING): Move implementation to...
33073 * config/pdp11/pdp11.c (pdp11_push_rounding): ...this new function.
33074 * config/stormy16/stormy16-protos.h (xstormy16_push_rounding): Declare.
33075 * config/stormy16/stormy16.h (PUSH_ROUNDING): Move implementation to...
33076 * config/stormy16/stormy16.c (xstormy16_push_rounding): ...this new
33078 * expr.c (emit_move_resolve_push): Treat the input and result
33079 of PUSH_ROUNDING as a poly_int64.
33080 (emit_move_complex_push, emit_single_push_insn_1): Likewise.
33081 (emit_push_insn): Likewise.
33082 * lra-eliminations.c (mark_not_eliminable): Likewise.
33083 * recog.c (push_operand): Likewise.
33084 * reload1.c (elimination_effects): Likewise.
33085 * rtlanal.c (nonzero_bits1): Likewise.
33086 * calls.c (store_one_arg): Likewise. Require the padding to be
33087 known at compile time.
33089 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33090 Alan Hayward <alan.hayward@arm.com>
33091 David Sherwood <david.sherwood@arm.com>
33093 * expr.c (emit_single_push_insn_1): Treat mode sizes as polynomial.
33094 Use plus_constant instead of gen_rtx_PLUS.
33096 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33097 Alan Hayward <alan.hayward@arm.com>
33098 David Sherwood <david.sherwood@arm.com>
33100 * auto-inc-dec.c (set_inc_state): Take the mode size as a poly_int64
33101 rather than an int.
33103 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33104 Alan Hayward <alan.hayward@arm.com>
33105 David Sherwood <david.sherwood@arm.com>
33107 * expr.c (expand_expr_real_1): Use tree_to_poly_uint64
33108 instead of int_size_in_bytes when handling VIEW_CONVERT_EXPRs
33109 via stack temporaries. Treat the mode size as polynomial too.
33111 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33112 Alan Hayward <alan.hayward@arm.com>
33113 David Sherwood <david.sherwood@arm.com>
33115 * expr.c (expand_expr_real_2): When handling conversions involving
33116 unions, apply tree_to_poly_uint64 to the TYPE_SIZE rather than
33117 multiplying int_size_in_bytes by BITS_PER_UNIT. Treat GET_MODE_BISIZE
33118 as a poly_uint64 too.
33120 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33121 Alan Hayward <alan.hayward@arm.com>
33122 David Sherwood <david.sherwood@arm.com>
33124 * rtlanal.c (subreg_get_info): Handle polynomial mode sizes.
33126 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33127 Alan Hayward <alan.hayward@arm.com>
33128 David Sherwood <david.sherwood@arm.com>
33130 * combine.c (can_change_dest_mode): Handle polynomial
33131 REGMODE_NATURAL_SIZE.
33132 * expmed.c (store_bit_field_1): Likewise.
33133 * expr.c (store_constructor): Likewise.
33134 * emit-rtl.c (validate_subreg): Operate on polynomial mode sizes
33135 and polynomial REGMODE_NATURAL_SIZE.
33136 (gen_lowpart_common): Likewise.
33137 * reginfo.c (record_subregs_of_mode): Likewise.
33138 * rtlanal.c (read_modify_subreg_p): Likewise.
33140 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33141 Alan Hayward <alan.hayward@arm.com>
33142 David Sherwood <david.sherwood@arm.com>
33144 * internal-fn.c (expand_vector_ubsan_overflow): Handle polynomial
33145 numbers of elements.
33147 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33148 Alan Hayward <alan.hayward@arm.com>
33149 David Sherwood <david.sherwood@arm.com>
33151 * match.pd: Cope with polynomial numbers of vector elements.
33153 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33154 Alan Hayward <alan.hayward@arm.com>
33155 David Sherwood <david.sherwood@arm.com>
33157 * fold-const.c (fold_indirect_ref_1): Handle polynomial offsets
33158 in a POINTER_PLUS_EXPR.
33160 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33161 Alan Hayward <alan.hayward@arm.com>
33162 David Sherwood <david.sherwood@arm.com>
33164 * omp-simd-clone.c (simd_clone_subparts): New function.
33165 (simd_clone_init_simd_arrays): Use it instead of TYPE_VECTOR_SUBPARTS.
33166 (ipa_simd_modify_function_body): Likewise.
33168 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33169 Alan Hayward <alan.hayward@arm.com>
33170 David Sherwood <david.sherwood@arm.com>
33172 * tree-vect-generic.c (nunits_for_known_piecewise_op): New function.
33173 (expand_vector_piecewise): Use it instead of TYPE_VECTOR_SUBPARTS.
33174 (expand_vector_addition, add_rshift, expand_vector_divmod): Likewise.
33175 (expand_vector_condition, vector_element): Likewise.
33176 (subparts_gt): New function.
33177 (get_compute_type): Use subparts_gt.
33178 (count_type_subparts): Delete.
33179 (expand_vector_operations_1): Use subparts_gt instead of
33180 count_type_subparts.
33182 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33183 Alan Hayward <alan.hayward@arm.com>
33184 David Sherwood <david.sherwood@arm.com>
33186 * tree-vect-data-refs.c (vect_no_alias_p): Replace with...
33187 (vect_compile_time_alias): ...this new function. Do the calculation
33188 on poly_ints rather than trees.
33189 (vect_prune_runtime_alias_test_list): Update call accordingly.
33191 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33192 Alan Hayward <alan.hayward@arm.com>
33193 David Sherwood <david.sherwood@arm.com>
33195 * tree-vect-slp.c (vect_build_slp_tree_1): Handle polynomial
33197 (vect_schedule_slp_instance): Likewise.
33199 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33200 Alan Hayward <alan.hayward@arm.com>
33201 David Sherwood <david.sherwood@arm.com>
33203 * tree-vect-slp.c (vect_get_and_check_slp_defs): Reject
33204 constant and extern definitions for variable-length vectors.
33205 (vect_get_constant_vectors): Note that the number of units
33206 is known to be constant.
33208 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33209 Alan Hayward <alan.hayward@arm.com>
33210 David Sherwood <david.sherwood@arm.com>
33212 * tree-vect-stmts.c (vectorizable_conversion): Treat the number
33213 of units as polynomial. Choose between WIDE and NARROW based
33216 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33217 Alan Hayward <alan.hayward@arm.com>
33218 David Sherwood <david.sherwood@arm.com>
33220 * tree-vect-stmts.c (simd_clone_subparts): New function.
33221 (vectorizable_simd_clone_call): Use it instead of TYPE_VECTOR_SUBPARTS.
33223 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33224 Alan Hayward <alan.hayward@arm.com>
33225 David Sherwood <david.sherwood@arm.com>
33227 * tree-vect-stmts.c (vectorizable_call): Treat the number of
33228 vectors as polynomial. Use build_index_vector for
33229 IFN_GOMP_SIMD_LANE.
33231 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33232 Alan Hayward <alan.hayward@arm.com>
33233 David Sherwood <david.sherwood@arm.com>
33235 * tree-vect-stmts.c (get_load_store_type): Treat the number of
33236 units as polynomial. Reject VMAT_ELEMENTWISE and VMAT_STRIDED_SLP
33237 for variable-length vectors.
33238 (vectorizable_mask_load_store): Treat the number of units as
33239 polynomial, asserting that it is constant if the condition has
33240 already been enforced.
33241 (vectorizable_store, vectorizable_load): Likewise.
33243 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33244 Alan Hayward <alan.hayward@arm.com>
33245 David Sherwood <david.sherwood@arm.com>
33247 * tree-vect-loop.c (vectorizable_live_operation): Treat the number
33248 of units as polynomial. Punt if we can't tell at compile time
33249 which vector contains the final result.
33251 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33252 Alan Hayward <alan.hayward@arm.com>
33253 David Sherwood <david.sherwood@arm.com>
33255 * tree-vect-loop.c (vectorizable_induction): Treat the number
33256 of units as polynomial. Punt on SLP inductions. Use an integer
33257 VEC_SERIES_EXPR for variable-length integer reductions. Use a
33258 cast of such a series for variable-length floating-point
33261 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33262 Alan Hayward <alan.hayward@arm.com>
33263 David Sherwood <david.sherwood@arm.com>
33265 * tree.h (build_index_vector): Declare.
33266 * tree.c (build_index_vector): New function.
33267 * tree-vect-loop.c (get_initial_defs_for_reduction): Treat the number
33268 of units as polynomial, forcibly converting it to a constant if
33269 vectorizable_reduction has already enforced the condition.
33270 (vect_create_epilog_for_reduction): Likewise. Use build_index_vector
33271 to create a {1,2,3,...} vector.
33272 (vectorizable_reduction): Treat the number of units as polynomial.
33273 Choose vectype_in based on the largest scalar element size rather
33274 than the smallest number of units. Enforce the restrictions
33277 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33278 Alan Hayward <alan.hayward@arm.com>
33279 David Sherwood <david.sherwood@arm.com>
33281 * tree-vect-data-refs.c (vector_alignment_reachable_p): Treat the
33282 number of units as polynomial.
33284 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33285 Alan Hayward <alan.hayward@arm.com>
33286 David Sherwood <david.sherwood@arm.com>
33288 * target.h (vector_sizes, auto_vector_sizes): New typedefs.
33289 * target.def (autovectorize_vector_sizes): Return the vector sizes
33290 by pointer, using vector_sizes rather than a bitmask.
33291 * targhooks.h (default_autovectorize_vector_sizes): Update accordingly.
33292 * targhooks.c (default_autovectorize_vector_sizes): Likewise.
33293 * config/aarch64/aarch64.c (aarch64_autovectorize_vector_sizes):
33295 * config/arc/arc.c (arc_autovectorize_vector_sizes): Likewise.
33296 * config/arm/arm.c (arm_autovectorize_vector_sizes): Likewise.
33297 * config/i386/i386.c (ix86_autovectorize_vector_sizes): Likewise.
33298 * config/mips/mips.c (mips_autovectorize_vector_sizes): Likewise.
33299 * omp-general.c (omp_max_vf): Likewise.
33300 * omp-low.c (omp_clause_aligned_alignment): Likewise.
33301 * optabs-query.c (can_vec_mask_load_store_p): Likewise.
33302 * tree-vect-loop.c (vect_analyze_loop): Likewise.
33303 * tree-vect-slp.c (vect_slp_bb): Likewise.
33304 * doc/tm.texi: Regenerate.
33305 * tree-vectorizer.h (current_vector_size): Change from an unsigned int
33307 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Take
33308 the vector size as a poly_uint64 rather than an unsigned int.
33309 (current_vector_size): Change from an unsigned int to a poly_uint64.
33310 (get_vectype_for_scalar_type): Update accordingly.
33311 * tree.h (build_truth_vector_type): Take the size and number of
33312 units as a poly_uint64 rather than an unsigned int.
33313 (build_vector_type): Add a temporary overload that takes
33314 the number of units as a poly_uint64 rather than an unsigned int.
33315 * tree.c (make_vector_type): Likewise.
33316 (build_truth_vector_type): Take the number of units as a poly_uint64
33317 rather than an unsigned int.
33319 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33320 Alan Hayward <alan.hayward@arm.com>
33321 David Sherwood <david.sherwood@arm.com>
33323 * target.def (get_mask_mode): Take the number of units and length
33324 as poly_uint64s rather than unsigned ints.
33325 * targhooks.h (default_get_mask_mode): Update accordingly.
33326 * targhooks.c (default_get_mask_mode): Likewise.
33327 * config/i386/i386.c (ix86_get_mask_mode): Likewise.
33328 * doc/tm.texi: Regenerate.
33330 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33331 Alan Hayward <alan.hayward@arm.com>
33332 David Sherwood <david.sherwood@arm.com>
33334 * omp-general.h (omp_max_vf): Return a poly_uint64 instead of an int.
33335 * omp-general.c (omp_max_vf): Likewise.
33336 * omp-expand.c (omp_adjust_chunk_size): Update call to omp_max_vf.
33337 (expand_omp_simd): Handle polynomial safelen.
33338 * omp-low.c (omplow_simd_context): Add a default constructor.
33339 (omplow_simd_context::max_vf): Change from int to poly_uint64.
33340 (lower_rec_simd_input_clauses): Update accordingly.
33341 (lower_rec_input_clauses): Likewise.
33343 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33344 Alan Hayward <alan.hayward@arm.com>
33345 David Sherwood <david.sherwood@arm.com>
33347 * tree-vectorizer.h (vect_nunits_for_cost): New function.
33348 * tree-vect-loop.c (vect_model_reduction_cost): Use it.
33349 * tree-vect-slp.c (vect_analyze_slp_cost_1): Likewise.
33350 (vect_analyze_slp_cost): Likewise.
33351 * tree-vect-stmts.c (vect_model_store_cost): Likewise.
33352 (vect_model_load_cost): Likewise.
33354 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33355 Alan Hayward <alan.hayward@arm.com>
33356 David Sherwood <david.sherwood@arm.com>
33358 * tree-vect-slp.c (vect_record_max_nunits, vect_build_slp_tree_1)
33359 (vect_build_slp_tree_2, vect_build_slp_tree): Change max_nunits
33360 from an unsigned int * to a poly_uint64_pod *.
33361 (calculate_unrolling_factor): New function.
33362 (vect_analyze_slp_instance): Use it. Track polynomial max_nunits.
33364 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33365 Alan Hayward <alan.hayward@arm.com>
33366 David Sherwood <david.sherwood@arm.com>
33368 * tree-vectorizer.h (_slp_instance::unrolling_factor): Change
33369 from an unsigned int to a poly_uint64.
33370 (_loop_vec_info::slp_unrolling_factor): Likewise.
33371 (_loop_vec_info::vectorization_factor): Change from an int
33373 (MAX_VECTORIZATION_FACTOR): Bump from 64 to INT_MAX.
33374 (vect_get_num_vectors): New function.
33375 (vect_update_max_nunits, vect_vf_for_cost): Likewise.
33376 (vect_get_num_copies): Use vect_get_num_vectors.
33377 (vect_analyze_data_ref_dependences): Change max_vf from an int *
33378 to an unsigned int *.
33379 (vect_analyze_data_refs): Change min_vf from an int * to a
33381 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
33382 than an unsigned HOST_WIDE_INT.
33383 * tree-vect-data-refs.c (vect_analyze_possibly_independent_ddr)
33384 (vect_analyze_data_ref_dependence): Change max_vf from an int *
33385 to an unsigned int *.
33386 (vect_analyze_data_ref_dependences): Likewise.
33387 (vect_compute_data_ref_alignment): Handle polynomial vf.
33388 (vect_enhance_data_refs_alignment): Likewise.
33389 (vect_prune_runtime_alias_test_list): Likewise.
33390 (vect_shift_permute_load_chain): Likewise.
33391 (vect_supportable_dr_alignment): Likewise.
33392 (dependence_distance_ge_vf): Take the vectorization factor as a
33393 poly_uint64 rather than an unsigned HOST_WIDE_INT.
33394 (vect_analyze_data_refs): Change min_vf from an int * to a
33396 * tree-vect-loop-manip.c (vect_gen_scalar_loop_niters): Take
33397 vfm1 as a poly_uint64 rather than an int. Make the same change
33398 for the returned bound_scalar.
33399 (vect_gen_vector_loop_niters): Handle polynomial vf.
33400 (vect_do_peeling): Likewise. Update call to
33401 vect_gen_scalar_loop_niters and handle polynomial bound_scalars.
33402 (vect_gen_vector_loop_niters_mult_vf): Assert that the vf must
33404 * tree-vect-loop.c (vect_determine_vectorization_factor)
33405 (vect_update_vf_for_slp, vect_analyze_loop_2): Handle polynomial vf.
33406 (vect_get_known_peeling_cost): Likewise.
33407 (vect_estimate_min_profitable_iters, vectorizable_reduction): Likewise.
33408 (vect_worthwhile_without_simd_p, vectorizable_induction): Likewise.
33409 (vect_transform_loop): Likewise. Use the lowest possible VF when
33410 updating the upper bounds of the loop.
33411 (vect_min_worthwhile_factor): Make static. Return an unsigned int
33412 rather than an int.
33413 * tree-vect-slp.c (vect_attempt_slp_rearrange_stmts): Cope with
33414 polynomial unroll factors.
33415 (vect_analyze_slp_cost_1, vect_analyze_slp_instance): Likewise.
33416 (vect_make_slp_decision): Likewise.
33417 (vect_supported_load_permutation_p): Likewise, and polynomial
33419 (vect_analyze_slp_cost): Handle polynomial vf.
33420 (vect_slp_analyze_node_operations): Likewise.
33421 (vect_slp_analyze_bb_1): Likewise.
33422 (vect_transform_slp_perm_load): Take the vf as a poly_uint64 rather
33423 than an unsigned HOST_WIDE_INT.
33424 * tree-vect-stmts.c (vectorizable_simd_clone_call, vectorizable_store)
33425 (vectorizable_load): Handle polynomial vf.
33426 * tree-vectorizer.c (simduid_to_vf::vf): Change from an int to
33428 (adjust_simduid_builtins, shrink_simd_arrays): Update accordingly.
33430 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33431 Alan Hayward <alan.hayward@arm.com>
33432 David Sherwood <david.sherwood@arm.com>
33434 * match.pd: Handle bit operations involving three constants
33435 and try to fold one pair.
33437 2018-01-03 Richard Sandiford <richard.sandiford@linaro.org>
33439 * tree-vect-loop-manip.c: Include gimple-fold.h.
33440 (slpeel_make_loop_iterate_ntimes): Add step, final_iv and
33441 niters_maybe_zero parameters. Handle other cases besides a step of 1.
33442 (vect_gen_vector_loop_niters): Add a step_vector_ptr parameter.
33443 Add a path that uses a step of VF instead of 1, but disable it
33445 (vect_do_peeling): Add step_vector, niters_vector_mult_vf_var
33446 and niters_no_overflow parameters. Update calls to
33447 slpeel_make_loop_iterate_ntimes and vect_gen_vector_loop_niters.
33448 Create a new SSA name if the latter choses to use a ste other
33449 than zero, and return it via niters_vector_mult_vf_var.
33450 * tree-vect-loop.c (vect_transform_loop): Update calls to
33451 vect_do_peeling, vect_gen_vector_loop_niters and
33452 slpeel_make_loop_iterate_ntimes.
33453 * tree-vectorizer.h (slpeel_make_loop_iterate_ntimes, vect_do_peeling)
33454 (vect_gen_vector_loop_niters): Update declarations after above changes.
33456 2018-01-02 Michael Meissner <meissner@linux.vnet.ibm.com>
33458 * config/rs6000/rs6000.md (floor<mode>2): Add support for IEEE
33459 128-bit round to integer instructions.
33460 (ceil<mode>2): Likewise.
33461 (btrunc<mode>2): Likewise.
33462 (round<mode>2): Likewise.
33464 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33466 * config/rs6000/rs6000-string.c (expand_block_move): Allow the use of
33467 unaligned VSX load/store on P8/P9.
33468 (expand_block_clear): Allow the use of unaligned VSX
33469 load/store on P8/P9.
33471 2018-01-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
33473 * config/rs6000/rs6000-p8swap.c (swap_feeds_both_load_and_store):
33475 (rs6000_analyze_swaps): Mark a web unoptimizable if it contains a
33476 swap associated with both a load and a store.
33478 2018-01-02 Andrew Waterman <andrew@sifive.com>
33480 * config/riscv/linux.h (ICACHE_FLUSH_FUNC): New.
33481 * config/riscv/riscv.md (clear_cache): Use it.
33483 2018-01-02 Artyom Skrobov <tyomitch@gmail.com>
33485 * web.c: Remove out-of-date comment.
33487 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33489 * expr.c (fixup_args_size_notes): Check that any existing
33490 REG_ARGS_SIZE notes are correct, and don't try to re-add them.
33491 (emit_single_push_insn_1): Move stack_pointer_delta adjustment to...
33492 (emit_single_push_insn): ...here.
33494 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33496 * rtl.h (CONST_VECTOR_ELT): Redefine to const_vector_elt.
33497 (const_vector_encoded_nelts): New function.
33498 (CONST_VECTOR_NUNITS): Redefine to use GET_MODE_NUNITS.
33499 (const_vector_int_elt, const_vector_elt): Declare.
33500 * emit-rtl.c (const_vector_int_elt_1): New function.
33501 (const_vector_elt): Likewise.
33502 * simplify-rtx.c (simplify_immed_subreg): Avoid taking the address
33503 of CONST_VECTOR_ELT.
33505 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33507 * expr.c: Include rtx-vector-builder.h.
33508 (const_vector_mask_from_tree): Use rtx_vector_builder and operate
33509 directly on the tree encoding.
33510 (const_vector_from_tree): Likewise.
33511 * optabs.c: Include rtx-vector-builder.h.
33512 (expand_vec_perm_var): Use rtx_vector_builder and create a repeating
33513 sequence of "u" values.
33514 * vec-perm-indices.c: Include rtx-vector-builder.h.
33515 (vec_perm_indices_to_rtx): Use rtx_vector_builder and operate
33516 directly on the vec_perm_indices encoding.
33518 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33520 * doc/rtl.texi (const_vector): Describe new encoding scheme.
33521 * Makefile.in (OBJS): Add rtx-vector-builder.o.
33522 * rtx-vector-builder.h: New file.
33523 * rtx-vector-builder.c: Likewise.
33524 * rtl.h (rtx_def::u2): Add a const_vector field.
33525 (CONST_VECTOR_NPATTERNS): New macro.
33526 (CONST_VECTOR_NELTS_PER_PATTERN): Likewise.
33527 (CONST_VECTOR_DUPLICATE_P): Likewise.
33528 (CONST_VECTOR_STEPPED_P): Likewise.
33529 (CONST_VECTOR_ENCODED_ELT): Likewise.
33530 (const_vec_duplicate_p): Check for a duplicated vector encoding.
33531 (unwrap_const_vec_duplicate): Likewise.
33532 (const_vec_series_p): Check for a non-duplicated vector encoding.
33533 Say that the function only returns true for integer vectors.
33534 * emit-rtl.c: Include rtx-vector-builder.h.
33535 (gen_const_vec_duplicate_1): Delete.
33536 (gen_const_vector): Call gen_const_vec_duplicate instead of
33537 gen_const_vec_duplicate_1.
33538 (const_vec_series_p_1): Operate directly on the CONST_VECTOR encoding.
33539 (gen_const_vec_duplicate): Use rtx_vector_builder.
33540 (gen_const_vec_series): Likewise.
33541 (gen_rtx_CONST_VECTOR): Likewise.
33542 * config/powerpcspe/powerpcspe.c: Include rtx-vector-builder.h.
33543 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
33544 Build a new vector rather than modifying a CONST_VECTOR in-place.
33545 (handle_special_swappables): Update call accordingly.
33546 * config/rs6000/rs6000-p8swap.c: Include rtx-vector-builder.h.
33547 (swap_const_vector_halves): Take an rtx pointer rather than rtx.
33548 Build a new vector rather than modifying a CONST_VECTOR in-place.
33549 (handle_special_swappables): Update call accordingly.
33551 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33553 * simplify-rtx.c (simplify_const_binary_operation): Use
33554 CONST_VECTOR_ELT instead of XVECEXP.
33556 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33558 * tree-cfg.c (verify_gimple_assign_ternary): Allow the size of
33559 the selector elements to be different from the data elements
33560 if the selector is a VECTOR_CST.
33561 * tree-vect-stmts.c (vect_gen_perm_mask_any): Use a vector of
33562 ssizetype for the selector.
33564 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33566 * optabs.c (shift_amt_for_vec_perm_mask): Try using series_p
33567 before testing each element individually.
33568 * tree-vect-generic.c (lower_vec_perm): Likewise.
33570 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33572 * selftest.h (selftest::vec_perm_indices_c_tests): Declare.
33573 * selftest-run-tests.c (selftest::run_tests): Call it.
33574 * vector-builder.h (vector_builder::operator ==): New function.
33575 (vector_builder::operator !=): Likewise.
33576 * vec-perm-indices.h (vec_perm_indices::series_p): Declare.
33577 (vec_perm_indices::all_from_input_p): New function.
33578 * vec-perm-indices.c (vec_perm_indices::series_p): Likewise.
33579 (test_vec_perm_12, selftest::vec_perm_indices_c_tests): Likewise.
33580 * fold-const.c (fold_ternary_loc): Use tree_to_vec_perm_builder
33581 instead of reading the VECTOR_CST directly. Detect whether both
33582 vector inputs are the same before constructing the vec_perm_indices,
33583 and update the number of inputs argument accordingly. Use the
33584 utility functions added above. Only construct sel2 if we need to.
33586 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33588 * optabs.c (expand_vec_perm_var): Use an explicit encoding for
33589 the broadcast of the low byte.
33590 (expand_mult_highpart): Use an explicit encoding for the permutes.
33591 * optabs-query.c (can_mult_highpart_p): Likewise.
33592 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
33593 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
33594 (vectorizable_bswap): Likewise.
33595 * tree-vect-data-refs.c (vect_grouped_store_supported): Use an
33596 explicit encoding for the power-of-2 permutes.
33597 (vect_permute_store_chain): Likewise.
33598 (vect_grouped_load_supported): Likewise.
33599 (vect_permute_load_chain): Likewise.
33601 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33603 * vec-perm-indices.h (vec_perm_indices_to_tree): Declare.
33604 * vec-perm-indices.c (vec_perm_indices_to_tree): New function.
33605 * tree-ssa-forwprop.c (simplify_vector_constructor): Use it.
33606 * tree-vect-slp.c (vect_transform_slp_perm_load): Likewise.
33607 * tree-vect-stmts.c (vectorizable_bswap): Likewise.
33608 (vect_gen_perm_mask_any): Likewise.
33610 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33612 * int-vector-builder.h: New file.
33613 * vec-perm-indices.h: Include int-vector-builder.h.
33614 (vec_perm_indices): Redefine as an int_vector_builder.
33615 (auto_vec_perm_indices): Delete.
33616 (vec_perm_builder): Redefine as a stand-alone class.
33617 (vec_perm_indices::vec_perm_indices): New function.
33618 (vec_perm_indices::clamp): Likewise.
33619 * vec-perm-indices.c: Include fold-const.h and tree-vector-builder.h.
33620 (vec_perm_indices::new_vector): New function.
33621 (vec_perm_indices::new_expanded_vector): Update for new
33622 vec_perm_indices class.
33623 (vec_perm_indices::rotate_inputs): New function.
33624 (vec_perm_indices::all_in_range_p): Operate directly on the
33625 encoded form, without computing elided elements.
33626 (tree_to_vec_perm_builder): Operate directly on the VECTOR_CST
33627 encoding. Update for new vec_perm_indices class.
33628 * optabs.c (expand_vec_perm_const): Create a vec_perm_indices for
33629 the given vec_perm_builder.
33630 (expand_vec_perm_var): Update vec_perm_builder constructor.
33631 (expand_mult_highpart): Use vec_perm_builder instead of
33632 auto_vec_perm_indices.
33633 * optabs-query.c (can_mult_highpart_p): Use vec_perm_builder and
33634 vec_perm_indices instead of auto_vec_perm_indices. Use a single
33635 or double series encoding as appropriate.
33636 * fold-const.c (fold_ternary_loc): Use vec_perm_builder and
33637 vec_perm_indices instead of auto_vec_perm_indices.
33638 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
33639 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
33640 (vect_permute_store_chain): Likewise.
33641 (vect_grouped_load_supported): Likewise.
33642 (vect_permute_load_chain): Likewise.
33643 (vect_shift_permute_load_chain): Likewise.
33644 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
33645 (vect_transform_slp_perm_load): Likewise.
33646 (vect_schedule_slp_instance): Likewise.
33647 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
33648 (vectorizable_mask_load_store): Likewise.
33649 (vectorizable_bswap): Likewise.
33650 (vectorizable_store): Likewise.
33651 (vectorizable_load): Likewise.
33652 * tree-vect-generic.c (lower_vec_perm): Use vec_perm_builder and
33653 vec_perm_indices instead of auto_vec_perm_indices. Use
33654 tree_to_vec_perm_builder to read the vector from a tree.
33655 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Take a
33656 vec_perm_builder instead of a vec_perm_indices.
33657 (have_whole_vector_shift): Use vec_perm_builder and
33658 vec_perm_indices instead of auto_vec_perm_indices. Leave the
33659 truncation to calc_vec_perm_mask_for_shift.
33660 (vect_create_epilog_for_reduction): Likewise.
33661 * config/aarch64/aarch64.c (expand_vec_perm_d::perm): Change
33662 from auto_vec_perm_indices to vec_perm_indices.
33663 (aarch64_expand_vec_perm_const_1): Use rotate_inputs on d.perm
33664 instead of changing individual elements.
33665 (aarch64_vectorize_vec_perm_const): Use new_vector to install
33666 the vector in d.perm.
33667 * config/arm/arm.c (expand_vec_perm_d::perm): Change
33668 from auto_vec_perm_indices to vec_perm_indices.
33669 (arm_expand_vec_perm_const_1): Use rotate_inputs on d.perm
33670 instead of changing individual elements.
33671 (arm_vectorize_vec_perm_const): Use new_vector to install
33672 the vector in d.perm.
33673 * config/powerpcspe/powerpcspe.c (rs6000_expand_extract_even):
33674 Update vec_perm_builder constructor.
33675 (rs6000_expand_interleave): Likewise.
33676 * config/rs6000/rs6000.c (rs6000_expand_extract_even): Likewise.
33677 (rs6000_expand_interleave): Likewise.
33679 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33681 * optabs-query.c (can_vec_perm_var_p): Check whether lowering
33682 to qimode could truncate the indices.
33683 * optabs.c (expand_vec_perm_var): Likewise.
33685 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33687 * Makefile.in (OBJS): Add vec-perm-indices.o.
33688 * vec-perm-indices.h: New file.
33689 * vec-perm-indices.c: Likewise.
33690 * target.h (vec_perm_indices): Replace with a forward class
33692 (auto_vec_perm_indices): Move to vec-perm-indices.h.
33693 * optabs.h: Include vec-perm-indices.h.
33694 (expand_vec_perm): Delete.
33695 (selector_fits_mode_p, expand_vec_perm_var): Declare.
33696 (expand_vec_perm_const): Declare.
33697 * target.def (vec_perm_const_ok): Replace with...
33698 (vec_perm_const): ...this new hook.
33699 * doc/tm.texi.in (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Replace with...
33700 (TARGET_VECTORIZE_VEC_PERM_CONST): ...this new hook.
33701 * doc/tm.texi: Regenerate.
33702 * optabs.def (vec_perm_const): Delete.
33703 * doc/md.texi (vec_perm_const): Likewise.
33704 (vec_perm): Refer to TARGET_VECTORIZE_VEC_PERM_CONST.
33705 * expr.c (expand_expr_real_2): Use expand_vec_perm_const rather than
33706 expand_vec_perm for constant permutation vectors. Assert that
33707 the mode of variable permutation vectors is the integer equivalent
33708 of the mode that is being permuted.
33709 * optabs-query.h (selector_fits_mode_p): Declare.
33710 * optabs-query.c: Include vec-perm-indices.h.
33711 (selector_fits_mode_p): New function.
33712 (can_vec_perm_const_p): Check whether targetm.vectorize.vec_perm_const
33713 is defined, instead of checking whether the vec_perm_const_optab
33714 exists. Use targetm.vectorize.vec_perm_const instead of
33715 targetm.vectorize.vec_perm_const_ok. Check whether the indices
33716 fit in the vector mode before using a variable permute.
33717 * optabs.c (shift_amt_for_vec_perm_mask): Take a mode and a
33718 vec_perm_indices instead of an rtx.
33719 (expand_vec_perm): Replace with...
33720 (expand_vec_perm_const): ...this new function. Take the selector
33721 as a vec_perm_indices rather than an rtx. Also take the mode of
33722 the selector. Update call to shift_amt_for_vec_perm_mask.
33723 Use targetm.vectorize.vec_perm_const instead of vec_perm_const_optab.
33724 Use vec_perm_indices::new_expanded_vector to expand the original
33725 selector into bytes. Check whether the indices fit in the vector
33726 mode before using a variable permute.
33727 (expand_vec_perm_var): Make global.
33728 (expand_mult_highpart): Use expand_vec_perm_const.
33729 * fold-const.c: Includes vec-perm-indices.h.
33730 * tree-ssa-forwprop.c: Likewise.
33731 * tree-vect-data-refs.c: Likewise.
33732 * tree-vect-generic.c: Likewise.
33733 * tree-vect-loop.c: Likewise.
33734 * tree-vect-slp.c: Likewise.
33735 * tree-vect-stmts.c: Likewise.
33736 * config/aarch64/aarch64-protos.h (aarch64_expand_vec_perm_const):
33738 * config/aarch64/aarch64-simd.md (vec_perm_const<mode>): Delete.
33739 * config/aarch64/aarch64.c (aarch64_expand_vec_perm_const)
33740 (aarch64_vectorize_vec_perm_const_ok): Fuse into...
33741 (aarch64_vectorize_vec_perm_const): ...this new function.
33742 (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
33743 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
33744 * config/arm/arm-protos.h (arm_expand_vec_perm_const): Delete.
33745 * config/arm/vec-common.md (vec_perm_const<mode>): Delete.
33746 * config/arm/arm.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
33747 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
33748 (arm_expand_vec_perm_const, arm_vectorize_vec_perm_const_ok): Merge
33750 (arm_vectorize_vec_perm_const): ...this new function. Explicitly
33751 check for NEON modes.
33752 * config/i386/i386-protos.h (ix86_expand_vec_perm_const): Delete.
33753 * config/i386/sse.md (VEC_PERM_CONST, vec_perm_const<mode>): Delete.
33754 * config/i386/i386.c (ix86_expand_vec_perm_const_1): Update comment.
33755 (ix86_expand_vec_perm_const, ix86_vectorize_vec_perm_const_ok): Merge
33757 (ix86_vectorize_vec_perm_const): ...this new function. Incorporate
33758 the old VEC_PERM_CONST conditions.
33759 * config/ia64/ia64-protos.h (ia64_expand_vec_perm_const): Delete.
33760 * config/ia64/vect.md (vec_perm_const<mode>): Delete.
33761 * config/ia64/ia64.c (ia64_expand_vec_perm_const)
33762 (ia64_vectorize_vec_perm_const_ok): Merge into...
33763 (ia64_vectorize_vec_perm_const): ...this new function.
33764 * config/mips/loongson.md (vec_perm_const<mode>): Delete.
33765 * config/mips/mips-msa.md (vec_perm_const<mode>): Delete.
33766 * config/mips/mips-ps-3d.md (vec_perm_constv2sf): Delete.
33767 * config/mips/mips-protos.h (mips_expand_vec_perm_const): Delete.
33768 * config/mips/mips.c (mips_expand_vec_perm_const)
33769 (mips_vectorize_vec_perm_const_ok): Merge into...
33770 (mips_vectorize_vec_perm_const): ...this new function.
33771 * config/powerpcspe/altivec.md (vec_perm_constv16qi): Delete.
33772 * config/powerpcspe/paired.md (vec_perm_constv2sf): Delete.
33773 * config/powerpcspe/spe.md (vec_perm_constv2si): Delete.
33774 * config/powerpcspe/vsx.md (vec_perm_const<mode>): Delete.
33775 * config/powerpcspe/powerpcspe-protos.h (altivec_expand_vec_perm_const)
33776 (rs6000_expand_vec_perm_const): Delete.
33777 * config/powerpcspe/powerpcspe.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK):
33779 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
33780 (altivec_expand_vec_perm_const_le): Take each operand individually.
33781 Operate on constant selectors rather than rtxes.
33782 (altivec_expand_vec_perm_const): Likewise. Update call to
33783 altivec_expand_vec_perm_const_le.
33784 (rs6000_expand_vec_perm_const): Delete.
33785 (rs6000_vectorize_vec_perm_const_ok): Delete.
33786 (rs6000_vectorize_vec_perm_const): New function.
33787 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
33788 an element count and rtx array.
33789 (rs6000_expand_extract_even): Update call accordingly.
33790 (rs6000_expand_interleave): Likewise.
33791 * config/rs6000/altivec.md (vec_perm_constv16qi): Delete.
33792 * config/rs6000/paired.md (vec_perm_constv2sf): Delete.
33793 * config/rs6000/vsx.md (vec_perm_const<mode>): Delete.
33794 * config/rs6000/rs6000-protos.h (altivec_expand_vec_perm_const)
33795 (rs6000_expand_vec_perm_const): Delete.
33796 * config/rs6000/rs6000.c (TARGET_VECTORIZE_VEC_PERM_CONST_OK): Delete.
33797 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
33798 (altivec_expand_vec_perm_const_le): Take each operand individually.
33799 Operate on constant selectors rather than rtxes.
33800 (altivec_expand_vec_perm_const): Likewise. Update call to
33801 altivec_expand_vec_perm_const_le.
33802 (rs6000_expand_vec_perm_const): Delete.
33803 (rs6000_vectorize_vec_perm_const_ok): Delete.
33804 (rs6000_vectorize_vec_perm_const): New function. Remove stray
33805 reference to the SPE evmerge intructions.
33806 (rs6000_do_expand_vec_perm): Take a vec_perm_builder instead of
33807 an element count and rtx array.
33808 (rs6000_expand_extract_even): Update call accordingly.
33809 (rs6000_expand_interleave): Likewise.
33810 * config/sparc/sparc.md (vec_perm_constv8qi): Delete in favor of...
33811 * config/sparc/sparc.c (sparc_vectorize_vec_perm_const): ...this
33813 (TARGET_VECTORIZE_VEC_PERM_CONST): Redefine.
33815 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33817 * optabs.c (expand_vec_perm_1): Assert that SEL has an integer
33818 vector mode and that that mode matches the mode of the data
33820 (expand_vec_perm): Split handling of non-CONST_VECTOR selectors
33821 out into expand_vec_perm_var. Do all CONST_VECTOR handling here,
33822 directly using expand_vec_perm_1 when forcing selectors into
33824 (expand_vec_perm_var): New function, split out from expand_vec_perm.
33826 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33828 * optabs-query.h (can_vec_perm_p): Delete.
33829 (can_vec_perm_var_p, can_vec_perm_const_p): Declare.
33830 * optabs-query.c (can_vec_perm_p): Split into...
33831 (can_vec_perm_var_p, can_vec_perm_const_p): ...these two functions.
33832 (can_mult_highpart_p): Use can_vec_perm_const_p to test whether a
33833 particular selector is valid.
33834 * tree-ssa-forwprop.c (simplify_vector_constructor): Likewise.
33835 * tree-vect-data-refs.c (vect_grouped_store_supported): Likewise.
33836 (vect_grouped_load_supported): Likewise.
33837 (vect_shift_permute_load_chain): Likewise.
33838 * tree-vect-slp.c (vect_build_slp_tree_1): Likewise.
33839 (vect_transform_slp_perm_load): Likewise.
33840 * tree-vect-stmts.c (perm_mask_for_reverse): Likewise.
33841 (vectorizable_bswap): Likewise.
33842 (vect_gen_perm_mask_checked): Likewise.
33843 * fold-const.c (fold_ternary_loc): Likewise. Don't take
33844 implementations of variable permutation vectors into account
33845 when deciding which selector to use.
33846 * tree-vect-loop.c (have_whole_vector_shift): Don't check whether
33847 vec_perm_const_optab is supported; instead use can_vec_perm_const_p
33848 with a false third argument.
33849 * tree-vect-generic.c (lower_vec_perm): Use can_vec_perm_const_p
33850 to test whether the constant selector is valid and can_vec_perm_var_p
33851 to test whether a variable selector is valid.
33853 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33855 * optabs-query.h (can_vec_perm_p): Take a const vec_perm_indices *.
33856 * optabs-query.c (can_vec_perm_p): Likewise.
33857 * fold-const.c (fold_vec_perm): Take a const vec_perm_indices &
33858 instead of vec_perm_indices.
33859 * tree-vectorizer.h (vect_gen_perm_mask_any): Likewise,
33860 (vect_gen_perm_mask_checked): Likewise,
33861 * tree-vect-stmts.c (vect_gen_perm_mask_any): Likewise,
33862 (vect_gen_perm_mask_checked): Likewise,
33864 2018-01-02 Richard Sandiford <richard.sandiford@linaro.org>
33866 * optabs-query.h (qimode_for_vec_perm): Declare.
33867 * optabs-query.c (can_vec_perm_p): Split out qimode search to...
33868 (qimode_for_vec_perm): ...this new function.
33869 * optabs.c (expand_vec_perm): Use qimode_for_vec_perm.
33871 2018-01-02 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
33873 * rtlanal.c (canonicalize_condition): Return 0 if final rtx
33874 does not have a conditional at the top.
33876 2018-01-02 Richard Biener <rguenther@suse.de>
33878 * ipa-inline.c (big_speedup_p): Fix expression.
33880 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
33883 * config/i386/x86-tune-costs.h: Increase cost of integer load costs
33886 2018-01-02 Jan Hubicka <hubicka@ucw.cz>
33890 * x86-tune-costs.h (generic_cost): Reduce cost of FDIV 20->17,
33891 cost of sqrt 20->14, DIVSS 18->13, DIVSD 32->17, SQRtSS 30->14
33892 and SQRTsD 58->18, cond_not_taken_branch_cost. 2->1. Increase
33893 cond_taken_branch_cost 3->4.
33895 2018-01-01 Jakub Jelinek <jakub@redhat.com>
33897 PR tree-optimization/83581
33898 * tree-loop-distribution.c (pass_loop_distribution::execute): Return
33899 TODO_cleanup_cfg if any changes have been made.
33901 PR middle-end/83608
33902 * expr.c (store_expr_with_bounds): Use simplify_gen_subreg instead of
33903 convert_modes if target mode has the right side, but different mode
33906 PR middle-end/83609
33907 * expr.c (expand_assignment): Fix up a typo in simplify_gen_subreg
33908 last argument when extracting from CONCAT. If either from_real or
33909 from_imag is NULL, use expansion through memory. If result is not
33910 a CONCAT and simplify_gen_subreg fails, try to simplify_gen_subreg
33911 the parts directly to inner mode, if even that fails, use expansion
33914 PR middle-end/83623
33915 * expmed.c (expand_shift_1): For 2-byte rotates by BITS_PER_UNIT,
33916 check for bswap in mode rather than HImode and use that in expand_unop
33919 Copyright (C) 2018 Free Software Foundation, Inc.
33921 Copying and distribution of this file, with or without modification,
33922 are permitted in any medium without royalty provided the copyright
33923 notice and this notice are preserved.