1 2017-08-28 Richard Biener <rguenther@suse.de>
4 2017-08-01 Richard Biener <rguenther@suse.de>
6 PR tree-optimization/81181
7 * tree-ssa-pre.c (compute_antic_aux): Defer clean() to ...
8 (compute_antic): ... end of iteration here.
10 2017-08-08 Richard Biener <rguenther@suse.de>
12 PR tree-optimization/81723
13 * tree-vect-slp.c (struct bst_traits): New hash traits.
14 (bst_fail): New global.
15 (vect_build_slp_tree_2): New worker, split out from ...
16 (vect_build_slp_tree): ... this now wrapping it with using
17 bst_fail set to cache SLP tree build fails. Properly handle
19 (vect_analyze_slp_instance): Allocate and free bst_fail.
21 2017-08-24 Richard Biener <rguenther@suse.de>
24 * config/i386/i386.c: Include symbol-summary.h, ipa-prop.h
26 (ix86_can_inline_p): When ix86_fpmath flags do not match
27 check whether the callee uses FP math at all.
29 2017-08-23 Peter Bergner <bergner@vnet.ibm.com>
31 Backport from mainline
32 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
35 * config/rs6000/vsx.md (*vsx_le_permute_<mode>): Add support for
36 operands residing in integer registers.
37 (*vsx_le_perm_load_<mode>): Likewise.
38 (*vsx_le_perm_store_<mode>): Likewise.
39 (define_peephole2): Add peepholes to optimize the above.
41 2017-08-22 Peter Bergner <bergner@vnet.ibm.com>
43 Backport from mainline
44 2017-08-17 Peter Bergner <bergner@vnet.ibm.com>
47 * config/rs6000/rs6000.c (rs6000_activate_target_options): New function.
48 (rs6000_set_current_function): Rewrite function to use it.
50 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
52 Backport from mainline
53 2017-08-22 Sebastian Huber <sebastian.huber@embedded-brains.de>
55 * config.gcc (powerpc-*-rtems*): Add rs6000/linux64.opt.
56 * config/rs6000/rtems.h (ASM_PREFERRED_EH_DATA_FORMAT): New define.
57 (DOT_SYMBOLS): Likewise.
58 (MINIMAL_TOC_SECTION_ASM_OP): Likewise.
59 (RELOCATABLE_NEEDS_FIXUP): Likewise.
60 (RS6000_ABI_NAME): Likewise.
61 (TARGET_CMODEL): Likewise.
62 (TOC_SECTION_ASM_OP): Likewise.
63 (SET_CMODEL): New macro.
64 (SUBSUBTARGET_OVERRIDE_OPTIONS): Evaluate cmodel options.
66 2017-08-22 Georg-Johann Lay <avr@gjlay.de>
68 Backport from 2017-08-22 trunk r251256.
71 * config/avr/avr.c (avr_handle_addr_attribute): Early return if
72 not VAR_P. Filter attribute warnings with OPT_Wattributes.
73 (avr_attribute_table) <io, io_low, address>: Initialize
74 .decl_required with true.
76 2017-08-21 Georg-Johann Lay <avr@gjlay.de>
79 * config/avr/avr.c (avr_set_current_function): Typo in diagnostic.
81 2017-08-19 Uros Bizjak <ubizjak@gmail.com>
84 * doc/extend.texi (x86 Built-in Functions): Correct the name of
85 __builtin_ia32_lzcnt_u16.
87 2017-08-17 Uros Bizjak <ubizjak@gmail.com>
89 Backport from mainline
90 2017-08-17 Maxim Ostapenko <m.ostapenko@samsung.com>
93 * config/i386/i386.c (ix86_option_override_internal): Save target
94 specific options after ix86_stack_protector_guard_reg was changed.
96 2017-08-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
98 Backport from mainline
99 2017-08-08 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
101 PR tree-optimization/81354
102 * gimple-ssa-strength-reduction.c (create_add_on_incoming_edge):
103 Insert on edges rather than explicitly creating landing pads.
104 (analyze_candidates_and_replace): Commit edge inserts.
106 2017-08-15 Joseph Myers <joseph@codesourcery.com>
110 * config/sh/sh-mem.cc (sh_expand_cmpnstr): Only unroll for
111 constant count if that count is less than 32.
113 2017-08-14 Richard Biener <rguenther@suse.de>
115 * BASE-VER: Set to 7.2.1.
117 2017-08-14 Release Manager
119 * GCC 7.2.0 released.
121 2017-08-08 Richard Biener <rguenther@suse.de>
124 * function.c (thread_prologue_and_epilogue_insns): Restore
125 behavior of always calling find_many_sub_basic_blocks on
126 the inserted prologue.
128 2017-08-02 Jakub Jelinek <jakub@redhat.com>
131 * function.c (thread_prologue_and_epilogue_insns): Determine blocks
132 for find_many_sub_basic_blocks bitmap by looking up BLOCK_FOR_INSN
133 of first NONDEBUG_INSN_P in each of the split_prologue_seq and
134 prologue_seq sequences - if any.
136 2017-08-01 Uros Bizjak <ubizjak@gmail.com>
139 * config/i386/i386.c (ix86_print_operand_address_as): For -masm=intel
140 print "ds:" only for immediates in generic address space.
142 2017-08-01 Jakub Jelinek <jakub@redhat.com>
145 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): For
146 __builtin_vec_cmpne verify both arguments are compatible vectors
147 before looking at TYPE_MODE on the element type. For __builtin_vec_ld
148 verify arg1_type is a pointer or array type. For __builtin_vec_st,
149 move computation of aligned to after checking the argument types.
152 2017-08-01 Martin Liska <mliska@suse.cz>
154 Backport from mainline
155 2017-07-26 Martin Liska <mliska@suse.cz>
157 PR gcov-profile/81561
158 * gcov.c (unblock): Make unblocking safe as we need to preserve
159 index correspondence of blocks and block_lists.
161 2017-08-01 Richard Biener <rguenther@suse.de>
163 PR tree-optimization/71752
164 PR tree-optimization/81633
165 * tree-vect-slp.c (vect_get_slp_defs): Handle null operands
166 in the original suggested way.
168 2017-08-01 Richard Sandiford <richard.sandiford@linaro.org>
170 PR tree-optimization/80769
171 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
172 for malloc and calloc. Document the new invariant that all related
173 strinfos have delayed lengths or none do.
174 (get_next_strinfo): New function.
175 (verify_related_strinfos): Move earlier in file.
176 (set_endptr_and_length): New function, split out from...
177 (get_string_length): ...here. Also set the lengths of related
180 2017-08-01 Jakub Jelinek <jakub@redhat.com>
182 PR tree-optimization/81588
183 * tree-ssa-reassoc.c (optimize_range_tests_var_bound): If
184 ranges[i].in_p, invert comparison code ccode. For >/>=,
185 swap rhs1 and rhs2 and comparison code unconditionally,
186 for </<= don't do that. Don't swap rhs1/rhs2 again if
187 ranges[i].in_p, instead invert comparison code ccode if
188 opcode or oe->rank is BIT_IOR_EXPR.
190 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
192 Backport from mainline
193 2017-07-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
195 * config.gcc: Add z14.
196 * config/s390/driver-native.c (s390_host_detect_local_cpu): Add
197 CPU model numbers for z13s and z14.
198 * config/s390/s390-c.c (s390_resolve_overloaded_builtin): Replace
200 * config/s390/s390-opts.h (enum processor_type): Rename
201 PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
202 * config/s390/s390.c (processor_table): Add field for CPU name to
203 be passed to Binutils.
204 (s390_asm_output_machine_for_arch): Use the new field in
205 processor_table for Binutils.
206 (s390_expand_builtin): Replace arch12 with z14.
207 (s390_issue_rate): Rename PROCESSOR_ARCH12 to PROCESSOR_3906_Z14.
208 (s390_get_sched_attrmask): Likewise.
209 (s390_get_unit_mask): Likewise.
210 * config/s390/s390.opt: Add z14 to processor_type enum.
212 2017-07-31 Jakub Jelinek <jakub@redhat.com>
215 * ubsan.c (ubsan_type_descriptor): For UBSAN_PRINT_ARRAY don't
216 change type to the element type, instead add eltype variable and
217 use it where we are interested in the element type.
219 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
221 Backport from mainline
222 2017-07-28 Peter Bergner <bergner@vnet.ibm.com>
224 * config/rs6000/ppc-auxv.h (PPC_FEATURE2_DARN): New define.
225 (PPC_FEATURE2_SCV): Likewise.
226 * config/rs6000/rs6000.c (cpu_supports_info): Use them.
228 2017-07-28 David Edelsohn <dje.gcc@gmail.com>
230 Backport from mainline
231 2017-07-25 David Edelsohn <dje.gcc@gmail.com>
233 * dwarf2asm.c (dw2_asm_output_nstring): Encode double quote
235 * dwarf2out.c (output_macinfo): Copy debug_line_section_label
236 to dl_section_ref. On AIX, append an expression to subtract
237 the size of the section length to dl_section_ref.
239 2017-07-28 Bin Cheng <bin.cheng@arm.com>
241 Backport from mainline r250496
242 2017-07-25 Bin Cheng <bin.cheng@arm.com>
245 * config/aarch64/cortex-a57-fma-steering.c (analyze): Skip fmul/fmac
246 instructions if no du chain is found.
248 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
250 Backport from mainline
251 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
253 * config.gcc (powerpc-*-rtems*): Remove rs6000/eabi.h. Add
255 * config/rs6000/rtems.h (ASM_DECLARE_FUNCTION_SIZE): New macro.
256 (ASM_OUTPUT_SPECIAL_POOL_ENTRY_P): Likewise.
257 (CRT_CALL_STATIC_FUNCTION): Likewise.
258 (ASM_DEFAULT_SPEC): New define.
259 (ASM_SPEC32): Likewise.
260 (ASM_SPEC64): Likewise.
261 (ASM_SPEC_COMMON): Likewise.
262 (ASM_SPEC): Likewise.
263 (INVALID_64BIT): Likewise.
264 (LINK_OS_DEFAULT_SPEC): Likewise.
265 (LINK_OS_SPEC32): Likewise.
266 (LINK_OS_SPEC64): Likewise.
267 (POWERPC_LINUX): Likewise.
268 (PTRDIFF_TYPE): Likewise.
269 (RESTORE_FP_PREFIX): Likewise.
270 (RESTORE_FP_SUFFIX): Likewise.
271 (SAVE_FP_PREFIX): Likewise.
272 (SAVE_FP_SUFFIX): Likewise.
273 (SIZE_TYPE): Likewise.
274 (SUBSUBTARGET_OVERRIDE_OPTIONS): Likewise.
275 (TARGET_64BIT): Likewise.
276 (TARGET_64BIT): Likewise.
277 (TARGET_AIX): Likewise.
278 (WCHAR_TYPE_SIZE): Likewise.
279 (WCHAR_TYPE): Undefine.
280 (TARGET_OS_CPP_BUILTINS): Add 64-bit PowerPC defines.
281 (CPP_OS_DEFAULT_SPEC): Use previous CPP_OS_RTEMS_SPEC.
282 (CPP_OS_RTEMS_SPEC): Delete.
283 (SUBSUBTARGET_EXTRA_SPECS): Remove cpp_os_rtems. Add
284 asm_spec_common, asm_spec32, asm_spec64, link_os_spec32, and
286 * config/rs6000/t-rtems: Add mcpu=e6500/m64 multilibs.
288 2017-07-28 Sebastian Huber <sebastian.huber@embedded-brains.de>
290 Backport from mainline
291 2017-07-27 Sebastian Huber <sebastian.huber@embedded-brains.de>
293 * config.gcc (riscv*-*-elf*): Add (riscv*-*-rtems*).
294 * config/riscv/rtems.h: New file.
296 2017-07-27 Eric Botcazou <ebotcazou@adacore.com>
298 * config/sparc/sparc.c (sparc_option_override): Set MASK_FSMULD flag
299 earlier and only if MASK_FPU is set. Adjust formatting.
301 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
303 Backport from mainline
304 2017-07-27 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
307 * config/s390/s390.md ("*atomic_compare_and_swap<mode>_1")
308 ("*atomic_compare_and_swapdi_2", "*atomic_compare_and_swapsi_3"):
309 Change s_operand to memory_operand.
311 2017-07-27 Jakub Jelinek <jakub@redhat.com>
313 PR tree-optimization/81555
314 PR tree-optimization/81556
315 * tree-ssa-reassoc.c (rewrite_expr_tree): Add NEXT_CHANGED argument,
316 if true, force CHANGED for the recursive invocation.
317 (reassociate_bb): Remember original length of ops array, pass
318 len != orig_len as NEXT_CHANGED in rewrite_expr_tree call.
320 2017-07-27 Martin Liska <mliska@suse.cz>
322 Backport from mainline
323 2017-07-17 Martin Liska <mliska@suse.cz>
326 * opts.c (finish_options): Do not allow -fgnu-tm
327 w/ -fsanitize={kernel-,}address. Say sorry.
329 2017-07-27 Martin Liska <mliska@suse.cz>
331 Backport from mainline
332 2017-07-26 Martin Liska <mliska@suse.cz>
335 * function.c (expand_function_start): Make expansion of
336 nonlocal_goto_save_area after parm_birth_insn.
338 2017-07-27 Martin Liska <mliska@suse.cz>
340 Backport from mainline
341 2017-06-30 Martin Liska <mliska@suse.cz>
344 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
345 before BUILT_IN_UNWIND_RESUME when ASAN is used.
347 2017-07-27 Martin Liska <mliska@suse.cz>
349 Backport from mainline
350 2017-06-28 Martin Liska <mliska@suse.cz>
353 * asan.c (instrument_derefs): Bail out inner references
354 that are hard register variables.
356 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
358 Backport from mainline
359 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
361 * config/sparc/sparc.c (dump_target_flag_bits): Dump MASK_FSMULD.
362 (sparc_option_override): Honour MASK_FSMULD.
363 * config/sparc/sparc.h (MASK_FEATURES): Add MASK_FSMULD.
364 * config/sparc/sparc.md (muldf3_extend): Use TARGET_FSMULD.
365 * config/sparc/sparc.opt (mfsmuld): New option.
366 * doc/invoke.texi (mfsmuld): Document option.
368 2017-07-26 Georg-Johann Lay <avr@gjlay.de>
370 Backport from 2017-07-25 trunk r250499.
373 * hsa-brig.c (brig_init): Use xasprintf instead of asprintf.
374 * gimple-pretty-print.c (dump_probability): Same.
375 * tree-ssa-structalias.c (alias_get_name): Same.
377 2017-07-26 Richard Biener <rguenther@suse.de>
379 Backport from mainline
380 2017-06-18 Richard Biener <rguenther@suse.de>
382 PR tree-optimization/81410
383 * tree-vect-stmts.c (vectorizable_load): Properly adjust for
384 the gap in the ! slp_perm SLP case after each group.
386 2017-07-25 Richard Biener <rguenther@suse.de>
388 PR tree-optimization/81455
389 * tree-ssa-loop-unswitch.c (find_loop_guard): Make sure to
390 not walk in cycles when looking for guards.
392 2017-07-25 Richard Biener <rguenther@suse.de>
395 * fold-const.c (fold_negate_const): TREE_OVERFLOW should be
398 2017-06-28 Jakub Jelinek <jakub@redhat.com>
401 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
402 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
404 2017-06-26 Richard Biener <rguenther@suse.de>
407 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
408 Use def_builtin_pure for all gather builtins.
410 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
412 * config/i386/i386.c (struct builtin_isa): New field pure_p.
413 Reorder for compactness.
414 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
415 (def_builtin_pure, def_builtin_pure2): New functions.
416 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
418 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
420 Backport from mainline
421 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
423 * config/sparc/sparc.c (sparc_option_override): Remove MASK_FPU
424 from all CPU target flags enable members.
426 2017-07-26 Sebastian Huber <sebastian.huber@embedded-brains.de>
428 Backport from mainline
429 2017-07-25 Sebastian Huber <sebastian.huber@embedded-brains.de>
432 * config/aarch64/rtems.h: Add GCC Runtime Library Exception.
434 * config/arm/rtems.h: Likewise.
435 * config/bfin/rtems.h: Likewise.
436 * config/i386/rtemself.h: Likewise.
437 * config/lm32/rtems.h: Likewise.
438 * config/m32c/rtems.h: Likewise.
439 * config/m68k/rtemself.h: Likewise.
440 * config/microblaze/rtems.h: Likewise.
441 * config/mips/rtems.h: Likewise.
442 * config/moxie/rtems.h: Likewise.
443 * config/nios2/rtems.h: Likewise.
444 * config/rs6000/rtems.h: Likewise.
445 * config/rtems.h: Likewise.
446 * config/sh/rtems.h: Likewise.
447 * config/sh/rtemself.h: Likewise.
448 * config/sparc/rtemself.h: Likewise.
450 2017-07-25 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
452 Backport from mainline
453 2017-07-14 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
455 PR tree-optimization/81162
456 * gimple-ssa-strength-reduction.c (replace_mult_candidate): Don't
457 replace a negate with an add.
459 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
461 Backport from 2017-07-12 trunk r250151.
464 * config/avr/avr.c (avr_encode_section_info)
465 [progmem && !TREE_READONLY]: Error if progmem object needs
468 2017-07-25 Wilco Dijkstra <wdijkstr@arm.com>
471 * config/aarch64/aarch64.c (aarch64_classify_symbol):
472 Avoid SYMBOL_SMALL_ABSOLUTE for literals with pc-relative literals.
474 2017-07-25 Georg-Johann Lay <avr@gjlay.de>
476 Backport from trunk r247719.
478 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
480 PR rtl-optimization/75964
481 * simplify-rtx.c (simplify_const_relational_operation): Remove
482 invalid handling of comparisons of integer ABS.
484 2017-07-25 Bin Cheng <bin.cheng@arm.com>
486 Backport from 2017-07-20 trunk r250384.
488 PR tree-optimization/81388
490 2016-07-21 Bin Cheng <bin.cheng@arm.com>
492 * tree-ssa-loop-niter.c (number_of_iterations_lt_to_ne): Clean up
493 by removing computation of may_be_zero.
495 2017-07-23 Uros Bizjak <ubizjak@gmail.com>
498 * config/i386/i386.c (ix86_option_override_internal): Disable
499 BMI, BMI2 and TBM instructions for -m16.
501 2017-07-19 Michael Meissner <meissner@linux.vnet.ibm.com>
504 2017-07-12 Michael Meissner <meissner@linux.vnet.ibm.com>
507 * config/rs6000/rs6000-c.c (rs6000_cpu_cpp_builtins): If GLIBC
508 provides the hardware capability bits, define the macro
509 __BUILTIN_CPU_SUPPORTS__.
510 * config/rs6000/rs6000.c (cpu_expand_builtin): Generate a warning
511 if GLIBC does not provide the hardware capability bits. Add a
512 gcc_unreachable call if the built-in cpu function is neither
513 __builtin_cpu_is nor __builtin_cpu_supports.
514 * doc/extend.texi (PowerPC built-in functions): Document that
515 GLIBC 2.23 or newer is needed by __builtin_cpu_is and
516 __builtin_cpu_supports. Document the macros defined by GCC if the
517 newer GLIBC is available.
519 2017-07-18 Uros Bizjak <ubizjak@gmail.com>
522 * config/i386/i386.md (rorx_immediate_operand): New mode attribute.
523 (*bmi2_rorx<mode>3_1): Use rorx_immediate_operand as
525 (*bmi2_rorxsi3_1_zext): Use const_0_to_31_operand as
527 (ror,rol -> rorx splitters): Use const_int_operand as
530 2017-07-18 Tom de Vries <tom@codesourcery.com>
532 backport from mainline:
534 2017-07-17 Tom de Vries <tom@codesourcery.com>
536 * config/nvptx/nvptx.c (nvptx_single): Insert diverging branch as late
539 2017-07-18 Georg-Johann Lay <avr@gjlay.de>
541 Backport from 2017-07-18 trunk r250301.
544 * config/avr/avr.c (avr_optimize_casesi): Don't use
545 INT8_MIN, INT8_MAX, UINT8_MAX, INT16_MIN, INT16_MAX, UINT16_MAX.
547 2017-07-17 Jakub Jelinek <jakub@redhat.com>
549 PR tree-optimization/81428
550 * match.pd (X / X -> one): Don't optimize _Fract divisions, as 1
551 can't be built for those types.
553 PR tree-optimization/81365
554 * tree-ssa-phiprop.c (propagate_with_phi): When considering hoisting
555 aggregate moves onto bb predecessor edges, make sure there are no
556 loads that could alias the lhs in between the start of bb and the
559 Backported from mainline
560 2017-06-30 Jakub Jelinek <jakub@redhat.com>
563 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For
564 V8FI, V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
565 of nonimmediate_operand and <store_mask_constraint> instead of m for
566 the input operand. For V8FI iterator, always split if input is a MEM.
567 For V16FI and V8SF_256 iterators, don't test if both operands are MEM
568 if <mask_applied>. For VI4F_256 iterator, use <store_mask_predicate>
569 instead of register_operand and <store_mask_constraint> instead of v for
570 the input operand. Make sure both operands aren't MEMs for if not
573 2017-07-17 Georg-Johann Lay <avr@gjlay.de>
575 Backport from 2017-07-17 trunk r250258.
578 * config/avr/avr.c (avr_mul_highpart_cost): New static function.
579 (avr_rtx_costs_1) [TRUNCATE]: Use it to compute mul_highpart cost.
580 [LSHIFTRT, outer_code = TRUNCATE]: Same.
582 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
584 Backport from mainline
585 2017-07-17 Sebastian Huber <sebastian.huber@embedded-brains.de>
587 * gcc/config/sparc/rtemself.h (TARGET_OS_CPP_BUILTINS): Add
588 conditional builtin define __FIX_LEON3FT_B2BST.
590 2017-07-17 Daniel Cederman <cederman@gaisler.com>
592 Backport from mainline
593 2017-07-17 Daniel Cederman <cederman@gaisler.com>
595 * config/sparc/t-rtems: Add mfix-gr712rc multilibs. Replace
596 MULTILIB_EXCEPTIONS with MULTILIB_REQUIRED. Match -mfix-gr712rc
599 2017-07-16 Eric Botcazou <ebotcazou@adacore.com>
601 PR rtl-optimization/81424
602 * optabs.c (prepare_cmp_insn): Use copy_to_reg instead of force_reg
603 to remove potential trapping from operands if -fnon-call-exceptions.
605 2017-07-16 Daniel Cederman <cederman@gaisler.com>
607 * config/sparc/sparc.md (divdf3_fix): Add NOP to prevent back
608 to back store errata sensitive sequence from being generated.
609 (sqrtdf2_fix): Likewise.
611 2017-07-12 Georg-Johann Lay <avr@gjlay.de>
613 Backport from 2017-07-12 trunk r250156.
616 * config/avr/avr.c (avr_set_current_function): In diagnostic
617 messages: Quote keywords and (parts of) identifiers.
618 [WITH_AVRLIBC]: Warn for functions named "ISR", "SIGNAL" or
621 2017-07-11 Daniel Cederman <cederman@gaisler.com>
623 * config/sparc/sparc.opt (mfix-ut700): New option.
624 (mfix-gr712rc): Likewise.
625 (sparc_fix_b2bst): New variable.
626 * doc/invoke.texi (SPARC options): Document them.
627 (ARM options): Fix warnings.
628 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
629 instructions to prevent sequences that can trigger the store-store
630 errata for certain LEON3FT processors.
631 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
632 (sparc_option_override): Set sparc_fix_b2bst appropriately.
633 * config/sparc/sparc.md (fix_b2bst): New attribute.
634 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
636 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
639 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
641 (*rsqrtsf2_sse): Ditto.
643 (div<mode>3): Macroize insn from divdf3 and divsf3
644 using MODEF mode iterator.
646 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
648 Backport from mainline
649 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
652 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
653 correct operand in doing the split.
655 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
657 * config/sparc/m8.md: New file.
658 * config/sparc/sparc.md: Include m8.md.
660 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
662 * config/sparc/sparc.opt: New option -mvis4b.
663 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
664 (sparc_option_override): Handle VIS4B.
665 (enum sparc_builtins): Define
666 SPARC_BUILTIN_DICTUNPACK{8,16,32},
667 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
668 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
669 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
670 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
671 (check_constant_argument): New function.
672 (sparc_vis_init_builtins): Define builtins
673 __builtin_vis_dictunpack{8,16,32},
674 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
675 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
676 __builtin_vis_fpcmpde{8,16,32}shl and
677 __builtin_vis_fpcmpur{8,16,32}shl.
678 (sparc_expand_builtin): Check that the constant operands to
679 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
680 constant and in range.
681 * config/sparc/sparc-c.c (sparc_target_macros): Handle
683 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
684 (SPARC_IMM5_P): Likewise.
685 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
686 (enabled): Handle vis4b.
687 (UNSPEC_DICTUNPACK): New unspec.
688 (UNSPEC_FPCMPSHL): Likewise.
689 (UNSPEC_FPUCMPSHL): Likewise.
690 (UNSPEC_FPCMPDESHL): Likewise.
691 (UNSPEC_FPCMPURSHL): Likewise.
692 (cpu_feature): New CPU feature `vis4b'.
693 (dictunpack{8,16,32}): New insns.
694 (FPCSMODE): New mode iterator.
695 (fpcscond): New code iterator.
696 (fpcsucond): Likewise.
697 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
698 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
699 (fpcmpde{8,16,32}{si,di}shl): Likewise.
700 (fpcmpur{8,16,32}{si,di}shl): Likewise.
701 * config/sparc/constraints.md: Define constraints `q' for unsigned
702 2-bit integer constants and `t' for unsigned 5-bit integer
704 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
706 (imm5_operand_dictunpack16): Likewise.
707 (imm5_operand_dictunpack32): Likewise.
708 (imm2_operand): Likewise.
709 * doc/invoke.texi (SPARC Options): Document -mvis4b.
710 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
711 ditunpack* and fpcmp*shl builtins.
713 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
715 * config.gcc: Handle m8 in --with-{cpu,tune} options.
716 * config.in: Add HAVE_AS_SPARC6 define.
717 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
719 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
721 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
722 (CPP_CPU_SPEC): Handle m8.
723 (ASM_CPU_SPEC): Likewise.
724 * config/sparc/sparc-opts.h (enum processor_type): Add
726 * config/sparc/sparc.c (m8_costs): New struct.
727 (sparc_option_override): Handle TARGET_CPU_m8.
728 (sparc32_initialize_trampoline): Likewise.
729 (sparc64_initialize_trampoline): Likewise.
730 (sparc_issue_rate): Likewise.
731 (sparc_register_move_cost): Likewise.
732 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
733 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
734 (ASM_CPU64_DEFAULT_SPEC): Likewise.
735 (CPP_CPU_SPEC): Handle M8.
736 (ASM_CPU_SPEC): Likewise.
737 (AS_M8_FLAG): Define.
738 * config/sparc/sparc.md: Add m8 to the cpu attribute.
739 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
740 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
742 * configure: Regenerate.
743 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
746 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
748 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
750 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
751 ("*movdi_insn_sp32"): Do not set v3pipe.
752 ("*movsi_insn"): Likewise.
753 ("*movdi_insn_sp64"): Likewise.
754 ("*movsf_insn"): Likewise.
755 ("*movdf_insn_sp32"): Likewise.
756 ("*movdf_insn_sp64"): Likewise.
757 ("*zero_extendsidi2_insn_sp64"): Likewise.
758 ("*sign_extendsidi2_insn"): Likewise.
759 ("*mov<VM32:mode>_insn"): Likewise.
760 ("*mov<VM64:mode>_insn_sp64"): Likewise.
761 ("*mov<VM64:mode>_insn_sp32"): Likewise.
762 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
763 ("<vlop:code><VL:mode>3"): Likewise.
764 ("*not_<vlop:code><VL:mode>3"): Likewise.
765 ("*nand<VL:mode>_vis"): Likewise.
766 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
767 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
768 ("one_cmpl<VL:mode>2"): Likewise.
769 ("faligndata<VM64:mode>_vis"): Likewise.
770 ("alignaddrsi_vis"): Likewise.
771 ("alignaddrdi_vis"): Likweise.
772 ("alignaddrlsi_vis"): Likewise.
773 ("alignaddrldi_vis"): Likewise.
774 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
775 ("bmaskdi_vis"): Likewise.
776 ("bmasksi_vis"): Likewise.
777 ("bshuffle<VM64:mode>_vis"): Likewise.
778 ("cmask8<P:mode>_vis"): Likewise.
779 ("cmask16<P:mode>_vis"): Likewise.
780 ("cmask32<P:mode>_vis"): Likewise.
781 ("pdistn<P:mode>_vis"): Likewise.
782 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
784 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
786 * config/sparc/sparc.md ("subtype"): New insn attribute.
787 ("*wrgsr_sp64"): Set insn subtype.
788 ("*rdgsr_sp64"): Likewise.
789 ("alignaddrsi_vis"): Likewise.
790 ("alignaddrdi_vis"): Likewise.
791 ("alignaddrlsi_vis"): Likewise.
792 ("alignaddrldi_vis"): Likewise.
793 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
794 ("fexpand_vis"): Likewise.
795 ("fpmerge_vis"): Likewise.
796 ("faligndata<VM64:mode>_vis"): Likewise.
797 ("bshuffle<VM64:mode>_vis"): Likewise.
798 ("cmask8<P:mode>_vis"): Likewise.
799 ("cmask16<P:mode>_vis"): Likewise.
800 ("cmask32<P:mode>_vis"): Likewise.
801 ("fchksm16_vis"): Likewise.
802 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
803 ("fmean16_vis"): Likewise.
804 ("fp<plusminus_insn>64_vis"): Likewise.
805 ("<plusminus_insn>v8qi3"): Likewise.
806 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
807 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
808 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
809 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
810 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
811 ("*movqi_insn"): Likewise.
812 ("*movhi_insn"): Likewise.
813 ("*movsi_insn"): Likewise.
814 ("movsi_pic_gotdata_op"): Likewise.
815 ("*movdi_insn_sp32"): Likewise.
816 ("*movdi_insn_sp64"): Likewise.
817 ("movdi_pic_gotdata_op"): Likewise.
818 ("*movsf_insn"): Likewise.
819 ("*movdf_insn_sp32"): Likewise.
820 ("*movdf_insn_sp64"): Likewise.
821 ("*zero_extendhisi2_insn"): Likewise.
822 ("*zero_extendqihi2_insn"): Likewise.
823 ("*zero_extendqisi2_insn"): Likewise.
824 ("*zero_extendqidi2_insn"): Likewise.
825 ("*zero_extendhidi2_insn"): Likewise.
826 ("*zero_extendsidi2_insn_sp64"): Likewise.
828 ("prefetch_64"): Likewise.
829 ("prefetch_32"): Likewise.
830 ("tie_ld32"): Likewise.
831 ("tie_ld64"): Likewise.
832 ("*tldo_ldub_sp32"): Likewise.
833 ("*tldo_ldub1_sp32"): Likewise.
834 ("*tldo_ldub2_sp32"): Likewise.
835 ("*tldo_ldub_sp64"): Likewise.
836 ("*tldo_ldub1_sp64"): Likewise.
837 ("*tldo_ldub2_sp64"): Likewise.
838 ("*tldo_ldub3_sp64"): Likewise.
839 ("*tldo_lduh_sp32"): Likewise.
840 ("*tldo_lduh1_sp32"): Likewise.
841 ("*tldo_lduh_sp64"): Likewise.
842 ("*tldo_lduh1_sp64"): Likewise.
843 ("*tldo_lduh2_sp64"): Likewise.
844 ("*tldo_lduw_sp32"): Likewise.
845 ("*tldo_lduw_sp64"): Likewise.
846 ("*tldo_lduw1_sp64"): Likewise.
847 ("*tldo_ldx_sp64"): Likewise.
848 ("*mov<VM32:mode>_insn"): Likewise.
849 ("*mov<VM64:mode>_insn_sp64"): Likewise.
850 ("*mov<VM64:mode>_insn_sp32"): Likewise.
852 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
854 * config/sparc/sparc.md ("type"): New insn type viscmp.
855 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
857 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
858 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
859 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
860 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
862 ("n7_vis_logical_11cycle"): Likewise.
863 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
864 * config/sparc/niagara2.md ("niag3_vis": Likewise.
865 * config/sparc/niagara.md ("niag_vis"): Likewise.
866 * config/sparc/ultra3.md ("us3_fga"): Likewise.
867 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
869 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
871 * config/sparc/sparc.md: New instruction type `bmask'.
872 (bmaskdi_vis): Use the `bmask' type.
873 (bmasksi_vis): Likewise.
874 * config/sparc/ultra3.md (us3_array): Likewise.
875 * config/sparc/niagara7.md (n7_array): Likewise.
876 * config/sparc/niagara4.md (n4_array): Likewise.
877 * config/sparc/niagara2.md (niag2_vis): Likewise.
878 (niag3_vis): Likewise.
879 * config/sparc/niagara.md (niag_vis): Likewise.
881 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
883 Backport from 2017-07-05 trunk r249995.
886 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
887 Don't depend on "optimize > 0".
888 (out_movhi_r_mr, out_movqi_mr_r): Same.
889 (out_movhi_mr_r, out_movqi_r_mr): Same.
890 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
891 io_address_operand on "optimize > 0".
893 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
896 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
897 Require dead FLAGS_REG at the beginning of a peephole.
899 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
902 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
903 arguments in the call to __builtin_ia32_sbb_u32.
904 (_subborrow_u64): Swap _X and _Y arguments in the call to
905 __builtin_ia32_sbb_u64.
907 2017-07-03 Segher Boessenkool <segher@kernel.crashing.org>
911 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
913 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
916 2017-07-03 Tom de Vries <tom@codesourcery.com>
918 backport from mainline:
919 PR tree-optimization/81192
920 2017-07-03 Tom de Vries <tom@codesourcery.com>
922 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
923 BB_SAME_SUCC (bb) == NULL.
925 2017-06-29 Michael Meissner <meissner@linux.vnet.ibm.com>
927 Backport from mainline
928 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
931 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
932 32-bit, since indexed is not valid for DImode.
933 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
934 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
935 (define_peephole2 for Altivec d-form load): Add 32-bit support.
936 (define_peephole2 for Altivec d-form store): Likewise.
938 Backport from mainline
939 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
942 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
943 for doing vector set of SFmode on ISA 3.0.
944 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
945 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
947 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
948 SFmode value into a V4SF variable that was extracted from another
949 V4SF variable without converting the element to double precision
950 and back to single precision vector format.
951 (vsx_insert_extract_v4sf_p9_2): Likewise.
953 2017-06-29 Richard Biener <rguenther@suse.de>
955 Backport from mainline
956 2017-06-19 Richard Biener <rguenther@suse.de>
959 * ipa-prop.c (find_constructor_constant_at_offset): Handle
960 RANGE_EXPR conservatively.
962 2017-06-28 Richard Biener <rguenther@suse.de>
964 Backport from mainline
965 2017-06-09 Richard Biener <rguenther@suse.de>
968 * ipa-polymorphic-call.c
969 (ipa_polymorphic_call_context::restrict_to_inner_class):
970 Skip FIELD_DECLs with error_mark_node type.
971 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
974 2017-06-14 Richard Biener <rguenther@suse.de>
976 PR tree-optimization/81083
977 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
980 2017-06-27 Segher Boessenkool <segher@kernel.crashing.org>
982 Backports from trunk:
984 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
986 * real.c (do_compare): Give decimal_do_compare preference over
987 comparing just the signs.
989 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
991 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
992 splitter result in the canonical way.
994 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
996 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
997 gen_add3_insn did not fail.
998 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
999 r0, construct that number in a temporary reg and add that reg to r0.
1000 If asked to put the result in r0 as well, fail.
1002 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
1004 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
1005 a call, force the call to not be a tail call.
1007 2017-06-27 Jakub Jelinek <jakub@redhat.com>
1010 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
1013 * gimple-fold.c (replace_call_with_call_and_fold): Handle
1014 gimple_vuse copying separately from gimple_vdef copying.
1016 2017-06-24 Jim Wilson <jim.wilson@linaro.org>
1018 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
1020 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
1021 * config/arm/arm-cpu-cdata.h: Regenerate.
1022 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
1023 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
1024 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
1025 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
1026 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
1028 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
1030 * config/arm/t-rmprofile: Likewise.
1031 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
1033 2017-06-24 Marek Polacek <polacek@redhat.com>
1035 Backport from mainline
1036 2017-05-04 Marek Polacek <polacek@redhat.com>
1038 PR tree-optimization/80612
1039 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
1041 2017-06-23 Thomas Preud'homme <thomas.preudhomme@arm.com>
1043 Backport from mainline
1044 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
1046 * config/arm/arm-builtins.c (arm_init_builtins): Rename
1047 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
1048 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
1050 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
1053 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
1056 2017-06-22 Martin Liska <mliska@suse.cz>
1058 Backport from mainline
1059 2017-06-19 Martin Liska <mliska@suse.cz>
1062 * gimplify.c (gimplify_switch_expr):
1063 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
1065 2017-06-22 Martin Liska <mliska@suse.cz>
1067 Backport from mainline
1068 2017-05-31 Martin Liska <mliska@suse.cz>
1071 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
1073 2017-06-22 Martin Liska <mliska@suse.cz>
1075 Backport from mainline
1076 2017-05-30 Martin Liska <mliska@suse.cz>
1079 * auto-profile.c (get_function_decl_from_block): Fix
1082 2017-06-22 Martin Liska <mliska@suse.cz>
1084 Backport from mainline
1085 2017-05-26 Martin Liska <mliska@suse.cz>
1088 * params.def: Bound partial-inlining-entry-probability param.
1090 2017-06-22 Martin Liska <mliska@suse.cz>
1092 Backport from mainline
1093 2017-05-16 Martin Liska <mliska@suse.cz>
1097 * ipa-devirt.c (warn_types_mismatch): Fix typo.
1098 (odr_types_equivalent_p): Likewise.
1100 2017-06-22 Martin Liska <mliska@suse.cz>
1102 Backport from mainline
1103 2017-05-15 Martin Liska <mliska@suse.cz>
1106 * gcc.c (process_command): Do not allow empty argument of -o option.
1108 2017-06-22 Martin Liska <mliska@suse.cz>
1110 Backport from mainline
1111 2017-05-02 Martin Liska <mliska@suse.cz>
1113 * doc/gcov.texi: Add missing preposition.
1114 * gcov.c (function_info::function_info): Properly fill up
1115 all member variables.
1117 2017-06-22 Martin Liska <mliska@suse.cz>
1119 Backport from mainline
1120 2017-05-02 Martin Liska <mliska@suse.cz>
1123 * common.opt: Fix typo.
1124 * doc/invoke.texi: Likewise.
1126 2017-06-22 Martin Liska <mliska@suse.cz>
1128 Backport from mainline
1129 2017-04-28 Martin Liska <mliska@suse.cz>
1131 * doc/gcov.texi: Enhance documentation of gcov.
1133 2017-06-22 Martin Liska <mliska@suse.cz>
1135 Backport from mainline
1136 2017-04-28 Martin Liska <mliska@suse.cz>
1138 * doc/gcov.texi: Sort options in alphabetic order.
1139 * doc/gcov-dump.texi: Likewise.
1140 * doc/gcov-tool.texi: Likewise.
1141 * gcov.c (print_usage): Likewise.
1142 * gcov-dump.c (print_usage): Likewise.
1143 * gcov-tool.c (print_merge_usage_message): Likewise.
1144 (print_rewrite_usage_message): Likewise.
1145 (print_overlap_usage_message): Likewise.
1147 2017-06-22 Martin Liska <mliska@suse.cz>
1149 Backport from mainline
1150 2017-04-28 Martin Liska <mliska@suse.cz>
1152 PR gcov-profile/53915
1153 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
1155 2017-06-22 Martin Liska <mliska@suse.cz>
1157 Backport from mainline
1158 2017-04-28 Martin Liska <mliska@suse.cz>
1161 * coverage.c (coverage_remove_note_file): New function.
1162 * coverage.h: Declare the function.
1163 * toplev.c (finalize): Clean if an error has been seen.
1165 2017-06-21 Jakub Jelinek <jakub@redhat.com>
1168 * config/i386/sse.md (round<mode>2): Renumber match_dup and
1169 operands indexes to avoid gap between operands and match_dups.
1172 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
1173 with ctors/dtors if GOVD_SHARED is set.
1175 Backported from mainline
1176 2017-06-20 Jakub Jelinek <jakub@redhat.com>
1179 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
1180 splitter): Require TARGET_SSE2 in the condition.
1183 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
1184 by removing enum keyword.
1185 (ubsan_type_descriptor): Likewise. Formatting fix.
1187 2017-06-19 Jakub Jelinek <jakub@redhat.com>
1190 * ubsan.h (enum ubsan_encode_value_phase): New.
1191 (ubsan_encode_value): Change second argument to
1192 enum ubsan_encode_value_phase with default value of
1193 UBSAN_ENCODE_VALUE_GENERIC.
1194 * ubsan.c (ubsan_encode_value): Change second argument to
1195 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
1196 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
1197 create_tmp_var_raw instead of create_tmp_var and use a
1199 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
1200 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
1201 ubsan_encode_value callers.
1204 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
1205 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
1206 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
1208 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
1210 Backport from Mainline
1211 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
1214 2017-06-20 Andreas Schwab <schwab@suse.de>
1217 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
1220 2017-06-19 James Greenhalgh <james.greenhalgh@arm.com>
1222 Backport from mainline
1223 2017-06-19 James Greenhalgh <james.greenhalgh@arm.com>
1226 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
1227 if given a non-constant argument for an intrinsic which requires a
1230 2017-06-19 Alexander Monakov <amonakov@ispras.ru>
1232 * doc/invoke.texi (mcx16): Rewrite.
1234 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
1236 PR rtl-optimization/80474
1237 * reorg.c (update_block): Do not ignore instructions in a delay slot.
1239 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
1241 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
1242 (MASK_FEATURES): New macro.
1243 * config/sparc/sparc.c (sparc_option_override): Remove the special
1244 handling of -mfpu and generalize it to all MASK_FEATURES switches.
1246 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
1248 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
1250 2017-06-12 David S. Miller <davem@davemloft.net>
1253 * config/sparc/sparc.md (return expander): Emit frame blockage if
1254 function uses alloca.
1256 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
1260 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
1263 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
1266 2017-06-08 David Edelsohn <dje.gcc@gmail.com>
1268 Backport from mainline
1269 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
1271 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
1272 (dl_section_ref): New.
1273 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
1274 On AIX, append an expression to subtract the size of the
1275 section length to dl_section_ref.
1277 2017-06-07 Richard Biener <rguenther@suse.de>
1279 Backport from mainline
1280 2017-05-02 Richard Biener <rguenther@suse.de>
1282 PR tree-optimization/80549
1283 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
1284 (cleanup_tree_cfg_noloop): Create forwarders to known loop
1285 headers if they do not have a preheader.
1287 2017-05-26 Richard Biener <rguenther@suse.de>
1289 PR tree-optimization/80842
1290 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
1293 2017-05-31 Richard Biener <rguenther@suse.de>
1295 PR tree-optimization/80906
1296 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
1297 and pass through iv_map.
1298 (copy_bb_and_scalar_dependences): Adjust.
1299 (translate_pending_phi_nodes): Likewise.
1300 (copy_loop_close_phi_args): Handle code-generating IVs instead
1303 2017-05-11 Richard Biener <rguenther@suse.de>
1305 PR tree-optimization/80705
1306 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
1307 bases are not vectorizable.
1309 2017-06-06 Michael Meissner <meissner@linux.vnet.ibm.com>
1311 Back port from mainline
1312 2017-05-19 Michael Meissner <meissner@linux.vnet.ibm.com>
1315 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Prefer
1316 VSX registers over GPRs, particularly on ISA 2.07 which does not
1317 have the MTVSRDD instruction.
1319 2017-06-06 David S. Miller <davem@davemloft.net>
1322 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
1323 blockage if function uses alloca.
1325 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
1327 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
1329 2017-06-02 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
1331 Backport from mainline
1332 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
1333 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
1336 * config/arm/arm.md (use_literal_pool): Remove.
1337 (64-bit immediate split): No longer takes cost into consideration
1338 if arm_disable_literal_pool is enabled.
1339 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
1340 used when arm_disable_literal_pool is enabled.
1341 (arm_max_const_double_inline_cost): Remove use of
1342 arm_disable_literal_pool.
1343 (push_minipool_fix): Add assert.
1344 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
1345 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
1346 (no_literal_pool_sf_immediate): New.
1348 2017-06-02 Jakub Jelinek <jakub@redhat.com>
1350 PR rtl-optimization/80903
1351 * loop-doloop.c (add_test): Unshare sequence.
1353 2017-05-31 Martin Jambor <mjambor@suse.cz>
1355 Backport from mainline
1356 2017-04-24 Martin Jambor <mjambor@suse.cz>
1358 PR tree-optimization/80293
1359 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
1360 char arrays not totally scalarizable if it is false.
1361 (analyze_all_variable_accesses): Pass correct value in the new
1362 parameter. Add a statistics counter.
1364 2017-05-30 Max Filippov <jcmvbkbc@gmail.com>
1366 Backport from mainline
1367 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
1369 * config/xtensa/xtensa.c (xtensa_emit_call): Use
1370 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
1371 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
1374 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
1376 * doc/install.texi (Options specification): Restore entry of
1377 --enable-sjlj-exceptions.
1379 2017-05-29 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1381 Backport from mainline
1382 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1385 * config/s390/s390.c (s390_check_qrst_address): Check incoming
1386 address against address_operand predicate.
1387 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
1389 2017-05-28 Uros Bizjak <ubizjak@gmail.com>
1391 Backport from mainline
1392 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
1394 * config/i386/i386.md (*movdi_internal): Remove SSE4
1395 alternative 18 (?r, *v). Update insn attributes.
1396 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
1397 Update insn attributes.
1398 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
1399 Update insn attributes.
1400 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
1401 alternative 1 (r, v). Remove isa attribute.
1402 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
1403 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
1404 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
1406 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
1408 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
1409 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
1410 and (?*y,m). Update insn attributes.
1412 2017-05-26 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
1414 Backported from mainline
1415 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
1417 * config/sparc/sparc.md (length): Return the correct value for -mflat
1418 sibcalls to match output_sibcall.
1420 2017-05-26 Marek Polacek <polacek@redhat.com>
1422 Backported from mainline
1423 2017-05-26 Marek Polacek <polacek@redhat.com>
1426 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
1429 2017-05-26 Jakub Jelinek <jakub@redhat.com>
1431 Backported from mainline
1432 2017-05-22 Jakub Jelinek <jakub@redhat.com>
1435 * omp-low.c (finish_taskreg_remap): New function.
1436 (finish_taskreg_scan): If unit size of ctx->record_type
1437 is non-constant, unshare the size expression and replace
1438 decls in it with possible outer var refs.
1441 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
1442 GOVD_SHARED rather than GOVD_PRIVATE with it.
1443 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
1444 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
1447 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
1448 as last argument to build_outer_var_ref for pointer bases of array
1451 2017-05-25 Michael Meissner <meissner@linux.vnet.ibm.com>
1454 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1457 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
1460 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
1461 (define_peephole2 for Altivec d-form load): Add peepholes to catch
1462 cases where the register allocator uses a move and an offsettable
1463 memory operation to/from a FPR register on ISA 2.06/2.07.
1464 (define_peephole2 for Altivec d-form store): Likewise.
1467 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
1470 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
1471 are now unused after splitting mov{sf,sd}_hardfloat.
1472 (f32_lr2): Likewise.
1474 (f32_lm2): Likewise.
1476 (f32_li2): Likewise.
1479 (f32_sr2): Likewise.
1481 (f32_sm2): Likewise.
1483 (f32_si2): Likewise.
1486 (f32_vsx): Likewise.
1488 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
1489 For movsf, order stores so the VSX stores occur before the GPR
1490 store which encourages the register allocator to use a traditional
1491 FPR instead of a GPR. For movsd, order the stores so that the GPR
1492 store comes before the VSX stores to allow the power6 to work.
1493 This is due to the power6 not having a 32-bit integer store
1494 instruction from a FPR.
1495 (movsf_hardfloat): Likewise.
1496 (movsd_hardfloat): Likewise.
1498 2017-05-25 Wilco Dijkstra <wdijkstr@arm.com>
1500 Backport from mainlin
1501 PR rtl-optimization/80754
1502 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
1504 2017-05-25 Wilco Dijkstra <wdijkstr@arm.com>
1506 Backport from mainline
1508 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
1509 Move member access before delete.
1511 2017-05-23 Sheldon Lobo <sheldon.lobo@oracle.com>
1513 Backport from mainline
1514 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
1516 * config/sparc/sparc.c (sparc_option_override): Set function
1517 alignment for -mcpu=niagara7 to 64 to match the I$ line.
1518 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
1520 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
1522 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
1524 2017-05-19 Uros Bizjak <ubizjak@gmail.com>
1526 Backport from mainline
1527 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
1530 * config/i386/mmx.md (*mov<mode>_internal): Enable
1531 alternatives 11, 12, 13 and 14 also for 32bit targets.
1532 Remove alternatives 15, 16, 17 and 18.
1533 * config/i386/sse.md (vec_concatv2di): Change
1534 alternative (!x, *y) to (x, ?!*Yn).
1536 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
1538 Backport from mainline
1539 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
1542 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
1543 (UNSPEC_STX_ATOMIC): Ditto.
1544 (loaddi_via_sse): New insn.
1545 (storedi_via_sse): Ditto.
1546 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
1547 Update corresponding peephole2 patterns.
1548 (atomic_storedi_fpu): Ditto.
1550 2017-05-13 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1552 Backport from mainline
1553 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1555 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
1556 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
1557 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
1558 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
1559 VF=2 that require versioning.
1561 2017-05-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1563 Backport from mainline
1564 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
1566 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
1567 built-ins for vec_xl and vec_xst with short and char pointer
1570 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
1573 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
1574 handle calling assemble_external ourself.
1577 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
1578 modes with zero size. Enhance comment.
1580 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
1582 Back port from mainline
1583 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
1588 * config/rs6000/rs6000.md (u code attribute): Add FIX and
1590 (extendsi<mode>2): Add support for doing sign extension via
1591 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
1592 don't have ISA 3.0 instructions.
1593 (extendsi<mode>2 splitter): Likewise.
1594 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
1595 generate the normal insns since SImode can now go in vector
1596 registers. Disallow the special UNSPECs needed for previous
1597 machines to hide SImode being used. Add new insns
1598 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
1599 (fix_trunc<mode>si2_stfiwx): Likewise.
1600 (fix_trunc<mode>si2_internal): Likewise.
1601 (fixuns_trunc<mode>si2): Likewise.
1602 (fixuns_trunc<mode>si2_stfiwx): Likewise.
1603 (fctiw<u>z_<mode>_smallint): Likewise.
1604 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
1605 of floating point to 32-bit integer from doing a direct move to
1606 the GPR registers to do a store.
1607 (fctiwz_<mode>): Break long line.
1609 2017-05-08 Tamar Christina <tamar.christina@arm.com>
1612 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
1613 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
1615 2017-05-03 Jan Beulich <jbeulich@suse.com>
1617 Backport from mainline
1618 2017-05-01 Jan Beulich <jbeulich@suse.com>
1620 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
1621 swapping, add (x,x,m,x,n) alternative.
1623 2017-05-03 Richard Biener <rguenther@suse.de>
1625 Backport from mainline
1626 2017-04-20 Richard Biener <rguenther@suse.de>
1628 PR tree-optimization/80453
1629 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
1630 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
1631 from the conditions.
1632 (vn_phi_eq): Pass them down.
1633 (vn_phi_lookup): Record them.
1634 (vn_phi_insert): Likewise.
1636 2017-04-25 Richard Biener <rguenther@suse.de>
1638 PR tree-optimization/80492
1639 * alias.c (compare_base_decls): Handle registers with asm
1640 specification conservatively.
1642 2017-04-27 Richard Biener <rguenther@suse.de>
1645 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
1646 being in loop-closed SSA form conservatively.
1647 (chrec_fold_multiply_poly_poly): Likewise.
1649 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
1651 Backport from mainline
1652 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
1655 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
1656 __get_cpuid_max returns 0.
1657 (__get_cpuid_count): Ditto.
1659 2017-05-02 Jakub Jelinek <jakub@redhat.com>
1661 Backported from mainline
1662 2017-04-25 Jakub Jelinek <jakub@redhat.com>
1664 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
1667 2017-05-02 Jakub Jelinek <jakub@redhat.com>
1669 * BASE-VER: Set to 7.1.1.
1671 2017-05-02 Release Manager
1673 * GCC 7.1.0 released.
1675 2017-05-02 Richard Biener <rguenther@suse.de>
1677 PR tree-optimization/80591
1679 2017-04-10 Richard Biener <rguenther@suse.de>
1681 * tree-ssa-structalias.c (find_func_aliases): Properly handle
1684 2017-04-28 Jakub Jelinek <jakub@redhat.com>
1687 * cgraph.h (symtab_node::debug_symtab): No longer inline.
1688 * symtab.c (symtab_node::debug_symtab): Move definition here.
1690 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
1693 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
1694 that the logic for permitting reciprocal estimates matches that
1697 2017-04-27 Jakub Jelinek <jakub@redhat.com>
1700 * tree.c (type_cache_hasher::equal): Only compare
1701 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
1702 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
1703 non-aggregate element types.
1704 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
1705 about the flag on ARRAY_TYPEs in the comment, formatting fix.
1708 * reg-stack.c (emit_swap_insn): If i1src mentions the stack pointer,
1709 punt if tmp contains autoinc of stack pointer.
1712 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
1713 (aarch64_function_arg_alignment): Return unsigned int again, but still
1714 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
1715 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
1716 Don't emit -Wpsabi note.
1717 (aarch64_function_arg_boundary): Likewise.
1718 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
1721 2017-04-25 Martin Sebor <msebor@redhat.com>
1723 PR tree-optimization/80497
1724 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
1725 constants are representable in HOST_WIDE_INT.
1726 (parse_directive): Ditto.
1728 2017-04-25 Marek Polacek <polacek@redhat.com>
1730 2017-04-25 Marek Polacek <polacek@redhat.com>
1731 Backport from mainline
1734 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
1735 first argument to type.
1737 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1738 Jakub Jelinek <jakub@redhat.com>
1741 * config/arm/arm.c: Include gimple.h.
1742 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
1743 returns negative, increment ncrn only if it returned positive.
1744 (arm_needs_doubleword_align): Return int instead of bool,
1745 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
1746 members, but if there is any such non-FIELD_DECL
1747 > PARM_BOUNDARY aligned decl, return -1 instead of false.
1748 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
1749 returns negative, increment nregs only if it returned positive.
1750 (arm_setup_incoming_varargs): Likewise.
1751 (arm_function_arg_boundary): Emit -Wpsabi note if
1752 arm_needs_doubleword_align returns negative, return
1753 DOUBLEWORD_ALIGNMENT only if it returned positive.
1755 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
1757 Backport from mainline
1759 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
1760 type checks to test for compatibility instead of equality.
1762 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
1763 Jakub Jelinek <jakub@redhat.com>
1766 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
1768 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
1769 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
1770 the alignment computation, but return their maximum in warn_alignment.
1771 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
1772 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
1774 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
1775 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
1778 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1780 Backport from mainline
1781 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1784 * config/s390/vector.md: Split MEM->GPR vector moves for
1785 non-s_operand addresses.
1787 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1789 Backport from mainline
1790 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1793 * config/s390/predicates.md (reload_const_wide_int_operand): New
1795 * config/s390/s390.md ("movti"): Remove d/P alternative.
1796 ("movti_bigconst"): New pattern definition.
1798 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1800 Backport from maineline
1801 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1804 * s390-protos.h (s390_expand_cs_hqi): Removed.
1805 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
1806 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
1807 modes as well as CCZ1mode and CCZmode.
1808 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
1809 signature of s390_emit_compare_and_swap.
1810 (s390_expand_cs_hqi): Likewise, make static.
1811 (s390_expand_cs_tdsi): Generate an explicit compare before trying
1812 compare-and-swap, in some cases.
1813 (s390_expand_cs): Wrapper function.
1814 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
1816 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
1817 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
1818 patterns for small and large integers. Forbid symref memory operands.
1819 Move expander to s390.c. Require cc register.
1820 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
1821 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
1822 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
1823 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
1824 symref memory operands. Remove CC mode and call s390_match_ccmode
1826 ("atomic_exchange<mode>"): Allow and implement all integer modes.
1828 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1830 Backport from mainline
1831 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1833 * config/s390/s390.md (define_peephole2): New peephole to help
1834 combining the load-and-test pattern with volatile memory.
1837 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1839 Backport from mainline
1840 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
1842 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
1843 with CCZmode for TARGET_Z196.
1845 2017-04-25 Jakub Jelinek <jakub@redhat.com>
1847 PR rtl-optimization/80501
1848 * combine.c (make_compound_operation_int): Set subreg_code to SET
1849 even for AND with mask of the sign bit of mode.
1851 PR rtl-optimization/80500
1852 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
1853 sum's initial value.
1855 2017-04-24 Martin Liska <mliska@suse.cz>
1857 Backport from mainline
1858 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
1861 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
1863 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
1865 Backport from mainline
1866 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
1868 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear
1869 and lastprivate clauses in SIMT case.
1871 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
1873 Backport from mainline
1874 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
1876 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
1877 uninitialized variable warning to avoid buffer overrun.
1879 2017-04-20 Jakub Jelinek <jakub@redhat.com>
1881 * DEV-PHASE: Set to prerelease.
1883 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
1885 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
1886 priority .init_array and .fini_array section with SECTION_NOTYPE
1889 2017-04-20 Jakub Jelinek <jakub@redhat.com>
1892 * tree.h (build_array_type): Add typeless_storage default argument.
1893 * tree.c (type_cache_hasher::equal): Also compare
1894 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
1895 (build_array_type): Add typeless_storage argument, set
1896 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
1898 (build_nonshared_array_type): Adjust build_array_type_1 caller.
1899 (build_array_type): Likewise. Add typeless_storage argument.
1901 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
1902 Jakub Jelinek <jakub@redhat.com>
1904 PR tree-optimization/80426
1905 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
1906 operation on symbolic operands, also compute the overflow for the
1907 invariant part when the operation degenerates into a negation.
1909 2017-04-19 Jakub Jelinek <jakub@redhat.com>
1912 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
1913 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
1916 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
1918 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
1921 * config/avr/avr.c (tree.h): Include it.
1922 (cgraph.h): Include it.
1923 (avr_encode_section_info): Don't warn for uninitialized progmem
1924 variable if it's just an alias.
1926 2017-04-19 Richard Biener <rguenther@suse.de>
1929 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
1930 when needed by AutoPGO.
1932 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
1935 * doc/lto.texi: Remove an extra 'that'.
1937 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
1939 PR rtl-optimization/80429
1940 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
1941 are only used in debug insns.
1943 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
1944 Vladimir Makarov <vmakarov@redhat.com>
1946 * config/sparc/predicates.md (input_operand): Add comment. Return
1947 true for any memory operand when LRA is in progress.
1948 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
1950 2017-04-18 Jeff Law <law@redhat.com>
1953 * mips.md ({return,simple_return}_internal): Do not overwrite
1956 2017-04-18 Jakub Jelinek <jakub@redhat.com>
1958 PR tree-optimization/80443
1959 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
1960 instead of adding 1, subtract -1 and similarly instead of subtracting
1963 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
1965 PR rtl-optimization/80357
1966 * haifa-sched.c (tmp_bitmap): New variable.
1967 (model_recompute): Handle duplicate use records.
1968 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
1969 (free_global_sched_pressure_data): Free it.
1971 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
1974 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
1975 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
1976 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
1977 instead of SYSTEM_HEADER_DIR.
1979 2017-04-18 Jeff Law <law@redhat.com>
1982 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
1983 predecessors after walking up the insn chain.
1985 2017-04-18 Jakub Jelinek <jakub@redhat.com>
1988 * dwarf2out.c (modified_type_die): Try harder not to emit internal
1989 sizetype type into debug info.
1991 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
1994 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
1995 unneeded test for TARGET_UPPER_REGS_SF.
1996 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
1998 2017-04-18 Jakub Jelinek <jakub@redhat.com>
2001 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
2002 instead of gsi_after_labels.
2004 2017-04-18 Jeff Law <law@redhat.com>
2006 * regcprop.c (maybe_mode_change): Avoid creating copies of the
2010 2017-04-13 Jeff Law <law@redhat.com>
2011 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
2012 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
2014 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
2017 * config/avr/avr.c (intl.h): Include it.
2018 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
2020 2017-04-18 Martin Liska <mliska@suse.cz>
2022 PR gcov-profile/78783
2023 * gcov-tool.c (gcov_output_files): Validate that destination
2024 file is either removed by the tool or by a user.
2026 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
2027 Guy Benyei <guybe@mellanox.com>
2029 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
2030 block, and do not negate it, the stored id is already negative.
2032 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
2034 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
2036 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
2039 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
2040 masks of options that should be turned off if the VSX vector
2041 options are turned off.
2042 (OTHER_P8_VECTOR_MASKS): Likewise.
2043 (OTHER_VSX_VECTOR_MASKS): Likewise.
2044 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
2045 rs6000_disable_incompatible_switches to validate no type switches
2047 (rs6000_incompatible_switch): New function to disallow turning on
2048 other vector options if -mno-vsx, -mno-power8-vector, or
2049 -mno-power9-vector are specified.
2051 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
2053 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
2055 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
2057 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
2058 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
2059 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
2060 (ARG_POINTER_CFA_OFFSET): Likewise.
2062 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
2064 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
2065 conditions to take advantage of various optimizations.
2067 2017-04-13 Jeff Law <law@redhat.com>
2069 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
2070 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
2071 (zero_extendsidi2_dext): Likewise.
2073 2017-04-13 Jakub Jelinek <jakub@redhat.com>
2076 * fold-const.c (fold_ternary_loc): Revert
2077 use op0 instead of fold_convert_loc (loc, type, arg0) part of
2080 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
2082 PR rtl-optimization/80343
2083 * lra-remat.c (update_scratch_ops): Assign original hard reg to
2086 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
2089 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
2090 to ubsan_encode_value.
2092 2017-04-13 Jeff Law <law@redhat.com>
2094 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
2095 appearing in DEBUG_INSNs.
2097 2017-04-13 Martin Liska <mliska@suse.cz>
2099 PR gcov-profile/80413
2100 * gcov-io.c (gcov_write_string): Copy to buffer just when
2101 allocated size is greater than zero.
2103 2017-04-13 Jakub Jelinek <jakub@redhat.com>
2106 * dwarf2out.c (decls_for_scope): Ignore declarations of
2107 current_function_decl in BLOCK_NONLOCALIZED_VARS.
2109 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
2112 * ipa-visibility.c (non_local_p): Fix typos.
2113 (localize_node): When localizing symbol in same comdat group,
2114 dissolve the group only when we know external symbols are going
2116 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
2118 2017-04-12 Jakub Jelinek <jakub@redhat.com>
2120 PR tree-optimization/79390
2121 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
2122 order does not result in usable sequence, retry with reversed operand
2128 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
2129 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
2130 op0 instead of fold_convert_loc (loc, type, arg0).
2132 2017-04-12 Jeff Law <law@redhat.com>
2134 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
2135 has a delay slot in the generated code.
2137 * config/cris/cris.md (cris_preferred_reload_class): Return
2138 GENNONACR_REGS rather than GENERAL_REGS.
2140 2017-04-12 Jakub Jelinek <jakub@redhat.com>
2143 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
2144 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
2145 signedness of the result type.
2147 2017-04-12 Richard Biener <rguenther@suse.de>
2148 Jeff Law <law@redhat.com>
2150 PR tree-optimization/80359
2151 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
2152 trim stores to TARGET_MEM_REFs.
2154 2017-04-12 Richard Biener <rguenther@suse.de>
2156 PR tree-optimization/79390
2157 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
2158 threading case even more.
2160 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
2163 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
2164 for quad_address_p for TImode, instead of just not indexed_address.
2166 2017-04-12 Richard Biener <rguenther@suse.de>
2167 Bernd Edlinger <bernd.edlinger@hotmail.de>
2170 * alias.c (component_uses_parent_alias_set_from): Handle
2171 TYPE_TYPELESS_STORAGE.
2172 (get_alias_set): Likewise.
2173 * tree-core.h (tree_type_common): Add typeless_storage flag.
2174 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
2175 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
2176 for types containing members with TYPE_TYPELESS_STORAGE.
2177 (place_field): Likewise.
2178 (layout_type): Likewise for ARRAY_TYPE.
2179 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
2180 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
2181 TYPE_TYPELESS_STORAGE.
2182 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
2184 2017-04-12 Jakub Jelinek <jakub@redhat.com>
2187 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
2188 first argument to type.
2190 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2194 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
2195 CONST0_RTX (mode) rather than const0_rtx where appropriate.
2196 (rs6000_expand_binop_builtin): Likewise.
2197 (rs6000_expand_ternop_builtin): Likewise; also add missing
2198 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
2199 vshasigma built-ins.
2200 * doc/extend.texi: Document that vec_xxpermdi's third argument
2203 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
2205 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
2206 Use shift_const cost parameter when calculating gain of STV shifts.
2208 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
2210 PR rtl-optimization/70478
2211 * lra-constraints.c (process_alt_operands): Check memory for
2212 disfavoring memory insn operand.
2214 2017-04-11 Jakub Jelinek <jakub@redhat.com>
2217 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
2218 left shift in unsigned HOST_WIDE_INT type.
2220 PR rtl-optimization/80385
2221 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
2222 (not (neg X)) into (plus X -1) for complex or non-integral modes.
2225 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
2226 if they have any depend clauses.
2228 2017-04-11 Martin Liska <mliska@suse.cz>
2231 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
2232 * ipa-split.c (split_function): Create a local comdat symbol
2233 if caller is in a comdat group.
2235 2017-04-11 Martin Liska <mliska@suse.cz>
2238 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
2241 2017-04-11 Martin Sebor <msebor@redhat.com>
2244 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
2245 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
2247 (directive::set_width, directive::set_precision, format_character):
2249 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
2252 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
2255 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
2256 conflict, set target->arch_name instead of target->cpu_name.
2258 2017-04-11 Richard Biener <rguenther@suse.de>
2260 PR tree-optimization/80374
2261 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
2262 build_zero_cst, remove fold_convertible_p check again.
2264 2017-04-11 Martin Liska <mliska@suse.cz>
2267 * ubsan.c (instrument_object_size): Do not instrument register
2270 2017-04-11 Jakub Jelinek <jakub@redhat.com>
2273 * config/i386/i386-builtin-types.def
2274 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
2275 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
2276 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
2277 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
2278 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
2279 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
2280 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
2281 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
2282 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
2283 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
2284 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
2285 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
2286 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
2287 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
2288 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
2289 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
2290 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
2291 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
2292 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
2293 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
2294 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
2295 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
2296 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
2297 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
2298 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
2299 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
2300 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
2301 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
2302 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
2303 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
2304 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
2305 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
2306 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
2307 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
2308 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
2309 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
2310 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
2311 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
2312 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
2313 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
2314 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
2315 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
2316 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
2317 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
2318 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
2319 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
2321 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
2322 flag to second_arg_count, handle 4 argument function type _COUNT
2323 aliases, handle second_arg_count on second argument rather than last.
2325 2017-04-10 Jeff Law <law@redhat.com>
2327 PR tree-optimization/80374
2328 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
2329 record anything if we can not convert integer_zero_node to the
2332 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
2335 * config/rs6000/rs6000.c (rs6000_option_override_internal):
2336 Enhance special handling given to the TARGET_P9_MINMAX option in
2337 relation to certain other options.
2339 2017-04-10 Bin Cheng <bin.cheng@arm.com>
2341 PR tree-optimization/80153
2342 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
2343 remove POINTER_PLUS_EXPR's base part directly, rather than through
2346 2017-04-10 Richard Biener <rguenther@suse.de>
2347 Bin Cheng <bin.cheng@arm.com>
2349 PR tree-optimization/80153
2350 * tree-affine.c (aff_combination_to_tree): Get base pointer from
2351 the first element of pointer type aff_tree. Build result expr in
2353 (add_elt_to_tree): Convert to type unconditionally. Remove other
2355 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
2356 (rewrite_use_nonlinear_expr): Check invariant using iv information.
2358 2017-04-10 Richard Biener <rguenther@suse.de>
2360 * tree-ssa-structalias.c (find_func_aliases): Properly handle
2363 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
2365 PR rtl-optimization/70478
2366 * lra-constraints.c (curr_small_class_check): New.
2367 (update_and_check_small_class_inputs): New.
2368 (process_alt_operands): Update curr_small_class_check. Disfavor
2369 alternative insn memory operands. Check available regs for small
2372 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
2375 * config/mips/mips.opt (-mvirt): Update description.
2376 * doc/invoke.texi (-mvirt): Likewise.
2378 2017-04-10 Richard Biener <rguenther@suse.de>
2381 * fold-const.c (fold_binary_loc): Look at unstripped ops when
2382 looking for NEGATE_EXPR in -A / -B to A / B folding.
2384 2017-04-10 Martin Liska <mliska@suse.cz>
2386 PR gcov-profile/80224
2387 * gcov.c (print_usage): Fix usage string.
2388 (get_gcov_intermediate_filename): Remove.
2389 (output_gcov_file): Use both for normal and intermediate format.
2390 (generate_results): Do not initialize special file for
2391 intermediate format.
2393 2017-04-10 Richard Biener <rguenther@suse.de>
2395 PR tree-optimization/80304
2396 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
2399 2017-04-10 Nathan Sidwell <nathan@acm.org>
2402 * config/rs6000/rs6000.c (rs6000_vector_type): New.
2403 (rs6000_init_builtins): Use it.
2405 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2407 * config/arm/arm.md (<mrc>): Add mode to SET source.
2410 2017-04-10 Richard Biener <rguenther@suse.de>
2413 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
2415 2017-04-10 Jakub Jelinek <jakub@redhat.com>
2418 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
2419 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
2420 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
2421 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
2422 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
2423 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
2424 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
2425 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
2426 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
2427 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
2428 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
2429 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
2430 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
2431 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
2432 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
2433 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
2434 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
2435 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
2436 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
2437 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
2438 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
2439 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
2440 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
2442 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
2444 PR rtl-optimization/70478
2445 * lra-constraints.c: Reverse the last patch.
2447 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
2449 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
2450 Add comment for WCHAR_T.
2452 2017-04-08 Martin Liska <mliska@suse.cz>
2455 2017-04-07 Martin Liska <mliska@suse.cz>
2458 * ipa-split.c (split_function): Add function part to a same comdat
2461 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2464 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
2466 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
2468 * rs6000/rs6000.c (vec_load_pendulum): Rename...
2469 (vec_pairing): ...to this.
2470 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
2471 (rs6000_sched_init): Adjust for name change.
2472 (struct rs6000_sched_context): Likewise.
2473 (rs6000_init_sched_context): Likewise.
2474 (rs6000_set_sched_context): Likewise.
2476 2017-04-07 Jakub Jelinek <jakub@redhat.com>
2482 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
2484 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
2485 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
2486 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
2488 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
2490 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
2492 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
2494 PR rtl-optimization/70703
2495 * ira-color.c (update_conflict_hard_regno_costs): Use
2496 int64_t instead of HOST_WIDE_INT.
2498 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
2500 PR rtl-optimization/70478
2501 * lra-constraints.c (process_alt_operands): Disfavor alternative
2502 insn memory operands.
2504 2017-04-07 Jeff Law <law@redhat.com>
2506 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
2507 CALL and NOTE_INSN_CALL_ARG_LOCATION.
2509 2017-04-07 Martin Liska <mliska@suse.cz>
2512 * config/aarch64/aarch64.c (aarch64_process_target_attr):
2513 Show error message instead of an ICE.
2515 2017-04-07 Martin Liska <mliska@suse.cz>
2518 * ipa-split.c (split_function): Add function part to a same comdat
2521 2017-04-07 Richard Biener <rguenther@suse.de>
2524 * tree.c (get_unwidened): Also handle ! for_type case for
2526 * convert.c (do_narrow): Split out from ...
2527 (convert_to_integer_1): ... here. Do not pass final truncation
2528 type to get_unwidened for TRUNC_DIV_EXPR.
2530 2017-04-07 Richard Biener <rguenther@suse.de>
2532 * tree-affine.c (wide_int_ext_for_comb): Take type rather
2534 (aff_combination_const): Adjust.
2535 (aff_combination_scale): Likewise.
2536 (aff_combination_add_elt): Likewise.
2537 (aff_combination_add_cst): Likewise.
2538 (aff_combination_convert): Likewise.
2539 (add_elt_to_tree): Likewise. Remove unused argument.
2540 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
2542 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
2544 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
2546 * config/arm/arm.c (arm_default_short_enums): Use
2547 ARM_DEFAULT_SHORT_ENUMS.
2548 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
2550 2017-04-06 Jakub Jelinek <jakub@redhat.com>
2553 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
2554 members with redundant out-of-class redeclaration.
2556 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
2559 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
2560 * config/i386/i386.md (*zero_extendsidi2):
2561 Add (?*x,*x) and (?*v,*v) alternatives.
2563 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
2566 * config/i386/i386.c (ix86_expand_builtin)
2567 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
2568 mode from insn data. Convert operands to insn operand mode.
2569 Copy operands that don't satisfy insn predicate to a register.
2571 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
2573 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
2576 2017-04-06 Richard Biener <rguenther@suse.de>
2578 PR tree-optimization/80334
2579 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
2580 preserve alignment of accesses.
2582 2017-04-06 Richard Biener <rguenther@suse.de>
2584 PR tree-optimization/80262
2585 * tree-sra.c (build_ref_for_offset): Preserve address-space
2587 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
2588 Drop useless address-space information on MEM_REF offsets.
2590 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
2592 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
2594 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
2596 PR rtl-optimization/70703
2597 * ira-color.c (update_conflict_hard_regno_costs): Use
2598 HOST_WIDE_INT instead of long.
2600 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
2603 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
2604 not defined for x86_64 target. Add -mmmx target option when __SSE2__
2606 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
2607 for x86_64 target. Handle -m3dnowa option.
2609 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
2611 PR rtl-optimization/70703
2612 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
2613 (update_conflict_hard_regno_costs): Use long instead of unsigned
2614 arithmetic for cost calculation.
2616 2017-04-05 Jakub Jelinek <jakub@redhat.com>
2617 Bernd Edlinger <bernd.edlinger@hotmail.de>
2620 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
2623 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
2626 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
2627 ptr_mode with Pmode throughout.
2628 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
2629 into probe_stack_range and use DImode.
2631 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
2634 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
2635 call_eh_return is true.
2637 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2639 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
2640 Initialize last_match_fntype_index.
2642 2017-04-05 Jakub Jelinek <jakub@redhat.com>
2645 * tree-nvr.c: Include internal-fn.h.
2646 (pass_return_slot::execute): Ignore internal calls without
2649 2017-04-04 Jakub Jelinek <jakub@redhat.com>
2650 Richard Biener <rguenther@suse.de>
2653 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
2654 captures used multiple times, except for the last use.
2655 * generic-match-head.c: Include gimplify.h.
2657 2017-04-04 Jakub Jelinek <jakub@redhat.com>
2659 PR tree-optimization/79390
2660 * target.h (struct noce_if_info): Declare.
2661 * targhooks.h (default_noce_conversion_profitable_p): Declare.
2662 * target.def (noce_conversion_profitable_p): New target hook.
2663 * ifcvt.h (struct noce_if_info): New type, moved from ...
2664 * ifcvt.c (struct noce_if_info): ... here.
2665 (noce_conversion_profitable_p): Renamed to ...
2666 (default_noce_conversion_profitable_p): ... this. No longer
2668 (noce_try_store_flag_constants, noce_try_addcc,
2669 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
2670 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
2671 instead of noce_conversion_profitable_p.
2672 * config/i386/i386.c: Include ifcvt.h.
2673 (ix86_option_override_internal): Don't override
2674 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
2675 (ix86_noce_conversion_profitable_p): New function.
2676 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
2677 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
2678 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
2679 * doc/tm.texi: Regenerated.
2681 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2683 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
2686 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
2689 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
2690 instructions for small multiply cores.
2692 2017-04-04 Jeff Law <law@redhat.com>
2694 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
2696 (mips_expand_vec_perm_const): Initialize elements in orig_perm
2697 that are not set by the loop over the elements.
2699 2017-04-04 Jakub Jelinek <jakub@redhat.com>
2702 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
2703 int mode, convert_modes it to mode as unsigned, otherwise use
2704 lowpart_subreg to mode rather than SImode.
2705 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
2706 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
2707 Use DImode instead of SImode for the shift count operand.
2708 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
2711 2017-04-04 Richard Biener <rguenther@suse.de>
2714 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
2715 arithmetic done for the negate or the plus. Simplify.
2716 (A - (-B) -> A + B): Likewise.
2717 * fold-const.c (split_tree): Make sure to not negate pointers.
2719 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
2721 PR rtl-optimization/60818
2722 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
2723 a compare of comparisons with the thing compared if this results
2724 in a different machine mode.
2726 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
2728 * alias.c (base_alias_check): Fix typo in comment.
2729 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
2730 * cgraphunit.c (symbol_table::compile): Likewise.
2731 * collect2.c (maybe_run_lto_and_relink): Likewise.
2732 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
2733 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
2734 * config/avr/avr.c (avr_map_op_t): Likewise.
2735 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
2736 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
2737 * config/epiphany/epiphany.md (movcc): Likewise.
2738 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
2739 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
2741 * config/mips/mips.c (mips_save_restore_reg): Likewise.
2742 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
2743 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
2744 * config/sh/sh.c (sh_rtx_costs): Likewise.
2745 * fold-const.c (fold_truth_andor): Likewise.
2746 * genautomata.c (collapse_flag): Likewise.
2747 * gengtype.h (struct type::u::s): Likewise.
2748 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
2749 * input.c (FORMAT_AMOUNT): Likewise.
2750 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
2751 (known_aggs_to_agg_replacement_list): Likewise.
2752 * ipa-inline-analysis.c: Likewise.
2753 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
2754 * ipa-polymorphic-call.c
2755 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
2756 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
2757 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
2759 * modulo-sched.c (apply_reg_moves): Likewise.
2760 * omp-expand.c (build_omp_regions_1): Likewise.
2761 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
2762 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
2763 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
2764 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
2765 * value-prof.c: Likewise.
2766 * var-tracking.c (val_reset): Likewise.
2768 2017-04-03 Richard Biener <rguenther@suse.de>
2770 PR tree-optimization/80275
2771 * fold-const.c (split_address_to_core_and_offset): Handle
2774 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
2776 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
2777 descriptors is at least equal to that of functions.
2779 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
2781 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
2783 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
2786 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
2787 (mov<IMOD4:mode>): New expander.
2788 (*mov<IMOD4:mode>_internal): New insn and split pattern.
2790 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
2792 PR rtl-optimization/79405
2793 * fwprop.c (propagations_left): New variable.
2794 (forward_propagate_into): Decrement it.
2795 (fwprop_init): Initialize it.
2796 (fw_prop): If the variable has reached zero, stop propagating.
2797 (fwprop_addr): Ditto.
2799 2017-03-31 Jakub Jelinek <jakub@redhat.com>
2802 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
2803 a FUNCTION_DECL, pass it as decl instead of origin to
2806 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
2808 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
2811 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
2814 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
2815 TARGET_VSX_SMALL_INTEGER.
2817 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2819 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
2820 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
2822 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
2824 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
2825 extraction from odd-numbered MSA register.
2827 2017-03-31 Jakub Jelinek <jakub@redhat.com>
2830 * expmed.c (store_bit_field_1): Don't attempt to create
2831 a word subreg out of hard registers wider than word if they
2832 have HARD_REGNO_NREGS of 1 for their mode.
2835 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
2836 conversions to integer types wider than word and pointer.
2839 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
2840 (rtx_equal_for_cselib_p): Pass 0 to it.
2841 * cselib.c (cselib_hasher::equal): Likewise.
2842 (rtx_equal_for_cselib_1): Add depth argument. If depth
2843 is 128, don't look up VALUE locs and punt. Increment
2844 depth in recursive calls when walking VALUE locs.
2846 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
2848 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
2849 (make_gcov_file_name): Use the canonical path name for generating
2851 (read_line): Fix handling of files with ascii null bytes.
2853 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
2855 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
2856 to initialise a vector register instead
2857 of using a const_int.
2859 2017-03-30 Jakub Jelinek <jakub@redhat.com>
2861 PR translation/80189
2862 * gimplify.c (omp_default_clause): Use %qs instead of %s in
2863 diagnostic messages.
2865 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
2868 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
2869 (dfp_diex_<mode>): Update mode of operand 1.
2870 * doc/extend.texi (dxex, dxexq): Document change to return type.
2871 (diex, diexq): Document change to argument type.
2873 2017-03-30 Martin Jambor <mjambor@suse.cz>
2876 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
2877 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
2878 it reflects the signature changes performed at the callee side.
2879 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
2880 to cgraph_build_function_type_skip_args.
2881 (build_function_decl_skip_args): Adjust call to the above function.
2883 2017-03-30 Jakub Jelinek <jakub@redhat.com>
2886 * config/i386/sse.md
2887 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
2888 register as dest whenever it is a MEM not rtx_equal_p to the
2889 corresponding dup operand, and when forcing into reg move the
2890 reg into the memory afterwards.
2891 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
2892 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
2893 for the force_reg mode.
2894 (avx512vl_vextractf128<mode>): Use register as dest either
2895 always when a MEM, or when it is a MEM not rtx_equal_p to the
2896 corresponding dup operand, or even not when it is a CONST_VECTOR
2897 depending on the mode and lo vs. hi.
2898 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
2900 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
2901 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
2902 Likewise. Require that operands[2] is even.
2903 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
2904 Remove extraneous parens. Require that operands[2] is a multiple
2906 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
2907 operands[0] is a MEM if <mask_applied>, the predicates/constraints
2908 disallow memory then.
2910 2017-03-30 Richard Biener <rguenther@suse.de>
2912 PR tree-optimization/77498
2913 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
2914 to non-constants over backedges.
2916 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
2918 PR rtl-optimization/80233
2919 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
2920 as last_combined_insn. Do not test for BARRIER_P separately.
2922 2017-03-29 Andreas Schwab <schwab@suse.de>
2925 * calls.c (prepare_call_address): Convert funexp to Pmode before
2926 copying to temp reg.
2928 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2930 PR tree-optimization/80158
2931 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
2932 Handle possible future case of more than one alternate
2934 (replace_rhs_if_not_dup): Likewise.
2935 (replace_one_candidate): Likewise.
2937 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
2939 PR rtl-optimization/80193
2940 * ira.c (ira): Do not check allocation for LRA.
2942 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
2944 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
2945 (nvptx_output_simt_exit): Declare.
2946 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
2947 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
2948 (init_softstack_frame): Move initialization of crtl->is_leaf to...
2949 (nvptx_declare_function_name): ...here. Emit declaration of local
2950 memory space buffer for omp_simt_enter insn.
2951 (nvptx_output_unisimt_switch): New.
2952 (nvptx_output_softstack_switch): New.
2953 (nvptx_output_simt_enter): New.
2954 (nvptx_output_simt_exit): New.
2955 * config/nvptx/nvptx.h (struct machine_function): New fields
2956 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
2957 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
2958 (UNSPECV_SIMT_EXIT): Ditto.
2959 (omp_simt_enter_insn): New insn.
2960 (omp_simt_enter): New expansion.
2961 (omp_simt_exit): New insn.
2962 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
2964 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
2965 (expand_GOMP_SIMT_ENTER_ALLOC): New.
2966 (expand_GOMP_SIMT_EXIT): New.
2967 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
2968 (GOMP_SIMT_ENTER_ALLOC): Ditto.
2969 (GOMP_SIMT_EXIT): Ditto.
2970 * target-insns.def (omp_simt_enter): New insn.
2971 (omp_simt_exit): Ditto.
2972 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
2974 (lower_rec_simd_input_clauses): Implement SIMT privatization.
2975 (lower_rec_input_clauses): Likewise.
2976 (lower_lastprivate_clauses): Handle SIMT privatization.
2978 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
2979 (ompdevlow_adjust_simt_enter): New.
2980 (find_simtpriv_var_op): New.
2981 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
2982 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
2984 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
2985 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
2986 (copy_decl_for_dup_finish): Ditto.
2988 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
2990 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
2993 * config/i386/i386.c (ix86_option_override_internal): Always
2994 allow -mpreferred-stack-boundary=3 for 64-bit targets.
2996 2017-03-28 Bin Cheng <bin.cheng@arm.com>
2998 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
3000 2017-03-28 Bin Cheng <bin.cheng@arm.com>
3002 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
3003 mark new edge's irreducible flag accordign to it.
3004 (vect_do_peeling): Check loop preheader edge's irreducible flag
3005 and pass it to function slpeel_add_loop_guard.
3007 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
3009 PR tree-optimization/80218
3010 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
3011 Update block frequencies and counts.
3013 2017-03-28 Richard Biener <rguenther@suse.de>
3015 PR tree-optimization/78644
3016 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
3017 of a simplification result we may not use it at all.
3019 2017-03-28 Richard Biener <rguenther@suse.de>
3022 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
3023 without arguments, generate default definition of a SSA name.
3025 2017-03-28 Richard Biener <rguenther@suse.de>
3028 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
3029 TYPE_REF_CAN_ALIAS_ALL references.
3030 * fold-const.c (fold_indirect_ref_1): Likewise.
3032 2017-03-28 Martin Liska <mliska@suse.cz>
3035 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
3036 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
3038 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
3039 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
3041 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
3042 (EXTRA_SPECS): Define.
3043 (SUBTARGET_EXTRA_SPECS): Likewise.
3044 (SUBTARGET_CPP_SPEC): Likewise.
3045 * config/arc/elf.h (EXTRA_SPECS): Renamed to
3046 SUBTARGET_EXTRA_SPECS.
3047 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
3049 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
3051 * config/arc/simdext.md (vst64_insn): Update pattern.
3052 (vld32wh_insn): Likewise.
3053 (vld32wl_insn): Likewise.
3054 (vld64_insn): Likewise.
3055 (vld32_insn): Likewise.
3057 2017-03-28 Marek Polacek <polacek@redhat.com>
3060 * fold-const.c (fold_comparison): Use protected_set_expr_location
3061 instead of SET_EXPR_LOCATION.
3063 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
3065 * tree.c (add_expr): Avoid name lookup warning.
3067 2017-03-27 Jeff Law <law@redhat.com>
3069 PR tree-optimization/80216
3070 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
3071 function name. Limit recursion depth.
3072 (record_temporary_equivalences): Corresponding changes.
3074 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
3076 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
3079 2017-03-27 Jakub Jelinek <jakub@redhat.com>
3082 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
3084 * cfgcleanup.c (reg_note_cfa_p): New array.
3085 (insns_have_identical_cfa_notes): New function.
3086 (old_insns_match_p): Don't cross-jump in between /f
3087 and non-/f instructions. If both i1 and i2 are frame related,
3088 verify all CFA notes, their order and content.
3090 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
3093 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
3094 HImode and SImode with zero extend to DImode to one insn.
3095 (bswap<mode>2_extenddi): Likewise.
3096 (bswapsi2_extenddi): Likewise.
3097 (bswaphi2_extendsi): Likewise.
3098 (bswaphi2): Combine bswap HImode and SImode into one insn.
3099 Separate memory insns from swapping register.
3100 (bswapsi2): Likewise.
3101 (bswap<mode>2): Likewise.
3102 (bswaphi2_internal): Delete, no longer used.
3103 (bswapsi2_internal): Likewise.
3104 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
3105 store, and gpr<-gpr swap insns.
3106 (bswap<mode>2_store): Likewise.
3107 (bswaphi2_reg): Register only splitter, combine with the splitter.
3108 (bswaphi2 splitter): Likewise.
3109 (bswapsi2_reg): Likewise.
3110 (bswapsi2 splitter): Likewise.
3111 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
3112 the insns into load, store, and register/register insns.
3113 (bswapdi2_ldbrx): Likewise.
3114 (bswapdi2_load): Likewise.
3115 (bswapdi2_store): Likewise.
3116 (bswapdi2_reg): Likewise.
3118 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
3120 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
3121 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
3123 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
3126 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
3128 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3129 special handling for target option conflicts between dform
3130 options (-mpower9-dform, -mpower9-dform-vector,
3131 -mpower9-dform-scalar) and -mno-direct-move.
3133 2017-03-27 Richard Biener <rguenther@suse.de>
3135 PR tree-optimization/80181
3136 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
3138 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
3140 * config/arc/predicates.md (move_double_src_operand): Replace the
3141 call to move_double_src_operand with a call to address_operand.
3143 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
3145 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
3146 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
3147 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
3149 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
3151 * config/arc/predicates.md (long_immediate_loadstore_operand):
3152 Consider scaled addresses cases.
3154 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
3156 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
3157 restored when in interrupt.
3158 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
3159 doesn't have delay slot.
3161 2017-03-27 Richard Biener <rguenther@suse.de>
3164 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
3165 inlined thunk clones.
3167 2017-03-27 Jakub Jelinek <jakub@redhat.com>
3170 * asan.c (instrument_derefs): Copy over last operand from
3171 original COMPONENT_REF to the new COMPONENT_REF with
3172 DECL_BIT_FIELD_REPRESENTATIVE.
3173 * ubsan.c (instrument_object_size): Likewise.
3175 2017-03-27 Richard Biener <rguenther@suse.de>
3177 PR tree-optimization/80170
3178 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
3179 sure DR/SCEV didnt fold in constants we do not see when looking
3180 at the reference base alignment.
3182 2017-03-27 Richard Biener <rguenther@suse.de>
3185 * gimple-fold.c (fold_ctor_reference): Properly guard against
3186 NULL return value from canonicalize_constructor_val.
3188 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
3191 * config/i386/i386.c (ix86_expand_builtin)
3192 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
3193 flags reg setting and flags reg using instructions.
3194 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
3195 clobbering instructions to zero extend op2.
3197 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
3199 * doc/install.texi (Configuration) <--with-aix-soname>:
3200 Update link to AIX ld.
3202 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
3204 PR rtl-optimization/80160
3205 PR rtl-optimization/80159
3206 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
3207 reg_alternate_class into account.
3209 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
3212 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
3213 to consider in curr_insn_transform.
3215 2017-03-24 Jakub Jelinek <jakub@redhat.com>
3217 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
3218 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
3219 and emit_mode_inner.
3221 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3223 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
3224 argument to the overloaded builtin variants. Use the new flag to
3225 deprecate certain builtin variants.
3226 * config/s390/s390-builtin-types.def: Add new builtin types.
3227 * config/s390/s390-builtins.h: Support new flags field for
3228 overloaded builtins.
3229 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
3230 (s390_macro_to_expand): Enable vector float data type.
3231 (s390_cpu_cpp_builtins_internal): Indicate support of the new
3232 builtins by incrementing the __VEC__ version number.
3233 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
3235 (s390_resolve_overloaded_builtin): Emit error messages depending
3236 on the builtin flags.
3237 * config/s390/s390.c (s390_expand_builtin): Support additional
3238 flags argument. Change error message to match the messages
3239 emitted in s390-c.c.
3240 * config/s390/s390.md: New UNSPEC_* constants.
3241 (op_type): Add new instruction types.
3242 * config/s390/vecintrin.h: Add new builtins and test data class
3244 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
3245 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
3246 (VEC_INEXACT, VEC_NOINEXACT): New constants.
3247 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
3248 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
3249 ("vec_mergel<mode>"): V_HW -> VEC_HW.
3251 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
3252 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
3253 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
3254 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
3256 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
3257 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
3258 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
3259 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
3261 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
3262 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
3263 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
3264 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
3265 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
3266 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
3267 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
3269 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
3270 ("vec_scatter_element<V_HW_4:mode>_DI")
3271 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
3272 ("vec_fpint<mode>", "vflls")
3273 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
3274 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
3275 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
3276 ("*vec_cmphe<mode>_cc"): ... these.
3278 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
3279 mode constant instead of magic value.
3281 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3283 * config/s390/s390.c (s390_expand_vec_compare): Support other
3284 vector floating point modes than just V2DF.
3285 (s390_expand_vcond): Likewise.
3286 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
3287 (s390_cannot_change_mode_class): Prevent mode changes between TF
3288 and V1TF in vector registers.
3289 * config/s390/s390.md (DF, SF): New mode attributes.
3290 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
3291 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
3292 SFmode support for VRs.
3293 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
3295 (VFT, VF_HW): New mode iterators.
3296 (vw, sdx): New mode attributes.
3297 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
3298 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
3299 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
3300 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
3301 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
3302 also the new vector floating point modes. Renaming to ...
3304 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
3305 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
3306 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
3307 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
3308 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
3309 ("vec_unordered<mode>"): ... these.
3311 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
3312 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
3313 ("*vec_extendv2df"): New insn definitions.
3315 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3317 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
3318 ("mulditi3_2", "*muldi3_sign"): New patterns.
3319 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
3320 rename the pattern definition.
3322 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3324 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
3326 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
3328 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3330 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
3331 instruction if possible.
3332 * config/s390/vector.md (vec_halfnumelts): New mode
3334 ("*vec_vllezlf<mode>"): New pattern.
3336 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3338 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
3339 ("popcountv4si2", "popcountv2di2"): Rename to ...
3340 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
3341 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
3343 ("popcount<mode>2_vxe"): New pattern.
3345 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3347 * common/config/s390/s390-common.c (processor_flags_table): Add
3349 * config.gcc: Add arch12.
3350 * config/s390/driver-native.c (s390_host_detect_local_cpu):
3351 Default to arch12 for unknown CPU model numbers.
3352 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
3353 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
3354 PROCESSOR_max sanity check.
3355 * config/s390/s390-opts.h (enum processor_type): Add
3357 * config/s390/s390.c (processor_table): Add arch12.
3358 (s390_expand_builtin): Add check for B_VXE flag.
3359 (s390_issue_rate): Add PROCESSOR_ARCH12.
3360 (s390_get_sched_attrmask): Likewise.
3361 (s390_get_unit_mask): Likewise.
3362 (s390_sched_score): Enable z13 scheduling for arch12.
3363 (s390_sched_reorder): Likewise.
3364 (s390_sched_variable_issue): Likewise.
3365 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
3367 (s390_tune_attr): Use z13 scheduling also for arch12.
3368 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
3369 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
3370 (TARGET_VXE_P): New macros.
3371 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
3372 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
3373 * config/s390/s390.opt: Add arch12 as processor_type.
3375 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3377 * config/s390/s390.md
3378 ("fixuns_truncdddi2", "fixuns_trunctddi2")
3379 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
3380 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
3382 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
3383 Rename expanders to ...
3385 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
3386 ("fixuns_truncdddi2_emu"): ... these.
3388 ("fixuns_trunc<mode>si2_emu"): New expander.
3390 ("*fixuns_truncdfdi2_z13"): Rename to ...
3391 ("*fixuns_truncdfdi2_vx"): ... this.
3393 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3395 * config/s390/2964.md: Remove the single element vector compare
3396 instructions which are no longer used.
3397 * config/s390/s390.c (s390_select_ccmode): Remove handling of
3399 (s390_canonicalize_comparison): Remove handling of DFmode
3401 (s390_expand_vec_compare_scalar): Remove function.
3402 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
3403 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
3405 ("*cmp<mode>_ccs"): Add wfcdb instruction.
3407 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3409 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
3411 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
3412 will anyway by matched by mov<mode>_64dfp.
3414 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3416 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
3417 vlef/vstef. Add missing operand to vleif.
3419 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3421 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
3422 pair for all vector types with 64 bit elements.
3423 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
3424 * config/s390/vector.md (V_HW_64): ... here.
3425 (V_128_NOSINGLE): New mode iterator.
3426 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
3427 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
3428 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
3429 ("*vec_load_pairv2di"): Change to ...
3430 ("*vec_load_pair<mode>"): ... this one.
3432 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3434 * config/s390/constraints.md: Add comments.
3435 (jKK): Reject element sizes > 8 bytes.
3436 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
3438 * config/s390/s390.md: Add the s_operand checks formerly in
3439 s390_split_ok_p to various splitters where they are still
3441 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
3442 for 128 bit vectors. Plus two splitters.
3444 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3446 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
3449 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3452 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
3453 error if the boundary argument is not constant.
3455 2017-03-24 Jakub Jelinek <jakub@redhat.com>
3457 PR rtl-optimization/80112
3458 * loop-doloop.c (doloop_condition_get): Don't check condition
3459 if cmp isn't SET with IF_THEN_ELSE src.
3461 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3463 PR tree-optimization/80158
3464 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
3465 replacing a candidate statement, also replace it for the
3466 candidate's alternate interpretation.
3467 (replace_rhs_if_not_dup): Likewise.
3468 (replace_one_candidate): Likewise.
3470 2017-03-24 Richard Biener <rguenther@suse.de>
3472 PR tree-optimization/80167
3473 * graphite-isl-ast-to-gimple.c
3474 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
3476 (translate_isl_ast_to_gimple::get_rename): Likewise.
3478 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
3480 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
3481 handling of certain combinations of target options, including the
3482 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
3483 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
3485 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
3488 * config/arm/arm.md (*load_multiple): Add reload_completed to
3491 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3492 Richard Biener <rguenth@suse.de>
3494 PR tree-optimization/79908
3495 PR tree-optimization/80136
3496 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
3497 been cast away, gimplify_and_add suffices.
3499 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
3501 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
3503 2017-03-23 Richard Biener <rguenther@suse.de>
3505 PR tree-optimization/80032
3506 * gimplify.c (gimple_push_cleanup): Forced unconditional
3507 cleanups still have to go to the conditional_cleanups
3510 2017-03-22 Jakub Jelinek <jakub@redhat.com>
3512 PR tree-optimization/80072
3513 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
3515 (next_operand_entry_id): Change type to unsigned int.
3516 (sort_by_operand_rank): Make sure to return the right return value
3517 even if unsigned fields are bigger than INT_MAX.
3518 (struct oecount): Change cnt and id type to unsigned int.
3519 (oecount_hasher::equal): Formatting fix.
3520 (oecount_cmp): Make sure to return the right return value
3521 even if unsigned fields are bigger than INT_MAX.
3522 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
3525 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
3526 TREE_READONLY on result if writing it more than once.
3529 * doc/invoke.texi (-fsanitize=thread): Document that with
3530 -fnon-call-exceptions atomics are not able to throw
3534 * tsan.c: Include tree-eh.h.
3535 (instrument_builtin_call): Call maybe_clean_eh_stmt or
3536 maybe_clean_or_replace_eh_stmt where needed.
3537 (instrument_memory_accesses): Add cfg_changed argument.
3538 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
3539 if it returned true.
3540 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
3542 PR rtl-optimization/63191
3543 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
3544 wrapper function, moved the whole old content into ...
3545 (ix86_delegitimize_address_1): ... this. New inline function.
3546 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
3547 true as last argument instead of ix86_delegitimize_address.
3549 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
3551 * config/aarch64/aarch64.c (generic_branch_cost): Copy
3552 cortexa57_branch_cost.
3554 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
3556 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
3558 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3561 * doc/md.texi (Constraints): Document wA constraint.
3562 * config/rs6000/constraints.md (wA): New.
3563 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
3564 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
3565 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
3566 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
3568 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
3571 * gimplify.c (is_oacc_declared): New function.
3572 (oacc_default_clause): Use it to set default flags for acc declared
3573 variables inside parallel regions.
3574 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
3576 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
3577 declare attribute to any decl as necessary.
3579 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
3582 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
3583 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
3584 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
3585 (arm_arch_lpae): This.
3586 * config/arm/arm.c (arm_arch7ve): Rename into ...
3587 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
3588 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
3591 2017-03-22 Martin Liska <mliska@suse.cz>
3594 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
3595 error message instead of an ICE.
3597 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3599 * doc/extend.texi (6.11 Additional Floating Types): Revise.
3601 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
3603 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
3605 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
3608 2017-03-21 Martin Sebor <msebor@redhat.com>
3610 * doc/extend.texi: Use "cannot" instead of "can't."
3611 * doc/hostconfig.texi: Same.
3612 * doc/install.texi: Same.
3613 * doc/invoke.texi: Same.
3614 * doc/loop.texi: Same.
3615 * doc/md.texi: Same.
3616 * doc/objc.texi: Same.
3617 * doc/rtl.texi: Same.
3618 * doc/tm.texi: Same.
3619 * doc/tm.texi.in: Same.
3620 * doc/trouble.texi: Same.
3622 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
3625 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
3626 (collect_checksum_attributes): Set it.
3627 (die_checksum_ordered): Use it.
3629 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3631 PR tree-optimization/79908
3632 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
3633 change: For a VA_ARG whose LHS has been cast away, use
3634 force_gimple_operand to construct the side effects.
3636 2017-03-21 David Malcolm <dmalcolm@redhat.com>
3638 PR translation/80001
3639 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
3640 more amenable to translation.
3641 (oacc_loop_auto_partitions): Likewise.
3643 2017-03-21 Marek Polacek <polacek@redhat.com>
3644 Martin Sebor <msebor@redhat.com>
3646 PR tree-optimization/80109
3647 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
3650 2017-03-21 Jakub Jelinek <jakub@redhat.com>
3651 Segher Boessenkool <segher@kernel.crashing.org>
3654 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
3655 check reg_used_between_p between insn and one of succ or succ2
3656 depending on if succ is artificial insn not inserted into insn
3659 2017-03-21 Martin Liska <mliska@suse.cz>
3661 PR gcov-profile/80081
3662 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
3663 * doc/gcc.texi: Include gcov-dump stuff.
3664 * doc/gcov-dump.texi: New file.
3666 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
3668 PR rtl-optimization/79150
3669 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
3670 conditional jump, if the jump is the last insn of the loop.
3672 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3673 Richard Biener <rguenth@suse.de>
3675 PR tree-optimization/79908
3676 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
3677 been cast away, use force_gimple_operand to construct the side
3680 2017-03-21 Martin Liska <mliska@suse.cz>
3683 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
3686 2017-03-21 Brad Spengler <spender@grsecurity.net>
3689 * plugin.c (htab_hash_plugin): New function.
3690 (add_new_plugin): Use it and adjust.
3691 (parse_plugin_arg_opt): Adjust.
3692 (init_one_plugin): Likewise.
3694 2017-03-21 Richard Biener <rguenther@suse.de>
3696 PR tree-optimization/80032
3697 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
3698 if set force the cleanup to happen unconditionally.
3699 (gimplify_target_expr): Push inserted clobbers with force_uncond
3700 to avoid them being removed by control-dependent DCE.
3702 2017-03-21 Richard Biener <rguenther@suse.de>
3704 PR tree-optimization/80122
3705 * tree-inline.c (copy_bb): Do not expans va-arg packs or
3706 va_arg_pack_len when the inlined call stmt requires pack
3708 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
3710 2017-03-21 Jakub Jelinek <jakub@redhat.com>
3713 * tsan.c (instrument_builtin_call): If the memory model argument
3714 is not a constant, assume it is valid.
3717 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
3720 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
3722 PR rtl-optimization/79910
3723 * combine.c (can_combine_p): Do not allow combining an I0 or I1
3724 if its dest is used by an insn before I2 (other than the combined
3725 insns themselves, which are properly handled already).
3727 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
3730 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
3732 * combine.c (record_used_regs): New static function.
3733 (try_combine): Handle situations where there is an additional
3734 instruction between I2 and I3 which needs to have a LOG_LINK
3738 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
3740 * combine.c (try_combine): Delete redundant i1 test. Call
3741 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
3743 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
3746 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
3749 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
3751 PR tree-optimization/80054
3752 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
3753 the optimization if a PHI or any of its arguments is not dominated
3754 by the candidate's basis. Use gphi* rather than gimple* as
3756 (replace_profitable_candidates): Clean up a gimple* variable that
3757 should be a gphi* variable.
3759 2017-03-20 Martin Sebor <msebor@redhat.com>
3762 * doc/extend.texi (attribute constructor): Document present limitation.
3764 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
3767 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
3768 __POWER9_VECTOR__ #ifdef control, change template definition to
3769 use Power9-specific built-in function.
3770 (vec_any_eq): Likewise.
3771 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
3772 to control outcomes from this test.
3773 (vector_ae_<mode>p): For VEC_F modes, likewise.
3775 2017-03-20 Ian Lance Taylor <iant@google.com>
3777 * config/i386/i386.c (ix86_function_regparm): Save an extra
3778 register for -fsplit-stack with DECL_STATIC_CHAIN.
3780 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
3783 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
3784 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
3786 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
3788 * config/riscv/riscv.c (riscv_print_operand): Use "fence
3790 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
3793 2017-03-20 Marek Polacek <polacek@redhat.com>
3796 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
3798 2017-03-20 Richard Biener <rguenther@suse.de>
3800 PR tree-optimization/80113
3801 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
3802 allocate extra SSA name for PHI def.
3803 (add_close_phis_to_outer_loops): Likewise.
3804 (add_close_phis_to_merge_points): Likewise.
3805 (copy_loop_close_phi_args): Likewise.
3806 (copy_cond_phi_nodes): Likewise.
3808 2017-03-20 Martin Liska <mliska@suse.cz>
3811 * tree-chkp.c (chkp_build_returned_bound): Do not build
3812 returned bounds for a LHS that's not a BOUNDED_P type.
3814 2017-03-20 Martin Liska <mliska@suse.cz>
3818 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
3819 COMPLEX_CST and VECTOR_CST.
3821 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
3824 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
3825 target operand. A new splitter adds the clobber statement in case
3826 the target operand is dead anyway.
3828 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
3830 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
3831 to age-old versions of binutils and glibc.
3833 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
3835 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
3837 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
3839 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
3841 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
3843 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
3844 requirement for binutils 2.13.
3846 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
3848 * combine.c (try_combine): Delete redundant i1 test. Call
3849 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
3851 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
3853 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
3854 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
3856 <riscv64-*-elf>: Re-arrange section
3857 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
3858 <riscv32-*-linux>: Likewise.
3859 <riscv64-*-elf>: Likewise
3860 <riscv64-*-linux>: Likewise.
3862 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
3865 * aarch64.opt(verbose-cost-dump): Fix typo.
3867 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
3870 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
3871 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
3873 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
3875 * reload.c (find_reloads): When reloading a nonoffsettable address,
3876 use RELOAD_OTHER for it and its address reloads.
3878 PR rtl-optimization/79910
3879 * combine.c (record_used_regs): New static function.
3880 (try_combine): Handle situations where there is an additional
3881 instruction between I2 and I3 which needs to have a LOG_LINK
3884 2017-03-17 Jeff Law <law@redhat.com>
3886 PR tree-optimization/71437
3887 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
3888 conditional in the hash table first.
3889 (vrp_dom_walker::before_dom_children): Extract condition from
3890 ASSERT_EXPR. Record condition, its inverion and any implied
3893 2017-03-17 Marek Polacek <polacek@redhat.com>
3894 Markus Trippelsdorf <markus@trippelsdorf.de>
3896 PR tree-optimization/80079
3897 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
3900 2017-03-17 Richard Biener <rguenther@suse.de>
3903 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
3904 Properly verify the LHS before the RHS possibly claims to be
3906 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
3909 2017-03-17 Martin Jambor <mjambor@suse.cz>
3911 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
3912 (List of -O2 options): Likewise.
3913 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
3916 2017-03-17 Tom de Vries <tom@codesourcery.com>
3918 * gcov-dump.c (print_usage): Print bug_report_url.
3920 2017-03-17 Richard Biener <rguenther@suse.de>
3923 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
3924 (parser::peek): Likewise.
3926 2017-03-17 Richard Biener <rguenther@suse.de>
3928 PR tree-optimization/80048
3929 * sese.c (free_sese_info): Properly release rename_map and
3930 copied_bb_map elements.
3932 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
3934 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
3935 Add linked-list forward and backlinks. Insert on
3936 construction, remove on destruction.
3937 (class pass_store_merging): Add m_stores_head field.
3938 (pass_store_merging::terminate_and_process_all_chains):
3939 Iterate over m_stores_head list.
3940 (pass_store_merging::terminate_all_aliasing_chains):
3942 (pass_store_merging::execute): Check for debug stmts first.
3943 Push new chains onto the m_stores_head stack.
3945 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
3948 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
3949 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
3950 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
3952 2017-03-16 Jeff Law <law@redhat.com>
3954 PR tree-optimization/71437
3955 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
3956 member function. Implementation moved into after_dom_children
3957 member function and into the threader's thread_outgoing_edges
3959 (dom_opt_dom_walker::after_dom_children): Simplify by moving
3960 some code into new thread_outgoing_edges.
3961 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
3962 definition. Simplify marker handling (do it here). Assume we always
3963 have the available expression and the const/copies tables.
3964 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
3966 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
3967 * tree-vrp.c (equiv_stack): No longer file scoped.
3968 (vrp_dom_walker): New class.
3969 (vrp_dom_walker::before_dom_children): New member function.
3970 (vrp_dom_walker::after_dom_children): Likewise.
3971 (identify_jump_threads): Setup domwalker. Use it rather than
3972 walking edges in a random order by hand. Simplify setup/finalization.
3973 (finalize_jump_threads): Remove.
3974 (vrp_finalize): Do not call identify_jump_threads here.
3975 (execute_vrp): Do it here instead and call thread_through_all_blocks
3978 PR tree-optimization/71437
3979 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
3981 (simplify_stmt_for_jump_threading): Add basic_block argument. All
3983 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
3984 (dom_opt_dom_walker::thread_across_edge): Remove
3985 handle_dominating_asserts argument. All callers changed.
3986 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
3987 changes. Remove calls to lhs_of_dominating_assert. Other
3988 uses of handle_dominating_asserts turn into unconditional code
3989 (simplify_control_stmt_condition_1): Likewise.
3990 (simplify_control_stmt_condition): Likewise.
3991 (thread_through_normal_block, thread_across_edge): Likewise.
3992 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
3993 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
3994 object if it is not an SSA_NAME.
3995 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
3996 before calling into the VRP specific simplifiers.
3997 (identify_jump_threads): Remove handle_dominating_asserts
4000 2017-03-16 Jakub Jelinek <jakub@redhat.com>
4003 * tree-diagnostic.c (default_tree_printer): No longer static.
4004 * tree-diagnostic.h (default_tree_printer): New prototype.
4006 2017-03-16 Tamar Christina <tamar.christina@arm.com>
4008 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
4009 Change ins into fmov.
4011 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4013 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
4014 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
4015 Use h_con constraint for operand 1.
4016 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
4017 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
4019 2017-03-15 Jeff Law <law@redhat.com>
4021 PR tree-optimization/71437
4022 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
4023 (record_temporary_equivalences): Use it.
4025 PR tree-optimization/71437
4026 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
4027 tree-ssa-scopedtables.
4028 (lookup_avail_expr, build_and_record_new_cond): Likewise.
4029 (record_conditions, record_cond, vuse_eq): Likewise.
4030 (record_edge_info): Adjust to API tweak of record_conditions.
4031 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
4032 (record_temporary_equivalences, optimize_stmt): Likewise.
4033 (eliminate_redundant_computations): Likewise.
4034 (record_equivalences_from_stmt): Likewise.
4035 * tree-ssa-scopedtables.c: Include options.h and params.h.
4036 (vuse_eq): New function, moved from tree-ssa-dom.c
4037 (build_and_record_new_cond): Likewise.
4038 (record_conditions): Likewise. Accept vector of conditions rather
4039 than edge_equivalence structure for first argument.
4040 for the first argument.
4041 (avail_exprs_stack::lookup_avail_expr): New member function, moved
4042 from tree-ssa-dom.c.
4043 (avail_exprs_stack::record_cond): Likewise.
4044 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
4045 from tree-ssa-dom.c.
4046 (avail_exprs_stack): Add new member functions lookup_avail_expr
4048 (record_conditions): Declare.
4050 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
4053 * lra-constraints.c (process_alt_operands): Increase reject for
4054 reloading an input/output operand.
4056 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
4059 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
4060 insns to convert from signed/unsigned char/short to IEEE 128-bit
4062 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
4064 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
4067 * config/i386/i386.c (ix86_vector_duplicate_value): Create
4068 subreg of inner mode for values already in registers.
4070 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
4072 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
4073 iteration reg is used after the loop.
4075 2017-03-14 Martin Sebor <msebor@redhat.com>
4077 PR tree-optimization/79800
4078 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
4079 precision in negative-positive range.
4080 (format_floating): Call non-const overload with adjusted precision.
4082 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
4085 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
4088 2017-03-14 Martin Sebor <msebor@redhat.com>
4091 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
4092 * builtins.def (aligned_alloc): Use it.
4095 * Makefile.in (GTFILES): Add calls.c.
4096 * calls.c: Include "gt-calls.h".
4098 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
4100 PR rtl-optimization/79728
4101 * regs.h (struct target_regs): New field
4102 x_contains_allocatable_regs_of_mode.
4103 (contains_allocatable_regs_of_mode): New macro.
4104 * reginfo.c (init_reg_sets_1): Initialize it, and change
4105 contains_reg_of_mode so it includes global regs as well.
4106 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
4107 rather than contains_regs_of_mode.
4109 2017-03-14 Martin Liska <mliska@suse.cz>
4111 * doc/invoke.texi: Document options that can't be combined with
4112 -fcheck-pointer-bounds.
4114 2017-03-14 Martin Liska <mliska@suse.cz>
4117 * doc/invoke.texi (-Wchkp): Document the option.
4119 2017-03-14 Martin Liska <mliska@suse.cz>
4121 * Makefile.in: Install gcov-dump.
4123 2017-03-14 Martin Liska <mliska@suse.cz>
4125 * multiple_target.c (expand_target_clones): Bail out for
4126 an invalid attribute.
4128 2017-03-14 Richard Biener <rguenther@suse.de>
4130 * alias.c (struct alias_set_entry): Pack properly.
4131 * cfgloop.h (struct loop): Likewise.
4132 * cse.c (struct set): Likewise.
4133 * ipa-utils.c (struct searchc_env): Likewise.
4134 * loop-invariant.c (struct invariant): Likewise.
4135 * lra-remat.c (struct cand): Likewise.
4136 * recog.c (struct change_t): Likewise.
4137 * rtl.h (struct address_info): Likewise.
4138 * symbol-summary.h (function_summary): Likewise.
4139 * tree-loop-distribution.c (struct partition): Likewise.
4140 * tree-object-size.c (struct object_size_info): Likewise.
4141 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
4142 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
4143 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
4144 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
4145 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
4146 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
4147 (struct _stmt_vec_info): Likewise.
4149 2017-03-14 Martin Liska <mliska@suse.cz>
4152 * multiple_target.c (create_dispatcher_calls): Check that
4153 a target can create a function dispatcher.
4155 2017-03-14 Martin Liska <mliska@suse.cz>
4158 * multiple_target.c (expand_target_clones): Drop local.local
4159 flag for default implementation.
4161 2017-03-14 Richard Biener <rguenther@suse.de>
4163 PR tree-optimization/80030
4164 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
4166 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
4168 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
4169 gcc_fallthrough() instead of __attribute__((fallthrough));
4171 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
4173 * doc/gcc.texi: Remove "up" link to (DIR).
4174 * doc/gccint.texi: Ditto.
4176 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
4178 * doc/install.texi (Specific) <avr>: Remove reference to
4181 2017-03-13 Jeff Law <law@redhat.com>
4183 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
4184 attribute rather than comments.
4186 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
4187 match_scratch operand is highest.
4189 2017-03-13 Martin Liska <mliska@suse.cz>
4192 * ipa-pure-const.c (warn_function_noreturn): If the declarations
4193 is a CHKP clone, use original declaration.
4195 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
4197 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
4198 (arc_conditional_register_usage): Use a different allocation order
4199 when optimizing for size.
4200 * common/config/arc/arc-common.c (arc_option_optimization_table):
4201 Section anchors default on when optimizing for size.
4203 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
4205 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
4207 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
4209 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
4210 * config/arc/arc.md (cpu_facility): Add cd variant.
4211 (*movqi_insn): Add code density variant.
4212 (*movhi_insn): Likewise.
4213 (*movqi_insn): Likewise.
4214 (*addsi3_mixed): Likewise.
4215 (subsi3_insn): Likewise.
4217 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
4219 * config/arc/arc.md (movsi_cond_exec): Update constraint.
4221 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
4223 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
4224 expressions with MINUS and UNARY ops.
4226 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4229 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
4231 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
4232 between vec_select and vector argument.
4233 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
4234 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
4235 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
4236 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
4237 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
4238 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
4240 2017-03-13 Richard Biener <rguenther@suse.de>
4243 * params.def (vect-max-peeling-for-alignment): Fix typo.
4245 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
4247 * doc/install.texi (Specific) <mips-*-*>: Remove description of
4248 issue that only occurred with binutils below 2.18.
4250 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
4252 * doc/install.texi (Specific) <cris-axis-elf>: No longer
4253 refer to binutils 2.11/2.12 minimum.
4255 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
4257 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
4258 ftp.kernel.org and simplify binutils requirement.
4260 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
4262 * doc/invoke.texi (Warning Options): Fix spelling of link-time
4264 (Optimize Options): Ditto. Also remove redundancy.
4266 2017-03-10 David Malcolm <dmalcolm@redhat.com>
4268 PR translation/79848
4269 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
4271 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
4272 to G_ to avoid double translation.
4274 2017-03-10 David Malcolm <dmalcolm@redhat.com>
4276 PR translation/79923
4277 * auto-profile.c (get_combined_location): Convert leading
4278 character of diagnostics to lower case and remove trailing period.
4279 (read_profile): Likewise for various diagnostics.
4280 * config/arm/arm.c (arm_option_override): Remove trailing period
4281 from various diagnostics.
4282 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
4283 (msp430_expand_delay_cycles): Likewise.
4285 2017-03-10 David Malcolm <dmalcolm@redhat.com>
4288 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
4289 full command-line argument, rather than just "str".
4290 (aarch64_validate_march): Likewise.
4291 (aarch64_validate_mtune): Likewise.
4293 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
4295 PR rtl-optimization/78911
4296 * lra-assigns.c (must_not_spill_p): New function.
4297 (spill_for): Use it.
4299 2017-03-10 Jakub Jelinek <jakub@redhat.com>
4301 PR tree-optimization/79981
4302 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
4303 ATOMIC_COMPARE_EXCHANGE ifn result.
4304 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
4305 IFN_ATOMIC_COMPARE_EXCHANGE.
4307 2017-03-10 David Malcolm <dmalcolm@redhat.com>
4310 * opts.c (parse_sanitizer_options): Add missing question mark to
4311 "did you mean" message.
4313 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4315 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
4317 (VMULEUH_UNS): Likewise.
4318 (VMULOUB_UNS): Likewise.
4319 (VMULOUH_UNS): Likewise.
4320 * config/rs6000/rs6000.c (builtin_function_type): Remove
4321 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
4323 2017-03-10 David Malcolm <dmalcolm@redhat.com>
4326 * read-rtl-function.c (function_reader::read_rtx_operand): Update
4327 x with result of extra_parsing_for_operand_code_0.
4328 (function_reader::extra_parsing_for_operand_code_0): Convert
4329 return type from void to rtx, returning x. When reading
4330 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
4331 larger size containing struct block_symbol.
4333 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
4335 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
4336 -mfloat128-hardware without -m64.
4338 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
4341 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
4342 entries to the case statement that marks unsigned arguments to
4343 overloaded functions.
4345 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
4347 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
4348 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
4350 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
4353 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
4354 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
4356 2017-03-10 Martin Liska <mliska@suse.cz>
4360 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
4361 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
4364 2017-03-10 Olivier Hainque <hainque@adacore.com>
4366 * tree-switch-conversion (array_value_type): Start by resetting
4367 candidate type to it's main variant.
4369 2017-03-10 Jakub Jelinek <jakub@redhat.com>
4371 PR rtl-optimization/79909
4372 * combine.c (try_combine): Use simplify_replace_rtx on individual
4373 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
4374 of the whole CALL_INSN_FUNCTION_USAGE.
4376 PR tree-optimization/79972
4377 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
4378 get_range_info on SSA_NAMEs. Formatting fixes.
4380 2017-03-10 Richard Biener <rguenther@suse.de>
4381 Jakub Jelinek <jakub@redhat.com>
4383 PR tree-optimization/77975
4384 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
4385 edge to be constant.
4386 (get_val_for): For constant x return it. Formatting fix.
4387 (loop_niter_by_eval): Avoid pointless looping if the next iteration
4388 would use the same bases as the current one.
4390 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4392 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
4393 instead of vec_select for V1TImode.
4394 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
4396 (VSX_LE_128): Add V1TI to this mode iterator.
4397 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
4398 (*vsx_le_perm_store_<mode>): Likewise.
4399 (pre-reload splitter for VSX stores): Likewise.
4400 (post-reload splitter for VSX stores): Likewise.
4401 (*vsx_xxpermdi2_le_<mode>): Likewise.
4402 (*vsx_lxvd2x2_le_<mode>): Likewise.
4403 (*vsx_stxvd2x2_le_<mode>): Likewise.
4405 2017-03-09 Michael Eager <eager@eagercon.com>
4407 Correct failures with --enable-checking=yes,rtl.
4409 * config/microblaze/microblaze.c (microblaze_expand_shift):
4410 Replace GET_CODE test with CONST_INT_P and INTVAL test with
4411 test for const0_rtx.
4412 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
4413 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
4415 2017-03-09 Richard Biener <rguenther@suse.de>
4417 PR tree-optimization/79977
4418 * graphite-scop-detection.c (scop_detection::merge_sese):
4419 Handle the case of extra exits to blocks dominating the entry.
4421 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
4423 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
4426 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
4428 PR rtl-optimization/79949
4429 * lra-constraints.c (process_alt_operands): Check memory when
4430 trying to predict a cycle. Print about the overall increase.
4432 2017-03-09 Richard Biener <rguenther@suse.de>
4435 * gimple-expr.c (useless_type_conversion_p): Preserve
4436 TYPE_SATURATING for fixed-point types.
4438 2017-03-09 Richard Biener <rguenther@suse.de>
4441 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
4442 alignment of BLKmode params.
4444 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4447 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
4448 (VALL_NO_V2Q): Likewise.
4450 * config/aarch64/aarch64-simd.md
4451 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
4453 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
4454 VALL_NO_V2Q mode iterator.
4455 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
4457 2017-03-09 Martin Liska <mliska@suse.cz>
4459 PR tree-optimization/79631
4460 * tree-chkp-opt.c (chkp_is_constant_addr): Call
4461 tree_int_cst_sign_bit just for INTEGER constants.
4463 2017-03-09 Martin Liska <mliska@suse.cz>
4467 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
4470 2017-03-09 Marek Polacek <polacek@redhat.com>
4473 * tree.c (inchash::add_expr): Handle TREE_VEC.
4475 2017-03-09 Martin Liska <mliska@suse.cz>
4478 (chkp_narrow_size_and_offset): New function.
4479 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
4480 (void chkp_parse_bit_field_ref): New function.
4481 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
4482 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
4484 2017-03-09 Martin Liska <mliska@suse.cz>
4487 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
4488 (chkp_find_bounds_1): Remove gcc_unreachable.
4490 2017-03-09 Jakub Jelinek <jakub@redhat.com>
4493 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
4494 BUILT_IN_SYNC*, determine the access type from the size suffix and
4495 always build a MEM_REF with that type. Handle forgotten
4496 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
4499 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
4500 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
4501 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
4502 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
4503 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
4504 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
4505 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
4506 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
4507 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
4508 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
4509 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
4510 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
4511 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
4512 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
4513 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
4514 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
4515 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
4516 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
4517 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
4518 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
4519 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
4520 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
4521 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
4522 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
4523 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
4524 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
4525 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
4526 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
4527 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
4528 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
4529 definitions outside of __OPTIMIZE__ guarded section.
4532 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
4533 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
4534 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
4535 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
4538 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
4540 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
4541 ("vfenez<mode>"): Add missing constraints.
4543 2017-03-08 Martin Sebor <msebor@redhat.com>
4546 * config/nds32/nds32.c (nds32_option_override):
4547 Fix misspelled diagnostic.
4549 2017-03-08 Jakub Jelinek <jakub@redhat.com>
4552 * gimplify.c (gimplify_omp_for): Replace index var in outer
4553 taskloop statement with an artificial variable and add
4554 OMP_CLAUSE_PRIVATE clause for it.
4556 2017-03-08 Richard Biener <rguenther@suse.de>
4558 PR tree-optimization/79955
4559 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
4560 for accesses that are completely outside of the variable.
4562 2017-03-08 Andrew Haley <aph@redhat.com>
4564 PR tree-optimization/79943
4565 * tree-ssa-loop-split.c (compute_new_first_bound): When
4566 calculating the new upper bound, (END-BEG) should be added, not
4569 2017-03-08 Jakub Jelinek <jakub@redhat.com>
4571 * config/avr/avr.md (setmemhi): Make sure match_dup
4572 operand number comes before match_scratch.
4574 2017-03-08 Richard Biener <rguenther@suse.de>
4576 PR tree-optimization/79920
4577 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
4578 with ncopies == 1 to ...
4579 (vect_transform_slp_perm_load): ... here. Properly compute
4580 all element loads by iterating VF times over the group. Do
4581 not handle ncopies (computed in a broken way) in
4582 vect_create_mask_and_perm.
4584 2017-03-08 Jakub Jelinek <jakub@redhat.com>
4587 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
4588 is a uniform vector, use uniform_vector_p return value instead of
4589 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
4591 2017-03-07 Marek Polacek <polacek@redhat.com>
4594 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
4595 (alloca_call_type): Likewise.
4597 2017-03-07 Martin Liska <mliska@suse.cz>
4599 * gcov.c (process_args): Put comment to correct location.
4601 2017-03-07 Martin Liska <mliska@suse.cz>
4604 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
4605 Use array_at_struct_end_p instead of DECL_CHAIN (field).
4606 (chkp_narrow_bounds_for_field): Likewise.
4607 (chkp_parse_array_and_component_ref): Pass one more argument to
4610 2017-03-07 Richard Biener <rguenther@suse.de>
4612 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
4615 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
4617 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
4618 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
4620 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4623 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
4624 to end of description.
4625 (PARAM_MAX_STORES_TO_MERGE): Likewise.
4627 2017-03-07 Jakub Jelinek <jakub@redhat.com>
4629 PR rtl-optimization/79901
4630 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
4632 (*avx512f_<code><mode>3<mask_name>): ... this.
4633 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
4634 iterator instead of VI8_AVX2_AVX512BW.
4636 PR rtl-optimization/79901
4637 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
4638 min/max expander, expand it using expand_vec_cond_expr.
4641 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
4644 2017-03-06 Jakub Jelinek <jakub@redhat.com>
4647 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
4648 to void * for PCH reasons.
4649 * dwarf2out.c (output_loc_operands, output_die): Cast
4650 v.val_vec.array to unsigned char *.
4652 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
4655 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
4658 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
4660 PR rtl-optimization/79571
4661 * lra-constraints.c (process_alt_operands): Calculate static
4662 reject and subtract it from overall when only addresses will be
4665 2017-03-06 Julia Koval <julia.koval@intel.com>
4668 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
4669 incoming stack boundary to 128 for 64-bit targets.
4671 2017-03-06 Richard Biener <rguenther@suse.de>
4673 PR tree-optimization/79894
4674 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
4675 to NULL after folding it.
4677 2017-03-06 Richard Biener <rguenther@suse.de>
4679 PR tree-optimization/79824
4680 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
4681 check disabling peeling for gaps.
4683 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
4685 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
4686 attributes): Document gettimeofday.
4688 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
4690 * config/s390/s390.c (s390_option_override_internal): Set
4691 PARAM_MIN_VECT_LOOP_BOUND
4693 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
4695 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
4696 * config/s390/s390.md: Likewise.
4698 2017-03-06 Jakub Jelinek <jakub@redhat.com>
4701 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
4702 (<avx2_avx512>_perm<mode>): Rename to ...
4703 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
4705 (<avx512>_perm<mode>_mask): Rename to ...
4706 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
4708 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
4709 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
4710 instead of VI8F_256_512.
4711 (avx512f_perm<mode>): New define_expand.
4712 (avx512f_perm<mode>_mask): Likewise.
4713 (avx512f_perm<mode>_1<mask_name>): New define_insn.
4714 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
4716 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
4718 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
4719 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
4721 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
4723 2017-03-06 Martin Liska <mliska@suse.cz>
4726 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
4727 when having a SSA NAME w/o VAR_DECL assigned to it.
4729 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
4731 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
4732 msa_dpsub_<su>_d): Fix MODE for vec_select.
4734 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
4736 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
4738 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
4740 2017-03-06 Richard Biener <rguenther@suse.de>
4742 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
4743 * plugin.c (register_plugin_info): Likewise.
4744 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
4746 2017-03-05 Jakub Jelinek <jakub@redhat.com>
4748 * config/i386/sse.md (sse_storehps, sse_storelps,
4749 avx_<castmode><avxsizesuffix>_<castmode>,
4750 avx512f_<castmode><avxsizesuffix>_<castmode>,
4751 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
4752 in condition that at least one operand is not a MEM.
4754 2017-03-03 Jakub Jelinek <jakub@redhat.com>
4757 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
4758 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
4760 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
4761 gimple_call_nothrow_p flag based on whether original builtin can throw.
4762 If it can, emit following stmts on the fallthrough edge.
4763 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
4764 don't create new bb if inserting just debug stmts on the edge, try to
4765 insert them on the fallthru bb or just reset debug stmts.
4767 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
4770 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
4771 restore recog_data (including the operand rtxes inside it) around
4772 the call to get_insn_template.
4774 2017-03-03 Martin Sebor <msebor@redhat.com>
4776 PR tree-optimization/79699
4777 * context.c (context::~context): Free MPFR caches to avoid
4778 a memory leak on program exit.
4780 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4782 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
4783 Use wide_int::ulow () instead of .elt (0).
4785 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
4787 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
4788 (*pushxf): Limit oF constraint to 32bit targets and add oC
4789 constraint for 64bit targets.
4790 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
4791 (*pushdf): Change rmF constraint to rmC.
4793 2017-03-03 Martin Liska <mliska@suse.cz>
4795 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
4796 Remove unused variable.
4798 2017-03-03 Jakub Jelinek <jakub@redhat.com>
4801 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
4802 is a memory operand, increase num_memory.
4803 (ix86_expand_args_builtin): Likewise.
4805 2017-03-03 Jan Hubicka <jh@suse.cz>
4808 * ipa-devirt.c (maybe_record_node): Properly handle
4809 __cxa_pure_virtual visibility.
4811 2017-03-03 Martin Liska <mliska@suse.cz>
4813 PR tree-optimization/79803
4814 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
4816 (pass_loop_prefetch::execute): Disabled optimization if an
4817 assumption about L1 cache size is not met.
4819 2017-03-03 Martin Liska <mliska@suse.cz>
4821 PR rtl-optimization/79574
4822 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
4823 (hash_scan_set): Likewise.
4824 (dump_hash_table): Likewise.
4825 (hoist_code): Likewise.
4827 2017-03-03 Richard Biener <rguenther@suse.de>
4829 * fixed-value.c (fixed_from_string): Restore use of elt (1)
4830 in place of uhigh ().
4831 (fixed_convert_from_real): Likewise.
4833 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
4836 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
4838 2017-03-03 Richard Biener <rguenther@suse.de>
4841 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
4842 TYPE_OVERFLOW_UNDEFINED check.
4844 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4846 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
4848 (vector_ae_<mode>_p): Likewise.
4849 (vector_nez_<mode>_p): Likewise.
4850 (vector_ne_v2di_p): Likewise.
4851 (vector_ae_v2di_p): Likewise.
4852 (vector_ne_<mode>_p): Likewise.
4853 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
4855 (vsx_tsqrt<mode>2_fe): Likewise.
4857 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
4860 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
4862 2017-03-02 Jakub Jelinek <jakub@redhat.com>
4864 PR rtl-optimization/79780
4865 * cprop.c (one_cprop_pass): When second and further conditional trap
4866 in a single basic block is turned into an unconditional trap, turn it
4867 into a deleted note to avoid RTL verification failures.
4869 2017-03-02 Richard Biener <rguenther@suse.de>
4871 * fold-const.c (const_binop): Use ulow () instead of elt (0).
4873 2017-03-02 Richard Biener <rguenther@suse.de>
4875 PR tree-optimization/79345
4877 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
4878 param and abort the walk, returning -1 if it is hit.
4879 (walk_aliased_vdefs): Take a limit param and pass it on.
4880 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
4881 defaulting to 0 and return a signed int.
4882 * tree-ssa-uninit.c (struct check_defs_data): New struct.
4883 (check_defs): New helper.
4884 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
4885 about uninitialized memory.
4886 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
4887 bogus uninitialized warning.
4888 (fixed_convert_from_real): Likewise.
4890 2017-03-02 Bin Cheng <bin.cheng@arm.com>
4892 PR tree-optimization/66768
4893 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
4894 iv_use if base object can't be determined.
4896 2017-03-02 Jakub Jelinek <jakub@redhat.com>
4898 PR tree-optimization/79345
4899 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
4900 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
4901 (get_pattern_stats): Initialize it.
4902 * genemit.c (gen_expand): Verify match_scratch numbers come after
4903 match_operand/match_dup numbers.
4904 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
4905 match_scratch numbers.
4906 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
4908 * config/s390/s390.md (trunctdsd2): Likewise.
4910 2017-03-02 Richard Biener <rguenther@suse.de>
4912 * wide-int.h (wide_int_storage::operator=): Implement in terms
4915 2017-03-02 Richard Biener <rguenther@suse.de>
4917 PR tree-optimization/79777
4918 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
4919 the to insert expression to sth existing.
4921 2017-03-01 Martin Sebor <msebor@redhat.com>
4924 * gimple-ssa-sprintf.c
4925 (directive::known_width_and_precision): New function.
4926 (format_integer): Use it.
4927 (get_mpfr_format_length): Consider the full range of precision
4928 when computing %g output with the # flag. Set the likely byte
4929 count to 3 rather than 1 when precision is indeterminate.
4930 (format_floating): Correct the lower bound of precision.
4932 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
4934 * doc/invoke.texi: Document default code model for 64-bit Linux.
4936 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
4939 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
4940 udiv rather than div since input pattern is unsigned.
4942 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
4944 * config/i386/i386.c (print_reg): Warn for values of
4945 unsupported size in integer register.
4947 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
4950 * config/rs6000/predicates.md (current_file_function_operand): Do
4951 not allow self calls to be local if the function is replaceable.
4953 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
4956 * config/rs6000/altivec.h (vec_ctz and others): Change the
4957 preprocessor macro that controls conditional compilation from
4958 _ARCH_PWR9 to __POWER9_VECTOR__.
4959 (vec_all_ne): Change parameterization of __altivec_scalar_pred
4960 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
4961 control (instead of _ARCH_PWR9 control) so that template
4962 definition uses power9-specific function.
4963 (vec_any_eq): Likewise.
4964 (vec_all_ne): Change macro definition to use a power9-specific
4965 expansion under #ifdef __POWER9_VECTOR__ control (instead of
4966 _ARCH_PWR9 control).
4967 (vec_any_eq) Likewise.
4968 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
4969 expansion for CMPNEF to remove support for xvcmpnesp instruction.
4970 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
4971 support for xvcmpnedp instruction.
4972 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
4973 macro expansion so that Power9 implementation of vec_all_ne does
4974 not use the AltiVec predicate framework.
4975 (VCMPNEH_P): Likewise.
4976 (VCMPNEW_P): Likewise.
4977 (VCMPNED_P): Likewise.
4978 (VCMPNEFP_P): Likewise.
4979 (VCMPNEDP_P): Likewise.
4980 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
4981 implementation of vec_any_eq to not use AltiVec predicate
4983 (VCMPAEH_P): Likewise.
4984 (VCMPAEW_P): Likewise.
4985 (VCMPAED_P): Likewise.
4986 (VCMPAEFP_P): Likewise.
4987 (VCMPAEDP_P): Likewise.
4988 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
4989 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
4990 not use the AltiVec predicate framework.
4991 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
4992 of vec_any_eq to not use AltiVec predicate framework.
4993 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
4994 support for predefined __POWER9_VECTOR__ macro to indicate that
4995 Power9 instruction selection is enabled.
4996 (altivec_overloaded_builtins): Remove extraneous
4997 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
4998 function argument types RS6000_BTI_bool_V16QI and
4999 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
5000 entry for overloaded function argument types RS6000_BTI_bool_V4SI
5001 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
5002 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
5003 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
5004 Power9 for implementations of vec_cmpne. Change the signature for
5005 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
5006 (representing vec_all_ne) to remove the previously described first
5007 argument of type RS6000_BTI_INTSI, as this was an artifact of
5008 reliance on the AltiVec predicate framework, which is no longer
5009 used in the implementation of these functions. Add
5010 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
5011 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
5012 since, unlike the AltiVec predicate framework implementation, we
5013 do not share function descriptors between vec_alle and vec_anyeq.
5014 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
5015 set of modes that receive special treatment even when
5016 TARGET_P9_VECTOR is true. The special treatment emits code that
5017 does not depend on Power9 instructions.
5018 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
5019 define_expand to not rely on AltiVec predicate framework.
5020 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
5022 (vector_ne_v2di_p): Change this define_expand to not rely on
5023 AltiVec predicate framework.
5024 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
5026 (vector_ne_<mode>_p): Change this define_expand to not rely on
5027 AltiVec predicate framework.
5028 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
5030 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
5031 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
5032 define_insn pattern.
5033 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
5034 define_insn pattern because the xvcmpne<VSs>. instruction is not
5036 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
5037 instruction is not supported.
5039 2017-03-01 Jakub Jelinek <jakub@redhat.com>
5041 * config/nvptx/nvptx.c: Include intl.h.
5043 2017-03-01 Martin Jambor <mjambor@suse.cz>
5046 * ipa-prop.h (ipa_bits): Removed field known.
5047 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
5048 to pointers. Adjusted their comments to warn about their sharing.
5049 (ipcp_transformation_summary): Change bits to a vector of pointers.
5050 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
5051 (ipa_get_ipa_bits_for_value): Declare.
5052 * tree-vrp.h (value_range): Mark as GTY((for_user)).
5053 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
5054 (ipa_bits_hash_table): Likewise.
5055 (ipa_vr_ggc_hash_traits): Likewise.
5056 (ipa_vr_hash_table): Likewise.
5057 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
5058 being pointers and vr_known being removed.
5059 (ipa_set_jf_unknown): Likewise.
5060 (ipa_get_ipa_bits_for_value): New function.
5061 (ipa_set_jfunc_bits): Likewise.
5062 (ipa_get_value_range): New overloaded functions.
5063 (ipa_set_jfunc_vr): Likewise.
5064 (ipa_compute_jump_functions_for_edge): Use the above functions to
5065 construct bits and vr parts of jump functions.
5066 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
5067 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
5069 (ipcp_grow_transformations_if_necessary): Also allocate
5070 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
5072 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
5073 them. Fix too long lines.
5074 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
5075 vr_known being removed.
5076 (ipa_read_jump_function): Use new setter functions to construct bits
5077 and vr parts of jump functions or set them to NULL.
5078 (write_ipcp_transformation_info): Adjust for bits being pointers.
5079 (read_ipcp_transformation_info): Likewise.
5080 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
5082 Include gt-ipa-prop.h.
5083 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
5085 (ipcp_store_bits_results): Likewise.
5086 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
5087 Do not write to existing jump functions but use a temporary instead.
5089 2017-03-01 Jakub Jelinek <jakub@redhat.com>
5092 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
5093 attempt to use its first operand as BIT_FIELD_REF base.
5095 2017-03-01 Richard Biener <rguenther@suse.de>
5098 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
5099 interpolating formula in wrapping arithmetic.
5100 (chrec_apply): Convert chrec_evaluate return value to wanted type.
5102 2017-03-01 Jakub Jelinek <jakub@redhat.com>
5104 PR tree-optimization/79734
5105 * tree-vect-generic.c (expand_vector_condition): Optimize
5106 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
5107 Handle VEC_COND_EXPR where comparison has different inner width from
5110 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
5112 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
5113 markup, and similar issues. Remove @opindex entries for things
5114 that aren't options. Add missing -mmpy-option entries.
5116 2017-02-28 Jakub Jelinek <jakub@redhat.com>
5118 PR tree-optimization/79737
5119 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
5120 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
5121 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
5122 instead of byte_size. Formatting fix.
5123 (shift_bytes_in_array_right): Formatting fix.
5125 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
5128 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
5129 condition on optimize for the leaf function test.
5131 2017-02-28 Martin Liska <mliska@suse.cz>
5134 * read-rtl-function.c (function_reader::handle_unknown_directive):
5135 Bail out when one uses -flto.
5137 2017-02-28 Martin Liska <mliska@suse.cz>
5139 * common.opt: Replace space with tabular for options of <number>
5141 * config/i386/i386.opt: Show <number> value for
5142 -mlarge-data-threshold.
5143 * opts.c (print_filtered_help): Do not display number in hexadecimal
5146 2017-02-28 Martin Liska <mliska@suse.cz>
5148 * common.opt: Fix --help=option -Q for options which are of
5151 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
5153 * config/i386/i386.c (print_reg): Error out for values
5154 of 8-bit size in invalid integer register.
5156 2017-02-28 Martin Sebor <msebor@redhat.com>
5158 PR tree-optimization/79691
5159 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
5161 2017-02-28 Jakub Jelinek <jakub@redhat.com>
5164 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
5165 gcc_unreachable with output_operand_lossage.
5167 2017-02-28 Richard Biener <rguenther@suse.de>
5169 PR tree-optimization/79740
5170 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
5172 (visit_nary_op): Insert the nary into the hashtable if we
5173 pattern-matched sth.
5174 * tree-ssa-pre.c (eliminate_insert): Robustify.
5176 2017-02-28 Richard Biener <rguenther@suse.de>
5179 * fold-const.c (decode_field_reference): Reject out-of-bound
5182 2017-02-28 Jakub Jelinek <jakub@redhat.com>
5184 * config/i386/i386.c: Include intl.h.
5185 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
5186 instead of just cond ? "..." : "...".
5187 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
5188 * coverage.c (read_counts_file): Likewise.
5189 * omp-offload.c: Include intl.h.
5190 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
5191 of just cond ? "..." : "...".
5192 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
5193 of just cond ? "..." : "...".
5195 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
5198 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
5200 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
5201 'tune for' CPU name.
5202 * config/arm/arm-cpu-data.h: Regenerated.
5204 2017-02-28 Richard Biener <rguenther@suse.de>
5206 PR tree-optimization/79732
5207 * tree-inline.c (expand_call_inline): Do not shadow var.
5209 2017-02-28 Richard Biener <rguenther@suse.de>
5211 PR tree-optimization/79723
5212 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
5213 address-space properly.
5215 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
5217 * doc/optinfo.texi (Optimization groups): Fix option used for
5219 * doc/invoke.texi (-fopt-info): Document "omp".
5220 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
5221 (OPTGROUP_ALL): Add OPTGROUP_OMP.
5222 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
5223 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
5224 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
5226 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
5228 * dumpfile.c (optgroup_options): Instead of "openmp", associate
5229 OPTGROUP_OMP with "omp".
5231 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
5234 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
5235 for arithmetic shift of unsigned V2DI.
5237 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
5239 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
5240 arc/linux.h headers.
5241 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
5242 (LINK_SPEC): Likewise.
5243 (ARC_TLS_EXTRA_START_SPEC): Likewise.
5244 (EXTRA_SPECS): Likewise.
5245 (STARTFILE_SPEC): Likewise.
5246 (ENDFILE_SPEC): Likewise.
5247 (LIB_SPEC): Likewise.
5248 (TARGET_SDATA_DEFAULT): Likewise.
5249 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
5250 (MULTILIB_DEFAULTS): Likewise.
5251 (DWARF2_UNWIND_INFO): Likewise.
5252 * config/arc/big.h: New file.
5253 * config/arc/elf.h: Likewise.
5254 * config/arc/linux.h: Likewise.
5255 * config/arc/t-uClibc: Remove.
5257 2017-02-27 Bin Cheng <bin.cheng@arm.com>
5259 PR tree-optimization/77536
5260 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
5261 (tree_transform_and_unroll_loop): Use above function to compute the
5262 estimated niter of unrolled loop and use it when scaling profile.
5263 Also use count info rather than frequency if it's non-zero.
5264 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
5265 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
5266 (vect_transform_loop): Call above function.
5268 2017-02-27 Richard Biener <rguenther@suse.de>
5270 PR tree-optimization/45397
5271 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
5272 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
5273 (visit_nary_op): Add pattern matching for CSEing sign-changed
5274 or truncated operations with wider ones.
5276 2017-02-27 Richard Biener <rguenther@suse.de>
5278 PR tree-optimization/79690
5279 * tree-vect-stmts.c (vectorizable_store): Use vector type
5280 built from the DR with address-space.
5282 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
5284 * doc/invoke.texi (Optimize Options): Refine the description
5285 of asan-use-after-return.
5287 2017-02-25 Alan Modra <amodra@gmail.com>
5289 PR rtl-optimization/79584
5290 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
5291 base, not ad->base_term, the reg within base. Remove assertion
5292 that ad->base == ad->base_term. Replace gen_int_mode using
5293 bogus mode with const0_rtx.
5295 2017-02-25 Jakub Jelinek <jakub@redhat.com>
5298 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
5299 FMA_EXPR like tcc_binary or tcc_unary.
5301 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
5304 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
5306 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
5307 (output_loc_operands): Handle DW_OP_call_ref and
5308 DW_OP_GNU_variable_value.
5309 (struct variable_value_struct): New type.
5310 (struct variable_value_hasher): Likewise.
5311 (variable_value_hash): New variable.
5312 (string_types): Remove.
5313 (copy_loc_descr): New function.
5314 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
5315 (prepend_loc_descr_to_each): New function.
5316 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
5317 instead of add_loc_descr_to_each if the first argument is single
5318 location list and the second has multiple.
5319 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
5320 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
5321 when looking for variable value which doesn't have other location info.
5322 (loc_list_from_tree): Formatting fix.
5323 (gen_array_type_die): Simplify DW_AT_string_length handling.
5324 (adjust_string_types): Remove.
5325 (gen_subprogram_die): Don't call adjust_string_types nor test/set
5326 string_types. Call resolve_variable_values.
5327 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
5328 (resolve_addr_in_expr): Likewise. Add A argument.
5329 (copy_deref_exprloc): Remove deref argument. Adjust for the
5330 original expression being DW_OP_GNU_variable_value with optionally
5331 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
5332 optionally after it.
5333 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
5334 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
5335 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
5336 (variable_value_hasher::hash, variable_value_hasher::equal): New
5338 (resolve_variable_value_in_expr, resolve_variable_value,
5339 resolve_variable_values, note_variable_value_in_expr,
5340 note_variable_value): New functions.
5341 (dwarf2out_early_finish): Call note_variable_value on all toplevel
5344 2017-02-24 Jakub Jelinek <jakub@redhat.com>
5347 * opts.h (handle_generated_option): Add GENERATED_P argument.
5348 * opts-common.c (handle_option): Adjust function comment.
5349 (handle_generated_option): Add GENERATED_P argument, pass it to
5351 (control_warning_option): Pass false to handle_generated_option
5353 * opts.c (maybe_default_option): Pass true to handle_generated_option
5355 * optc-gen.awk: Likewise.
5357 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
5359 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
5360 a REG, look at the REG it is a SUBREG of.
5361 (splitter for cmpeqsi_t): Ditto.
5363 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
5365 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
5366 the special USEs with the pattern of the insn, not the insn itself.
5368 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
5371 * doc/invoke.texi: Document -mload-store-pairs.
5373 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
5374 Sandra Loosemore <sandra@codesourcery.com>
5376 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
5377 argument isn't a CONST_INT.
5378 (nios2_alternate_compare_const): Assert op is a CONST_INT.
5379 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
5380 (nios2_validate_compare): Bypass alternate compare logic if *op2
5382 (ldstwm_operation_p): Return false if first_base is not a REG or
5383 if first_offset is not a CONST_INT.
5385 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
5387 * config/cris/cris.md: Use correct operand in a define_peephole2.
5389 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
5391 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
5393 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
5395 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
5396 this_insn if it is an INSN or JUMP_INSN.
5397 (force_offsettable): Look at base, not at addr.
5398 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
5399 on things that aren't necessarily CONST_INTs.
5401 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
5403 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
5404 -mfpmath=sse is the default also for x86-32 targets with SSE2
5405 instruction set when @option{-ffast-math} is enabled
5407 2017-02-24 Jeff Law <law@redhat.com>
5409 PR rtl-optimizatoin/79286
5410 * ira.c (update_equiv_regs): Drop may_trap_p exception to
5413 2017-02-24 Richard Biener <rguenther@suse.de>
5415 PR tree-optimization/79389
5416 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
5419 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
5421 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
5422 function comment to reflect reality.
5423 (loop_exits_before_overflow): Fix typo in function description.
5425 2017-02-24 Richard Biener <rguenther@suse.de>
5427 PR tree-optimization/79389
5428 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
5429 properly that a threading opportunity exists. Detect conditional
5430 copy/constant propagation opportunities.
5432 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
5434 * config/visium/visium.md (type): Add trap.
5435 (b): New mode attribute.
5436 (*btst): Rename into...
5437 (*btst<mode>): ...this and adjust.
5438 (*cbranchsi4_btst_insn): Rename into...
5439 (*cbranch<mode>4_btst_insn): ...this and adjust.
5440 (trap): New define_insn.
5442 2017-02-23 Jakub Jelinek <jakub@redhat.com>
5444 PR tree-optimization/79389
5445 * ifcvt.c (struct noce_if_info): Add rev_cond field.
5446 (noce_reversed_cond_code): New function.
5447 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
5448 reversed_comparison_code. Formatting fix.
5449 (noce_try_store_flag): Test rev_cond != NULL in addition to
5450 reversed_comparison_code.
5451 (noce_try_store_flag_constants): Likewise.
5452 (noce_try_store_flag_mask): Likewise.
5453 (noce_try_addcc): Use rev_cond if non-NULL instead of
5454 reversed_comparison_code.
5455 (noce_try_cmove_arith): Likewise. Formatting fixes.
5456 (noce_try_minmax, noce_try_abs): Clear rev_cond.
5457 (noce_find_if_block): Initialize rev_cond.
5458 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
5459 instead of false as last argument never attempt to reverse it
5462 2017-02-23 Bin Cheng <bin.cheng@arm.com>
5464 PR tree-optimization/79663
5465 * tree-predcom.c (combine_chains): Process refs in reverse order
5466 only for ZERO length chains, and add explaining comment.
5468 2017-02-23 Jeff Law <law@redhat.com>
5470 PR tree-optimization/79578
5471 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
5472 in call to operand_equal_p.
5474 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
5477 * config/i386/cpuid.h: Fix another undefined behavior.
5479 2017-02-23 Richard Biener <rguenther@suse.de>
5481 PR tree-optimization/79683
5482 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
5483 vector types for data-refs.
5485 2017-02-23 Martin Liska <mliska@suse.cz>
5487 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
5489 2017-02-23 Jakub Jelinek <jakub@redhat.com>
5492 * internal-fn.c (get_range_pos_neg): Moved to ...
5493 * tree.c (get_range_pos_neg): ... here. No longer static.
5494 * tree.h (get_range_pos_neg): New prototype.
5495 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
5496 are known to be in between 0 and signed maximum inclusive, try to
5497 expand both unsigned and signed divmod and use the cheaper one from
5500 2017-02-22 Jeff Law <law@redhat.com>
5502 PR tree-optimization/79578
5503 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
5504 to compare base operands.
5506 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
5509 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
5510 gpc_reg_operand instead of fpr_reg_operand.
5512 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
5514 * config/mips/mips.c (mips_return_in_memory): Force FP
5515 vector types to be returned in memory for o32 ABI.
5517 2017-02-22 Jakub Jelinek <jakub@redhat.com>
5519 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
5520 instead of DW_TAG_member for static data member declarations and don't
5521 set no_linkage_name for static inline data members.
5522 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
5525 2017-02-22 Martin Liska <mliska@suse.cz>
5527 * doc/invoke.texi: Replace inequality signs with square brackets
5530 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5532 PR tree-optimization/68644
5533 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
5535 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
5538 * lra-constraints.c (simplify_operand_subreg): Handle
5539 WORD_REGISTER_OPERATIONS targets.
5541 2017-02-22 Jakub Jelinek <jakub@redhat.com>
5544 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
5545 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
5546 elimination by swapping fld*.
5548 2017-02-22 Richard Biener <rguenther@suse.de>
5550 PR tree-optimization/79673
5551 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
5552 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
5553 irrelevant address-space qualifiers and avoiding a
5554 ADDR_SPACE_CONVERT_EXPR from fold_convert.
5556 2017-02-22 Richard Biener <rguenther@suse.de>
5558 PR tree-optimization/79666
5559 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
5560 to not symbolically negate if that may introduce undefined
5563 2017-02-22 Martin Liska <mliska@suse.cz>
5566 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
5567 * data-streamer-out.c (streamer_write_gcov_count_stream):
5569 * value-prof.c (stream_out_histogram_value): Make assert more
5570 precise based on type of counter.
5572 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
5575 * config/i386/i386.md (standard_x87sse_constant_load splitter):
5576 Use nonimmediate_operand instead of memory_operand for operand 1.
5577 (float-extend standard_x87sse_constant_load splitter): Ditto.
5579 2017-02-21 Jeff Law <law@redhat.com>
5581 PR tree-optimization/79621
5582 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
5583 blocks with edges to themselves.
5585 2017-02-21 Jakub Jelinek <jakub@redhat.com>
5588 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
5589 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
5590 Use gimple_call_builtin_p.
5593 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
5594 on temporarily removed DEBUG_INSNs.
5596 PR tree-optimization/79649
5597 * tree-loop-distribution.c (classify_partition): Give up on
5598 non-generic address space loads/stores.
5600 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
5602 * doc/loop.texi (Loop manipulation): Remove nonexistent
5603 tree_ssa_loop_version from the documentation.
5604 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
5606 2017-02-21 Jakub Jelinek <jakub@redhat.com>
5609 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
5610 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
5611 * config/rs6000/rs6000.c: Include except.h.
5612 (rs6000_expand_split_stack_prologue): Call
5613 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
5615 2017-02-21 Martin Jambor <mjambor@suse.cz>
5618 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
5621 2017-02-21 Martin Jambor <mjambor@suse.cz>
5623 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
5624 for backward compatibility only.
5625 * doc/invoke.texi (Option Summary): Remove all references to
5628 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
5632 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
5634 * lra-constraints.c (curr_insn_transform): Handle
5635 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
5637 2017-02-21 Martin Liska <mliska@suse.cz>
5639 * config/i386/i386.opt: Replace -masm-dialect with -masm.
5641 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
5643 PR translation/79638
5644 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
5646 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
5649 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
5650 (arm_function_ok_for_sibcall): Return false for an indirect call by
5651 descriptor if all the argument registers are used.
5652 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
5653 alignment of the function.
5655 2017-02-21 Jakub Jelinek <jakub@redhat.com>
5657 PR tree-optimization/61441
5658 * simplify-rtx.c (simplify_const_unary_operation): For
5659 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
5660 the sNaN unmodified.
5662 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
5664 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
5665 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
5666 instead of SYSTEM_HEADER_DIR.
5668 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
5669 Martin Liška <mliska@suse.cz>
5671 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
5672 Fix typos and grammar, use active voice, and clarify.
5674 2017-02-20 Marek Polacek <polacek@redhat.com>
5677 * gimplify.c (gimplify_expr): Handle unused *&&L;.
5680 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
5682 2017-02-20 Jakub Jelinek <jakub@redhat.com>
5685 * config/i386/i386.c (ix86_expand_builtin): Handle
5686 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
5687 ix86_builtins_isa[fcode].isa as a requirement of those
5688 flags and any other flag in the bitmask.
5689 (ix86_init_mmx_sse_builtins): Use 0 instead of
5690 ~OPTION_MASK_ISA_64BIT as mask.
5691 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
5692 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
5693 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
5694 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
5696 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
5699 * lra-constraints.c (split_reg): Check requested split mode
5700 is supported by the register.
5702 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
5704 * lra-constraints.c (simplify_operand_subreg): Remove early
5707 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
5710 * lra-constraints.c (curr_insn_transform): Tighten condition
5711 for converting SUBREG reloads from OP_OUT to OP_INOUT.
5713 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
5716 * lra-constraints.c (curr_insn_transform): Handle
5717 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
5719 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
5722 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
5724 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
5726 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
5729 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
5732 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
5734 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
5737 2017-02-18 Jakub Jelinek <jakub@redhat.com>
5739 * final.c (last_columnnum, override_columnnum): New variables.
5740 (final_start_function): Set last_columnnum, pass it to begin_prologue
5741 hook and pass 0 to dwarf2out_begin_prologue.
5742 (final_scan_insn): Update override_columnnum. Pass last_columnnum
5743 to source_line debug hook.
5744 (notice_source_line): Compute last_columnnum and for debug_column_info
5745 return true on column changes.
5746 * debug.h (struct gcc_debug_hooks): Add column argument to
5747 source_line and begin_prologue hooks.
5748 (debug_nothing_int_charstar_int_bool): Remove prototype.
5749 (debug_nothing_int_int_charstar,
5750 debug_nothing_int_int_charstar_int_bool): New prototypes.
5751 (dwarf2out_begin_prologue): Add column argument.
5752 * debug.c (do_nothing_debug_hooks): Adjust source_line and
5753 begin_prologue hooks.
5754 (debug_nothing_int_charstar_int_bool): Remove.
5755 (debug_nothing_int_int_charstar,
5756 debug_nothing_int_int_charstar_int_bool): New functions.
5757 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
5758 through to dwarf2out_source_line.
5759 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
5760 (dwarf2out_source_line): Add column argument, emit it if requested.
5761 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
5763 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
5764 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
5765 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
5766 through to dwarf2out_begin_prologue.
5767 (vmsdbgout_source_line): Add column argument, pass it through to
5768 dwarf2out_source_line.
5769 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
5770 dbxout_source_line caller.
5771 (dbxout_source_line): Add column argument.
5773 * common.opt (gno-column-info, gcolumn-info): New options.
5774 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
5775 (check_die): Also test for multiple DW_AT_decl_column attributes.
5776 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
5777 DW_AT_decl_column if requested.
5778 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
5780 (gen_variable_die): Likewise.
5781 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
5782 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
5785 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
5786 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
5787 (ix86_handle_option): Handle OPT_m3dnowa.
5788 * doc/invoke.texi (-m3dnowa): Document.
5789 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
5790 -m3dnowa instead of -m3dnow -march=athlon.
5793 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
5794 instead of gcc_assert for K, r and R code checks. Formatting fixes.
5796 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
5799 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
5800 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
5801 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
5802 generator for vsx_xxpermdi_<mode>_be.
5803 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
5804 force big-endian semantics.
5805 (vsx_xxpermdi_<mode>_be): New define_expand with same
5806 implementation as previous version of vsx_xxpermdi_<mode>.
5808 2017-02-17 Jakub Jelinek <jakub@redhat.com>
5810 PR tree-optimization/79327
5811 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
5812 variable, its initialization and use.
5814 2017-02-17 Julia Koval <julia.koval@intel.com>
5816 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
5817 (OPTION_MASK_ISA_PKU_UNSET): New.
5818 (ix86_handle_option): Handle -mrdpid.
5819 * config/i386/cpuid.h (bit_RDPID): New.
5820 * config/i386/driver-i386.c (host_detect_local_cpu):
5821 Detect RDPID feature.
5822 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
5823 * config/i386/i386-c.c (ix86_target_macros_internal):
5825 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
5826 (ix86_valid_target_attribute_inner_p): Add "rdpid".
5827 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
5828 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
5829 * config/i386/i386.md (define_insn "rdpid"): New.
5830 * config/i386/i386.opt Add -mrdpid.
5831 * config/i386/immintrin.h (_rdpid_u32): New.
5833 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
5835 PR rtl-optimization/79541
5836 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
5837 instead of transforming it into USE.
5839 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
5841 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
5842 If HONOR_SNANS (SFmode) force the input to a register.
5843 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
5844 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
5845 an frsp or similar insn.
5847 2017-02-17 Martin Liska <mliska@suse.cz>
5849 PR rtl-optimization/79577
5850 * params.def (selsched-max-sched-times): Increase minimum to 1.
5852 2017-02-17 Martin Liska <mliska@suse.cz>
5854 PR rtl-optimization/79574
5855 * gcse.c (want_to_gcse_p): Prevent integer overflow.
5857 2017-02-17 Martin Liska <mliska@suse.cz>
5859 PR tree-optimization/79529
5860 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
5861 ssa_defined_default_def_p to handle cases which are implicitly
5863 * tree-ssa.c (ssa_defined_default_def_p): New function.
5864 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
5865 which are implicitly defined.
5866 * tree-ssa.h (ssa_defined_default_def_p): Declare.
5868 2017-02-17 Richard Biener <rguenther@suse.de>
5871 * params.def (max-ssa-name-query-depth): Limit to 10.
5873 2017-02-17 Richard Biener <rguenther@suse.de>
5875 PR tree-optimization/79552
5876 * tree-ssa-structalias.c (visit_loadstore): Properly verify
5879 2017-02-17 Richard Biener <rguenther@suse.de>
5882 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
5884 2017-02-17 Marek Polacek <polacek@redhat.com>
5887 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
5888 (fold_negate_expr): New wrapper.
5890 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
5892 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
5893 Correct terminology and de-emphasize pre-standard behavior.
5895 2017-02-16 Alan Modra <amodra@gmail.com>
5897 PR rtl-optimization/79286
5898 * ira.c (def_dominates_uses): New function.
5899 (update_equiv_regs): Don't create an equivalence for insns that
5900 may trap where the register def does not dominate the use.
5902 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
5904 PR rtl-optimization/78127
5905 * lra.c (lra): Call lra_eliminate before finish the loop after
5908 2017-02-16 Richard Biener <rguenther@suse.de>
5910 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
5912 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
5913 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
5914 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
5915 (isl_val_int_from_wi): New function.
5916 (extract_affine_gmp): Rename to ...
5917 (extract_affine_wi): ... this, take a widest_int.
5918 (extract_affine_int): Just wrap extract_affine_wi.
5919 (add_param_constraints): Use isl_val_int_from_wi.
5920 (add_loop_constraints): Likewise, and extract_affine_wi.
5922 2017-02-15 Jeff Law <law@redhat.com>
5925 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
5926 ira_init_register_move_cost_if_necessary.
5928 2017-02-15 Martin Sebor <msebor@redhat.com>
5931 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
5932 removed in a prior commit.
5934 2017-02-15 Bin Cheng <bin.cheng@arm.com>
5936 PR tree-optimization/79347
5937 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
5938 counters during peeling.
5940 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
5942 * Makefile.in (site.exp): Remove "set ISLVER".
5944 2017-02-15 Jakub Jelinek <jakub@redhat.com>
5947 * real.c (real_from_integer): Call real_convert even for decimal.
5949 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
5952 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
5954 2017-02-14 Andrew Pinski <apinski@cavium.com>
5956 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
5957 cores and change the partno/implementer to be correct.
5958 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
5959 the 'B" as the implementer.
5960 * config/aarch64/aarch64-tune.md: Regenerate.
5962 2017-02-14 Carl Love <cel@us.ibm.com>
5964 * config/rs6000/rs6000.c: Add case statement entry to make the
5965 xvcvuxdsp built-in argument unsigned.
5966 * config/rs6000/vsx.md: Fix the source and return operand types so they
5967 match the instruction definitions from the ISA document. Fix typo
5968 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
5971 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
5974 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
5975 member early_clobber_alts.
5976 * lra-lives.c (reg_early_clobber_p): New.
5977 (process_bb_lives): Use it.
5978 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
5979 (debug_operand_data): Initialize early_clobber_alts.
5980 (setup_operand_alternative): Set up early_clobber_alts.
5981 (collect_non_operand_hard_regs): Ditto. Pass early clobber
5982 alternatives to new_insn_reg.
5983 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
5985 (lra_update_insn_regno_info): Pass the new arg.
5987 2017-02-14 Jakub Jelinek <jakub@redhat.com>
5990 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
5991 (new_oacc_loop_raw): Don't clear already cleared fields.
5994 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
5995 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
5996 _mm512_prefetch_i64gather_ps): New inline functions and macros.
5998 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
6001 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
6003 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
6006 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
6007 the extra instruction to the right place to store 128-bit constant
6010 2017-02-14 Martin Sebor <msebor@redhat.com>
6013 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
6014 warning for strings of unknown length.
6016 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
6018 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
6020 2017-02-14 Jeff Law <law@redhat.com>
6023 * ira-costs.c (scan_one_insn): Initialize register move costs
6024 for pseudos seen in USE/CLOBBER insns.
6026 PR tree-optimization/79095
6027 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
6028 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
6029 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
6030 if the operands are known to be not equal, then the resulting range
6032 (intersect_ranges): If the new range is ~[0,0] and the old range is
6033 wide, then prefer ~[0,0].
6034 * tree-vrp.c (overflow_comparison_p_1): New function.
6035 (overflow_comparison_p): New function.
6036 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
6037 if NAME is used in an overflow test.
6038 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
6039 overflow check that can be expressed as an equality test, then adjust
6040 ops to be that equality test.
6042 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6044 * config/s390/s390-builtin-types.def: Remove flags argument.
6045 * config/s390/s390.c (s390_init_builtins): Likewise.
6047 2017-02-14 Martin Liska <mliska@suse.cz>
6049 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
6050 vector. Fix trailing white spaces.
6052 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
6054 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
6057 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6059 PR rtl-optimization/68664
6060 * config/arm/arm.c (arm_sched_can_speculate_insn):
6061 New function. Declare prototype.
6062 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
6064 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6066 PR rtl-optimization/68664
6067 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
6069 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
6071 2017-02-14 Amit Pawar <amit.pawar@amd.com>
6073 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
6074 max skip bytes for function, loop and jump.
6076 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
6078 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
6079 ABS_EXPR for gimple dump.
6081 2017-02-14 Jakub Jelinek <jakub@redhat.com>
6084 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
6086 PR tree-optimization/79408
6087 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
6088 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
6089 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
6090 also if rhs1 is INTEGER_CST.
6092 2017-02-14 Richard Biener <rguenther@suse.de>
6095 * tree-into-ssa.c (insert_phi_nodes): When the function can
6096 have abnormal edges rewrite SSA names with broken use-def
6097 dominance out of SSA and register them for PHI insertion.
6099 2017-02-13 Martin Sebor <msebor@redhat.com>
6102 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
6103 clearing info.nowrite flag when snprintf size argument is a range.
6105 2017-02-13 Jakub Jelinek <jakub@redhat.com>
6107 * cprop.c (cprop_jump): Add missing space in string literal.
6108 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
6109 (get_constraint_for_component_ref): Likewise.
6110 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
6111 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
6112 * lra-constraints.c (process_alt_operands): Likewise.
6113 * ipa-inline.c (inline_small_functions): Likewise.
6114 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
6115 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
6116 * trans-mem.c (diagnose_tm_1_op): Likewise.
6117 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
6118 (grid_parallel_clauses_gridifiable): Likewise.
6120 * config/nvptx/mkoffload.c (process): Add space in between
6123 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
6124 "MOD4_SSE_REGS" and "ALL_REGS".
6126 * spellcheck.c (test_data): Add , in between "foo" and "food".
6128 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6131 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
6132 boundary crossing check and subsequent code generation agree.
6134 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6136 * config/aarch64/aarch64.c (has_memory_op): Delete.
6137 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
6140 2017-02-13 Jakub Jelinek <jakub@redhat.com>
6142 PR rtl-optimization/79388
6143 PR rtl-optimization/79450
6144 * combine.c (distribute_notes): When removing TEM_INSN for which
6145 corresponding dest has last value recorded, invalidate that last
6148 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6150 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
6151 of explicit '@'. Add missing assembly comment marker on branch costs
6154 2017-02-13 Nathan Sidwell <nathan@acm.org>
6156 * gengtype-lex.l (<in_struct>): Add '/'.
6158 2017-02-13 Martin Liska <mliska@suse.cz>
6161 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
6163 2017-02-13 Richard Biener <rguenther@suse.de>
6165 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
6167 * configure: Re-generate.
6168 * config.in: Likewise.
6169 * graphite-dependences.c: Simplify as if
6170 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
6171 * graphite-isl-ast-to-gimple.c: Likewise.
6172 * graphite-optimize-isl.c: Likewise.
6173 * graphite-poly.c: Likewise.
6174 * graphite-sese-to-poly.c: Likewise.
6175 * graphite.h: Likewise.
6176 * toplev.c: Include isl/version.h and use isl_version () for
6177 printing the ISL version.
6178 * doc/install.texi: Update ISL requirement.
6180 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
6182 * doc/standards.texi (Standards): Update reference to
6185 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
6187 * doc/extend.texi (Named Address Spaces): sourceware.org now
6189 * doc/install.texi (Binaries): Ditto.
6192 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
6194 * doc/cpp.texi: Replace "stringify"/"stringification" with C
6195 standard terminology "stringize"/"stringizing" throughout.
6196 * doc/cppinternals.texi: Likewise.
6198 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
6200 * doc/extend.texi: Fix some spelling mistakes and typos.
6201 * doc/invoke.texi: Likewise.
6203 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
6206 * params.def (inline-min-speedup) Change from 10 to 8.
6208 2017-02-11 Jakub Jelinek <jakub@redhat.com>
6210 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
6213 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
6216 * ipa-inline-analysis.c (get_minimal_bb): New function.
6217 (record_modified): Use it.
6218 (remap_edge_change_prob): Handle also ancestor functions.
6220 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
6222 * doc/contrib.texi (Contributors): Remove broken link into
6223 the Mauve CVS repository.
6225 2017-02-11 Jakub Jelinek <jakub@redhat.com>
6228 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
6229 result computation whenever lhs doesn't have vector mode, not
6230 just when it has BLKmode.
6232 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
6234 * doc/makefile.texi (profiledbootstrap): Refer to the
6235 installation instructions only in textual form.
6237 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
6240 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
6242 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
6244 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
6245 (Specific): Update mingw-w64 reference.
6247 (Specific): Remove broken link to Renesas RX processor.
6249 2017-02-10 Richard Biener <rguenther@suse.de>
6251 * toplev.c (process_options): Do not mention obsolete graphite
6252 options when printing sorry message about missing graphite support.
6253 Mention -floop-nest-optimize.
6255 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
6257 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
6258 (vtst_p16): Likewise.
6259 (vtstq_p8): Likewise.
6260 (vtstq_p16): Likewise.
6262 (vtstq_p64): Likewise.
6263 * config/arm/arm_neon.h (vgetq_lane_p64): New.
6264 (vset_lane_p64): New.
6265 (vsetq_lane_p64): New.
6267 2017-02-10 Jakub Jelinek <jakub@redhat.com>
6269 PR tree-optimization/79411
6270 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
6271 stmt operands are SSA_NAMEs used in abnormal phis.
6272 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
6275 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
6278 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
6281 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
6283 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
6285 2017-02-09 Jakub Jelinek <jakub@redhat.com>
6287 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
6291 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
6292 not arbitrary TREE_CONSTANT.
6295 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
6296 "omp declare target link" attribute unless is_global_var.
6297 * omp-offload.c (find_link_var_op): Likewise.
6299 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
6300 Chung-Lin Tang <cltang@codesourcery.com>
6302 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
6304 (gimplify_adjust_omp_clauses): Don't delete TILE.
6305 (gimplify_omp_for): Deal with TILE.
6306 * internal-fn.c (expand_GOACC_TILE): New function.
6307 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
6309 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
6310 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
6312 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
6313 avoid DIV for outermost collapse var.
6314 (expand_oacc_for): Insert tile element loop as needed. Adjust.
6315 Remove out of date comments, fix whitespace.
6316 * omp-general.c (omp_extract_for_data): Deal with tiling.
6317 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
6318 adjust OLF_DIM_BASE value.
6319 (struct omp_for_data): Add tiling field.
6320 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
6321 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
6322 for auto loops. Remove default auto determining, moved to
6323 oacc_loop_fixed_partitions.
6324 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
6325 stmts, add e_mask field.
6326 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
6327 (oacc_thread_numbers): Use oacc_dim_call.
6328 (oacc_xform_tile): New.
6329 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
6330 (finish_oacc_loop): Adjust for ifns vector.
6331 (oacc_loop_discover_walk): Append loop abstraction sites to list,
6332 add case for GOACC_TILE fns.
6333 (oacc_loop_xform_loop): Delete.
6334 (oacc_loop_process): Iterate over call list directly, and add
6335 handling for GOACC_TILE fns.
6336 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
6338 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
6339 vector partitioning to outer loops. Assign 2 partitions to loops
6340 when available. Add TILE handling.
6341 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
6342 (execite_oacc_device_lower): Process GOACC_TILE fns, ignore unknown specs.
6343 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
6344 * tree.c (omp_clause_num_ops): Adjust TILE ops.
6345 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
6347 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
6349 * configure.ac (ACX_BUGURL): Update.
6350 * configure: Regenerate.
6352 2017-02-09 Richard Biener <rguenther@suse.de>
6354 PR tree-optimization/69823
6355 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
6356 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
6358 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
6360 * config/arc/arc-c.def: Add __NPS400__ definition.
6361 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
6362 (TARGET_NPS400): Define.
6364 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
6366 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
6368 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
6370 (arc_cpu_types): Fill the arch_info field with a pointer into the
6371 arc_arch_types table.
6372 (arc_selected_cpu): Declare.
6373 * config/arc/arc.c (arc_selected_cpu): Make global.
6374 (arc_selected_arch): Delete.
6375 (arc_base_cpu): Delete.
6376 (arc_override_options): Remove references to deleted variables,
6377 update access to arch information.
6378 (ARC_OPT): Update access to arch information.
6379 (ARC_OPTX): Likewise.
6380 * config/arc/arc.h (arc_base_cpu): Remove declaration.
6381 (TARGET_ARC600): Update access to arch information.
6382 (TARGET_ARC601): Likewise.
6383 (TARGET_ARC700): Likewise.
6384 (TARGET_EM): Likewise.
6385 (TARGET_HS): Likewise.
6386 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
6389 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
6392 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
6393 condition/operands for integer GE/LE/GEU/LEU operations.
6395 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
6397 PR translation/79397
6398 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
6401 2017-02-08 Martin Jambor <mjambor@suse.cz>
6404 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
6405 whether allocation happened.
6406 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
6407 nothing was allocated.
6409 2017-02-08 Jakub Jelinek <jakub@redhat.com>
6411 PR tree-optimization/79408
6412 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
6413 constant, but SSA_NAME with a known integer range, use the minimum
6414 of that range instead of op1 to determine if modulo can be replaced
6415 with its first operand.
6417 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
6419 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
6421 2017-02-08 Richard Biener <rguenther@suse.de>
6423 PR tree-optimization/71824
6424 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
6425 Check all loops contained in the merged region.
6427 2017-02-07 Andrew Pinski <apinski@cavium.com>
6429 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
6431 2017-02-07 Andrew Pinski <apinski@cavium.com>
6433 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
6434 (thunderxt88): Likewise.
6435 (thunderxt81): Disable LSE and change v8.1 to v8.
6436 (thunderxt83): Likewise.
6438 2017-02-07 Jakub Jelinek <jakub@redhat.com>
6439 Richard Biener <rguenther@suse.de>
6442 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
6443 type from int to size_t.
6444 * ira-costs.c (struct_costs_size): Change type from int to size_t.
6446 2017-02-07 Jakub Jelinek <jakub@redhat.com>
6448 PR rtl-optimization/79386
6449 * cprop.c (bypass_conditional_jumps): Initialize
6450 bypass_last_basic_block already before splitting bbs after
6451 unconditional traps...
6452 (bypass_conditional_jumps): ... rather than here.
6455 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
6456 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
6457 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
6458 fix -masm=intel patterns.
6460 2017-02-07 Richard Biener <rguenther@suse.de>
6462 PR tree-optimization/79256
6464 * builtins.c (get_object_alignment_2): Use min_align_of_type
6465 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
6466 and ADJUST_FIELD_ALIGN.
6468 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
6470 * doc/tm.texi: Regenerate.
6471 * stor-layout.c (layout_decl): Adjust.
6472 (update_alignment_for_field): Likewise.
6473 (place_field): Likewise.
6474 (min_align_of_type): Likewise.
6475 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
6476 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
6477 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
6478 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
6479 * config/frv/frv.c (frv_adjust_field_align): Likewise.
6480 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
6481 * config/i386/i386.c (x86_field_alignment): Likewise.
6482 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
6483 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
6484 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
6485 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
6486 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
6487 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
6491 2017-01-30 Richard Biener <rguenther@suse.de>
6493 PR tree-optimization/79256
6494 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
6495 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
6498 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
6500 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
6501 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
6502 builtins to SImode and emit a zero-extend, if necessary.
6504 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
6506 * docs/invoke.texi (RISC-V Options): Alphabetize.
6508 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
6510 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
6513 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
6515 * config/riscv/riscv.c: New file.
6516 * gcc/common/config/riscv/riscv-common.c: Likewise.
6517 * config.gcc: Likewise.
6518 * config/riscv/constraints.md: Likewise.
6519 * config/riscv/elf.h: Likewise.
6520 * config/riscv/generic.md: Likewise.
6521 * config/riscv/linux.h: Likewise.
6522 * config/riscv/multilib-generator: Likewise.
6523 * config/riscv/peephole.md: Likewise.
6524 * config/riscv/pic.md: Likewise.
6525 * config/riscv/predicates.md: Likewise.
6526 * config/riscv/riscv-builtins.c: Likewise.
6527 * config/riscv/riscv-c.c: Likewise.
6528 * config/riscv/riscv-ftypes.def: Likewise.
6529 * config/riscv/riscv-modes.def: Likewise.
6530 * config/riscv/riscv-opts.h: Likewise.
6531 * config/riscv/riscv-protos.h: Likewise.
6532 * config/riscv/riscv.h: Likewise.
6533 * config/riscv/riscv.md: Likewise.
6534 * config/riscv/riscv.opt: Likewise.
6535 * config/riscv/sync.md: Likewise.
6536 * config/riscv/t-elf-multilib: Likewise.
6537 * config/riscv/t-linux: Likewise.
6538 * config/riscv/t-linux-multilib: Likewise.
6539 * config/riscv/t-riscv: Likewise.
6540 * configure.ac: Likewise.
6541 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
6542 Waterman as RISC-V maintainers.
6543 * doc/install.texi: Add RISC-V entries.
6544 * doc/invoke.texi: Add RISC-V options section.
6545 * doc/md.texi: Add RISC-V constraints section.
6546 * configure: Regenerated.
6548 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
6551 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
6552 false values to be constant vectors with all 0 or all 1 bits set.
6553 (vcondu<mode><mode>): Likewise.
6554 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
6556 (fpmask_comparison_operator): Update comment.
6557 (vecint_comparison_operator): New predicate.
6558 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
6559 vector conditionals when the true and false values are constant
6560 vectors with all 0 bits or all 1 bits set.
6562 2017-02-06 Martin Sebor <msebor@redhat.com>
6564 PR tree-optimization/79376
6565 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
6567 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
6569 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
6570 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
6571 to simplify split condition.
6573 2017-02-06 Jakub Jelinek <jakub@redhat.com>
6575 * omp-expand.c (oxpand_omp_atomic_fetch_op,
6576 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
6579 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
6581 PR rtl-optimization/68664
6582 * target.def (can_speculate_insn): New hook.
6583 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
6584 * doc/tm.texi: Regenerate.
6585 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
6586 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
6587 (rs6000_sched_can_speculate_insn): New function.
6589 2017-02-06 Jakub Jelinek <jakub@redhat.com>
6591 PR tree-optimization/79284
6592 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
6593 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
6594 vectorizable_mask_load_store, vectorizable_operation,
6595 vect_is_simple_cond, get_same_sized_vectype): Use it instead
6596 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
6597 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
6598 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
6599 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
6600 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
6601 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
6602 is_gimple_assign (stmt). Replace another such test with
6603 is_gimple_assign (stmt).
6605 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
6608 * config/avr/avr.c (rtl-iter.h): Include it.
6609 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
6610 (avr_legitimate_combined_insn): ...and implementation.
6612 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
6614 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
6615 * config/s390/s390.c (s390_const_operand_ok)
6616 (s390_canonicalize_comparison, s390_extract_part)
6617 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
6618 (s390_contiguous_bitmask_p, s390_rtx_costs)
6619 (legitimize_pic_address): Likewise.
6620 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
6621 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
6622 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
6623 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
6624 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
6626 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
6628 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
6629 REGNO($0) == REGNO($1).
6631 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6633 * config/s390/linux.h(SIZE_TYPE): Add comment.
6635 2017-02-06 Julian Brown <julian@codesourcery.com>
6636 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
6637 Virendra Pathak <virendra.pathak@broadcom.com>
6639 * config/aarch64/aarch64-cores.def: Change the scheduler
6641 * config/aarch64/aarch64.md: Include thunderx2t99.md.
6642 * config/aarch64/thunderx2t99.md: New file.
6644 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
6646 * doc/standards.texi (Go Language): Update link to language
6649 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
6651 * tree-eh.c (lower_resx): Sanitize profile.
6652 (cleanup_empty_eh_move_lp): Likewise.
6654 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
6657 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
6659 * cfgloopmanip.h (loop_version): Update prototype.
6660 * modulo-sched.c (sms_schedule): Update call of loop_version.
6661 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
6662 * tree-parloops.c (gen_parallel_loop): Likewise.
6663 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
6664 * tree-ssa-loop-split.c (split_loop): Likewise.
6665 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
6666 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
6668 2017-02-05 Martin Liska <mliska@suse.cz>
6671 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
6673 (print_operand_address): Initialize a struct to zero.
6675 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
6677 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
6678 garbage collector only in textual form.
6680 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
6682 * doc/extend.texi (x86 specific memory model extensions for
6683 transactional memory): Simplify a phrase.
6685 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
6688 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
6689 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
6690 (atomic_storedi_1): Likewise.
6692 2017-02-04 Jakub Jelinek <jakub@redhat.com>
6694 PR tree-optimization/79338
6695 * tree-parloops.c (gather_scalar_reductions): Don't call
6696 vect_analyze_loop_form for loop->inner before destroying loop's
6699 2017-02-03 Martin Sebor <msebor@redhat.com>
6701 PR tree-optimization/79327
6702 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
6703 when precision has resulted in leading zeros.
6704 (format_integer): Adjust the likely counter to assume an unknown
6705 argument that may be zero is non-zero.
6707 2017-02-03 Jason Merrill <jason@redhat.com>
6710 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
6711 avoid copying non-taken branch.
6713 2017-02-03 Jakub Jelinek <jakub@redhat.com>
6715 PR tree-optimization/79340
6716 * tree-vect-loop.c (vectorizable_reduction): Release
6717 vec_defs elements after safe_splicing them into other vectors.
6720 PR tree-optimization/79327
6721 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
6722 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
6724 (format_integer): Use wide_int_to_tree instead of build_int_cst
6725 + to_?hwi. If argmin is NULL, just set argmin and argmax to
6726 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
6727 of shortest and longest sequence.
6729 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
6731 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
6732 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
6734 2017-02-03 Walter Lee <walt@tilera.com>
6737 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
6738 after initial stackframe link reg save.
6739 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
6741 2017-02-03 Jakub Jelinek <jakub@redhat.com>
6744 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
6745 wu for stxssp alternative.
6747 2017-02-03 Martin Sebor <msebor@redhat.com>
6749 PR tree-optimization/79352
6750 * gimple-fold.c (get_range_strlen): Add argument.
6751 (get_range_strlen): Change return type to bool.
6752 (get_maxval_strlen): Pass in a dummy argument.
6753 * gimple-fold.h (get_range_strlen): Change return type to bool.
6754 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
6755 * tree.h (array_at_struct_end_p): Add argument.
6756 * tree.c (array_at_struct_end_p): Handle it.
6758 2017-02-03 Martin Liska <mliska@suse.cz>
6761 * multiple_target.c (create_dispatcher_calls): Redirect edge
6762 from a caller of a dispatcher.
6763 (expand_target_clones): Make the clones local.
6764 (ipa_target_clone): Do both target clones and resolvers.
6765 (ipa_dispatcher_calls): Remove the pass.
6766 (pass_dispatcher_calls::gate): Likewise.
6767 (make_pass_dispatcher_calls): Likewise.
6768 * passes.def (pass_target_clone): Put as very first IPA early
6771 2017-02-03 Martin Liska <mliska@suse.cz>
6773 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
6774 in case of a function with ifunc attribute.
6776 2017-02-03 Martin Liska <mliska@suse.cz>
6778 * cgraph.c (cgraph_node::dump): Dump function version info.
6779 * symtab.c (symtab_node::dump_base): Add missing new line.
6781 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
6783 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
6784 (ifcombine_ifandif): Use it.
6786 2017-02-03 Martin Liska <mliska@suse.cz>
6788 * doc/invoke.texi: Document default value for
6789 use-after-scope-direct-emission-threshold.
6791 2017-02-03 Martin Liska <mliska@suse.cz>
6793 PR tree-optimization/79339
6794 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
6795 (format_floating): Likewise.
6797 2017-02-03 Martin Liska <mliska@suse.cz>
6800 * ipa-prop.c (ipa_node_params_t::insert): Remove current
6802 (ipa_node_params_t::remove): Likewise.
6803 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
6804 initialization from removed ipa_node_params_t::insert.
6805 (ipa_node_params::~ipa_node_params): Move from removed
6806 ipa_node_params_t::release.
6807 * symbol-summary.h (symbol_summary::m_released): New member.
6808 Do not release a summary twice. Do not allow to call finalizer
6809 for types of a summary that live in GGC memory.
6811 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
6813 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
6816 2017-02-02 Martin Sebor <msebor@redhat.com>
6819 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
6820 (format_string): Tighten up the range of output for non-constant
6821 strings and correct the expected range for wide non-constant strings.
6823 2017-02-02 Martin Sebor <msebor@redhat.com>
6825 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
6828 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
6830 (-fdump-tree-@var): Add to index and document how to come up
6831 with pass-specific option and dump file names.
6832 (-fdump-passes): Clarify where to look for output.
6834 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
6837 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
6838 statistics of the analyzed path; allow threading for speed when
6839 any of BBs along the path are optimized for speed.
6841 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
6844 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
6845 settings of the virtual registers.
6848 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
6850 * explow.c (get_dynamic_stack_size): Take known alignment of stack
6851 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
6854 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6856 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
6857 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
6859 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6861 * config/s390/s390.md: Add missing comments with the expanded
6863 * config/s390/vector.md: Likewise.
6864 * config/s390/vx-builtins.md: Likewise.
6866 2017-02-02 Jakub Jelinek <jakub@redhat.com>
6869 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
6870 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
6871 conditions on a single line.
6873 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
6875 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
6876 __S390_VX__ to __VX__.
6878 2017-02-01 Andrew Pinski <apinski@cavium.com>
6880 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
6881 stmt_info to record_stmt_cost.
6882 (vect_get_known_peeling_cost): Pass stmt_info if known to
6884 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
6885 cpu_vector_cost field into
6886 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
6887 field into vec_int_stmt_cost and vec_fp_stmt_cost.
6888 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
6889 splitting of scalar_stmt_cost and vec_stmt_cost.
6890 (thunderx_vector_cost): Likewise.
6891 (cortexa57_vector_cost): LIkewise.
6892 (exynosm1_vector_cost): Likewise.
6893 (xgene1_vector_cost): Likewise.
6894 (thunderx2t99_vector_cost): Improve after the splitting of the two
6896 (aarch64_builtin_vectorization_cost): Update for the splitting of
6897 scalar_stmt_cost and vec_stmt_cost.
6899 2017-02-01 Torvald Riegel <triegel@redhat.com>
6900 Richard Henderson <rth@redhat.com>
6902 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
6903 conditional on existance of a fast atomic load.
6904 * optabs-query.c (can_atomic_load_p): New function.
6905 * optabs-query.h (can_atomic_load_p): Declare it.
6906 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
6907 no fast atomic load is available for the particular size of access.
6908 (expand_atomic_compare_and_swap): Likewise.
6909 (expand_atomic_load): Likewise.
6910 (expand_atomic_store): Likewise.
6911 (expand_atomic_fetch_op): Likewise.
6912 * testsuite/lib/target-supports.exp
6913 (check_effective_target_sync_int_128): Remove x86 because it provides
6914 no fast atomic load.
6915 (check_effective_target_sync_int_128_runtime): Likewise.
6917 2017-02-01 Richard Biener <rguenther@suse.de>
6919 * graphite.c: Include tree-vectorizer.h for find_loop_location.
6920 (graphite_transform_loops): Provide opt-info for optimized nests.
6921 * tree-parloop.c (parallelize_loops): Provide opt-info for
6924 2017-02-01 Richard Biener <rguenther@suse.de>
6927 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
6930 2017-02-01 Richard Biener <rguenther@suse.de>
6932 PR tree-optimization/71824
6933 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
6934 Verify the loops are valid in the merged SESE region.
6935 (scop_detection::can_represent_loop_1): Check analyzing the
6936 evolution of the number of iterations in the region succeeds.
6938 2017-01-31 Ian Lance Taylor <iant@golang.org>
6940 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
6941 REG_ARGS_SIZE note to 32-bit push insns and call insn.
6943 2017-01-31 David Malcolm <dmalcolm@redhat.com>
6945 PR preprocessor/79210
6946 * input.c (get_substring_ranges_for_loc): Replace line_width
6947 assertion with error-handling.
6949 2017-01-31 Richard Biener <rguenther@suse.de>
6951 PR tree-optimization/77318
6952 * graphite-sese-to-poly.c (extract_affine): Fix assert.
6953 (create_pw_aff_from_tree): Take loop parameter.
6954 (add_condition_to_pbb): Pass loop of the condition to
6955 create_pw_aff_from_tree.
6957 2017-01-31 Jakub Jelinek <jakub@redhat.com>
6959 * config/s390/s390.c (s390_asan_shadow_offset): New function.
6960 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
6962 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
6966 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
6968 (convert_int_to_float128): Likewise.
6969 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
6970 (convert_int_to_float128): Likewise.
6971 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
6972 (UNSPEC_IEEE128_CONVERT): Likewise.
6973 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
6974 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
6975 Use local variables for IBM extended format.
6976 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
6977 (fix_trunc<mode>si2_fprs): Likewise.
6978 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
6979 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
6980 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
6981 to know that we can now have integers of all sizes in vector
6983 (fix<uns>_<mode>di2_hw): Likewise.
6984 (float<uns>_<mode>si2_hw): Likewise.
6985 (fix_<mode>si2_hw): Likewise.
6986 (fixuns_<mode>si2_hw): Likewise.
6987 (float<uns>_<mode>di2_hw): Likewise.
6988 (float_<mode>di2_hw): Likewise.
6989 (float_<mode>si2_hw): Likewise.
6990 (floatuns_<mode>di2_hw): Likewise.
6991 (floatuns_<mode>si2_hw): Likewise.
6992 (xscvqp<su>wz_<mode>): Delete, no longer used.
6993 (xscvqp<su>dz_<mode>): Likewise.
6994 (xscv<su>dqp_<mode>): Likewise.
6995 (ieee128_mfvsrd_64bit): Likewise.
6996 (ieee128_mfvsrd_32bit): Likewise.
6997 (ieee128_mfvsrwz): Likewise.
6998 (ieee128_mtvsrw): Likewise.
6999 (ieee128_mtvsrd_64bit): Likewise.
7000 (ieee128_mtvsrd_32bit): Likewise.
7002 2017-01-31 Martin Liska <mliska@suse.cz>
7005 * ipa-prop.c (ipa_free_all_node_params): Call release method
7006 instead of ~sumbol_summary to not to trigger double times
7009 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
7011 PR tree-optimization/71691
7012 * bitmap.h (class auto_bitmap): New.
7013 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
7014 is_maybe_undefined instead of ssa_undefined_value_p.
7016 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
7018 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
7019 __S390_ARCH_LEVEL__ to __ARCH__.
7021 2017-01-31 Jakub Jelinek <jakub@redhat.com>
7023 PR tree-optimization/79267
7024 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
7025 if should_remove_lhs_p is true.
7027 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
7030 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
7031 (add_alignment_attribute): New.
7032 (base_type_die): Add alignment attribute.
7033 (subrange_type_die): Likewise.
7034 (modified_type_die): Likewise.
7035 (gen_array_type_die): Likewise.
7036 (gen_descr_array_type_die: Likewise.
7037 (gen_enumeration_type_die): Likewise.
7038 (gen_subprogram_die): Likewise.
7039 (gen_variable_die): Likewise.
7040 (gen_field_die): Likewise.
7041 (gen_ptr_to_mbr_type_die): Likewise.
7042 (gen_struct_or_union_type_die): Likewise.
7043 (gen_subroutine_type_die): Likewise.
7044 (gen_typedef_die): Likewise.
7045 (base_type_cmp): Compare alignment attribute.
7047 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
7050 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
7051 (setb_unsigned) New pattern for setb with CCUNS.
7052 * config/rs6000/rs6000.c (expand_block_compare): Use a different
7053 subfc./subfe sequence to avoid overflow problems. Generate a
7054 shorter sequence with cmpld/setb for power9.
7055 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
7056 for generating subfc. instruction.
7057 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
7058 now uses this instruction.
7060 2017-01-30 Ian Lance Taylor <iant@google.com>
7063 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
7064 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
7066 2017-01-30 Martin Sebor <msebor@redhat.com>
7068 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
7069 Move constant to the right of a relational operator.
7070 (get_mpfr_format_length, format_character, format_string): Ditto.
7071 (should_warn_p, maybe_warn): Same.
7073 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
7075 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
7078 * asan.c (get_translation_unit_decl): Remove function.
7079 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
7081 2017-01-30 Martin Liska <mliska@suse.cz>
7083 PR gcov-profile/79259
7084 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
7087 2017-01-30 Martin Liska <mliska@suse.cz>
7090 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
7091 Initialize variables with NULL value.
7093 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
7096 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
7098 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
7100 2017-01-30 Richard Biener <rguenther@suse.de>
7102 PR tree-optimization/79276
7103 * tree-vrp.c (process_assert_insertions): Properly adjust common
7104 when removing a duplicate.
7106 * gcc.dg/torture/pr79276.c: New testcase.
7108 2017-01-30 Richard Biener <rguenther@suse.de>
7110 PR tree-optimization/79256
7111 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
7112 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
7114 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
7116 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
7119 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
7120 ("*r<noxa>sbg_<mode>_sll_bitmask")
7121 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
7122 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
7123 Use contiguous_bitmask_nowrap_operand.
7125 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7128 * config/rs6000/altivec.h (vec_xl): Revise #define.
7129 (vec_xst): Likewise.
7131 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
7133 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
7135 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
7137 PR rtl-optimization/79194
7138 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
7139 traps before call to bypass_conditional_jumps.
7141 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
7143 PR tree-optimization/71374
7144 * lra-constraints.c (check_conflict_input_operands): New.
7145 (match_reload): Use it.
7147 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
7150 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
7151 account to calculate conflict_set.
7153 2017-01-27 Bin Cheng <bin.cheng@arm.com>
7155 PR rtl-optimization/78559
7156 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
7157 other_insn in combine.
7159 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
7161 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
7162 uint16_type_node for BT_UINT16.
7164 2017-01-27 David Malcolm <dmalcolm@redhat.com>
7166 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
7167 "RTL Tests" to menu.
7168 (GIMPLE Tests): New node.
7169 (RTL Tests): New node.
7171 2017-01-27 Richard Biener <rguenther@suse.de>
7173 PR tree-optimization/79245
7174 * tree-loop-distribution.c (distribute_loop): Apply cost
7175 modeling also to detected patterns.
7177 2017-01-27 Richard Biener <rguenther@suse.de>
7179 PR tree-optimization/71433
7180 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
7181 (compare_assert_loc): New function.
7182 (process_assert_insertions): Sort and optimize assert locations
7183 to remove duplicates and push down identical assertions on
7184 edges to their destination block.
7186 2017-01-27 Richard Biener <rguenther@suse.de>
7188 PR tree-optimization/79244
7189 * tree-vrp.c (remove_range_assertions): Forcefully propagate
7190 out SSA names even if abnormal.
7192 2017-01-27 Jakub Jelinek <jakub@redhat.com>
7194 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
7195 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
7196 instead of MPFR_RNDN.
7198 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
7201 * arm.c (arm_option_override): Don't call build_target_option_node
7202 until after doing all option overrides.
7203 (arm_valid_target_attribute_tree): Likewise.
7205 2017-01-27 Martin Liska <mliska@suse.cz>
7207 * doc/invoke.texi (-fprofile-arcs): Document profiling support
7208 for {cd}tors and C++ {cd}tors.
7210 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
7212 * config/s390/s390.md ("*setmem_long_and")
7213 ("*setmem_long_and_31z"): Use zero_extend instead of and.
7215 2017-01-26 Martin Sebor <msebor@redhat.com>
7217 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
7220 2017-01-26 Martin Sebor <msebor@redhat.com>
7222 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
7223 HAVE_DFmode before using XFmode or DFmode.
7224 (parse_directive): Avoid using the z length modifier to avoid
7225 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
7228 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
7229 to accept adjustment as an array.
7230 (get_int_range): New function.
7231 (struct directive): Make width and prec arrays.
7232 (directive::set_width, directive::set_precision): Call get_int_range.
7233 (format_integer, format_floating): Handle width and precision ranges.
7234 (format_string, parse_directive): Same.
7236 2017-01-26 Jakub Jelinek <jakub@redhat.com>
7239 * dwarf2out.c (generate_skeleton_bottom_up): For children with
7240 comdat_type_p set, just clone them, but keep the children in the
7244 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
7245 which have direct callers with -fvar-tracking-assignments enabled
7247 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
7248 inside of type units.
7250 2017-01-26 Martin Sebor <msebor@redhat.com>
7253 * gimple-ssa-sprintf.c (struct result_range): Add likely and
7255 (struct format_result): Replace number_chars, number_chars_min,
7256 and number_chars_max with a single member of struct result_range.
7258 (format_result::operator+=): Adjust.
7259 (struct fmtresult): Remove bounded. Handle likely and unlikely
7261 (fmtresult::adjust_for_width_or_precision): New function.
7262 (fmtresult:type_max_digits): New function.
7263 (bytes_remaining): Handle likely and unlikely counters.
7264 (min_bytes_remaining): Remove.
7265 (format_percent): Simplify.
7266 (format_integer, format_floating): Set likely and unlikely counters.
7267 (get_string_length, format_character, format_string): Same.
7268 (format_plain, should_warn_p): New function.
7269 (maybe_warn): Call should_warn_p. Update diagnostic messages
7270 and handle those for all directives, including plain strings.
7271 (format_directive): Handle likely and unlikely counters.
7272 Remove unnecessary quoting from diagnostics. Add an informational
7274 (add_bytes): Remove.
7275 (pass_sprintf_length::compute_format_length): Simplify.
7276 (try_substitute_return_value): Handle likely and unlikely counters.
7278 2017-01-26 Carl Love <cel@us.ibm.com>
7280 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
7281 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
7283 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
7286 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
7287 endianess for subregs into account.
7288 * lra-constraints.c (lra_constraints): Do risky transformations
7289 always on the first iteration.
7290 * lra-lives.c (check_pseudos_live_through_calls): Add arg
7291 last_call_used_reg_set.
7292 (process_bb_lives): Define and use last_call_used_reg_set.
7293 * lra.c (lra): Always continue after lra_constraints on the first
7296 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
7298 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
7300 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
7302 2017-01-26 Jakub Jelinek <jakub@redhat.com>
7304 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
7305 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
7306 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
7307 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
7308 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
7309 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
7310 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
7311 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
7312 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
7314 2017-01-26 Marek Polacek <polacek@redhat.com>
7317 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
7318 for the third operand.
7320 2017-01-26 Jakub Jelinek <jakub@redhat.com>
7323 * omp-low.c (struct omp_context): Add simt_stmt field.
7324 (scan_omp_for): Return omp_context *.
7325 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
7326 context to the _simt_ SIMD stmt.
7327 (lower_omp_for): For combined SIMD with sibling _simt_
7328 SIMD, make sure to use the same decls in _looptemp_
7329 clauses as in the sibling.
7331 2017-01-26 David Sherwood <david.sherwood@arm.com>
7334 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
7337 2017-01-26 Jakub Jelinek <jakub@redhat.com>
7340 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
7341 emit fld b; fld a; if possible.
7343 * brig-builtins.def: Update copyright years.
7344 * config/arm/arm_acle_builtins.def: Update copyright years.
7346 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
7349 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
7350 constraint instead of o for the stxsd instruction.
7352 2017-01-25 Carl Love <cel@us.ibm.com>
7354 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
7355 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
7357 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
7359 * doc/invoke.texi (C++ Dialect Options): Fix typo.
7361 2017-01-25 Richard Biener <rguenther@suse.de>
7363 PR tree-optimization/69264
7364 * target.def (vector_alignment_reachable): Improve documentation.
7365 * doc/tm.texi: Regenerate.
7366 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
7368 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
7369 earlier changes with respect to TYPE_USER_ALIGN.
7370 (vector_alignment_reachable_p): Likewise. Improve dumping.
7372 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7375 * config/arm/arm.md (xordi3): Force constant operand into a register
7378 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7380 * doc/invoke.texi (-fstore-merging): Correct default optimization
7381 levels at which it is enabled.
7382 (-O): Move -fstore-merging from list to...
7385 2017-01-25 Richard Biener <rguenther@suse.de>
7388 * omp-expand.c: Include debug.h.
7389 (expand_omp_taskreg): Make sure to generate early debug before
7390 outlining anything from a function.
7391 (expand_omp_target): Likewise.
7392 (grid_expand_target_grid_body): Likewise.
7394 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
7397 * asan.c (get_translation_unit_decl): New function.
7398 (asan_add_global): Extract modules file name from globals
7399 TRANSLATION_UNIT_DECL name.
7401 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
7404 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
7405 for long calls with APCS frame and VFP.
7407 2017-01-24 David Malcolm <dmalcolm@redhat.com>
7409 * cfg.c (original_copy_tables_initialized_p): New function.
7410 * cfg.h (original_copy_tables_initialized_p): New decl.
7411 * cfgrtl.c (relink_block_chain): Guard the call to
7412 free_original_copy_tables with a call to
7413 original_copy_tables_initialized_p.
7414 * cgraph.h (symtab_node::native_rtl_p): New decl.
7415 * cgraphunit.c (symtab_node::native_rtl_p): New function.
7416 (symtab_node::needed_p): Don't assert for early assembly output
7417 for __RTL functions.
7418 (cgraph_node::finalize_function): Set "force_output" for __RTL
7420 (cgraph_node::analyze): Bail out early for __RTL functions.
7421 (analyze_functions): Update assertion to support __RTL functions.
7422 (cgraph_node::expand): Bail out early for __RTL functions.
7423 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
7425 * function.h (struct function): Update comment for field
7427 * gimple-expr.c: Include "tree-pass.h".
7428 (gimple_has_body_p): Return false for __RTL functions.
7429 * Makefile.in (OBJS): Add run-rtl-passes.o.
7430 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
7432 (gcc::pass_manager::get_clean_slate): New accessor.
7433 * passes.c: Include "insn-addr.h".
7434 (should_skip_pass_p): Add logging. Update logic for running
7435 "expand" to be compatible with both __GIMPLE and __RTL. Guard
7436 property-provider override so it is only done for gimple passes.
7438 (skip_pass): New function.
7439 (execute_one_pass): Call skip_pass when skipping passes.
7440 * read-md.c (md_reader::read_char): Support filtering
7441 the input to a subset of line numbers.
7442 (md_reader::md_reader): Initialize fields
7443 m_first_line and m_last_line.
7444 (md_reader::read_file_fragment): New function.
7445 * read-md.h (md_reader::read_file_fragment): New decl.
7446 (md_reader::m_first_line): New field.
7447 (md_reader::m_last_line): New field.
7448 * read-rtl-function.c (function_reader::create_function): Only
7449 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
7450 curr_properties. Set DECL_INITIAL to a dummy block.
7451 (read_rtl_function_body_from_file_range): New function.
7452 * read-rtl-function.h (read_rtl_function_body_from_file_range):
7454 * run-rtl-passes.c: New file.
7455 * run-rtl-passes.h: New file.
7457 2017-01-24 Jeff Law <law@redhat.com>
7459 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
7462 2017-01-24 Bin Cheng <bin.cheng@arm.com>
7464 PR tree-optimization/79159
7465 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
7466 (record_nonwrapping_iv): Improve boundary using above function if no
7467 value range information.
7469 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
7470 Martin Jambor <mjambor@suse.cz>
7472 * brig-builtins.def: New file.
7473 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
7474 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
7475 (DEF_HSAIL_SAT_BUILTIN): Likewise.
7476 (DEF_HSAIL_INTR_BUILTIN): Likewise.
7477 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
7478 * builtin-types.def (BT_INT8): New.
7479 (BT_INT16): Likewise.
7480 (BT_UINT8): Likewise.
7481 (BT_UINT16): Likewise.
7482 (BT_FN_ULONG): Likewise.
7483 (BT_FN_UINT_INT): Likewise.
7484 (BT_FN_UINT_ULONG): Likewise.
7485 (BT_FN_UINT_LONG): Likewise.
7486 (BT_FN_UINT_PTR): Likewise.
7487 (BT_FN_ULONG_PTR): Likewise.
7488 (BT_FN_INT8_FLOAT): Likewise.
7489 (BT_FN_INT16_FLOAT): Likewise.
7490 (BT_FN_UINT32_FLOAT): Likewise.
7491 (BT_FN_UINT16_FLOAT): Likewise.
7492 (BT_FN_UINT8_FLOAT): Likewise.
7493 (BT_FN_UINT64_FLOAT): Likewise.
7494 (BT_FN_UINT16_UINT32): Likewise.
7495 (BT_FN_UINT32_UINT16): Likewise.
7496 (BT_FN_UINT16_UINT16_UINT16): Likewise.
7497 (BT_FN_INT_PTR_INT): Likewise.
7498 (BT_FN_UINT_PTR_UINT): Likewise.
7499 (BT_FN_LONG_PTR_LONG): Likewise.
7500 (BT_FN_ULONG_PTR_ULONG): Likewise.
7501 (BT_FN_VOID_UINT64_UINT64): Likewise.
7502 (BT_FN_UINT8_UINT8_UINT8): Likewise.
7503 (BT_FN_INT8_INT8_INT8): Likewise.
7504 (BT_FN_INT16_INT16_INT16): Likewise.
7505 (BT_FN_INT_INT_INT): Likewise.
7506 (BT_FN_UINT_FLOAT_UINT): Likewise.
7507 (BT_FN_FLOAT_UINT_UINT): Likewise.
7508 (BT_FN_ULONG_UINT_UINT): Likewise.
7509 (BT_FN_ULONG_UINT_PTR): Likewise.
7510 (BT_FN_ULONG_ULONG_ULONG): Likewise.
7511 (BT_FN_UINT_UINT_UINT): Likewise.
7512 (BT_FN_VOID_UINT_PTR): Likewise.
7513 (BT_FN_UINT_UINT_PTR: Likewise.
7514 (BT_FN_UINT32_UINT64_PTR): Likewise.
7515 (BT_FN_INT_INT_UINT_UINT): Likewise.
7516 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
7517 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
7518 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
7519 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
7520 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
7521 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
7522 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
7523 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
7524 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
7525 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
7526 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
7527 * doc/frontends.texi: List BRIG FE.
7528 * doc/install.texi (Testing): Add BRIG tesring requirements.
7529 * doc/invoke.texi (Overall Options): Mention BRIG.
7530 * doc/standards.texi (Standards): Doucment BRIG HSA version.
7532 2017-01-24 Richard Biener <rguenther@suse.de>
7534 PR translation/79208
7535 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
7537 2017-01-24 Martin Jambor <mjambor@suse.cz>
7540 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
7541 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
7544 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
7547 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
7548 casts from signed to unsigned really don't have a range.
7550 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
7552 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
7553 GMP_RNDx for compatiblity.
7555 2017-01-24 Martin Liska <mliska@suse.cz>
7558 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
7559 that would prevent us to call alloca with -1 as argument.
7561 2017-01-24 Jakub Jelinek <jakub@redhat.com>
7563 * dwarf2out.c (output_compilation_unit_header, output_file_names):
7564 Avoid -Wformat-security warning.
7566 2017-01-23 Andrew Pinski <apinski@cavium.com>
7568 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
7571 2017-01-23 Martin Sebor <msebor@redhat.com>
7574 * gimple-ssa-sprintf.c (warn_level): New global.
7575 (format_integer): Use it here and throughout the rest of the file.
7576 Use the same switch to compute sign as base.
7577 (maybe_warn): New function.
7578 (format_directive): Factor out warnings into maybe_warn.
7579 Add debugging output. Use warn_level.
7580 (add_bytes): Use warn_level.
7581 (pass_sprintf_length::compute_format_length): Add debugging output.
7582 (try_substitute_return_value): Same.
7583 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
7586 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
7587 (struct fmtresult, format_integer, format_floating): Adjust.
7588 (fmtresult::fmtresult): Set max correctly in two argument ctor.
7589 (get_string_length, format_string,format_directive): Same.
7590 (pass_sprintf_length::compute_format_length): Same.
7591 (try_substitute_return_value): Simplify slightly.
7594 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
7595 (fmtresult::operator+=): Outlined.
7596 (struct fmtresult): Add ctors.
7597 (struct conversion_spec): Rename...
7598 (struct directive): ...to this. Add and remove data members.
7599 (directive::set_width, directive::set_precision): New functions.
7600 (format_percent): Use fmtresult ctor.
7601 (get_width_and_precision): Remove.
7602 (format_integer): Make naming changes. Avoid computing width and
7604 (format_floating): Same. Adjust indentation.
7605 (format_character, format_none): New functions.
7606 (format_string): Moved character handling to format_character.
7607 (format_directive): Remove arguments, change return type.
7608 (parse_directive): New function.
7609 (pass_sprintf_length::compute_format_length): Move directive
7610 parsing to parse_directive.
7612 2017-01-23 Jakub Jelinek <jakub@redhat.com>
7614 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
7615 (assign_assembler_name_if_needed): ... this.
7616 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
7617 (assign_assembler_name_if_needed): ... this.
7618 (free_lang_data_in_cgraph): Adjust callers.
7619 * cgraphunit.c (cgraph_node::analyze): Likewise.
7620 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
7623 2017-01-23 Richard Biener <rguenther@suse.de>
7625 PR tree-optimization/79088
7626 PR tree-optimization/79188
7627 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
7628 resetting loop bounds after last path deletion. Reset loop
7629 bounds of the target loop, make code match the comments.
7630 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
7631 Make sure loops need no fixups.
7633 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
7635 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
7636 exponent support with double type for first argument.
7637 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
7638 type returned by __builtin_vec_extract_sig,
7639 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
7640 functions from "vector int" to "vector unsigned int" or from
7641 "vector long long int" to "vector unsigned long long int".
7642 Changed type returned by __builtin_vec_extract_exp,
7643 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
7644 functions from "vector int" to "vector unsigned int" or from
7645 "vector long long int" to "vector unsigned long long int".
7646 Changed return type of __builtin_vec_test_data_class,
7647 __builtin_vec_test_data_class_sp, and
7648 __builtin_vec_test_data_class_dp from "vector int" to
7649 "vector bool int" or from "vector long long int" to "vector bool
7650 long long int" and changed second argument type from "unsigned
7651 int" to "int". Added new overloaded function forms "vector float
7652 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
7653 "vector float __builtin_vec_insert_exp_sp (vector float, vector
7654 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
7655 double, vector unsigned long long int)" and "vector double
7656 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
7657 long int)". Changed return type of
7658 __builtin_scalar_test_data_class and
7659 __builtin_scalar_test_data_class_sp and
7660 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
7661 int" and changed second argument from "unsigned int" to "int".
7662 Changed type returned by __builtin_scalar_test_neg,
7663 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
7664 from "int" to "bool int". Added new overloaded function form
7665 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
7666 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
7667 exponent double-precision with floating point first argument.
7668 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
7669 documentation of scalar_test_data_class, scalar_test_neg,
7670 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
7671 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
7672 vec_test_data_class built-in functions to reflect refinements in
7673 their type signatures.
7675 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
7677 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
7679 (aarch64_elf_asm_destructor): Likewise.
7681 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
7683 PR rtl-optimization/78634
7684 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
7685 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
7686 * ifcvt.c (noce_try_cmove): Add missing cost check.
7688 PR rtl-optimization/71724
7689 * combine.c (if_then_else_cond): Look for situations where it is
7690 beneficial to undo the work of one of the recursive calls.
7692 2017-01-23 Bin Cheng <bin.cheng@arm.com>
7694 PR tree-optimization/70754
7695 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
7696 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
7697 combined stmt before it if not NULL.
7698 (combine_chains): Process refs reversely and compute dominance point
7701 2017-01-23 Martin Liska <mliska@suse.cz>
7703 PR tree-optimization/79196
7704 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
7705 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
7707 (strlen_optimize_stmt): Call the renamed function.
7709 2017-01-23 Michael Matz <matz@suse.de>
7711 PR tree-optimization/78384
7712 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
7714 2017-01-23 Richard Biener <rguenther@suse.de>
7716 PR tree-optimization/79186
7717 * tree-vrp.c (register_new_assert_for): Make sure we've seen
7718 both incoming edges before moving an assert.
7720 2017-01-23 Martin Jambor <mjambor@suse.cz>
7722 * ipa-prop.c (load_from_param_1): Removed.
7723 (load_from_unmodified_param): Bits from load_from_param_1 put back
7725 (load_from_param): Removed.
7726 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
7727 with stmt. Reverted back to use of load_from_unmodified_param.
7729 2017-01-23 Martin Jambor <mjambor@suse.cz>
7732 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
7733 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
7734 field a pointer to garbage collected vector, mark lattices and
7735 ipcp_orig_node with GTY((skip)).
7736 (ipa_get_param_count): Adjust to descriptors being a pointer.
7737 (ipa_get_param): Likewise.
7738 (ipa_get_type): Likewise.
7739 (ipa_get_param_move_cost): Likewise.
7740 (ipa_set_param_used): Likewise.
7741 (ipa_get_controlled_uses): Likewise.
7742 (ipa_set_controlled_uses): Likewise.
7743 (ipa_is_param_used): Likewise.
7744 (ipa_node_params_t): Move into garbage collector. New methods insert
7746 (ipa_node_params_sum): Annotate wth GTY(()).
7747 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
7749 (ipa_load_from_parm_agg): Adjust declaration.
7750 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
7751 * ipa-profile.c (ipa_profile): Likewise.
7752 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
7753 (ipa_populate_param_decls): Make descriptors parameter garbage
7755 (ipa_dump_param): Adjust to descriptors being a pointer.
7756 (ipa_alloc_node_params): Likewise.
7757 (ipa_initialize_node_params): Likewise.
7758 (load_from_param_1): Make descriptors parameter garbage collected.
7759 (load_from_unmodified_param): Likewise.
7760 (load_from_param): Likewise.
7761 (ipa_load_from_parm_agg): Likewise.
7762 (ipa_node_params::~ipa_node_params): Removed.
7763 (ipa_free_all_node_params): Remove call to delete operator.
7764 (ipa_node_params_t::insert): New.
7765 (ipa_node_params_t::remove): Likewise.
7766 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
7767 copy known_csts and known_contexts vectors.
7768 (ipa_read_node_info): Adjust to descriptors being a pointer.
7769 (ipcp_modif_dom_walker): Make m_descriptors field garbage
7771 (ipcp_transform_function): Make descriptors variable garbage
7774 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
7776 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
7777 * config/i386/avx512dqintrin.h: Ditto.
7778 * config/i386/avx512fintrin.h: Ditto.
7779 * gcc/config/i386/i386.c: Handle new builtins.
7780 * config/i386/i386-builtin.def: Add new builtins.
7781 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
7782 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
7784 2017-01-23 Jakub Jelinek <jakub@redhat.com>
7785 Martin Liska <mliska@suse.cz>
7787 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
7788 * asan.c (asan_expand_poison_ifn): Support stores and use
7789 appropriate ASAN report function.
7790 * internal-fn.c (expand_ASAN_POISON_USE): New function.
7791 * internal-fn.def (ASAN_POISON_USE): Declare.
7792 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
7793 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
7794 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
7795 ASAN_POISON calls w/o LHS.
7796 * tree-ssa.c (execute_update_addresses_taken): Create clobber
7797 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
7798 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
7799 * gimplify.c (asan_poison_variables): Add attribute
7800 use_after_scope_memory to variables that really needs to live
7802 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
7803 having the attribute.
7805 2017-01-23 Martin Liska <mliska@suse.cz>
7807 * asan.c (create_asan_shadow_var): New function.
7808 (asan_expand_poison_ifn): Likewise.
7809 * asan.h (asan_expand_poison_ifn): New declaration.
7810 * internal-fn.c (expand_ASAN_POISON): Likewise.
7811 * internal-fn.def (ASAN_POISON): New builtin.
7812 * sanopt.c (pass_sanopt::execute): Expand
7813 asan_expand_poison_ifn.
7814 * tree-inline.c (copy_decl_for_dup_finish): Make function
7816 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
7817 * tree-ssa.c (is_asan_mark_p): New function.
7818 (execute_update_addresses_taken): Rewrite local variables
7819 (identified just by use-after-scope as addressable) into SSA.
7821 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
7823 * doc/install.texi (Specific): opensource.apple.com uses https
7824 now. Remove trailing slash.
7826 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
7828 * README.Portability: Remove note on an Irix compatibility issue.
7830 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
7832 * gcov.c (INCLUDE_ALGORITHM): Define.
7833 (INCLUDE_VECTOR): Define.
7834 No longer include <vector> and <algorithm> directly.
7836 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
7838 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
7840 * doc/invoke.texi (Code Gen Options): Ditto.
7842 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
7845 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
7847 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
7849 rtl-optimization/79125
7850 * cprop.c (local_cprop_pass): Handle cases where we make an
7853 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
7857 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
7858 read from, for big endian.
7860 2017-01-20 Jiong Wang <jiong.wang@arm.com>
7862 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
7863 register pauth builtins for LP64 only.
7865 2017-01-20 Marek Polacek <polacek@redhat.com>
7868 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
7871 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
7873 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
7875 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
7876 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
7877 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
7879 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
7882 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
7883 in big-endian mode when they are not a single duplicated value.
7885 2017-01-20 Richard Biener <rguenther@suse.de>
7887 * BASE-VER: Bump to 7.0.1.
7889 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
7891 * omp-low.c (omplow_simd_context): New struct. Use it...
7892 (lower_rec_simd_input_clauses): ...here and...
7893 (lower_rec_input_clauses): ...here to hold common data. Adjust all
7894 references to idx, lane, max_vf, is_simt.
7896 2017-01-20 Graham Markall <graham.markall@embecosm.com>
7898 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
7901 2017-01-20 Martin Jambor <mjambor@suse.cz>
7903 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
7904 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
7906 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
7907 (GTFILES): Rename hsa.c to hsa-common.c.
7908 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
7909 * hsa-dump.c: Likewise.
7910 * hsa-gen.c: Likewise.
7911 * hsa-regalloc.c: Likewise.
7912 * ipa-hsa.c: Likewise.
7913 * omp-expand.c: Likewise.
7914 * omp-low.c: Likewise.
7915 * toplev.c: Likewise.
7917 2017-01-20 Marek Polacek <polacek@redhat.com>
7920 * doc/invoke.texi: Document -Wduplicated-branches.
7921 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
7922 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
7923 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
7924 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
7925 return 0 only when not OEP_LEXICOGRAPHIC.
7926 (fold_build_cleanup_point_expr): Use the expression
7927 location when building CLEANUP_POINT_EXPR.
7928 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
7929 * tree.c (add_expr): Handle error_mark_node.
7931 2017-01-20 Martin Liska <mliska@suse.cz>
7934 * tree-profile.c (init_ic_make_global_vars): Do not call
7936 (gimple_init_gcov_profiler): Likewise.
7938 2017-01-20 Martin Liska <mliska@suse.cz>
7941 * cgraph.h (maybe_create_reference): Remove argument and
7943 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
7945 * ipa-cp.c (create_specialized_node): Likewise.
7946 * symtab.c (symtab_node::maybe_create_reference): Handle
7947 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
7949 2017-01-20 Martin Liska <mliska@suse.cz>
7951 * read-rtl-function.c (function_reader::create_function): Use
7952 build_decl instread of build_decl_stat.
7954 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
7956 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
7957 * config/i386/avx512dqintrin.h: Ditto.
7958 * config/i386/avx512fintrin.h: Ditto.
7959 * config/i386/i386-builtin-types.def: Add new types.
7960 * gcc/config/i386/i386.c: Handle new types.
7961 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
7962 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
7963 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
7964 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
7965 (__builtin_ia32_kshiftridi): New.
7966 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
7968 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
7972 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
7973 define to rs6000_init_stack_protect_guard.
7974 (rs6000_init_stack_protect_guard): New function.
7976 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
7977 Yunqiang Su <yunqiang.su@imgtec.com>
7979 * config.gcc (supported_defaults): Add madd4.
7980 (with_madd4): Add validation.
7981 (all_defaults): Add madd4.
7982 * config/mips/mips.opt (mmadd4): New option.
7983 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
7985 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
7987 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
7988 (ISA_HAS_FUSED_MADD4): Likewise.
7989 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
7990 * gcc/doc/install.texi (--with-madd4): Document the new option.
7992 2017-01-19 Jiong Wang <jiong.wang@arm.com>
7994 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
7995 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
7996 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
7997 (aarch64_init_pauth_hint_builtins): New.
7998 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
7999 (aarch64_expand_builtin): Expand new builtins.
8001 2017-01-19 Jiong Wang <jiong.wang@arm.com>
8003 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
8004 * combine-stack-adj.c (no_unhandled_cfa): Handle
8005 REG_CFA_TOGGLE_RA_MANGLE.
8006 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
8007 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
8008 info for return address signing.
8009 (aarch64_expand_epilogue): Likewise.
8011 2017-01-19 Jiong Wang <jiong.wang@arm.com>
8013 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
8014 * config/aarch64/aarch64-protos.h
8015 (aarch64_return_address_signing_enabled): New declaration.
8016 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
8018 (aarch64_expand_prologue): Sign return address before it's pushed onto
8020 (aarch64_expand_epilogue): Authenticate return address fetched from
8022 (aarch64_override_options): Sanity check for ILP32 and ISA level.
8023 (aarch64_attributes): New function attributes for "sign-return-address".
8024 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
8025 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
8026 ("*do_return"): Generate combined instructions according to key index.
8027 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
8028 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
8030 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
8031 * config/aarch64/aarch64.opt (msign-return-address=): New.
8032 * doc/extend.texi (AArch64 Function Attributes): Documents
8033 "sign-return-address=".
8034 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
8036 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
8038 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
8039 overall option summary.
8041 2017-01-19 Jiong Wang <jiong.wang@arm.com>
8043 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
8044 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
8045 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
8046 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
8048 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
8050 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
8051 -mpower9-minmax by default for -mcpu=power9.
8052 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
8053 128-bit floating point.
8055 2017-01-20 Alan Modra <amodra@gmail.com>
8057 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
8058 optimizing for size.
8060 2017-01-20 Alan Modra <amodra@gmail.com>
8063 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
8064 for strcmp and strncmp from corresponding builtin decl.
8066 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
8068 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
8069 instead of i386/rtems-64.h.
8070 * config/i386/rtems-64.h: Remove.
8072 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
8076 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
8078 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
8080 2017-01-19 Tamar Christina <tamar.christina@arm.com>
8082 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
8083 Change int to HOST_WIDE_INT.
8084 * config/aarch64/aarch64-protos.h
8085 (aarch64_simd_gen_const_vector_dup): Likewise.
8086 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
8088 2017-01-19 David Malcolm <dmalcolm@redhat.com>
8090 * langhooks-def.h (lhd_type_for_size): New decl.
8091 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
8092 * langhooks.c (lhd_type_for_size): New function, taken from
8095 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
8097 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
8098 define_bypass for CR latency.
8099 (power9-cracked-alu): Update bypass latency and remove power9-branch.
8100 (power9-alu2): Add define_bypass for CR latency.
8102 (power9-mul): Update insn latency.
8103 (power9-mul-compare): Update insn latency, bypass latency and remove
8106 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8108 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
8110 * config/aarch64/aarch64.md
8111 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
8112 aarch64_nopcrelative_literal_loads.
8113 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
8115 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
8117 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
8119 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
8121 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
8124 * config.gcc (supported_defaults): Add lxc1-sxc1.
8125 (with_lxc1_sxc1): Add validation.
8126 (all_defaults): Add lxc1-sxc1.
8127 * config/mips/mips.opt (mlxc1-sxc1): New option.
8128 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
8130 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
8131 __mips_no_lxc1_sxc1.
8132 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
8133 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
8134 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
8136 2017-01-19 Richard Biener <rguenther@suse.de>
8138 PR tree-optimization/72488
8139 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
8140 sure to restore SSA info.
8141 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
8143 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
8145 PR rtl-optimization/79121
8146 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
8147 of the inner type when shifting an extended value.
8149 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
8152 * symtab.c (symtab_node::equal_address_to): Fix comparing of
8153 interposable aliases.
8155 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
8158 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
8159 Use the evmergelohi instruction.
8160 (mov_si<mode>_e500_subreg4_2_le): Likewise.
8161 (mov_sitf_e500_subreg8_2_be): Likewise.
8162 (mov_sitf_e500_subreg12_2_le): Likewise.
8163 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
8164 (mov_si<mode>_e500_subreg4_2_be): Likewise.
8165 (mov_sitf_e500_subreg8_2_le): Likewise.
8166 (mov_sitf_e500_subreg12_2_be): Likewise.
8168 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8170 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
8171 attribute from vecsimple to vecperm.
8172 (altivec_vbpermq2): Likewise.
8174 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8177 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
8179 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
8180 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
8181 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
8182 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
8183 case where N arg is SIZE_MAX.
8184 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
8185 (cmpstrsi): Add pattern.
8187 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
8189 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
8190 __builtin_vec_revb builtins.
8191 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
8192 built-in functions to support generation of the ISA 3.0 XXBR<x>
8193 vector byte reverse instructions.
8194 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
8195 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
8196 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
8197 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
8198 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
8199 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
8200 (P9V_BUILTIN_VEC_REVB): Likewise.
8201 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
8202 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
8203 (p9_xxbrq_v16qi): Likewise.
8204 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
8205 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
8206 (p9_xxbrh_v8hi): Likewise.
8207 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
8208 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
8209 vec_revb built-in functions.
8211 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
8213 PR rtl-optimization/78952
8214 * config/i386/i386.md (any_extract): New code iterator.
8215 (*insvqi_2): Use any_extract for source operand.
8216 (*insvqi_3): Use any_shiftrt for source operand.
8218 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
8220 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
8222 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
8224 2017-01-18 Matthias Klose <doko@ubuntu.com>
8226 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
8228 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8230 * config/rs6000/altivec.h (vec_bperm): Change #define.
8231 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
8232 (altivec_vbpermq2): New define_insn.
8233 (altivec_vbpermd): Likewise.
8234 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
8236 (VBPERMD): Likewise.
8237 (VBPERM): New polymorphic function interface.
8238 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
8239 Add entries for P9V_BUILTIN_VEC_VBPERM.
8240 * doc/extend.texi: Add interfaces for vec_bperm.
8242 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
8244 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
8245 first letter of error messages.
8246 (s390_resolve_overloaded_builtin): Likewise.
8247 * config/s390/s390.c (s390_expand_builtin): Likewise.
8248 (s390_invalid_arg_for_unprototyped_fn): Likewise.
8249 (s390_valid_target_attribute_inner_p): Likewise.
8250 * config/s390/s390.md ("tabort"): Likewise.
8252 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
8254 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
8255 (ISA_AVOID_DIV_HILO): New macro.
8256 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
8257 (ISA_HAS_DDIV): Likewise.
8259 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
8261 * doc/invoke.texi (fabi-version): Correct number of occurrences.
8263 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
8265 * doc/invoke.texi (fabi-version): Spelling fix.
8267 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
8270 * doc/invoke.texi (fabi-version): Mention mangling fix for
8273 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
8276 * doc/invoke.texi (fabi-version): Document discriminator mangling.
8278 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
8281 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
8282 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
8284 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
8286 (stack_protect_test): Ditto.
8287 * config/rs6000/rs6000.opt (mstack-protector-guard=,
8288 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
8290 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
8291 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
8292 -mstack-protector-guard-offset=.
8293 (RS/6000 and PowerPC Options): Ditto.
8295 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
8297 * config/i386/i386.h (MASK_CLASS_P): New define.
8298 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
8299 there are no registers from different register sets also when
8300 mask registers are used. Update function comment.
8301 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
8302 to (*k/*r) and (*k/*km) alternatives.
8304 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
8306 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
8307 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
8308 (EH_RETURN_HANDLER_RTX): New define.
8309 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
8310 Force frame pointer in EH return functions.
8311 (aarch64_expand_epilogue): Add barrier for eh_return.
8312 (aarch64_final_eh_return_addr): Remove.
8313 (aarch64_eh_return_handler_rtx): New function.
8314 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
8316 (aarch64_eh_return_handler_rtx): New prototype.
8318 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8320 * config/rs6000/altivec.h (vec_rlmi): New #define.
8321 (vec_vrlnm): Likewise.
8322 (vec_rlnm): Likewise.
8323 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
8324 (UNSPEC_VRLNM): Likewise.
8325 (VIlong): New mode iterator.
8326 (altivec_vrl<VI_char>mi): New define_insn.
8327 (altivec_vrl<VI_char>nm): Likewise.
8328 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
8331 (RLNM): New polymorphic function entry.
8332 (VRLWMI): New monomorphic function entry.
8334 (RLMI): New polymorphic function entry.
8335 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
8336 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
8337 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
8340 2017-01-17 Jakub Jelinek <jakub@redhat.com>
8343 * dwarf2out.c (field_byte_offset): Restore the
8344 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
8345 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
8346 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
8349 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
8352 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
8354 2017-01-17 Jakub Jelinek <jakub@redhat.com>
8357 * dwarf2out.c (add_data_member_location_attribute): For constant
8358 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
8359 instead of DW_AT_data_member_location, DW_AT_bit_offset and
8360 DW_AT_byte_size attributes.
8362 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
8364 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
8365 after forcing to constant memory when the code model is medium.
8367 2017-01-17 Julia Koval <julia.koval@intel.com>
8370 * config/i386/avx512fintrin.h
8371 (_mm512_i32gather_ps): Change __addr type to void const*.
8372 (_mm512_mask_i32gather_ps): Ditto.
8373 (_mm512_i32gather_pd): Ditto.
8374 (_mm512_mask_i32gather_pd): Ditto.
8375 (_mm512_i64gather_ps): Ditto.
8376 (_mm512_mask_i64gather_ps): Ditto.
8377 (_mm512_i64gather_pd): Ditto.
8378 (_mm512_mask_i64gather_pd): Ditto.
8379 (_mm512_i32gather_epi32): Ditto.
8380 (_mm512_mask_i32gather_epi32): Ditto.
8381 (_mm512_i32gather_epi64): Ditto.
8382 (_mm512_mask_i32gather_epi64): Ditto.
8383 (_mm512_i64gather_epi32): Ditto.
8384 (_mm512_mask_i64gather_epi32): Ditto.
8385 (_mm512_i64gather_epi64): Ditto.
8386 (_mm512_mask_i64gather_epi64): Ditto.
8387 (_mm512_i32scatter_ps): Change __addr type to void*.
8388 (_mm512_mask_i32scatter_ps): Ditto.
8389 (_mm512_i32scatter_pd): Ditto.
8390 (_mm512_mask_i32scatter_pd): Ditto.
8391 (_mm512_i64scatter_ps): Ditto.
8392 (_mm512_mask_i64scatter_ps): Ditto.
8393 (_mm512_i64scatter_pd): Ditto.
8394 (_mm512_mask_i64scatter_pd): Ditto.
8395 (_mm512_i32scatter_epi32): Ditto.
8396 (_mm512_mask_i32scatter_epi32): Ditto.
8397 (_mm512_i32scatter_epi64): Ditto.
8398 (_mm512_mask_i32scatter_epi64): Ditto.
8399 (_mm512_i64scatter_epi32): Ditto.
8400 (_mm512_mask_i64scatter_epi32): Ditto.
8401 (_mm512_i64scatter_epi64): Ditto.
8402 (_mm512_mask_i64scatter_epi64): Ditto.
8403 * config/i386/avx512pfintrin.h
8404 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
8405 (_mm512_mask_prefetch_i32gather_ps): Ditto.
8406 (_mm512_mask_prefetch_i64gather_pd): Ditto.
8407 (_mm512_mask_prefetch_i64gather_ps): Ditto.
8408 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
8409 (_mm512_prefetch_i32scatter_ps): Ditto.
8410 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
8411 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
8412 (_mm512_prefetch_i64scatter_pd): Ditto.
8413 (_mm512_prefetch_i64scatter_ps): Ditto.
8414 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
8415 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
8416 * config/i386/avx512vlintrin.h
8417 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
8418 (_mm_mmask_i32gather_ps): Ditto.
8419 (_mm256_mmask_i32gather_pd): Ditto.
8420 (_mm_mmask_i32gather_pd): Ditto.
8421 (_mm256_mmask_i64gather_ps): Ditto.
8422 (_mm_mmask_i64gather_ps): Ditto.
8423 (_mm256_mmask_i64gather_pd): Ditto.
8424 (_mm_mmask_i64gather_pd): Ditto.
8425 (_mm256_mmask_i32gather_epi32): Ditto.
8426 (_mm_mmask_i32gather_epi32): Ditto.
8427 (_mm256_mmask_i32gather_epi64): Ditto.
8428 (_mm_mmask_i32gather_epi64): Ditto.
8429 (_mm256_mmask_i64gather_epi32): Ditto.
8430 (_mm_mmask_i64gather_epi32): Ditto.
8431 (_mm256_mmask_i64gather_epi64): Ditto.
8432 (_mm_mmask_i64gather_epi64): Ditto.
8433 (_mm256_i32scatter_ps): Change __addr type to void*.
8434 (_mm256_mask_i32scatter_ps): Ditto.
8435 (_mm_i32scatter_ps): Ditto.
8436 (_mm_mask_i32scatter_ps): Ditto.
8437 (_mm256_i32scatter_pd): Ditto.
8438 (_mm256_mask_i32scatter_pd): Ditto.
8439 (_mm_i32scatter_pd): Ditto.
8440 (_mm_mask_i32scatter_pd): Ditto.
8441 (_mm256_i64scatter_ps): Ditto.
8442 (_mm256_mask_i64scatter_ps): Ditto.
8443 (_mm_i64scatter_ps): Ditto.
8444 (_mm_mask_i64scatter_ps): Ditto.
8445 (_mm256_i64scatter_pd): Ditto.
8446 (_mm256_mask_i64scatter_pd): Ditto.
8447 (_mm_i64scatter_pd): Ditto.
8448 (_mm_mask_i64scatter_pd): Ditto.
8449 (_mm256_i32scatter_epi32): Ditto.
8450 (_mm256_mask_i32scatter_epi32): Ditto.
8451 (_mm_i32scatter_epi32): Ditto.
8452 (_mm_mask_i32scatter_epi32): Ditto.
8453 (_mm256_i32scatter_epi64): Ditto.
8454 (_mm256_mask_i32scatter_epi64): Ditto.
8455 (_mm_i32scatter_epi64): Ditto.
8456 (_mm_mask_i32scatter_epi64): Ditto.
8457 (_mm256_i64scatter_epi32): Ditto.
8458 (_mm256_mask_i64scatter_epi32): Ditto.
8459 (_mm_i64scatter_epi32): Ditto.
8460 (_mm_mask_i64scatter_epi32): Ditto.
8461 (_mm256_i64scatter_epi64): Ditto.
8462 (_mm256_mask_i64scatter_epi64): Ditto.
8463 (_mm_i64scatter_epi64): Ditto.
8464 (_mm_mask_i64scatter_epi64): Ditto.
8465 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
8466 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
8467 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
8468 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
8469 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
8470 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
8471 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
8472 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
8473 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
8474 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
8475 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
8476 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
8477 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
8478 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
8479 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
8480 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
8481 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
8482 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
8483 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
8484 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
8485 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
8486 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
8487 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
8488 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
8489 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
8490 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
8491 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
8492 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
8493 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
8494 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
8495 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
8496 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
8497 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
8498 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
8499 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
8500 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
8501 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
8502 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
8503 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
8504 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
8505 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
8506 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
8507 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
8508 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
8509 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
8510 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
8511 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
8512 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
8513 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
8514 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
8515 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
8516 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
8517 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
8518 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
8519 definitions accordingly.
8521 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
8522 Kuan-Lin Chen <kuanlinchentw@gmail.com>
8525 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
8528 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
8531 * ira-conflicts.c (ira_build_conflicts): Update total conflict
8532 hard regs for inner regno.
8534 2017-01-17 Martin Liska <mliska@suse.cz>
8537 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
8538 assumption and add comment.
8540 2017-01-17 Nathan Sidwell <nathan@acm.org>
8542 * ipa-visibility.c (localize_node): New function, broken out of ...
8543 (function_and_variable_visibility): ... here. Call it.
8545 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
8548 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
8549 correctly set frequency of oudgoing edge.
8550 (duplicate_thread_path): Fix profile updating.
8552 2017-01-17 Jakub Jelinek <jakub@redhat.com>
8555 * configure.ac: Add GCC_BASE_VER.
8556 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
8557 version from BASE-VER file.
8558 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
8559 (gcc.o): Depend on $(BASEVER).
8560 * common.opt (dumpfullversion): New option.
8561 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
8562 * doc/invoke.texi: Document -dumpfullversion.
8563 * doc/install.texi: Document --with-gcc-major-version-only.
8564 * configure: Regenerated.
8566 2017-01-17 Richard Biener <rguenther@suse.de>
8568 PR tree-optimization/71433
8569 * tree-vrp.c (register_new_assert_for): Merge same asserts
8570 on all incoming edges.
8571 (process_assert_insertions_for): Handle insertions at the
8574 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
8576 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
8577 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
8579 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
8582 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
8585 2017-01-17 Alan Modra <amodra@gmail.com>
8588 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
8589 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
8590 symbolic stack limit when pic.
8592 2017-01-16 Martin Sebor <msebor@redhat.com>
8594 PR tree-optimization/78608
8595 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
8597 2017-01-16 Jeff Law <law@redhat.com>
8600 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
8601 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
8602 for several include directories that may be relative to sysroot.
8603 * config/i386/x-mingw32 (gplus_includedir): Define.
8604 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
8605 (native_system_includedir): Likewise.
8606 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
8607 override if TARGET_SYSTEM_ROOT is defined.
8608 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
8610 PR tree-optimization/79090
8611 PR tree-optimization/33562
8612 PR tree-optimization/61912
8613 PR tree-optimization/77485
8614 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
8615 and computed trims into the dump file.
8617 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
8619 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
8621 2017-01-16 Jakub Jelinek <jakub@redhat.com>
8624 * gimplify.c (gimplify_init_constructor): If want_value and
8625 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
8629 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
8630 sequence. Formatting fixes.
8631 (doloop_optimize): Formatting fixes.
8634 * gcc.c (debug_level_greater_than_spec_func): New function.
8635 (static_spec_functions): Add debug-level-gt spec function.
8636 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
8638 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
8639 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
8640 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
8641 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
8642 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
8643 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
8645 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
8647 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
8648 QImode fixups to general and mask registers only.
8650 2017-01-16 Carl Love <cel@us.ibm.com>
8652 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
8653 for built-in functions
8654 vector signed char vec_nabs (vector signed char)
8655 vector signed short vec_nabs (vector signed short)
8656 vector signed int vec_nabs (vector signed int)
8657 vector signed long long vec_nabs (vector signed long long)
8658 vector float vec_nabs (vector float)
8659 vector double vec_nabs (vector double)
8660 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
8662 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
8663 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
8664 * doc/extend.texi: Update the documentation file for the new built-in
8667 2017-01-16 Martin Sebor <msebor@redhat.com>
8669 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
8672 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8674 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
8675 UNSPEC_VSX__XXSPLTD to require special splat handling.
8677 2017-01-16 David Malcolm <dmalcolm@redhat.com>
8680 * system.h: Poison strndup.
8682 2017-01-16 Alan Modra <amodra@gmail.com>
8685 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
8688 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
8690 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
8692 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
8694 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
8695 call recog here. Assert that INSN_CODE (insn) is non-negative.
8697 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
8700 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
8702 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
8703 in the currently scheduled RTL fragment.
8705 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
8707 PR rtl-optimization/78751
8708 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
8711 2017-01-14 Jeff Law <law@redhat.com>
8713 PR tree-optimization/79090
8714 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
8715 variable length stores.
8716 (compute_trims): Delete dead assignment to *trim_tail.
8717 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
8720 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
8722 PR rtl-optimization/78626
8723 PR rtl-optimization/78727
8724 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
8725 of a block, and split such blocks after everything else is finished.
8727 2017-01-14 Alan Modra <amodra@gmail.com>
8730 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
8731 target legitimate_combined_insn.
8732 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
8733 (rs6000_legitimate_combined_insn): New function.
8734 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
8736 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
8737 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
8738 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
8740 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
8742 * doc/frontends.texi (G++ and GCC): Remove references to Java.
8744 2017-01-13 Jeff Law <law@redhat.com>
8746 PR tree-optimization/33562
8747 PR tree-optimization/61912
8748 PR tree-optimization/77485
8749 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
8751 (delete_dead_assignment): Likewise.
8752 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
8753 statement to delete_dead_call and delete_dead_assignment.
8755 2017-01-13 David Malcolm <dmalcolm@redhat.com>
8758 * substring-locations.c (format_warning_va): Strengthen case 1 so
8759 that both endpoints of the substring must be within the format
8760 range for just the substring to be printed.
8762 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
8764 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
8765 * config/i386/i386.c (ix86_target_string): Add missing options
8766 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
8767 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
8768 flags_other and ix86_target_other to flags2_other. Display unknown
8770 (ix86_valid_target_attribute_inner_p): Add missing options and
8771 reorder options by implied ISAs, as in ix86_target_string.
8773 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
8775 * hash-table.h (hash_table::too_empty_p): New function.
8776 (hash_table::expand): Use it.
8777 (hash_table::traverse): Likewise.
8778 (hash_table::empty_slot): Use sizeof (value_type) instead of
8779 sizeof (PTR) to convert bytes to elements. Shrink the table
8780 if the current size is excessive for the current number of
8783 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
8785 * ira-costs.c (record_reg_classes): Break from the inner loop
8786 early once alt_fail is known to be true. Update outer loop
8787 handling accordingly.
8789 2017-01-13 Jeff Law <law@redhat.com>
8791 * tree-ssa-dse.c (decrement_count): New function.
8792 (increment_start_addr, maybe_trim_memstar_call): Likewise.
8793 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
8794 when we know the partially dead statement is a mem* function.
8796 PR tree-optimization/61912
8797 PR tree-optimization/77485
8798 * tree-ssa-dse.c: Include expr.h.
8799 (maybe_trim_constructor_store): New function.
8800 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
8802 PR tree-optimization/33562
8803 PR tree-optimization/61912
8804 PR tree-optimization/77485
8805 * doc/invoke.texi: Document new dse-max-object-size param.
8806 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
8807 * tree-ssa-dse.c: Include params.h.
8808 (dse_store_status): New enum.
8809 (initialize_ao_ref_for_dse): New, partially extracted from
8811 (valid_ao_ref_for_dse, normalize_ref): New.
8812 (setup_live_bytes_from_ref, compute_trims): Likewise.
8813 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
8814 (maybe_trim_partially_dead_store): Likewise.
8815 (maybe_trim_complex_store): Likewise.
8816 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
8817 Track what bytes live from the original store. Return tri-state
8818 for dead, partially dead or live.
8819 (dse_dom_walker): Add constructor, destructor and new private members.
8820 (delete_dead_call, delete_dead_assignment): New extracted from
8822 (dse_optimize_stmt): Make a member of dse_dom_walker.
8823 Use initialize_ao_ref_for_dse.
8825 PR tree-optimization/33562
8826 PR tree-optimization/61912
8827 PR tree-optimization/77485
8828 * sbitmap.h (bitmap_count_bits): Prototype.
8829 (bitmap_clear_range, bitmap_set_range): Likewise.
8830 * sbitmap.c (bitmap_clear_range): New function.
8831 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
8833 2017-01-13 Martin Liska <mliska@suse.cz>
8836 * function.c (set_cfun): Add new argument force.
8837 * function.h (set_cfun): Likewise.
8838 * ipa-inline-transform.c (inline_call): Use the function when
8839 strict alising from is dropped for function we inline to.
8841 2017-01-13 Richard Biener <rguenther@suse.de>
8843 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
8844 for dumping GIMPLE INTEGER_CSTs.
8846 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8848 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
8849 to 201112L since C++17.
8851 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
8854 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
8855 if -fsanitize=kernel-address is present.
8857 2017-01-13 Richard Biener <rguenther@suse.de>
8859 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
8860 as _Literal ( type ) number in case usual suffixes do not
8861 preserve all information.
8863 2017-01-13 Richard Biener <rguenther@suse.de>
8865 PR tree-optimization/77283
8866 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
8867 and ssa-iterators.h.
8868 (is_feasible_trace): Implement a cost model based on joiner
8871 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
8874 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
8875 char or short to __float128/_Float128 directly.
8877 2017-01-12 Martin Sebor <msebor@redhat.com>
8879 to -Wformat-overflow.
8880 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
8881 (min_bytes_remaining): Same.
8882 (get_string_length): Same.
8883 (format_string): Same.
8884 (format_directive): Same.
8886 (pass_sprintf_length::handle_gimple_call): Same.
8888 2017-01-12 Jakub Jelinek <jakub@redhat.com>
8890 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
8891 info.nowrite calls with no lhs that can't throw. Return bool
8892 whether gsi_remove has been called or not.
8893 (pass_sprintf_length::handle_gimple_call): Return bool whether
8894 try_substitute_return_value called gsi_remove. Formatting fix.
8895 (pass_sprintf_length::execute): Don't use gsi_remove if
8896 handle_gimple_call returned true.
8899 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
8900 be removed due to side-effects, don't remove following barrier nor
8901 turn the successor edge into fallthru edge.
8903 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
8906 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
8907 element-reversing loads and stores as not swappable.
8909 2017-01-12 Nathan Sidwell <nathan@acm.org>
8910 Nicolai Stange <nicstange@gmail.com>
8912 * combine.c (try_combine): Don't ignore result of overlap checking
8913 loop. Combine overlap & asm check into single loop.
8915 2017-01-12 Richard Biener <rguenther@suse.de>
8917 * tree-pretty-print.c (dump_generic_node): Provide -gimple
8918 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
8920 2017-01-12 Richard Biener <rguenther@suse.de>
8922 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
8923 and TS_TARGET_OPTION directly derive from TS_BASE.
8924 * tree-core.h (tree_optimization_option): Derive from tree_base.
8925 (tree_target_option): Likewise.
8927 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
8929 * config/i386/i386.c (memory_address_length): Increase len
8930 only when rip_relative_addr_p returns false.
8932 2017-01-11 Julia Koval <julia.koval@intel.com>
8934 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
8935 (OPTION_MASK_ISA_SGX_SET): New.
8936 (ix86_handle_option): Handle OPT_msgx.
8937 * config.gcc: Added sgxintrin.h.
8938 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
8939 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
8940 * config/i386/i386.c (ix86_target_string): Add -msgx.
8942 (ix86_option_override_internal): Handle new options.
8943 (ix86_valid_target_attribute_inner_p): Add sgx.
8944 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
8945 * config/i386/i386.opt: Add msgx.
8946 * config/i386/sgxintrin.h: New file.
8947 * config/i386/x86intrin.h: Add sgxintrin.h.
8949 2017-01-11 Jakub Jelinek <jakub@redhat.com>
8952 * fold-const.c (maybe_nonzero_address): Return 1 for function
8954 (tree_single_nonzero_warnv_p): Don't handle function local objects
8958 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
8961 2017-01-11 David Malcolm <dmalcolm@redhat.com>
8964 * opts.c: Include "spellcheck.h"
8965 (struct string_fragment): New struct.
8966 (struct edit_distance_traits<const string_fragment &>): New
8968 (get_closest_sanitizer_option): New function.
8969 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
8971 2017-01-11 Jakub Jelinek <jakub@redhat.com>
8973 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
8975 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
8976 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
8977 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
8978 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
8979 for initial die_offset if dwarf_split_debug_info.
8980 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
8981 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
8983 (output_skeleton_debug_sections): Formatting fix. Use
8984 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
8985 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
8987 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
8989 * config/arm/cortex-a53.md: Add bypasses for
8991 (cortex_a53_r2f): Only use for transfers.
8992 (cortex_a53_f2r): Likewise.
8993 (cortex_a53_r2f_cvt): Add reservation for conversions.
8994 (cortex_a53_f2r_cvt): Likewise.
8996 2017-01-11 Tamar Christina <tamar.christina@arm.com>
8998 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
8999 to all inlined functions, change static to extern.
9001 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
9004 * config/arm/arm.c (legitimize_pic_address): Handle reference to
9006 (arm_assemble_integer): Likewise.
9008 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
9010 * config.gcc: Use new awk script to check CPU, FPU and architecture
9011 parameters for --with-... options.
9012 * config/arm/parsecpu.awk: New file
9013 * config/arm/arm-cpus.in: New file.
9014 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
9016 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
9018 * config/arm/t-arm: Update dependency rules.
9019 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
9020 of processing .def files.
9021 * config/arm/genopt.sh: Deleted.
9022 * config/arm/gentune.sh: Deleted.
9023 * config/arm/arm-cores.def: Deleted.
9024 * config/arm/arm-arches.def: Deleted.
9025 * config/arm/arm-fpus.def: Deleted.
9026 * config/arm/arm-tune.md: Regenerated.
9027 * config/arm/arm-tables.opt: Regenerated.
9028 * config/arm/arm-cpu.h: New generated file.
9029 * config/arm/arm-cpu-data.h: New generated file.
9030 * config/arm/arm-cpu-cdata.h: New generated file.
9032 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
9035 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
9037 (input_varpool_node): Unpack dynamically_initialized bit.
9039 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
9041 PR rtl-optimization/79032
9042 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
9043 the alignment of the adjusted memory reference against that of MODE,
9044 instead of the alignment of the original memory reference.
9046 2017-01-11 Martin Jambor <mjambor@suse.cz>
9048 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
9050 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
9051 decorated functions.
9053 2017-01-11 Richard Biener <rguenther@suse.de>
9055 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
9056 set range/nonnull info for PHI results. Do not set it on
9057 stmts marked for removal.
9059 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
9061 * expr.c (store_field): In the bitfield case, fetch the return value
9062 from the registers before applying a single big-endian adjustment.
9063 Always do a final load for a BLKmode value not larger than a word.
9065 2017-01-10 David Malcolm <dmalcolm@redhat.com>
9068 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
9069 that we correctly handle column numbers greater than
9070 LINE_MAP_MAX_COLUMN_NUMBER.
9072 2017-01-10 Martin Sebor <msebor@redhat.com>
9075 * gimple-ssa-sprintf.c (get_destination_size): Call
9076 {init,fini}object_sizes.
9077 * tree-object-size.c (addr_object_size): Adjust.
9078 (pass_through_call): Adjust.
9079 (pass_object_sizes::execute): Adjust.
9080 * tree-object-size.h (fini_object_sizes): Declare.
9082 2017-01-10 Martin Sebor <msebor@redhat.com>
9084 PR tree-optimization/78775
9085 * builtins.c (get_size_range): Move...
9086 * calls.c: ...to here.
9087 (alloc_max_size): Accept zero argument.
9088 (operand_signed_p): Remove.
9089 (maybe_warn_alloc_args_overflow): Call get_size_range.
9090 * calls.h (get_size_range): Declare.
9092 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
9094 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
9095 from TI's devices.csv file as of September 2016.
9096 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
9098 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
9100 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
9101 * doc/invoke.texi: Likewise.
9102 * doc/md.texi: Likewise.
9103 * doc/objc.texi: Likewise.
9105 2017-01-10 Joshua Conner <joshconner@google.com>
9107 * config/arm/fuchsia-elf.h: New file.
9108 * config/fuchsia.h: New file.
9109 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
9110 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
9112 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
9114 2016-01-10 Richard Biener <rguenther@suse.de>
9116 PR tree-optimization/79034
9117 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
9118 Propagate out degenerate PHIs in the joiner.
9120 2017-01-10 Martin Liska <mliska@suse.cz>
9122 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
9123 (sort_congruence_classes_by_decl_uid): Likewise.
9124 (sort_congruence_class_groups_by_decl_uid): Likewise.
9125 (sem_item_optimizer::merge_classes): Sort class, groups in these
9126 classes and members in the groups by DECL_UID of declarations.
9127 This would make merge operations stable.
9129 2017-01-10 Martin Liska <mliska@suse.cz>
9131 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
9132 usage of m_classes_vec.
9133 (sem_item_optimizer::~sem_item_optimizer): Likewise.
9134 (sem_item_optimizer::get_group_by_hash): Likewise.
9135 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
9136 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
9137 (sem_item_optimizer::verify_classes): Likewise.
9138 (sem_item_optimizer::process_cong_reduction): Likewise.
9139 (sem_item_optimizer::dump_cong_classes): Likewise.
9140 (sem_item_optimizer::merge_classes): Likewise.
9141 * ipa-icf.h (congruence_class_hash): Rename from
9142 congruence_class_group_hash. Remove declaration of m_classes_vec.
9144 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
9146 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
9147 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
9148 * config.gcc: Add avx512vpopcntdqintrin.h.
9149 * config/i386/avx512vpopcntdqintrin.h: New.
9150 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
9151 * config/i386/i386-builtin-types.def: Add new types.
9152 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
9153 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
9154 __builtin_ia32_vpopcountq_v8di_mask): New.
9155 * config/i386/i386-c.c (ix86_target_macros_internal): Define
9156 __AVX512VPOPCNTDQ__.
9157 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
9158 (PTA_AVX512VPOPCNTDQ): Define.
9159 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
9160 TARGET_AVX512VPOPCNTDQ_P): Define.
9161 * config/i386/i386.opt: Add mavx512vpopcntdq.
9162 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
9163 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
9165 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
9168 * predict.def (PRED_CALL): Set to 67.
9170 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
9172 * expr.c (store_field): In the bitfield case, if the value comes from
9173 a function call and is of an aggregate type returned in registers, do
9174 not modify the field mode; extract the value in all cases if the mode
9175 is BLKmode and the size is not larger than a word.
9177 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
9180 * config/i386/cpuid.h: Fix undefined behavior.
9182 2017-01-04 Jeff Law <law@redhat.com>
9184 PR tree-optimization/79007
9185 PR tree-optimization/67955
9186 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
9187 conservative for pt.null when flag_non_call_exceptions is on.
9189 2017-01-09 Jakub Jelinek <jakub@redhat.com>
9191 PR translation/79019
9192 PR translation/79020
9193 * params.def (PARAM_INLINE_MIN_SPEEDUP,
9194 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
9195 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
9197 * config/avr/avr.opt (maccumulate-args): Likewise.
9198 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
9199 * common.opt (freport-bug): Likewise.
9200 * cif-code.def (CIF_FINAL_ERROR): Likewise.
9201 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
9202 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
9203 translatable string.
9204 * config/i386/i386.c (function_value_32): Likewise.
9205 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
9206 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
9208 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
9209 * common/config/msp430/msp430-common.c (msp430_handle_option):
9211 * symtab.c (symtab_node::verify_base): Likewise.
9212 * opts.c (set_debug_level): Likewise.
9213 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
9214 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
9215 missing whitespace to translatable strings.
9216 * config/avr/avr.md (bswapsi2): Fix typo in comment.
9217 * config/sh/superh.h: Likewise.
9218 * config/i386/xopintrin.h: Likewise.
9219 * config/i386/znver1.md: Likewise.
9220 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
9221 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
9222 * double-int.h (struct double_int): Likewise.
9223 * double-int.c (div_and_round_double): Likewise.
9224 * wide-int.cc: Likewise.
9225 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
9226 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
9227 * cfgcleanup.c (crossjumps_occured): Renamed to ...
9228 (crossjumps_occurred): ... this.
9229 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
9232 PR tree-optimization/78899
9233 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
9234 returning bool return struct loop *, NULL for failure and the new
9236 (versionable_outer_loop_p): Don't version outer loop if it has
9237 dont_vectorized bit set.
9238 (tree_if_conversion): When versioning outer loop, ensure
9239 tree_if_conversion is performed also on the inner loop of the
9240 non-vectorizable outer loop copy.
9241 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
9242 LOOP_VECTORIZED in inner loop of the scalar outer loop and
9243 prevent vectorization of it.
9244 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
9245 the outer loop vectorization of the non-scalar version is attempted
9246 before vectorization of the inner loop in scalar version. If
9247 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
9248 vectorization of its inner loop.
9249 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
9250 has 2 inner loops, rename also on edges from bb whose single pred
9251 is outer_loop->header. Fix typo in function comment.
9253 2017-01-09 Martin Sebor <msebor@redhat.com>
9256 * asan.c (asan_emit_stack_protection): Increase local buffer size
9257 to avoid snprintf truncation warning.
9259 2017-01-09 Andrew Pinski <apinski@cavium.com>
9261 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
9262 to reference thunderx2t99 for the tuning structure
9263 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
9265 (thunderx2t99_extra_costs): This.
9266 * config/aarch64/aarch64-tune.md: Regenerate.
9267 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
9268 (vulcan_addrcost_table): This.
9269 (vulcan_regmove_cost): Rename to ...
9270 (thunderx2t99_regmove_cost): This.
9271 (vulcan_vector_cost): Rename to ...
9272 (thunderx2t99_vector_cost): this.
9273 (vulcan_branch_cost): Rename to ...
9274 (thunderx2t99_branch_cost): This.
9275 (vulcan_tunings): Rename to ...
9276 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
9277 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
9279 2017-01-09 Martin Jambor <mjambor@suse.cz>
9283 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
9284 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
9285 (propagate_vr_accross_jump_function): Use the above function for all
9286 value range computations for pass-through jump functions and type
9287 converasion from explicit value range values.
9288 (ipcp_propagate_stage): Do not attempt to deduce types of formal
9289 parameters from TYPE_ARG_TYPES.
9290 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
9291 (ipa_write_node_info): Stream type of the actual argument.
9292 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
9294 2017-01-09 Martin Liska <mliska@suse.cz>
9297 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
9298 (lookup_compiler): Do not show error message with have_E.
9300 2017-01-09 Jakub Jelinek <jakub@redhat.com>
9302 PR tree-optimization/78938
9303 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
9304 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
9305 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
9306 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
9309 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
9311 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
9314 2017-01-09 Richard Biener <rguenther@suse.de>
9316 PR tree-optimization/78997
9317 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
9318 name condition properly.
9320 2017-01-09 Richard Biener <rguenther@suse.de>
9323 * dwarf2out.c (is_cxx): New overload with context.
9324 (is_naming_typedef_decl): Use it.
9326 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
9328 * invoke.texi (Option Summary): Correct spacing in option lists
9329 and add line breaks to fix over-long lines.
9331 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
9335 * extend.texi (Common Variable Attributes): Add xref to GCC
9336 Internals manual to explain mode attribute keywords.
9338 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
9341 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
9342 and Preprocessor Options.
9343 (Options for Linking): Document -pthread here....
9344 (RS/6000 and PowerPC Options): ...not here.
9345 (Solaris 2 Options): ...or here.
9346 * doc/cppopts.texi: Document -pthread.
9348 2017-01-08 Martin Sebor <msebor@redhat.com>
9351 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
9352 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
9353 New member functions.
9354 (format_directive): Used them.
9356 (pass_sprintf_length::handle_gimple_call): Same.
9357 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
9358 to avoid truncation for any argument.
9359 (extract_affine_mul): Same.
9360 * tree.c (get_file_function_name): Same.
9362 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
9365 * predict.def (PRED_INDIR_CALL): Set to 86.
9367 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
9369 PR preprocessor/54124
9370 * doc/cppopts.texi: Reformat -d subtable to list the full name
9371 of the options. Add cross-reference to the docs for the general
9372 compiler -d options.
9373 * doc/invoke.texi (Developer Options): Add cross-reference to the
9374 preprocessor-specific -d option documentation.
9376 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
9378 PR preprocessor/13498
9379 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
9380 redudant material, and reflect new command-line options.
9381 (System Headers): Likewise.
9383 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
9385 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
9386 -isystem, and -idirafter. Copy-edit.
9387 * doc/cppopts.texi: Copy-edit. Remove contradiction about
9388 default for -ftrack-macro-expansion. Delete obsolete and
9389 badly-formatted implementation details about -fdebug-cpp output.
9390 * doc/cppwarnopts.texi: Copy-edit.
9392 2017-01-07 David Malcolm <dmalcolm@redhat.com>
9395 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
9396 that the transition from a max line width >= 1<<10 to narrower
9397 lines works correctly.
9399 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
9401 * doc/options.texi (PerFunction): New.
9402 * opt-functions.awk (switch_flags): Map both Optimization and
9403 PerFunction to CL_OPTIMIZATION.
9404 * opth-gen.awk: Test for PerFunction flag along with
9406 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
9407 it only when the latter is present. Skip those that don't in
9408 the hash function generator.
9409 * common.opt (fvar-tracking): Mark as PerFunction instead of
9411 (fvar-tracking-assignments): Likewise.
9412 (fvar-tracking-assignments-toggle): Likewise.
9413 (fvar-tracking-uninit): Likewise.
9415 2017-01-07 Jakub Jelinek <jakub@redhat.com>
9417 PR translation/79018
9418 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
9421 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
9424 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
9425 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
9426 TARGET_LONG_JUMP_TABLE_OFFSETS.
9427 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
9428 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
9429 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
9430 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
9431 * config/m68k/m68k.md (tablejump expander): Likewise.
9432 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
9433 TARGET_LONG_JUMP_TABLE_OFFSETS.
9434 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
9435 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
9437 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
9438 David Holsgrove <david.holsgrove@xilinx.com>
9440 * common/config/microblaze/microblaze-common.c
9441 (TARGET_EXCEPT_UNWIND_INFO): Remove.
9442 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
9444 * config/microblaze/microblaze.c (microblaze_must_save_register)
9445 (microblaze_expand_epilogue, microblaze_return_addr): Handle
9447 (microblaze_eh_return): New function.
9448 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
9449 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
9450 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
9451 * config/microblaze/microblaze.md (eh_return): New pattern.
9453 2017-01-06 Jakub Jelinek <jakub@redhat.com>
9455 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
9456 GCC_DIAGNOSTIC_STRINGIFY): Define.
9458 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
9460 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
9462 * config/arm/arm.md (<mcrr>): New.
9464 * config/arm/arm.c (arm_arch5te): New.
9465 (arm_option_override): Set arm_arch5te.
9466 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
9468 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
9469 (arm_mcrr_qualifiers): ... this. New.
9470 (MRRC_QUALIFIERS): Define to...
9471 (arm_mrrc_qualifiers): ... this. New.
9472 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
9474 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
9475 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
9476 (MRRCI, mrrc, MRRC): New.
9477 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
9478 VUNSPEC_MRRC2): New.
9480 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
9482 * config/arm/arm.md (<mcr>): New.
9484 * config/arm/arm.c (arm_coproc_builtin_available): Add
9485 support for mcr, mrc, mcr2 and mrc2.
9486 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
9487 (arm_mcr_qualifiers): ... this. New.
9488 (MRC_QUALIFIERS): Define to ...
9489 (arm_mrc_qualifiers): ... this. New.
9490 (MCR_QUALIFIERS): Define to ...
9491 (arm_mcr_qualifiers): ... this. New.
9492 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
9494 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
9495 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
9496 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
9499 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
9501 * config/arm/arm.md (*ldc): New.
9505 * config/arm/arm.c (arm_coproc_builtin_available): Add
9506 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
9507 (arm_coproc_ldc_stc_legitimate_address): New.
9508 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
9509 'qualifier_const_pointer'.
9510 (LDC_QUALIFIERS): Define to...
9511 (arm_ldc_qualifiers): ... this. New.
9512 (STC_QUALIFIERS): Define to...
9513 (arm_stc_qualifiers): ... this. New.
9514 * config/arm/arm-protos.h
9515 (arm_coproc_ldc_stc_legitimate_address): New.
9516 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
9517 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
9518 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
9519 stc2, stcl, stc2l): New.
9520 * config/arm/constraints.md (Uz): New.
9521 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
9522 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
9523 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
9524 VUNSPEC_STC2L): New.
9526 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
9528 * config/arm/arm.md (<cdp>): New.
9529 * config/arm/arm.c (neon_const_bounds): Rename this ...
9530 (arm_const_bounds): ... this.
9531 (arm_coproc_builtin_available): New.
9532 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
9533 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
9534 (CDP_QUALIFIERS): Define to...
9535 (arm_cdp_qualifiers): ... this. New.
9537 (arm_expand_builtin_args): Add case for 6 arguments.
9538 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
9539 (arm_const_bounds): ... this.
9540 (arm_coproc_builtin_available): New.
9541 * config/arm/arm_acle.h (__arm_cdp): New.
9543 * config/arm/arm_acle_builtins.def (cdp): New.
9545 * config/arm/iterators.md (CDPI,CDP,cdp): New.
9546 * config/arm/neon.md: Rename all 'neon_const_bounds' to
9548 * config/arm/types.md (coproc): New.
9549 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
9550 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
9551 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
9552 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
9554 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
9556 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
9557 (UBINOP_QUALIFIERS): New.
9559 (acle_builtin_data): New. Change comment.
9560 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
9561 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
9562 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
9563 arm_acle_builtins.def.
9564 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
9565 (arm_init_acle_builtins): New.
9566 (CRC32_BUILTIN): Remove.
9567 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
9568 crc32cb, crc32ch and crc32cw.
9569 (arm_init_crc32_builtins): Remove.
9570 (arm_init_builtins): Use arm_init_acle_builtins rather
9571 than arm_init_crc32_builtins.
9572 (arm_expand_acle_builtin): New.
9573 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
9574 * config/arm/arm_acle_builtins.def: New.
9576 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
9578 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
9579 (arm_builtin_datum): ... this.
9580 (arm_init_neon_builtin): Rename to ...
9581 (arm_init_builtin): ... this. Add a new parameters PREFIX
9582 and USE_SIG_IN_NAME.
9583 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
9584 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
9585 'arm_builtin_datum'.
9586 (arm_init_vfp_builtins): Likewise.
9587 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
9588 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
9589 (arm_expand_neon_args): Rename to ...
9590 (arm_expand_builtin_args): ... this. Rename builtin_arg
9591 enum values and differentiate between ARG_BUILTIN_MEMORY
9592 and ARG_BUILTIN_NEON_MEMORY.
9593 (arm_expand_neon_builtin_1): Rename to ...
9594 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
9595 values, arm_expand_builtin_args and add bool parameter NEON.
9596 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
9597 (arm_expand_vfp_builtin): Likewise.
9598 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
9600 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
9603 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
9604 * predict.c (tree_estimate_probability_bb): Reverse direction of
9605 polymorphic call predictor.
9607 2017-01-06 David Malcolm <dmalcolm@redhat.com>
9609 * passes.c (execute_one_pass): Split out pass-skipping logic into...
9610 (determine_pass_name_match): ...this new function and...
9611 (should_skip_pass_p): ...this new function.
9613 2017-01-06 Nathan Sidwell <nathan@acm.org>
9615 * ipa-visibility.c (function_and_variable_visibility): Reformat
9616 comments and long lines. Remove extrneous if.
9617 * symtab.c (symtab_node::make_decl_local): Fix code format.
9618 (symtab_node::set_section_for_node): Fix comment typo.
9620 2017-01-06 Martin Liska <mliska@suse.cz>
9623 * lra-constraints.c: Rename invariant to lra_invariant.
9624 * predict.c (set_even_probabilities): Initialize e to NULL.
9626 2017-01-05 Martin Sebor <msebor@redhat.com>
9628 PR tree-optimization/78910
9629 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
9630 (format_integer): Correct off-by-one error in the handling
9631 of precision with negative numbers in signed conversions..
9633 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
9635 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
9637 2017-01-05 Jakub Jelinek <jakub@redhat.com>
9639 PR tree-optimization/71016
9640 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
9641 factor_out_conditional_conversion. Formatting fix.
9642 (factor_out_conditional_conversion): Add cond_stmt argument.
9643 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
9644 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
9647 2017-01-05 David Malcolm <dmalcolm@redhat.com>
9649 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
9650 read-rtl-function.o, and selftest-rtl.o.
9651 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
9652 (selftest::aarch64_test_loading_full_dump): New function.
9653 (selftest::aarch64_run_selftests): New function.
9654 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
9655 selftest::aarch64_run_selftests.
9656 * config/i386/i386.c
9657 (selftest::ix86_test_loading_dump_fragment_1): New function.
9658 (selftest::ix86_test_loading_call_insn): New function.
9659 (selftest::ix86_test_loading_full_dump): New function.
9660 (selftest::ix86_test_loading_unspec): New function.
9661 (selftest::ix86_run_selftests): Call the new functions.
9662 * emit-rtl.c (maybe_set_max_label_num): New function.
9663 * emit-rtl.h (maybe_set_max_label_num): New decl.
9664 * function.c (instantiate_decls): Guard call to
9665 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
9666 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
9668 * gensupport.c (gen_reader::gen_reader): Pass "false"
9669 for new "compact" param of rtx_reader.
9670 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
9671 rather than an empty string for NULL strings.
9672 * read-md.c: Potentially include config.h rather than bconfig.h.
9673 Wrap include of errors.h with #ifdef GENERATOR_FILE.
9674 (have_error): New global, copied from errors.c.
9675 (md_reader::read_name): Rename to...
9676 (md_reader::read_name_1): ...this, adding "out_loc" param,
9677 and converting "missing name or number" to returning false, rather
9679 (md_reader::read_name): Reimplement in terms of read_name_1.
9680 (md_reader::read_name_or_nil): New function.
9681 (md_reader::read_string): Handle "(nil)" by returning NULL.
9682 (md_reader::md_reader): Add new param "compact".
9683 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
9684 (md_reader::read_file): New method.
9685 * read-md.h (md_reader::md_reader): Add new param "compact".
9686 (md_reader::read_file): New method.
9687 (md_reader::is_compact): New accessor.
9688 (md_reader::read_name): Convert return type from void to file_location.
9689 (md_reader::read_name_or_nil): New decl.
9690 (md_reader::read_name_1): New decl.
9691 (md_reader::m_compact): New field.
9692 (noop_reader::noop_reader): Pass "false" for new "compact" param
9694 (rtx_reader::rtx_reader): Add new "compact" param.
9695 (rtx_reader::read_rtx_operand): Make virtual and convert return
9696 type from void to rtx.
9697 (rtx_reader::read_until): New decl.
9698 (rtx_reader::handle_any_trailing_information): New virtual function.
9699 (rtx_reader::postprocess): New virtual function.
9700 (rtx_reader::finalize_string): New virtual function.
9701 (rtx_reader::m_in_call_function_usage): New field.
9702 (rtx_reader::m_reuse_rtx_by_id): New field.
9703 * read-rtl-function.c: New file.
9704 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
9705 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
9706 (selftest::verify_three_block_rtl_cfg): New decl.
9707 * read-rtl-function.h: New file.
9708 * read-rtl.c: Potentially include config.h rather than bconfig.h.
9709 For host, include function.h, memmodel.h, and emit-rtl.h.
9710 (one_time_initialization): New function.
9711 (struct compact_insn_name): New struct.
9712 (compact_insn_names): New array.
9713 (find_code): Handle insn codes in compact dumps.
9714 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
9715 (bind_subst_iter_and_attr): Likewise.
9716 (add_condition_to_string): Likewise.
9717 (add_condition_to_rtx): Likewise.
9718 (apply_attribute_uses): Likewise.
9719 (add_current_iterators): Likewise.
9720 (apply_iterators): Likewise.
9721 (initialize_iterators): Guard usage of apply_subst_iterator with
9722 #ifdef GENERATOR_FILE.
9723 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
9724 (md_reader::read_mapping): Likewise.
9725 (add_define_attr_for_define_subst): Likewise.
9726 (add_define_subst_attr): Likewise.
9727 (read_subst_mapping): Likewise.
9728 (check_code_iterator): Likewise.
9729 (rtx_reader::read_rtx): Likewise. Move one-time initialization
9731 (one_time_initialization): New function.
9732 (rtx_reader::read_until): New method.
9733 (read_flags): New function.
9734 (parse_reg_note_name): New function.
9735 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
9736 Handle reuse_rtx ids.
9737 Wrap iterator lookup within #ifdef GENERATOR_FILE.
9738 Add parsing support for RTL dumps, mirroring the special-cases in
9739 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
9740 values, and calling handle_any_trailing_information.
9741 (rtx_reader::read_rtx_operand): Convert return type from void
9742 to rtx, returning return_rtx. Handle case 'e'. Call
9743 finalize_string on XSTR and XTMPL fields.
9744 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
9745 "(nil)" values were omitted. Call the postprocess vfunc on the
9747 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
9748 class ctor. Initialize m_in_call_function_usage. Call
9749 one_time_initialization.
9750 * rtl-tests.c (selftest::test_uncond_jump): Call
9751 set_new_first_and_last_insn.
9752 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
9753 * selftest-rtl.c: New file.
9754 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
9755 (selftest::get_insn_by_uid): New decl.
9756 * selftest-run-tests.c (selftest::run_tests): Call
9757 read_rtl_function_c_tests.
9758 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
9759 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
9762 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
9764 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
9765 operands in a special way. Assert that pos+len <= mode precision.
9767 2017-01-05 Jakub Jelinek <jakub@redhat.com>
9769 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
9770 3 argument Alias with unlimited for the negative form.
9771 (fno-vect-cost-model): Removed.
9773 2017-01-05 Martin Liska <mliska@suse.cz>
9775 * hsa-gen.c (gen_hsa_divmod): New function.
9776 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
9778 2017-01-05 Martin Liska <mliska@suse.cz>
9781 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
9784 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9786 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
9787 small constant length operands.
9789 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9791 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
9792 between loop iterations.
9794 2017-01-05 Martin Liska <mliska@suse.cz>
9797 * gimplify.c (gimplify_decl_expr): Compare to
9798 asan_poisoned_variables instread of checking flags.
9799 (gimplify_target_expr): Likewise.
9800 (gimplify_expr): Likewise.
9801 (gimplify_function_tree): Conditionally initialize
9802 asan_poisoned_variables.
9804 2017-01-04 Jeff Law <law@redhat.com>
9806 PR tree-optimizatin/78812
9807 * rtl.h (contains_mem_rtx_p): Prototype.
9808 * ifcvt.c (containts_mem_rtx_p): Move from here to...
9809 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
9810 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
9811 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
9812 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
9814 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
9816 * input.c (assert_char_at_range): Default-initialize actual_range.
9818 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
9820 * df-scan.c (df_ref_create_structure): Make regno unsigned,
9821 to match the caller.
9823 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
9825 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
9826 insns after final jump in test to emit dummy move.
9828 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
9830 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
9831 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
9833 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
9835 * multiple_target.c (create_dispatcher_calls): Init e_next.
9836 * tree-ssa-loop-split.c (split_loop): Init border.
9837 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
9840 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
9845 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
9846 (altivec_register_operand): Do not return true if the operand
9847 contains a SUBREG mixing SImode and SFmode.
9848 (vsx_register_operand): Likewise.
9849 (vsx_reg_sfsubreg_ok): New predicate.
9850 (vfloat_operand): Do not return true if the operand contains a
9851 SUBREG mixing SImode and SFmode.
9852 (vint_operand): Likewise.
9853 (vlogical_operand): Likewise.
9854 (gpc_reg_operand): Likewise.
9855 (int_reg_operand): Likewise.
9856 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
9857 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
9858 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
9860 (rs6000_emit_move_si_sf_subreg): New helper function.
9861 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
9862 fixup SUBREGs involving SImode and SFmode.
9863 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
9864 numbers for the new peephole2 optimization.
9865 (peephole2 for SFmode unions): New peephole2 to optimize cases in
9866 the GLIBC math library that do AND/IOR/XOR operations on single
9867 precision floating point.
9868 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
9869 target macros to say whether we need to avoid SUBREGs mixing
9871 (TARGET_ALLOW_SF_SUBREG): Likewise.
9872 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
9873 (UNSPEC_SI_FROM_SF): Likewise.
9874 (iorxor): Change spacing.
9875 (and_ior_xor): New iterator for AND, IOR, and XOR.
9876 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
9877 (movdi_from_sf_zero_ext): Likewise.
9878 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
9879 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
9880 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
9881 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
9882 (fms<mode>4): Likewise.
9883 (fnma<mode>4): Likewise.
9884 (fnms<mode>4): Likewise.
9885 (nfma<mode>4): Likewise.
9886 (nfms<mode>4): Likewise.
9888 2017-01-04 Marek Polacek <polacek@redhat.com>
9891 * doc/invoke.texi: Document -Wpointer-compare.
9893 2017-01-04 Jakub Jelinek <jakub@redhat.com>
9895 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
9898 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
9899 descriptions for -gdwarf-5 and emit them as uleb128 instead of
9902 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
9905 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
9906 documentation of the powerpc_popcntb_ok attribute.
9907 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
9908 code to issue warning messages if a requested CPU configuration is
9909 not supported by the binary (assembler and loader) toolchain.
9910 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
9911 made to define a built-in function that has been disabled.
9912 (paired_init_builtins): Add assertion to prevent ICE if attempt is
9913 made to define a built-in function that has been disabled.
9914 (altivec_init_builtins): Add comment explaining why definition
9915 of the DST built-in functions is not preceded by an assertion
9916 check. Add assertions to prevent ICE if attempts are made to
9917 define an altivec predicate or an abs* built-in function that has
9919 (htm_init_builtins): Add comment explaining why definition of the
9920 htm built-in functions is not preceded by an assertion check.
9922 2017-01-04 Jeff Law <law@redhat.com>
9924 PR tree-optimizatin/67955
9925 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
9926 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
9927 the points-to solution does not include pt_null. Use DECL_PT_UID
9930 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
9932 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
9933 Use gen_int_mode instead of gen_lopwart for const_int operands.
9935 2017-01-04 Jakub Jelinek <jakub@redhat.com>
9937 PR tree-optimization/71563
9938 * match.pd: Simplify X << Y into X if Y is known to be 0 or
9939 out of range value - has low bits known to be zero.
9941 2017-01-04 Alan Modra <amodra@gmail.com>
9943 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
9944 * configure: Regenerate.
9945 * config.in: Regenerate.
9947 2017-01-04 Jakub Jelinek <jakub@redhat.com>
9950 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
9951 a substring of the message, but strcmp with the whole message. Ifdef
9952 ENABLE_NLS, translate the message first using dgettext.
9954 2017-01-03 Jeff Law <law@redhat.com>
9956 PR tree-optimizatin/78856
9957 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
9958 (mark_threaded_blocks): Remove code to truncate thread paths that
9959 cross multiple loop headers. Instead invalidate the cached loop
9960 iteration information and handle case of a thread path walking
9961 into an irreducible region.
9963 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
9966 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
9967 assertions. Add support for doing the signbit if the IEEE 128-bit
9968 floating point value is in a GPR.
9969 * config/rs6000/rs6000.md (Fsignbit): Delete.
9970 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
9971 Update the length attribute if the value is in a GPR.
9972 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
9973 the sign or zero extension instruction, since the value is always 0/1.
9974 (signbit<mode>2_dm2): Delete using <Fsignbit>.
9977 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
9978 extracting SImode to a GPR register so that we can generate a
9979 store, limit the vector to be in a traditional Altivec register
9980 for the vextuwrx instruction.
9982 2017-01-03 Ian Lance Taylor <iant@google.com>
9984 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
9986 2017-01-03 Martin Sebor <msebor@redhat.com>
9988 PR tree-optimization/78696
9989 * gimple-ssa-sprintf.c (format_floating): Correct handling of
9990 precision. Use MPFR for %f for greater fidelity. Correct handling
9992 (pass_sprintf_length::compute_format_length): Set width and precision
9993 specified by asrerisk to void_node for vararg functions.
9994 (try_substitute_return_value): Adjust dump output.
9996 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
9998 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
10000 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
10002 * doc/invoke.texi (SPARC options): Document -mlra as the default.
10003 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
10004 -mlra/-mno-lra was passed to the compiler.
10006 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
10008 PR rtl-optimization/65618
10009 * emit-rtl.c (try_split): Move initialization of "before" and
10010 "after" to just before the call to emit_insn_after_setloc.
10012 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
10014 * doc/md.texi (Standard Names): Remove reference to Java frontend.
10016 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
10018 * dwarf2out.c (gen_enumeration_type_die): When
10019 -gno-strict-dwarf, add a DW_AT_encoding attribute.
10021 2017-01-03 Jakub Jelinek <jakub@redhat.com>
10023 PR tree-optimization/78965
10024 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
10025 Change first argument from const call_info & to call_info &. For %n
10026 set info.nowrite to false.
10028 PR middle-end/78901
10029 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
10030 possibly throwing calls.
10032 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
10033 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
10034 and fns handling, rather than in a separate case SSA_NAME.
10036 2017-01-02 Jeff Law <law@redhat.com>
10038 * config/darwin-driver.c (darwin_driver_init): Const-correctness
10039 fixes for first_period and second_period variables.
10041 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
10044 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
10045 (*insvqi_1): New insn pattern.
10046 (*insvqi_1_mem_rex64): Ditto.
10047 (*insvqi_2): Ditto.
10048 (*insvqi_3): Rename from *insvqi.
10050 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
10052 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
10054 * doc/cfg.texi (Edges): Remove reference to Java.
10055 (Maintaining the CFG): Ditto.
10057 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
10059 PR middle-end/77674
10060 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
10061 transparent aliases.
10063 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
10065 PR middle-end/77484
10066 * predict.def (PRED_CALL): Update hitrate.
10067 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
10068 * predict.c (tree_estimate_probability_bb): Split CALL predictor
10069 into direct/indirect/polymorphic variants.
10071 2017-01-01 Jakub Jelinek <jakub@redhat.com>
10073 Update copyright years.
10075 * gcc.c (process_command): Update copyright notice dates.
10076 * gcov-dump.c (print_version): Ditto.
10077 * gcov.c (print_version): Ditto.
10078 * gcov-tool.c (print_version): Ditto.
10079 * gengtype.c (create_file): Ditto.
10080 * doc/cpp.texi: Bump @copying's copyright year.
10081 * doc/cppinternals.texi: Ditto.
10082 * doc/gcc.texi: Ditto.
10083 * doc/gccint.texi: Ditto.
10084 * doc/gcov.texi: Ditto.
10085 * doc/install.texi: Ditto.
10086 * doc/invoke.texi: Ditto.
10088 Copyright (C) 2017 Free Software Foundation, Inc.
10090 Copying and distribution of this file, with or without modification,
10091 are permitted in any medium without royalty provided the copyright
10092 notice and this notice are preserved.