1 2017-07-11 Michael Collison <michael.collison@arm.com>
3 * config/aarch64/aarch64-simd.md (aarch64_sub<mode>_compare0):
6 2017-07-11 Carl Love <cel@us.ibm.com>
8 * config/rs6000/rs6000-c.c: Add support for builtins
9 vector unsigned int vec_parity_lsbb (vector signed int);
10 vector unsigned int vec_parity_lsbb (vector unsigned int);
11 vector unsigned __int128 vec_parity_lsbb (vector signed __int128);
12 vector unsigned __int128 vec_parity_lsbb (vector unsigned __int128);
13 vector unsigned long long vec_parity_lsbb (vector signed long long);
14 vector unsigned long long vec_parity_lsbb (vector unsigned long long);
15 * config/rs6000/rs6000-builtin.def (VPARITY_LSBB): Add BU_P9V_OVERLOAD1.
16 * config/rs6000/altivec.h (vec_parity_lsbb): Add define.
17 * doc/extend.texi: Update the built-in documentation file for the
18 new built-in functions.
20 2017-07-11 David Malcolm <dmalcolm@redhat.com>
22 * diagnostic-show-locus.c: Include "gcc-rich-location.h".
23 (layout::m_primary_loc): New field.
24 (layout::layout): Initialize new field. Move location filtering
26 (layout::maybe_add_location_range): ...this new method. Add
27 support for filtering to just the lines already specified by other
29 (layout::will_show_line_p): New method.
30 (gcc_rich_location::add_location_if_nearby): New method.
31 (selftest::test_add_location_if_nearby): New test function.
32 (selftest::diagnostic_show_locus_c_tests): Call it.
33 * gcc-rich-location.h (gcc_rich_location::add_location_if_nearby):
36 2017-07-11 Tom de Vries <tom@codesourcery.com>
38 * config/nvptx/nvptx.c (WORKAROUND_PTXJIT_BUG): New macro.
39 (bb_first_real_insn): New function.
40 (nvptx_single): Add extra initialization of broadcasted condition
43 2017-07-11 Nathan Sidwell <nathan@acm.org>
45 * dwarf2out.c (gen_member_die): Remove useless check for anon ctors.
47 2017-07-11 Georg-Johann Lay <avr@gjlay.de>
49 * doc/extend.texi (AVR Function Attributes): Remove weblink to
50 Binutils doc as TEXI will mess them up.
51 * doc/invoke.texi (AVR Options): Same here.
53 2017-07-11 Daniel Cederman <cederman@gaisler.com>
55 * config/sparc/sparc.opt (mfix-ut700): New option.
56 (mfix-gr712rc): Likewise.
57 (sparc_fix_b2bst): New variable.
58 * doc/invoke.texi (SPARC options): Document them.
59 (ARM options): Fix warnings.
60 * config/sparc/sparc.c (sparc_do_work_around_errata): Insert NOP
61 instructions to prevent sequences that can trigger the store-store
62 errata for certain LEON3FT processors.
63 (pass_work_around_errata::gate): Also test sparc_fix_b2bst.
64 (sparc_option_override): Set sparc_fix_b2bst appropriately.
65 * config/sparc/sparc.md (fix_b2bst): New attribute.
66 (in_branch_delay): Prevent stores in delay slot if fix_b2bst.
68 2017-07-10 Uros Bizjak <ubizjak@gmail.com>
71 * config/i386/i386.md (divsf3): Add TARGET_SSE to TARGET_SSE_MATH.
73 (*rsqrtsf2_sse): Ditto.
75 (div<mode>3): Macroize insn from divdf3 and divsf3
76 using MODEF mode iterator.
78 2017-07-10 Martin Sebor <msebor@redhat.com>
80 PR tree-optimization/80397
81 * gimple-ssa-sprintf.c (format_integer): Use INTEGRAL_TYPE_P()
82 instead of testing for equality to INTEGER_TYPE.
84 2017-07-10 Vineet Gupta <vgupta@synopsys.com>
86 * config.gcc: Remove uclibc from arc target spec.
88 2017-07-10 Claudiu Zissulescu <claziss@synopsys.com>
90 * config/arc/arc.h (ADDITIONAL_REGISTER_NAMES): Define.
92 2017-07-07 Jan Hubicka <hubicka@ucw.cz>
95 * lto-wrapper.c (remove_option): New function.
96 (merge_and_complain): Merge PIC/PIE options more realistically.
98 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
100 Better ISR prologues by supporting GASes __gcc_isr pseudo insn.
104 * configure.ac [target=avr]: Add GAS check for -mgcc-isr.
105 (HAVE_AS_AVR_MGCCISR_OPTION): If so, AC_DEFINE it.
106 * config.in: Regenerate.
107 * configure: Regenerate.
108 * doc/extend.texi (AVR Function Attributes) <no_gccisr>: Document it.
109 * doc/invoke.texi (AVR Options) <-mgas-isr-prologues>: Document it.
110 * config/avr/avr.opt (-mgas-isr-prologues): New option and...
111 (TARGET_GASISR_PROLOGUES): ...target mask.
112 * common/config/avr/avr-common.c
113 (avr_option_optimization_table) [OPT_LEVELS_1_PLUS_NOT_DEBUG]:
114 Set -mgas-isr-prologues.
115 * config/avr/avr-passes.def (avr_pass_pre_proep): Add
116 INSERT_PASS_BEFORE for it.
117 * config/avr/avr-protos.h (make_avr_pass_pre_proep): New proto.
118 * config/avr/avr.c (avr_option_override)
119 [!HAVE_AS_AVR_MGCCISR_OPTION]: Unset TARGET_GASISR_PROLOGUES.
120 (avr_no_gccisr_function_p, avr_hregs_split_reg): New static functions.
121 (avr_attribute_table) <no_gccisr>: Add new function attribute.
122 (avr_set_current_function) <is_no_gccisr>: Init machine field.
123 (avr_pass_data_pre_proep, avr_pass_pre_proep): New pass data
125 (make_avr_pass_pre_proep): New function.
126 (emit_push_sfr) <treg>: Add argument to function and use it
128 (avr_expand_prologue) [machine->gasisr.maybe]: Emit gasisr insn
129 and set machine->gasisr.yes.
130 (avr_expand_epilogue) [machine->gasisr.yes]: Similar.
131 (avr_asm_function_end_prologue) [machine->gasisr.yes]: Add
132 __gcc_isr.n_pushed to .L__stack_usage.
133 (TARGET_ASM_FINAL_POSTSCAN_INSN): Define to...
134 (avr_asm_final_postscan_insn): ...this new static function.
135 * config/avr/avr.h (machine_function)
136 <is_no_gccisr, use_L__stack_usage>: New fields.
137 <gasisr, gasisr.yes, gasisr.maybe, gasisr.regno>: New fields.
138 * config/avr/avr.md (UNSPECV_GASISR): Add unspecv enum.
139 (GASISR_Prologue, GASISR_Epilogue, GASISR_Done): New define_constants.
140 (gasisr, *gasisr): New expander and insn.
141 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
142 [HAVE_AS_AVR_MGCCISR_OPTION]: Print asm_gccisr spec.
143 * config/avr/specs.h (ASM_SPEC) <asm_gccisr>: Add sub spec.
145 2017-07-10 Richard Earnshaw <rearnsha@arm.com>
147 * config/arm/parsecpu.awk (gen_comm_data): Do not escape single quotes
150 2017-07-10 Georg-Johann Lay <avr@gjlay.de>
152 Move jump-tables out of .text again.
155 * config/avr/avr.c (ASM_OUTPUT_ADDR_VEC_ELT): Remove function.
156 (ASM_OUTPUT_ADDR_VEC): New function.
157 (avr_adjust_insn_length) [JUMP_TABLE_DATA_P]: Return 0.
158 (avr_final_prescan_insn) [avr_log.insn_addresses]: Dump
159 INSN_ADDRESSes as asm comment.
160 * config/avr/avr.h (JUMP_TABLES_IN_TEXT_SECTION): Adjust comment.
161 (ASM_OUTPUT_ADDR_VEC_ELT): Remove define.
162 (ASM_OUTPUT_ADDR_VEC): Define to avr_output_addr_vec.
163 * config/avr/avr.md (*tablejump): Adjust comment.
164 * config/avr/elf.h (ASM_OUTPUT_BEFORE_CASE_LABEL): Remove.
165 * config/avr/avr-log.c (avr_log_set_avr_log) <insn_addresses>:
167 * config/avr/avr-protos.h (avr_output_addr_vec_elt): Remove proto.
168 (avr_output_addr_vec): New proto.
169 (avr_log_t) <insn_addresses>: New field.
171 2017-07-09 H.J. Lu <hongjiu.lu@intel.com>
174 * config/i386/i386.c (ix86_function_arg_advance): Set
175 outgoing_args_on_stack to true if there are outgoing arguments
177 (ix86_function_arg): Likewise.
178 (ix86_get_drap_rtx): Use DRAP only if there are outgoing
179 arguments on stack and ACCUMULATE_OUTGOING_ARGS is false.
180 * config/i386/i386.h (machine_function): Add
181 outgoing_args_on_stack.
183 2017-07-09 Krister Walfridsson <krister.walfridsson@gmail.com>
185 * config.gcc (*-*-netbsd*): Remove check for NetBSD versions not
187 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always enable pthreads.
189 2017-07-08 Richard Sandiford <richard.sandiford@linaro.org>
191 * Makefile.in (HOOKS_H, RTL_BASE_H, FUNCTION_H, EXPR_H, REGS_H)
192 (REAL_H): Remove $(MACHMODE_H).
193 (FIXED_VALUE_H, TREE_CORE_H, CFGLOOP_H): Remove $(MACHMODE_H) and
195 (CORETYPES_H): Add signop.h, wide-int.h, wide-int-print.h,
196 $(MACHMODE_H) and double-int.h.
197 (build/min-insn-modes.o): Depend on $(CORETYPES_H) rather than
199 (gengtype-state.o, gengtype.o, build/gengtype.o): Don't depend on
202 2017-07-07 Andrew Pinski <apinski@cavium.com>
204 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Check
205 prev_set and curr_set for AARCH64_FUSE_ALU_BRANCH.
207 2017-07-07 Michael Meissner <meissner@linux.vnet.ibm.com>
209 * config/rs6000/rs6000.c (rs6000_get_function_versions_dispatcher):
210 Add warning if GCC was not configured to link against a GLIBC that
211 exports the hardware capability bits.
212 (make_resolver_func): Make resolver function private and not a
213 COMDAT function. Create the name with clone_function_name instead
217 * config/rs6000/rs6000.md (HI sign_extend splitter): Use the
218 correct operand in doing the split.
220 2017-07-07 Carl Love <cel@us.ibm.com>
222 * config/rs6000/rs6000-c: Add support for built-in function
223 vector unsigned short vec_pack_to_short_fp32 (vector float,
225 * config/rs6000/rs6000-builtin.def (CONVERT_4F32_8I16): Add
226 BU_P9V_AV_2 and BU_P9V_OVERLOAD_2 definitions.
227 * config/rs6000/altivec.h (vec_pack_to_short_fp32): Add define.
228 * config/rs6000/altivec.md(UNSPEC_CONVERT_4F32_8I16): Add UNSPEC.
229 (convert_4f32_8i16): Add define_expand.
230 * doc/extend.texi: Update the built-in documentation file for the
231 new built-in function.
233 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
235 * config/sparc/m8.md: New file.
236 * config/sparc/sparc.md: Include m8.md.
238 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
240 * config/sparc/sparc.opt: New option -mvis4b.
241 * config/sparc/sparc.c (dump_target_flag_bits): Handle MASK_VIS4B.
242 (sparc_option_override): Handle VIS4B.
243 (enum sparc_builtins): Define
244 SPARC_BUILTIN_DICTUNPACK{8,16,32},
245 SPARC_BUILTIN_FPCMP{LE,GT,EQ,NE}{8,16,32}SHL,
246 SPARC_BUILTIN_FPCMPU{LE,GT}{8,16,32}SHL,
247 SPARC_BUILTIN_FPCMPDE{8,16,32}SHL and
248 SPARC_BUILTIN_FPCMPUR{8,16,32}SHL.
249 (check_constant_argument): New function.
250 (sparc_vis_init_builtins): Define builtins
251 __builtin_vis_dictunpack{8,16,32},
252 __builtin_vis_fpcmp{le,gt,eq,ne}{8,16,32}shl,
253 __builtin_vis_fpcmpu{le,gt}{8,16,32}shl,
254 __builtin_vis_fpcmpde{8,16,32}shl and
255 __builtin_vis_fpcmpur{8,16,32}shl.
256 (sparc_expand_builtin): Check that the constant operands to
257 __builtin_vis_fpcmp*shl and _builtin_vis_dictunpack* are indeed
258 constant and in range.
259 * config/sparc/sparc-c.c (sparc_target_macros): Handle
261 * config/sparc/sparc.h (SPARC_IMM2_P): Define.
262 (SPARC_IMM5_P): Likewise.
263 * config/sparc/sparc.md (cpu_feature): Add new feagure "vis4b".
264 (enabled): Handle vis4b.
265 (UNSPEC_DICTUNPACK): New unspec.
266 (UNSPEC_FPCMPSHL): Likewise.
267 (UNSPEC_FPUCMPSHL): Likewise.
268 (UNSPEC_FPCMPDESHL): Likewise.
269 (UNSPEC_FPCMPURSHL): Likewise.
270 (cpu_feature): New CPU feature `vis4b'.
271 (dictunpack{8,16,32}): New insns.
272 (FPCSMODE): New mode iterator.
273 (fpcscond): New code iterator.
274 (fpcsucond): Likewise.
275 (fpcmp{le,gt,eq,ne}{8,16,32}{si,di}shl): New insns.
276 (fpcmpu{le,gt}{8,16,32}{si,di}shl): Likewise.
277 (fpcmpde{8,16,32}{si,di}shl): Likewise.
278 (fpcmpur{8,16,32}{si,di}shl): Likewise.
279 * config/sparc/constraints.md: Define constraints `q' for unsigned
280 2-bit integer constants and `t' for unsigned 5-bit integer
282 * config/sparc/predicates.md (imm5_operand_dictunpack8): New
284 (imm5_operand_dictunpack16): Likewise.
285 (imm5_operand_dictunpack32): Likewise.
286 (imm2_operand): Likewise.
287 * doc/invoke.texi (SPARC Options): Document -mvis4b.
288 * doc/extend.texi (SPARC VIS Built-in Functions): Document the
289 ditunpack* and fpcmp*shl builtins.
291 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
293 * config.gcc: Handle m8 in --with-{cpu,tune} options.
294 * config.in: Add HAVE_AS_SPARC6 define.
295 * config/sparc/driver-sparc.c (cpu_names): Add entry for the SPARC
297 * config/sparc/sol2.h (CPP_CPU64_DEFAULT_SPEC): Define for
299 (ASM_CPU32_DEFAUILT_SPEC): Likewise.
300 (CPP_CPU_SPEC): Handle m8.
301 (ASM_CPU_SPEC): Likewise.
302 * config/sparc/sparc-opts.h (enum processor_type): Add
304 * config/sparc/sparc.c (m8_costs): New struct.
305 (sparc_option_override): Handle TARGET_CPU_m8.
306 (sparc32_initialize_trampoline): Likewise.
307 (sparc64_initialize_trampoline): Likewise.
308 (sparc_issue_rate): Likewise.
309 (sparc_register_move_cost): Likewise.
310 * config/sparc/sparc.h (TARGET_CPU_m8): Define.
311 (CPP_CPU64_DEFAULT_SPEC): Define for M8.
312 (ASM_CPU64_DEFAULT_SPEC): Likewise.
313 (CPP_CPU_SPEC): Handle M8.
314 (ASM_CPU_SPEC): Likewise.
315 (AS_M8_FLAG): Define.
316 * config/sparc/sparc.md: Add m8 to the cpu attribute.
317 * config/sparc/sparc.opt: New option -mcpu=m8 for sparc targets.
318 * configure.ac (HAVE_AS_SPARC6): Check for assembler support for
320 * configure: Regenerate.
321 * doc/invoke.texi (SPARC Options): Document -mcpu=m8 and
324 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
326 * config/sparc/niagara7.md: Rework the DFA scheduler to use insn
328 * config/sparc/sparc.md: Remove the `v3pipe' insn attribute.
329 ("*movdi_insn_sp32"): Do not set v3pipe.
330 ("*movsi_insn"): Likewise.
331 ("*movdi_insn_sp64"): Likewise.
332 ("*movsf_insn"): Likewise.
333 ("*movdf_insn_sp32"): Likewise.
334 ("*movdf_insn_sp64"): Likewise.
335 ("*zero_extendsidi2_insn_sp64"): Likewise.
336 ("*sign_extendsidi2_insn"): Likewise.
337 ("*mov<VM32:mode>_insn"): Likewise.
338 ("*mov<VM64:mode>_insn_sp64"): Likewise.
339 ("*mov<VM64:mode>_insn_sp32"): Likewise.
340 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
341 ("<vlop:code><VL:mode>3"): Likewise.
342 ("*not_<vlop:code><VL:mode>3"): Likewise.
343 ("*nand<VL:mode>_vis"): Likewise.
344 ("*<vlnotop:code>_not1<VL:mode>_vis"): Likewise.
345 ("*<vlnotop:code>_not2<VL:mode>_vis"): Likewise.
346 ("one_cmpl<VL:mode>2"): Likewise.
347 ("faligndata<VM64:mode>_vis"): Likewise.
348 ("alignaddrsi_vis"): Likewise.
349 ("alignaddrdi_vis"): Likweise.
350 ("alignaddrlsi_vis"): Likewise.
351 ("alignaddrldi_vis"): Likewise.
352 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
353 ("bmaskdi_vis"): Likewise.
354 ("bmasksi_vis"): Likewise.
355 ("bshuffle<VM64:mode>_vis"): Likewise.
356 ("cmask8<P:mode>_vis"): Likewise.
357 ("cmask16<P:mode>_vis"): Likewise.
358 ("cmask32<P:mode>_vis"): Likewise.
359 ("pdistn<P:mode>_vis"): Likewise.
360 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
362 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
364 * config/sparc/sparc.md ("subtype"): New insn attribute.
365 ("*wrgsr_sp64"): Set insn subtype.
366 ("*rdgsr_sp64"): Likewise.
367 ("alignaddrsi_vis"): Likewise.
368 ("alignaddrdi_vis"): Likewise.
369 ("alignaddrlsi_vis"): Likewise.
370 ("alignaddrldi_vis"): Likewise.
371 ("<plusminus_insn><VADDSUB:mode>3"): Likewise.
372 ("fexpand_vis"): Likewise.
373 ("fpmerge_vis"): Likewise.
374 ("faligndata<VM64:mode>_vis"): Likewise.
375 ("bshuffle<VM64:mode>_vis"): Likewise.
376 ("cmask8<P:mode>_vis"): Likewise.
377 ("cmask16<P:mode>_vis"): Likewise.
378 ("cmask32<P:mode>_vis"): Likewise.
379 ("fchksm16_vis"): Likewise.
380 ("v<vis3_shift_patname><GCM:mode>3"): Likewise.
381 ("fmean16_vis"): Likewise.
382 ("fp<plusminus_insn>64_vis"): Likewise.
383 ("<plusminus_insn>v8qi3"): Likewise.
384 ("<vis3_addsub_ss_patname><VASS:mode>3"): Likewise.
385 ("<vis4_minmax_patname><VMMAX:mode>3"): Likewise.
386 ("<vis4_uminmax_patname><VMMAX:mode>3"): Likewise.
387 ("<vis3_addsub_ss_patname>v8qi3"): Likewise.
388 ("<vis4_addsub_us_patname><VAUS:mode>3"): Likewise.
389 ("*movqi_insn"): Likewise.
390 ("*movhi_insn"): Likewise.
391 ("*movsi_insn"): Likewise.
392 ("movsi_pic_gotdata_op"): Likewise.
393 ("*movdi_insn_sp32"): Likewise.
394 ("*movdi_insn_sp64"): Likewise.
395 ("movdi_pic_gotdata_op"): Likewise.
396 ("*movsf_insn"): Likewise.
397 ("*movdf_insn_sp32"): Likewise.
398 ("*movdf_insn_sp64"): Likewise.
399 ("*zero_extendhisi2_insn"): Likewise.
400 ("*zero_extendqihi2_insn"): Likewise.
401 ("*zero_extendqisi2_insn"): Likewise.
402 ("*zero_extendqidi2_insn"): Likewise.
403 ("*zero_extendhidi2_insn"): Likewise.
404 ("*zero_extendsidi2_insn_sp64"): Likewise.
406 ("prefetch_64"): Likewise.
407 ("prefetch_32"): Likewise.
408 ("tie_ld32"): Likewise.
409 ("tie_ld64"): Likewise.
410 ("*tldo_ldub_sp32"): Likewise.
411 ("*tldo_ldub1_sp32"): Likewise.
412 ("*tldo_ldub2_sp32"): Likewise.
413 ("*tldo_ldub_sp64"): Likewise.
414 ("*tldo_ldub1_sp64"): Likewise.
415 ("*tldo_ldub2_sp64"): Likewise.
416 ("*tldo_ldub3_sp64"): Likewise.
417 ("*tldo_lduh_sp32"): Likewise.
418 ("*tldo_lduh1_sp32"): Likewise.
419 ("*tldo_lduh_sp64"): Likewise.
420 ("*tldo_lduh1_sp64"): Likewise.
421 ("*tldo_lduh2_sp64"): Likewise.
422 ("*tldo_lduw_sp32"): Likewise.
423 ("*tldo_lduw_sp64"): Likewise.
424 ("*tldo_lduw1_sp64"): Likewise.
425 ("*tldo_ldx_sp64"): Likewise.
426 ("*mov<VM32:mode>_insn"): Likewise.
427 ("*mov<VM64:mode>_insn_sp64"): Likewise.
428 ("*mov<VM64:mode>_insn_sp32"): Likewise.
430 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
432 * config/sparc/sparc.md ("type"): New insn type viscmp.
433 ("fcmp<gcond:code><GCM:gcm_name><P:mode>_vis"): Set insn type to
435 ("fpcmp<gcond:code>8<P:mode>_vis"): Likewise.
436 ("fucmp<gcond:code>8<P:mode>_vis"): Likewise.
437 ("fpcmpu<gcond:code><GCM:gcm_name><P:mode>_vis"): Likewise.
438 * config/sparc/niagara7.md ("n7_vis_logical_v3pipe"): Handle
440 ("n7_vis_logical_11cycle"): Likewise.
441 * config/sparc/niagara4.md ("n4_vis_logical"): Likewise.
442 * config/sparc/niagara2.md ("niag3_vis": Likewise.
443 * config/sparc/niagara.md ("niag_vis"): Likewise.
444 * config/sparc/ultra3.md ("us3_fga"): Likewise.
445 * config/sparc/ultra1_2.md ("us1_fga_double"): Likewise.
447 2017-07-07 Jose E. Marchesi <jose.marchesi@oracle.com>
449 * config/sparc/sparc.md: New instruction type `bmask'.
450 (bmaskdi_vis): Use the `bmask' type.
451 (bmasksi_vis): Likewise.
452 * config/sparc/ultra3.md (us3_array): Likewise.
453 * config/sparc/niagara7.md (n7_array): Likewise.
454 * config/sparc/niagara4.md (n4_array): Likewise.
455 * config/sparc/niagara2.md (niag2_vis): Likewise.
456 (niag3_vis): Likewise.
457 * config/sparc/niagara.md (niag_vis): Likewise.
459 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
461 * ipa-comdats.c: Remove optimize check from gate.
462 * ipa-fnsummary.c (ipa_fn_summary_generate): do not generate summary
463 for functions not optimized.
464 (ipa_fn_summary_read): Skip optimize check.
465 (ipa_fn_summary_write): Likewise.
466 * ipa-inline-analysis.c (do_estimate_growth_1): Check that caller
468 * ipa-inline.c (can_inline_edge_p): Not optimized functions are
470 (can_inline_edge_p): Check flag_pcc_struct_return for match.
471 (check_callers): Give up on caller which is not optimized.
472 (inline_small_functions): Likewise.
473 (ipa_inline): Do not give up when not optimizing.
474 * ipa-visbility.c (function_and_variable_visibility): Do not optimize
475 away unoptimizes cdtors.
476 (whole_program_function_and_variable_visibility): Do
477 ipa_discover_readonly_nonaddressable_vars in LTO mode.
478 * ipa.c (process_references): Do not check optimize.
479 (symbol_table::remove_unreachable_nodes): Update optimize check.
480 (set_writeonly_bit): Update optimize check.
481 (pass_ipa_cdtor_merge::gate): Do not check optimize.
482 (pass_ipa_single_use::gate): Remove.
484 2017-07-06 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
486 * config/rs6000/rs6000.c (union_defs, union_uses, insn_is_load_p,
487 insn_is_store_p, insn_is_swap_p, const_load_sequence_p, v2df_reduction_p,
488 rtx_is_swappable_p, insn_is_swappable_p, chain_contains_only_swaps,
489 mark_swaps_for_removal, swap_const_vector_halves, adjust_subreg_index,
490 permute_load, permute_store, adjust_extract, adjust_splat,
491 adjust_xxpermdi, adjust_concat, adjust_vperm, handle_special_swappables,
492 replace_swap_with_copy, dump_swap_insn_table,
493 alignment_with_canonical_addr, alignment_mask, find_alignment_op,
494 recombine_lvx_pattern, recombine_stvx_pattern,
495 recombine_lvx_stvx_patterns, rs6000_analyze_swaps,
496 make_pass_analyze_swaps): Move all code related to p8 swap optimizations
497 to file rs6000-p8swap.c.
498 * config/rs6000/rs6000-p8swap.c: New file.
499 * config/rs6000/t-rs6000: Add rule to build rs6000-p8swap.o.
500 * config.gcc: Add rs6000-p8swap.o to extra_objs for powerpc*-*-*
501 and rs6000*-*-* targets.
503 2017-07-06 David Malcolm <dmalcolm@redhat.com>
505 * Makefile.in (selftest): Remove dependency on s-selftest-c++.
507 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
509 * lto-wrapper.c (merge_and_complain): Do not merge
510 fexceptions, fnon_call_exceptions, ftrapv, ffp_contract_, fmath_errno,
511 fsigned_zeros, ftrapping_math, fwrapv.
512 (append_compiler_options): Do not track these options.
513 (append_linker_options): Likewie
515 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
517 * cgraphunit.c (cgraph_node::finalize_function): When
518 !flag_toplevel_reorde set no_reorder flag.
519 (varpool_node::finalize_decl): Likewise.
520 (symbol_table::compile): Drop no toplevel reorder path.
522 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
524 * bb-reorder.c (better_edge_p): Do not build traces across abnormal/eh
525 edges; zero probability is not better than uninitialized.
527 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
529 * asan.h (asan_sanitize_allocas_p): Declare.
530 * asan.c (asan_sanitize_allocas_p): New function.
531 (handle_builtin_stack_restore): Bail out if !asan_sanitize_allocas_p.
532 (handle_builtin_alloca): Likewise.
533 * cfgexpand.c (expand_used_vars): Do not add allocas unpoisoning stuff
534 if !asan_sanitize_allocas_p.
535 * params.def (asan-instrument-allocas): Add new option.
536 * params.h (ASAN_PROTECT_ALLOCAS): Define.
537 * opts.c (common_handle_option): Disable allocas sanitization for
540 2017-07-06 Maxim Ostapenko <m.ostapenko@samsung.com>
542 * asan.c: Include gimple-fold.h.
543 (get_last_alloca_addr): New function.
544 (handle_builtin_stackrestore): Likewise.
545 (handle_builtin_alloca): Likewise.
546 (asan_emit_allocas_unpoison): Likewise.
547 (get_mem_refs_of_builtin_call): Add new parameter, remove const
548 quallifier from first paramerer. Handle BUILT_IN_ALLOCA,
549 BUILT_IN_ALLOCA_WITH_ALIGN and BUILT_IN_STACK_RESTORE builtins.
550 (instrument_builtin_call): Pass gimple iterator to
551 get_mem_refs_of_builtin_call.
552 (last_alloca_addr): New global.
553 * asan.h (asan_emit_allocas_unpoison): Declare.
554 * builtins.c (expand_asan_emit_allocas_unpoison): New function.
555 (expand_builtin): Handle BUILT_IN_ASAN_ALLOCAS_UNPOISON.
556 * cfgexpand.c (expand_used_vars): Call asan_emit_allocas_unpoison
557 if function calls alloca.
558 * gimple-fold.c (replace_call_with_value): Remove static keyword.
559 * gimple-fold.h (replace_call_with_value): Declare.
560 * internal-fn.c: Include asan.h.
561 * sanitizer.def (BUILT_IN_ASAN_ALLOCA_POISON,
562 BUILT_IN_ASAN_ALLOCAS_UNPOISON): New builtins.
564 2017-07-06 David Malcolm <dmalcolm@redhat.com>
566 * Makefile.in (SELFTEST_FLAGS): Drop "-x c", moving it to...
567 (C_SELFTEST_FLAGS): New.
568 (CPP_SELFTEST_FLAGS): New.
569 (SELFTEST_DEPS): New, from deps of s-selftest.
570 (C_SELFTEST_DEPS): New, from deps of s-selftest.
571 (CPP_SELFTEST_DEPS): New.
572 (selftest): Add dependency on s-selftest-c++.
573 (s-selftest): Rename to...
574 (s-selftest-c): ...this, moving deps to SELFTEST_DEPS
575 and C_SELFTEST_DEPS, and using C_SELFTEST_FLAGS rather
577 (selftest-gdb): Rename to...
578 (selftest-c-gdb): ...this, using C_SELFTEST_DEPS and
580 (selftest-gdb): Reintroduce as an alias for selftest-c-gdb.
581 (selftest-valgrind): Rename to...
582 (selftest-c-valgrind): ...this, using C_SELFTEST_DEPS and
584 (selftest-valgrind): Reintroduce as an alias for
586 (s-selftest-c++): New.
587 (selftest-c++-gdb): New.
588 (selftest-c++-valgrind): New.
590 2017-07-06 Olivier Hainque <hainque@adacore.com>
592 * gcc.c (process_command): When deciding if undefined variables
593 should be ignored when processing specs, accept "gcc -v" as well.
595 2017-07-06 Jan Hubicka <hubicka@ucw.cz>
597 * auto-profile.c (afdo_set_bb_count, afdo_propagate_edge,
598 afdo_annotate_cfg): Set counts/probabilities as determined by afdo.
600 2017-07-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
602 * config/arm/arm-cpus.in (armv8-r): Add new entry.
603 * config/arm/arm-isa.h (ISA_ARMv8r): Define macro.
604 * config/arm/arm-tables.opt: Regenerate.
605 * config/arm/arm.h (enum base_architecture): Add BASE_ARCH_8R
607 * doc/invoke.texi: Mention -march=armv8-r and its extensions.
609 2017-07-06 Carl Love <cel@us.ibm.com>
611 * ChangeLog: Clean up from mid air collision
613 2017-07-06 Carl Love <cel@us.ibm.com>
615 * config/rs6000/rs6000-c.c: Add support for built-in functions
616 vector signed int vec_subc (vector signed int, vector signed int);
617 vector signed __int128 vec_subc (vector signed __int128,
618 vector signed __int128);
619 vector unsigned __int128 vec_subc (vector unsigned __int128,
620 vector unsigned __int128);
621 vector signed int vec_sube (vector signed int, vector signed int,
623 vector unsigned int vec_sube (vector unsigned int,
625 vector unsigned int);
626 vector signed __int128 vec_sube (vector signed __int128,
627 vector signed __int128,
628 vector signed__int128);
629 vector unsigned __int128 vec_sube (vector unsigned __int128,
630 vector unsigned __int128,
631 vector unsigned __int128);
632 vector signed int vec_subec (vector signed int, vector signed int,
634 vector unsigned int vec_subec (vector unsigned int,
636 vector unsigned int);
637 vector signed __int128 vec_subec (vector signed __int128,
638 vector signed __int128,
639 vector signed__int128);
640 vector unsigned __int128 vec_subec (vector unsigned __int128,
641 vector unsigned __int128,
642 vector unsigned __int128);
643 * config/rs6000/rs6000.c (ALTIVEC_BUILTIN_VEC_SUBE,
644 ALTIVEC_BUILTIN_VEC_SUBEC): Add ef_builtins.
645 * config/rs6000/rs6000-builtin.def (SUBE, SUBEC): Add
646 BU_ALTIVEC_OVERLOAD_X definitions.
647 * config/rs6000/altivec.h (vec_sube, vec_subec): Add builtin defines.
648 * doc/extend.texi: Update the built-in documentation file for the new
651 2017-07-06 David Malcolm <dmalcolm@redhat.com>
654 * diagnostic-show-locus.c (layout::layout): Use start and finish
655 spelling location for the start and finish of each range.
656 * genmatch.c (linemap_client_expand_location_to_spelling_point):
657 Add unused aspect param.
658 * input.c (expand_location_1): Add "aspect" param, and use it
659 to access the correct part of the location.
660 (expand_location): Pass LOCATION_ASPECT_CARET to new param of
662 (expand_location_to_spelling_point): Likewise.
663 (linemap_client_expand_location_to_spelling_point): Add "aspect"
664 param, and pass it to expand_location_1.
666 2017-07-06 Sebastian Peryt <sebastian.peryt@intel.com>
668 * config/i386/avx512fintrin.h (_mm_mask_getexp_round_ss,
669 _mm_maskz_getexp_round_ss, _mm_mask_getexp_round_sd,
670 _mm_maskz_getexp_round_sd, _mm_mask_getmant_round_sd,
671 _mm_maskz_getmant_round_sd, _mm_mask_getmant_round_ss,
672 _mm_maskz_getmant_round_ss, _mm_mask_getexp_ss, _mm_maskz_getexp_ss,
673 _mm_mask_getexp_sd, _mm_maskz_getexp_sd, _mm_mask_getmant_sd,
674 _mm_maskz_getmant_sd, _mm_mask_getmant_ss,
675 _mm_maskz_getmant_ss): New intrinsics.
676 (__builtin_ia32_getexpss128_mask): Changed to ...
677 __builtin_ia32_getexpss128_round ... this.
678 (__builtin_ia32_getexpsd128_mask): Changed to ...
679 __builtin_ia32_getexpsd128_round ... this.
680 * config/i386/i386-builtin-types.def
681 ((V2DF, V2DF, V2DF, INT, V2DF, UQI, INT),
682 (V4SF, V4SF, V4SF, INT, V4SF, UQI, INT)): New function type aliases.
683 * config/i386/i386-builtin.def (__builtin_ia32_getexpsd_mask_round,
684 __builtin_ia32_getexpss_mask_round, __builtin_ia32_getmantsd_mask_round,
685 __builtin_ia32_getmantss_mask_round): New builtins.
686 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_INT_V2DF_UQI_INT,
687 V4SF_FTYPE_V4SF_V4SF_INT_V4SF_UQI_INT): Handle new types.
688 (CODE_FOR_avx512f_vgetmantv2df_mask_round,
689 CODE_FOR_avx512f_vgetmantv4sf_mask_round): New cases.
691 (avx512f_sgetexp<mode><round_saeonly_name>): Changed to ...
692 avx512f_sgetexp<mode><mask_scalar_name>
693 <round_saeonly_scalar_name> ... this.
694 (vgetexp<ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|
695 %0, %1, %2<round_saeonly_op3>}): Changed to ...
696 vgetexp<ssescalarmodesuffix>
697 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
698 %0<mask_scalar_operand3>, %1, %2<round_saeonly_scalar_mask_op3>} ... this.
699 (avx512f_vgetmant<mode><round_saeonly_name>): Changed to ...
700 avx512f_vgetmant<mode><mask_scalar_name>
701 <round_saeonly_scalar_name> ... this.
702 (vgetmant<ssescalarmodesuffix>\t{%3, <round_saeonly_op4>%2, %1, %0|
703 %0, %1, %2<round_saeonly_op4>, %3}): Changed to ...
704 vgetmant<ssescalarmodesuffix>
705 \t{%3, <round_saeonly_scalar_mask_op4>%2, %1, %0<mask_scalar_operand4>|
706 %0<mask_scalar_operand4>, %1, %2
707 <round_saeonly_scalar_mask_op4>, %3} ... this.
708 * config/i386/subst.md (mask_scalar_operand4,
709 round_saeonly_scalar_mask_operand4, round_saeonly_scalar_mask_op4,
710 round_saeonly_scalar_nimm_predicate): New subst attributes.
712 2017-07-06 Julia Koval <julia.koval@intel.com>
714 * gcc/config/i386/i386.c (ix86_erase_embedded_rounding):
715 Remove code for old rounding pattern.
717 2017-07-06 Richard Earnshaw <rearnsha@arm.com>
719 * config/arm/t-arm (GTM_H): Add arm-cpu.h.
721 2017-07-06 Christophe Lyon <christophe.lyon@linaro.org>
723 * doc/sourcebuild.texi (Test Directives, Variants of
724 dg-require-support): Add documentation for dg-require-stack-check.
726 2017-07-05 Sebastian Peryt <sebastian.peryt@intel.com>
728 * config/i386/subst.md (mask_scalar, round_scalar,
729 round_saeonly_scalar): New meta-templates.
730 (mask_scalar_name, mask_scalar_operand3, round_scalar_name,
731 round_scalar_mask_operand3, round_scalar_mask_op3,
732 round_scalar_constraint, round_scalar_prefix, round_saeonly_scalar_name,
733 round_saeonly_scalar_mask_operand3, round_saeonly_scalar_mask_op3,
734 round_saeonly_scalar_constraint,
735 round_saeonly_scalar_prefix): New subst attribute.
737 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): Renamed to ...
738 <sse>_vm<plusminus_insn><mode>3<mask_scalar_name>
739 <round_scalar_name> ... this.
740 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): Renamed to ...
741 <sse>_vm<multdiv_mnemonic><mode>3<mask_scalar_name>
742 <round_scalar_name> ... this.
743 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): Renamed to ...
744 <sse>_vm<code><mode>3<mask_scalar_name>
745 <round_saeonly_scalar_name> ... this.
746 (v<plusminus_mnemonic><ssescalarmodesuffix>
747 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
748 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
749 v<plusminus_mnemonic><ssescalarmodesuffix>
750 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
751 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
752 (v<multdiv_mnemonic><ssescalarmodesuffix>
753 \t{<round_mask_op3>%2, %1, %0<mask_operand3>|
754 %0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}): Changed to ...
755 v<multdiv_mnemonic><ssescalarmodesuffix>
756 \t{<round_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
757 %0<mask_scalar_operand3>, %1, %<iptr>2<round_scalar_mask_op3>} ... this.
758 (v<maxmin_float><ssescalarmodesuffix>
759 \t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|
760 %0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}): Changed to ...
761 v<maxmin_float><ssescalarmodesuffix>
762 \t{<round_saeonly_scalar_mask_op3>%2, %1, %0<mask_scalar_operand3>|
763 %0<mask_scalar_operand3>, %1, %<iptr>2
764 <round_saeonly_scalar_mask_op3>} ... this.
766 2017-07-05 Richard Earnshaw <rearnsha@arm.com>
768 * config/arm/arm.c (arm_fixed_condition_code_regs): New function.
769 (TARGET_FIXED_CONDITION_CODE_REGS): Redefine.
771 2017-07-05 Richard Sandiford <richard.sandiford@linaro.org>
772 Alan Hayward <alan.hayward@arm.com>
773 David Sherwood <david.sherwood@arm.com>
775 * combine.c (simplify_if_then_else): Remove "enum" before
777 * compare-elim.c (can_eliminate_compare): Likewise.
778 * config/aarch64/aarch64-builtins.c (aarch64_simd_builtin_std_type):
780 (aarch64_lookup_simd_builtin_type): Likewise.
781 (aarch64_simd_builtin_type): Likewise.
782 (aarch64_init_simd_builtin_types): Likewise.
783 (aarch64_simd_expand_args): Likewise.
784 * config/aarch64/aarch64-protos.h (aarch64_simd_attr_length_rglist):
786 (aarch64_reverse_mask): Likewise.
787 (aarch64_simd_emit_reg_reg_move): Likewise.
788 (aarch64_gen_adjusted_ldpstp): Likewise.
789 (aarch64_ccmp_mode_to_code): Likewise.
790 (aarch64_operands_ok_for_ldpstp): Likewise.
791 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
792 * config/aarch64/aarch64.c (aarch64_ira_change_pseudo_allocno_class):
794 (aarch64_min_divisions_for_recip_mul): Likewise.
795 (aarch64_reassociation_width): Likewise.
796 (aarch64_get_condition_code_1): Likewise.
797 (aarch64_simd_emit_reg_reg_move): Likewise.
798 (aarch64_simd_attr_length_rglist): Likewise.
799 (aarch64_reverse_mask): Likewise.
800 (aarch64_operands_ok_for_ldpstp): Likewise.
801 (aarch64_operands_adjust_ok_for_ldpstp): Likewise.
802 (aarch64_gen_adjusted_ldpstp): Likewise.
803 * config/aarch64/cortex-a57-fma-steering.c (fma_node::rename):
805 * config/arc/arc.c (legitimate_offset_address_p): Likewise.
806 * config/arm/arm-builtins.c (arm_simd_builtin_std_type): Likewise.
807 (arm_lookup_simd_builtin_type): Likewise.
808 (arm_simd_builtin_type): Likewise.
809 (arm_init_simd_builtin_types): Likewise.
810 (arm_expand_builtin_args): Likewise.
811 * config/arm/arm-protos.h (arm_expand_builtin): Likewise.
812 * config/ft32/ft32.c (ft32_libcall_value): Likewise.
813 (ft32_setup_incoming_varargs): Likewise.
814 (ft32_function_arg): Likewise.
815 (ft32_function_arg_advance): Likewise.
816 (ft32_pass_by_reference): Likewise.
817 (ft32_arg_partial_bytes): Likewise.
818 (ft32_valid_pointer_mode): Likewise.
819 (ft32_addr_space_pointer_mode): Likewise.
820 (ft32_addr_space_legitimate_address_p): Likewise.
821 * config/i386/i386-protos.h (ix86_operands_ok_for_move_multiple):
823 * config/i386/i386.c (ix86_setup_incoming_vararg_bounds): Likewise.
824 (ix86_emit_outlined_ms2sysv_restore): Likewise.
825 (iamcu_alignment): Likewise.
826 (canonicalize_vector_int_perm): Likewise.
827 (ix86_noce_conversion_profitable_p): Likewise.
828 (ix86_mpx_bound_mode): Likewise.
829 (ix86_operands_ok_for_move_multiple): Likewise.
830 * config/microblaze/microblaze-protos.h
831 (microblaze_expand_conditional_branch_reg): Likewise.
832 * config/microblaze/microblaze.c
833 (microblaze_expand_conditional_branch_reg): Likewise.
834 * config/powerpcspe/powerpcspe.c (rs6000_init_hard_regno_mode_ok):
836 (rs6000_reassociation_width): Likewise.
837 (rs6000_invalid_binary_op): Likewise.
838 (fusion_p9_p): Likewise.
839 (emit_fusion_p9_load): Likewise.
840 (emit_fusion_p9_store): Likewise.
841 * config/riscv/riscv-protos.h (riscv_regno_mode_ok_for_base_p):
843 (riscv_hard_regno_mode_ok_p): Likewise.
844 (riscv_address_insns): Likewise.
845 (riscv_split_symbol): Likewise.
846 (riscv_legitimize_move): Likewise.
847 (riscv_function_value): Likewise.
848 (riscv_hard_regno_nregs): Likewise.
849 (riscv_expand_builtin): Likewise.
850 * config/riscv/riscv.c (riscv_build_integer_1): Likewise.
851 (riscv_build_integer): Likewise.
852 (riscv_split_integer): Likewise.
853 (riscv_legitimate_constant_p): Likewise.
854 (riscv_cannot_force_const_mem): Likewise.
855 (riscv_regno_mode_ok_for_base_p): Likewise.
856 (riscv_valid_base_register_p): Likewise.
857 (riscv_valid_offset_p): Likewise.
858 (riscv_valid_lo_sum_p): Likewise.
859 (riscv_classify_address): Likewise.
860 (riscv_legitimate_address_p): Likewise.
861 (riscv_address_insns): Likewise.
862 (riscv_load_store_insns): Likewise.
863 (riscv_force_binary): Likewise.
864 (riscv_split_symbol): Likewise.
865 (riscv_force_address): Likewise.
866 (riscv_legitimize_address): Likewise.
867 (riscv_move_integer): Likewise.
868 (riscv_legitimize_const_move): Likewise.
869 (riscv_legitimize_move): Likewise.
870 (riscv_address_cost): Likewise.
871 (riscv_subword): Likewise.
872 (riscv_output_move): Likewise.
873 (riscv_canonicalize_int_order_test): Likewise.
874 (riscv_emit_int_order_test): Likewise.
875 (riscv_function_arg_boundary): Likewise.
876 (riscv_pass_mode_in_fpr_p): Likewise.
877 (riscv_pass_fpr_single): Likewise.
878 (riscv_pass_fpr_pair): Likewise.
879 (riscv_get_arg_info): Likewise.
880 (riscv_function_arg): Likewise.
881 (riscv_function_arg_advance): Likewise.
882 (riscv_arg_partial_bytes): Likewise.
883 (riscv_function_value): Likewise.
884 (riscv_pass_by_reference): Likewise.
885 (riscv_setup_incoming_varargs): Likewise.
886 (riscv_print_operand): Likewise.
887 (riscv_elf_select_rtx_section): Likewise.
888 (riscv_save_restore_reg): Likewise.
889 (riscv_for_each_saved_reg): Likewise.
890 (riscv_register_move_cost): Likewise.
891 (riscv_hard_regno_mode_ok_p): Likewise.
892 (riscv_hard_regno_nregs): Likewise.
893 (riscv_class_max_nregs): Likewise.
894 (riscv_memory_move_cost): Likewise.
895 * config/rl78/rl78-protos.h (rl78_split_movsi): Likewise.
896 * config/rl78/rl78.c (rl78_split_movsi): Likewise.
897 (rl78_addr_space_address_mode): Likewise.
898 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin):
900 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Likewise.
901 (rs6000_reassociation_width): Likewise.
902 (rs6000_invalid_binary_op): Likewise.
903 (fusion_p9_p): Likewise.
904 (emit_fusion_p9_load): Likewise.
905 (emit_fusion_p9_store): Likewise.
906 * config/visium/visium-protos.h (prepare_move_operands): Likewise.
907 (ok_for_simple_move_operands): Likewise.
908 (ok_for_simple_move_strict_operands): Likewise.
909 (ok_for_simple_arith_logic_operands): Likewise.
910 (visium_legitimize_reload_address): Likewise.
911 (visium_select_cc_mode): Likewise.
912 (output_cbranch): Likewise.
913 (visium_split_double_move): Likewise.
914 (visium_expand_copysign): Likewise.
915 (visium_expand_int_cstore): Likewise.
916 (visium_expand_fp_cstore): Likewise.
917 * config/visium/visium.c (visium_pass_by_reference): Likewise.
918 (visium_function_arg): Likewise.
919 (visium_function_arg_advance): Likewise.
920 (visium_libcall_value): Likewise.
921 (visium_setup_incoming_varargs): Likewise.
922 (visium_legitimate_constant_p): Likewise.
923 (visium_legitimate_address_p): Likewise.
924 (visium_legitimize_address): Likewise.
925 (visium_secondary_reload): Likewise.
926 (visium_register_move_cost): Likewise.
927 (visium_memory_move_cost): Likewise.
928 (prepare_move_operands): Likewise.
929 (ok_for_simple_move_operands): Likewise.
930 (ok_for_simple_move_strict_operands): Likewise.
931 (ok_for_simple_arith_logic_operands): Likewise.
932 (visium_function_value_1): Likewise.
933 (rtx_ok_for_offset_p): Likewise.
934 (visium_legitimize_reload_address): Likewise.
935 (visium_split_double_move): Likewise.
936 (visium_expand_copysign): Likewise.
937 (visium_expand_int_cstore): Likewise.
938 (visium_expand_fp_cstore): Likewise.
939 (visium_split_cstore): Likewise.
940 (visium_select_cc_mode): Likewise.
941 (visium_split_cbranch): Likewise.
942 (output_cbranch): Likewise.
943 (visium_print_operand_address): Likewise.
944 * expmed.c (flip_storage_order): Likewise.
945 * expmed.h (emit_cstore): Likewise.
946 (flip_storage_order): Likewise.
947 * genrecog.c (validate_pattern): Likewise.
948 * hsa-gen.c (gen_hsa_addr): Likewise.
949 * internal-fn.c (expand_arith_overflow): Likewise.
950 * ira-color.c (allocno_copy_cost_saving): Likewise.
951 * lra-assigns.c (find_hard_regno_for_1): Likewise.
952 * lra-constraints.c (prohibited_class_reg_set_mode_p): Likewise.
953 (process_invariant_for_inheritance): Likewise.
954 * lra-eliminations.c (move_plus_up): Likewise.
955 * omp-low.c (lower_oacc_reductions): Likewise.
956 * simplify-rtx.c (simplify_subreg): Likewise.
957 * target.def (TARGET_SETUP_INCOMING_VARARG_BOUNDS): Likewise.
958 (TARGET_CHKP_BOUND_MODE): Likewise..
959 * targhooks.c (default_chkp_bound_mode): Likewise.
960 (default_setup_incoming_vararg_bounds): Likewise.
961 * targhooks.h (default_chkp_bound_mode): Likewise.
962 (default_setup_incoming_vararg_bounds): Likewise.
963 * tree-ssa-math-opts.c (divmod_candidate_p): Likewise.
964 * tree-vect-loop.c (calc_vec_perm_mask_for_shift): Likewise.
965 (have_whole_vector_shift): Likewise.
966 * tree-vect-stmts.c (vectorizable_load): Likewise.
967 * doc/tm.texi: Regenerate.
969 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
971 Graceful degrade if Binutils PR21472 is not available.
974 * configure.ac [target=avr]: WARN instead of ERROR if avrxmega3
975 .rodata in flash test fails.
976 (HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH): Define it if test passes.
977 * confgure: Regenerate.
978 * config.in: Regenerate.
979 * config/avr/avr.c (avr_asm_named_section)
980 [HAVE_LD_AVR_AVRXMEGA3_RODATA_IN_FLASH]: Only trigger
981 __do_copy_data for stuff in .rodata if flash_pm_offset = 0.
982 (avr_asm_init_sections): Same.
984 2017-07-05 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
986 * config/arm/neon.md (fma<VCVTF:mode>4): Remove %?.
987 (fma<VH:mode>4_intrinsic): Likewise.
988 (*fmsub<VCVTF:mode>4): Likewise.
989 (*fmsub<VH:mode>4_intrinsic): Likewise.
991 2017-07-05 Georg-Johann Lay <avr@gjlay.de>
994 * config/avr/avr.c (avr_out_movhi_mr_r_xmega) [CONSTANT_ADDRESS_P]:
995 Don't depend on "optimize > 0".
996 (out_movhi_r_mr, out_movqi_mr_r): Same.
997 (out_movhi_mr_r, out_movqi_r_mr): Same.
998 (avr_address_cost) [CONSTANT_ADDRESS_P]: Don't depend cost for
999 io_address_operand on "optimize > 0".
1001 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1003 * tree-loop-distribution.c: Add general explanantion on the pass.
1004 (generate_loops_for_partition): Mark distributed loop.
1005 (pg_add_dependence_edges): New parameter. Handle alias data
1006 dependence specially and record it in the parameter if asked.
1007 (struct pg_vdata, pg_edata, pg_edge_callback_data): New structs.
1008 (init_partition_graph_vertices, add_partition_graph_edge): New.
1009 (pg_skip_alias_edge, free_partition_graph_edata_cb): New.
1010 (free_partition_graph_vdata, build_partition_graph): New.
1011 (sort_partitions_by_post_order, merge_dep_scc_partitions): New.
1012 (pg_collect_alias_ddrs, break_alias_scc_partitions): New.
1013 (data_ref_segment_size, latch_dominated_by_data_ref): New.
1014 (compute_alias_check_pairs, version_loop_by_alias_check): New.
1015 (version_for_distribution_p, finalize_partitions): New.
1016 (distribute_loop): Handle alias data dependence specially. Factor
1017 out loop fusion code as functions and call these functions.
1019 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1021 * tree-loop-distribution.c (classify_partition): New parameter and
1022 better handle reduction statement.
1023 (rdg_build_partitions): Revise comment.
1024 (distribute_loop): Compute statements in all partitions and pass it
1025 to classify_partition.
1027 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1029 * tree-loop-distribution.c (enum partition_type): New.
1030 (struct partition): New field type.
1031 (partition_merge_into): Add parameter. Update partition type.
1032 (data_dep_in_cycle_p, update_type_for_merge): New functions.
1033 (build_rdg_partition_for_vertex): Compute partition type.
1034 (rdg_build_partitions): Dump partition type.
1035 (distribute_loop): Update calls to partition_merge_into.
1037 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1039 * tree-loop-distribution.c (struct ddr_hasher): New.
1040 (ddr_hasher::hash, ::equal, get_data_dependence): New function.
1042 (classify_partition): Call get_data_dependence.
1043 (pg_add_dependence_edges): Ditto.
1044 (distribute_loop): Release data dependence hash table.
1046 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1048 * tree-loop-distribution.c (ref_base_address): Delete.
1049 (similar_memory_accesses): Rename ...
1050 (share_memory_accesses): ... to this. Check if partitions access
1051 the same memory reference.
1052 (distribute_loop): Call share_memory_accesses.
1054 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1056 * tree-loop-distribution.c (struct partition): New field recording
1058 (partition_alloc, partition_free): Init and release data refs.
1059 (partition_merge_into): Merge data refs.
1060 (build_rdg_partition_for_vertex): Collect data refs for partition.
1061 (pg_add_dependence_edges): Change parameters from vector to bitmap.
1063 (distribute_loop): Remve data refs from vertice data of partition
1066 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1068 * tree-loop-distribution.c (params.h): Include header file.
1069 (MAX_DATAREFS_NUM, DR_INDEX): New macro.
1070 (datarefs_vec): New global var.
1071 (create_rdg_vertices): Use datarefs_vec directly.
1072 (free_rdg): Don't free data references.
1073 (build_rdg): Update use. Don't free data references.
1074 (distribute_loop): Compute global variable for data references.
1075 Bail out if there are too many data references.
1077 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1079 * tree-loop-distribution.c (loop_nest): New global var.
1080 (build_rdg): Use loop directly, rather than loop nest.
1081 (pg_add_dependence_edges): Remove loop nest parameter. Use global
1083 (distribute_loop): Compute global variable loop nest. Update use.
1085 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1087 * tree-loop-distribution.c (enum fuse_type, fuse_message): New.
1088 (partition_merge_into): New parameter. Dump reason for fusion.
1089 (distribute_loop): Update use of partition_merge_into.
1091 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1093 * tree-loop-distribution.c (bb_top_order_index): New.
1094 (bb_top_order_index_size, bb_top_order_cmp): New.
1095 (stmts_from_loop): Use topological order.
1096 (pass_loop_distribution::execute): Compute and release topological
1097 order for basic blocks.
1099 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1101 * tree-loop-distribution.c (pass_loop_distribution::execute): Skip
1104 2017-07-05 Bin Cheng <bin.cheng@arm.com>
1106 * cfgloop.h (struct loop): Add comment. New field orig_loop_num.
1107 * cfgloopmanip.c (lv_adjust_loop_entry_edge): Comment change.
1108 * internal-fn.c (expand_LOOP_DIST_ALIAS): New function.
1109 * internal-fn.def (LOOP_DIST_ALIAS): New.
1110 * tree-vectorizer.c (fold_loop_vectorized_call): Rename to ...
1111 (fold_loop_internal_call): ... this.
1112 (vect_loop_dist_alias_call): New function.
1113 (set_uid_loop_bbs): Call fold_loop_internal_call.
1114 (vectorize_loops): Fold IFN_LOOP_VECTORIZED and IFN_LOOP_DIST_ALIAS
1117 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
1120 * config/i386/i386.md (setcc + movzbl/and to xor + setcc peepholes):
1121 Require dead FLAGS_REG at the beginning of a peephole.
1123 2017-07-04 Uros Bizjak <ubizjak@gmail.com>
1126 * config/i386/adxintrin.h (_subborrow_u32): Swap _X and _Y
1127 arguments in the call to __builtin_ia32_sbb_u32.
1128 (_subborrow_u64): Swap _X and _Y arguments in the call to
1129 __builtin_ia32_sbb_u64.
1131 2017-07-04 Jakub Jelinek <jakub@redhat.com>
1134 * tree-vrp.c (compare_assert_loc): Turn into a function template
1135 with stable template parameter. Only test if a->e is NULL,
1136 !a->e == !b->e has been verified already. Use e == NULL or
1137 e != NULL instead of e or ! e tests. If stable is true, don't use
1138 iterative_hash_expr, on the other side allow a or b or both NULL
1139 and sort the NULLs last.
1140 (process_assert_insertions): Sort using compare_assert_loc<false>
1141 instead of compare_assert_loc, later sort using
1142 compare_assert_loc<true> before calling process_assert_insertions_for
1143 in a loop. Use break instead of continue once seen NULL pointer.
1145 2017-07-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
1147 * config/arm/driver-arm.c (arm_cpu_table): Add entry for ARM
1148 Cortex-R7 and Cortex-R8 processors.
1150 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
1152 * ipa-utils.c (ipa_merge_profiles): Fix merging when dst is
1153 uninitialized while src is not.
1155 2017-07-04 Richard Earnshaw <rearnsha@arm.com>
1157 * common/config/arm/arm-common.c: Adjust include path for
1159 * config/arm/t-arm (TM_H): Adjust path for arm-cpu.h.
1160 (arm-cpu.h): Create in build directory. Adjust dependency rules.
1161 (arm-cpu-data.h): Likewise.
1162 (arm-cpu-cdata.h): Likewise.
1163 * config/arm/arm-cpu.h: Delete.
1164 * config/arm/arm-cpu-cdata.h: Delete.
1165 * config/arm/arm-cpu-data.h: Delete.
1167 2017-07-04 James Greenhalgh <james.greenhalgh@arm.com>
1169 * config/arm/arm-cpus.in (cortex-a55): New.
1170 (cortex-a75): Likewise.
1171 (cortex-a75.cortex-a55): Likewise.
1172 * config/arm/driver-arm.c (arm_cpu_table): Add cortex-a55 and
1174 * doc/invoke.texi (-mcpu): Document cortex-a55 and cortex-a75.
1175 * config/arm/arm-cpu-cdata.h: Regenerate.
1176 * config/arm/arm-cpu-data.h: Regenerate.
1177 * config/arm/arm-cpu.h: Regenerate.
1178 * config/arm/arm-tables.opt: Regenerate.
1179 * config/arm/arm-tune.md: Regenerate.
1181 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
1183 * haifa-sched.c (sched_create_recovery_edges): Update profile.
1185 2017-07-04 Jan Hubicka <hubicka@ucw.cz>
1187 * bb-reorder.c (better_edge_p): Fix handling of uninitialized
1190 2017-07-04 Richard Sandiford <richard.sandiford@linaro.org>
1192 PR tree-optimization/81292
1193 * tree-ssa-strlen.c (handle_builtin_strlen): When setting
1194 full_string_p, also call adjust_related_strinfos if the adjustment
1195 is simple, otherwise invalidate related strinfos.
1197 2017-07-04 Martin Liska <mliska@suse.cz>
1200 * sanopt.c (sanitize_rewrite_addressable_params): Mark the
1201 newly created variable as DECL_IGNORED_P.
1203 2017-07-04 Martin Liska <mliska@suse.cz>
1206 * ipa-inline.c (inline_small_functions):
1207 Use xstrdup_for_dump.
1209 2017-07-04 Tom de Vries <tom@codesourcery.com>
1211 * graph.c (draw_cfg_edges): Save and restore EDGE_DFS_BACK.
1213 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
1216 * config/darwin.c (darwin_function_switched_text_sections):
1219 2017-07-03 Jan Hubicka <hubicka@ucw.cz>
1221 * tree-vect-loop-manip.c (vect_do_peeling): Fix scaling up.
1223 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
1225 * doc/invoke.texi (ARM Options): Add -mbe8 and -mbe32 to option summary.
1227 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
1229 * tree-vect-loop.c (vect_analyze_loop_2): Treat min_scalar_loop_bound,
1230 min_profitable_iters, and th as inclusive lower bounds.
1231 Fix LOOP_VINFO_PEELING_FOR_GAPS condition.
1232 (vect_estimate_min_profitable_iters): Return inclusive lower bounds
1233 for min_profitable_iters and min_profitable_estimate.
1234 (vect_transform_loop): Treat th as an inclusive lower bound.
1235 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
1237 2017-07-03 Dominique d'Humieres <dominiq@lps.ens.fr>
1240 * config/darwin.c (darwin_function_switched_text_sections):
1241 Replace DECL_NAME with DECL_ASSEMBLER_NAME, split assemble_name_raw
1242 in two pieces, and suppress the use of buf.
1244 2017-07-03 Nathan Sidwell <nathan@acm.org>
1246 * hash-table.h (hash_table_mod1): Fix indentation.
1248 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
1251 * predict.c (force_edge_cold): Be more careful about propagation
1253 * profile-count.h (profile_probability::guessed,
1254 profile_probability::fdo, profile_count::guessed, profile_count::fdo):
1256 * tree-ssa-threadupdate.c (recompute_probabilities): Result is guessed.
1258 2017-07-03 James Greenhalgh <james.greenhalgh@arm.com>
1260 * doc/invoke.texi (rcpc architecture extension): Document it.
1262 2017-07-03 Richard Biener <rguenther@suse.de>
1264 PR tree-optimization/60510
1265 * tree-vect-loop.c (vect_create_epilog_for_reduction): Pass in
1266 the scalar reduction PHI and use it.
1267 (vectorizable_reduction): Properly guard the single_defuse_cycle
1268 path for non-SLP reduction chains where we cannot use it.
1269 Rework reduc_def/index and vector type deduction. Rework
1270 vector operand gathering during reduction op code-gen.
1271 * tree-vect-slp.c (vect_analyze_slp): For failed SLP reduction
1272 chains dissolve the chain and leave it to non-SLP reduction
1275 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
1277 * tree-data-ref.h (dr_alignment): Declare.
1278 * tree-data-ref.c (dr_alignment): New function.
1279 * tree-vectorizer.h (dataref_aux): Remove base_element_aligned.
1280 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Don't
1282 * tree-vect-stmts.c (vectorizable_store): Use dr_alignment.
1284 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
1286 * tree-data-ref.h (innermost_loop_behavior): Add base_alignment
1287 and base_misalignment fields.
1288 (DR_BASE_ALIGNMENT, DR_BASE_MISALIGNMENT): New macros.
1289 * tree-data-ref.c: Include builtins.h.
1290 (dr_analyze_innermost): Set up the new innmost_loop_behavior fields.
1291 * tree-vectorizer.h (STMT_VINFO_DR_BASE_ALIGNMENT): New macro.
1292 (STMT_VINFO_DR_BASE_MISALIGNMENT): Likewise.
1293 * tree-vect-data-refs.c: Include tree-cfg.h.
1294 (vect_compute_data_ref_alignment): Use the new innermost_loop_behavior
1295 fields instead of calculating an alignment here.
1296 (vect_analyze_data_refs): Use dr_analyze_innermost. Dump the new
1297 innermost_loop_behavior fields.
1299 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
1301 * tree-data-ref.h (innermost_loop_behavior): Add a step_alignment
1303 (DR_STEP_ALIGNMENT): New macro.
1304 * tree-vectorizer.h (STMT_VINFO_DR_STEP_ALIGNMENT): Likewise.
1305 * tree-data-ref.c (dr_analyze_innermost): Initalize step_alignment.
1306 (create_data_ref): Print it.
1307 * tree-vect-stmts.c (vectorizable_load): Use the step alignment
1308 to tell whether the step preserves vector (mis)alignment.
1309 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
1310 Move the check for an integer step and generalise to all INTEGER_CST.
1311 (vect_analyze_data_refs): Set DR_STEP_ALIGNMENT when setting DR_STEP.
1312 Print the outer step alignment.
1314 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
1316 * tree-data-ref.h (innermost_loop_behavior): Replace aligned_to
1317 with offset_alignment.
1318 (DR_ALIGNED_TO): Delete.
1319 (DR_OFFSET_ALIGNMENT): New macro.
1320 * tree-vectorizer.h (STMT_VINFO_DR_ALIGNED_TO): Delete.
1321 (STMT_VINFO_DR_OFFSET_ALIGNMENT): New macro.
1322 * tree-data-ref.c (dr_analyze_innermost): Update after above changes.
1323 (create_data_ref): Likewise.
1324 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Likewise.
1325 (vect_analyze_data_refs): Likewise.
1326 * tree-if-conv.c (if_convertible_loop_p_1): Use memset before
1327 creating dummy innermost behavior.
1329 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
1331 * tree-data-ref.h (dr_analyze_innermost): Replace the dr argument
1332 with a "innermost_loop_behavior *" and refeence tree.
1333 * tree-data-ref.c (dr_analyze_innermost): Likewise.
1334 (create_data_ref): Update call accordingly.
1335 * tree-predcom.c (find_looparound_phi): Likewise.
1337 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
1339 * tree-vectorizer.h (_stmt_vec_info): Replace individual dr_*
1340 fields with dr_wrt_vec_loop.
1341 (STMT_VINFO_DR_BASE_ADDRESS, STMT_VINFO_DR_INIT, STMT_VINFO_DR_OFFSET)
1342 (STMT_VINFO_DR_STEP, STMT_VINFO_DR_ALIGNED_TO): Update accordingly.
1343 (STMT_VINFO_DR_WRT_VEC_LOOP): New macro.
1344 (vect_dr_behavior): New function.
1345 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
1346 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Use
1347 vect_dr_behavior. Use a step_preserves_misalignment_p boolean to
1348 track whether the step preserves the misalignment.
1349 (vect_create_addr_base_for_vector_ref): Remove loop parameter.
1350 Use vect_dr_behavior.
1351 (vect_setup_realignment): Update call accordingly.
1352 (vect_create_data_ref_ptr): Likewise. Use vect_dr_behavior.
1353 * tree-vect-loop-manip.c (vect_gen_prolog_loop_niters): Update
1354 call to vect_create_addr_base_for_vector_ref.
1355 (vect_create_cond_for_align_checks): Likewise.
1356 * tree-vect-patterns.c (vect_recog_bool_pattern): Copy
1357 STMT_VINFO_DR_WRT_VEC_LOOP as a block.
1358 (vect_recog_mask_conversion_pattern): Likewise.
1359 * tree-vect-stmts.c (compare_step_with_zero): Use vect_dr_behavior.
1360 (new_stmt_vec_info): Remove redundant zeroing.
1362 2017-07-03 Richard Earnshaw <rearnsha@arm.com>
1364 * common/config/arm/arm-common.c (arm_be8_option): New function.
1365 * config/arm/arm-isa.h (isa_feature): Add new feature bit isa_bit_be8.
1366 (ISA_ARMv6): Add isa_bit_be8.
1367 * config/arm/arm.h (arm_be8_option): Add prototype.
1368 (BE8_SPEC_FUNCTION): New define.
1369 (EXTRA_SPEC_FUNCTIONS): Add BE8_SPEC_FUNCTION.
1370 * config/arm/arm.opt (mbig-endian): Mark as Negative of mlittle-endian.
1371 (mlittle-endian): Similarly.
1372 (mbe8, mbe32): New options.
1373 * config/arm/bpabi.h (BE8_LINK_SPEC): Call arm_be8_option.
1374 * doc/invoke.texi (ARM Options): Document -mbe8 and -mbe32.
1376 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
1378 * tree-cfg.c (gimple_find_sub_bbs): Fix profile updating.
1380 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
1382 * tree-cfgcleanup.c (want_merge_blocks_p): New function.
1383 (cleanup_tree_cfg_bb): Use it.
1384 * profile-count.h (profile_count::of_for_merging, profile_count::merge):
1386 * tree-cfg.c (gimple_merge_blocks): Use profile_count::merge.
1388 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
1391 * loop-doloop.c (add_test): Update profile.
1393 2017-07-03 Martin Liska <mliska@suse.cz>
1396 * sanopt.c (rewrite_usage_of_param): New function.
1397 (sanitize_rewrite_addressable_params): Likewise.
1398 (pass_sanopt::execute): Call rewrite_usage_of_param.
1400 2017-07-03 Richard Biener <rguenther@suse.de>
1402 * tree-vect-loop.c (vect_create_epilog_for_reduction): Revert
1403 back to using VIEW_CONVERT_EXPR.
1405 2017-07-03 Martin Liska <mliska@suse.cz>
1408 * doc/extend.texi: Document when a resolver function is
1409 generated for target_clones.
1411 2017-07-03 Martin Liska <mliska@suse.cz>
1413 * asan.c (asan_emit_stack_protection): Unpoison just red zones
1414 and shadow memory of auto variables which are subject of
1415 use-after-scope sanitization.
1416 (asan_expand_mark_ifn): Add do set only when is_poison.
1418 2016-07-03 Richard Biener <rguenther@suse.de>
1420 * tree-vect-loop.c (vect_analyze_loop_operations): Also analyze
1422 (vect_force_simple_reduction): Record reduction def -> phi mapping.
1423 (vectorizable_reduction): Perform reduction PHI creation when
1424 visiting a reduction PHI and adjust and simplify code generation
1425 phase of the reduction op. Cache dts, use fold_binary, not fold_build2.
1426 (vect_transform_loop): Visit reduction PHIs.
1427 * tree-vect-slp.c (vect_get_and_check_slp_defs): Record reduction
1428 defs into the SLP tree.
1429 (vect_build_slp_tree): Reduction defs terminate the recursion.
1430 * tree-vect-stmts.c (vect_get_vec_def_for_operand_1): Allow lookup
1432 (vect_get_vec_defs_for_stmt_copy): Export.
1433 (vect_get_vec_defs): Likewise.
1434 * tree-vectorizer.h (struct _stmt_vec_info): Amend reduc_def
1436 (vect_get_vec_defs_for_stmt_copy): Declare.
1437 (vect_get_vec_defs): Likewise.
1439 2017-07-03 Richard Sandiford <richard.sandiford@linaro.org>
1441 * tree-data-ref.c (dr_analyze_innermost): Replace the "nest"
1442 parameter with a "loop" parameter and use it instead of the
1443 loop containing DR_STMT. Don't check simple_iv when doing
1444 BB analysis. Describe the two analysis modes in the comment.
1446 2017-07-03 Tom de Vries <tom@codesourcery.com>
1448 PR tree-optimization/69468
1449 * tree-ssa-tail-merge.c (ignore_edge_flags): New constant.
1450 (find_same_succ_bb): Handle ignore_edge_flags.
1452 2017-07-03 Tom de Vries <tom@codesourcery.com>
1454 PR tree-optimization/81192
1455 * tree-ssa-tail-merge.c (same_succ_hash): Use bb->loop_father->num in
1457 (same_succ::equal): Don't find bbs to be equal if bb->loop_father
1459 (find_same_succ_bb): Remove obsolete test on bb->loop_father->latch.
1461 2017-07-03 Tom de Vries <tom@codesourcery.com>
1463 PR tree-optimization/81192
1464 * tree-ssa-tail-merge.c (same_succ_flush_bb): Handle
1465 BB_SAME_SUCC (bb) == NULL.
1467 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
1469 * cfgrtl.c (rtl_verify_edges): Enable checking of profile_probability
1472 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
1474 * dumpfile.c: Include profile-count.h
1475 * tree-cfg.c (gimple_duplicate_sese_tail): Drop UNUSED attributes;
1477 (insert_cond_bb): Update profile.
1478 * tree-cfg.h (insert_cond_bb): Update prototype.
1479 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Update.
1480 * tree-dump.c: Do not include tree-cfg.
1482 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
1484 * bb-reorder.c (fix_up_crossing_landing_pad): Update profile.
1486 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
1488 * expect.c (dw2_build_landing_pads): Update profile of the landing pad
1491 2017-07-02 Jan Hubicka <hubicka@ucw.cz>
1493 * tree-complex.c (expand_complex_div_wide): update profile.
1495 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
1496 Alan Hayward <alan.hayward@arm.com>
1497 David Sherwood <david.sherwood@arm.com>
1499 * Makefile.in (MACHMODE_H): Remove insn-modes.h
1500 (CORETYPES_H): New define.
1501 (MOSTLYCLEANFILES): Add insn-modes-inline.h.
1502 (insn-modes-inline.h, s-modes-inline-h): New rules.
1503 (generated_files): Add insn-modes-inline.h.
1504 (RTL_BASE_H, TREE_CORE_H): Use CORETYPES_H instead of coretypes.h.
1505 (build/gensupport.o, build/ggc-none.o, build/print-rtl.o): Likewise.
1506 (build/read-md.o, build/read-rtl.o, build/rtl.o): Likewise.
1507 (build/vec.o, build/hash-table.o, build/inchash.o): Likewise.
1508 (build/gencondmd.o, build/genattr.o, build/genattr-common.o): Likewise.
1509 (build/genattrtab.o, build/genautomata.o, build/gencheck.o): Likewise.
1510 (build/gencodes.o, build/genconditions.o): Likewise.
1511 (build/genconfig.o, build/genconstants.o, build/genemit.o): Likewise.
1512 (build/genenums.o, build/genextract.o, build/genflags.o): Likewise.
1513 (build/gentarget-def.o, build/genmddeps.o, build/genopinit.o)
1514 (build/genoutput.o, build/genpeep.o, build/genpreds.o): Likewise.
1515 (build/genrecog.o, build/genmddump.o, build/genmatch.o): Likewise.
1516 (build/gencfn-macros.o, build/gcov-iov.o): Likewise.
1517 * coretypes.h: Include everything up to real.h for generators.
1518 Include insn-modes.h first. Include wide-int-print.h after
1519 wide-int.h. Include insn-modes-inline.h and then machmode.h.
1520 * machmode.h: Don't include insn-modes.h here.
1521 * function-tests.c: Remove includes of signop.h, machmode.h,
1522 double-int.h and wide-int.h.
1524 * gcc-rich-location.c: Remove includes of machmode.h, double-int.h
1526 * optc-save-gen.awk: Likewise.
1527 * gencheck.c (BITS_PER_UNIT): Delete dummy definition.
1528 * godump.c: Remove include of wide-int-print.h.
1529 * pretty-print.h: Likewise.
1530 * wide-int-print.cc: Likewise.
1531 * wide-int.cc: Likewise.
1532 * hash-map-tests.c: Remove include of signop.h.
1533 * hash-set-tests.c: Likewise.
1534 * rtl-tests.c: Likewise.
1535 * mkconfig.sh: Remove include of machmode.h.
1536 * genmodes.c (emit_insn_modes_h): Split emission of inline functions
1538 (emit_insn_modes_inline_h): ...this new function. Emit the code
1539 into an insn-modes-inline.h header file, adding appropriate
1540 include guards and end comments.
1541 (emit_insn_modes_c_header): Remove include of machmode.h.
1542 (emit_min_insn_modes_c_header): Include coretypes.h rather than
1544 (main): Handle -i flag and call emit_insn_modes_inline_h when
1547 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
1549 * tree-ssa-strlen.c (strinfo): Rename the length field to
1550 nonzero_chars. Add a full_string_p field.
1551 (compare_nonzero_chars, zero_length_string_p): New functions.
1552 (get_addr_stridx): Add an offset_out parameter.
1553 Use compare_nonzero_chars.
1554 (get_stridx): Update accordingly. Use compare_nonzero_chars.
1555 (new_strinfo): Update after above changes to strinfo.
1556 (set_endptr_and_length): Set full_string_p.
1557 (get_string_length): Update after above changes to strinfo.
1558 (unshare_strinfo): Update call to new_strinfo.
1559 (maybe_invalidate): Likewise.
1560 (get_stridx_plus_constant): Change off to unsigned HOST_WIDE_INT.
1561 Use compare_nonzero_chars and zero_string_p. Treat nonzero_chars
1562 as a uhwi instead of an shwi. Update after above changes to
1563 strinfo and new_strinfo.
1564 (zero_length_string): Assert that chainsi contains full strings.
1565 Use zero_length_string_p. Update call to new_strinfo.
1566 (adjust_related_strinfos): Update after above changes to strinfo.
1567 Copy full_string_p from origsi.
1568 (adjust_last_stmt): Use zero_length_string_p.
1569 (handle_builtin_strlen): Update after above changes to strinfo and
1570 new_strinfo. Install the lhs as the string length if the previous
1571 entry didn't describe a full string.
1572 (handle_builtin_strchr): Update after above changes to strinfo
1574 (handle_builtin_strcpy): Likewise.
1575 (handle_builtin_strcat): Likewise.
1576 (handle_builtin_malloc): Likewise.
1577 (handle_pointer_plus): Likewise.
1578 (handle_builtin_memcpy): Likewise. Track nonzero characters
1579 that aren't necessarily followed by a nul terminator.
1580 (handle_char_store): Likewise.
1582 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
1584 PR tree-optimization/80769
1585 * tree-ssa-strlen.c (strinfo): Document that "stmt" is also used
1586 for malloc and calloc. Document the new invariant that all related
1587 strinfos have delayed lengths or none do.
1588 (verify_related_strinfos): Move earlier in file.
1589 (set_endptr_and_length): New function, split out from...
1590 (get_string_length): ...here. Also set the lengths of related
1592 (zero_length_string): Assert that chainsi has known (rather than
1594 (adjust_related_strinfos): Likewise.
1596 2017-07-02 Richard Sandiford <richard.sandiford@linaro.org>
1598 PR tree-optimization/81136
1599 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Only
1600 assert that two references with the same misalignment have the same
1601 compile-time misalignment if those compile-time misalignments
1604 2017-07-01 Andi Kleen <ak@linux.intel.com>
1606 * print-tree.c (print_node): Print all attributes.
1608 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
1610 * cfg.c (scale_bbs_frequencies): New function.
1611 * cfg.h (scale_bbs_frequencies): Declare it.
1612 * cfgloopanal.c (single_likely_exit): Cleanup.
1613 * cfgloopmanip.c (scale_loop_frequencies): Take profile_probability
1615 (scale_loop_profile): Likewise.
1616 (loop_version): Likewise.
1617 (create_empty_loop_on_edge): Update.
1618 * cfgloopmanip.h (scale_loop_frequencies, scale_loop_profile,
1619 scale_loop_frequencies, scale_loop_profile, loopify,
1620 loop_version): Update prototypes.
1621 * modulo-sched.c (sms_schedule): Update.
1622 * predict.c (unlikely_executed_edge_p): Also check probability.
1623 (probably_never_executed_edge_p): Fix typo.
1624 * tree-if-conv.c (version_loop_for_if_conversion): Update.
1625 * tree-parloops.c (gen_parallel_loop): Update.
1626 * tree-ssa-loop-ivcanon.c (try_peel_loop): Update.
1627 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
1628 * tree-ssa-loop-split.c (split_loop): Update.
1629 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
1630 * tree-vect-loop-manip.c (vect_do_peeling): Update.
1631 (vect_loop_versioning): Update.
1632 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
1634 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
1636 * trans-mem.c (split_bb_make_tm_edge): Update profile.
1638 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
1640 * tree-if-conv.c (combine_blocks): Use make_single_succ_edge
1641 to keep profile consistent.
1643 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
1645 * cfgrtl.c (rtl_flow_call_edges_add): Update profile.
1646 * tree-cfg.c (gimple_flow_call_edges_add): Likewise.
1647 * profile-count.h (max_safe_multiplier): Make unsigned.
1648 (profile_count::guessed_zero): New.
1650 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
1652 * bb-reorder.c (fix_up_crossing_landing_pad,
1653 fix_crossing_conditional_branches): Use make_single_succ_edge
1654 to keep profile consistent.
1656 2017-07-01 Jan Hubicka <hubicka@ucw.cz>
1658 * tree-vect-loop.c (optimize_mask_stores): Use make_single_succ_edge
1661 2017-07-01 Jakub Jelinek <jakub@redhat.com>
1664 * bb-reorder.c (fix_up_fall_thru_edges): Move variable declarations to
1665 the right scopes, make sure cond_jump isn't preserved between multiple
1666 iterations. Search for fallthru edge whenever there are 3+ edges and
1667 use find_fallthru_edge for it.
1669 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
1671 Patch by Alexander Monakov <amonakov@ispras.ru>
1672 * sel-sched-ir.c (compute_succs_info): Handle uninitialized
1673 probabilities consistently.
1675 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
1677 * pa.c (pa_expand_compare_and_swap_loop): Update call of
1678 emit_cmp_and_jump_insns.
1680 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
1683 * tree-inline.c (expand_call_inline): Combine profile statuses.
1685 2017-06-30 Andrew Pinski <apinski@cavium.com>
1687 * tree-if-conv.c (predicate_scalar_phi): Update new_stmt if
1688 fold_stmt returned true.
1690 2017-06-30 Nathan Sidwell <nathan@acm.org>
1692 * ggc.h (empty_string): Delete.
1693 * cfgexpand.c (expand_asm_stmt): Use plain "".
1694 * optabs.c (expand_asm_memory_barrier): Likewise.
1695 * stringpool.c (empty_string): Delete.
1696 (digit_vector, digit_string): Delete.
1697 (ggc_alloc_string): Use plain "", don't optimize single digit
1698 strings. Use ggc_alloc_atomic.
1700 2017-06-30 Richard Earnshaw <rearnsha@arm.com>
1702 * rtlanal.c (insn_rtx_cost): If a parallel contains exactly one
1703 comparison set and one other set, use the cost of the non-comparison
1706 2017-06-30 Nathan Sidwell <nathan@acm.org>
1708 * ggc.h: Replace all 'static inline' with plain 'inline'. Fix
1711 2017-06-30 Peter Bergner <bergner@vnet.ibm.com>
1713 * tree-cfg.c (group_case_labels_stmt): Merge scanning and compressing
1714 loops. Remove now unneeded calls to gimple_switch_set_label() that
1715 just set removed labels to NULL_TREE.
1717 2017-06-30 Aldy Hernandez <aldyh@redhat.com>
1719 * tree-ssanames.c (set_range_info_raw): Abstract from ...
1720 (set_range_info): ...here. Only call set_range_info_raw if domain
1722 (set_nonzero_bits): Call set_range_info_raw.
1723 * tree-ssanames.h (set_range_info_raw): New.
1725 2017-06-30 Jakub Jelinek <jakub@redhat.com>
1728 * config/i386/sse.md (vec_extract_lo_<mode><mask_name>): For V8FI,
1729 V16FI and VI8F_256 iterators, use <store_mask_predicate> instead
1730 of nonimmediate_operand and <store_mask_constraint> instead of m
1731 for the input operand. For V8FI iterator, always split if input
1732 is a MEM. For V16FI and V8SF_256 iterators, don't test if both
1733 operands are MEM if <mask_applied>. For VI4F_256 iterator, use
1734 <store_mask_predicate> instead of register_operand and
1735 <store_mask_constraint> instead of v for the input operand. Make
1736 sure both operands aren't MEMs for if not <mask_applied>.
1738 2017-06-30 Sylvestre Ledru <sylvestre@debian.org>
1740 * lto-wrapper.c (copy_file) Close both file descriptors before
1743 2017-06-30 Martin Liska <mliska@suse.cz>
1746 * multiple_target.c (create_dispatcher_calls): Make ifunc
1747 also for function that don't have calls or are not referenced.
1749 2017-06-30 Richard Biener <rguenther@suse.de>
1751 * tree-vect-slp.c (vect_slp_analyze_node_operations): Only
1752 analyze the first scalar stmt. Move vector type computation
1753 for the BB case here from ...
1754 * tree-vect-stmts.c (vect_analyze_stmt): ... here. Guard
1755 live operation processing in the SLP case properly.
1757 2017-06-30 Richard Biener <rguenther@suse.de>
1759 * graph.c (draw_cfg_node_succ_edges): Fix broken dot syntax.
1761 2017-06-30 Martin Liska <mliska@suse.cz>
1764 * tree-eh.c (lower_resx): Call BUILT_IN_ASAN_HANDLE_NO_RETURN
1765 before BUILT_IN_UNWIND_RESUME when ASAN is used.
1767 2017-06-30 Yvan Roux <yvan.roux@linaro.org>
1769 * doc/invoke.texi (AArch64): Add missing options and remove redundant
1772 2017-06-30 Richard Biener <rguenther@suse.de>
1774 PR tree-optimization/81249
1775 * tree-vect-loop.c (vect_create_epilog_for_reduction): Convert
1776 condition reduction result to original scalar type.
1778 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1780 * profile-count.h (enum profile_quality): Fix typos and whitespace
1783 2017-06-30 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
1785 * config/s390/s390.c (s390_expand_setmem): Adjust to the new data
1786 type for branch probabilities.
1788 2017-06-29 Julian Brown <julian@codesourcery.com>
1789 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1791 * config/aarch64/aarch64-fusion-pairs.def: Add ALU_BRANCH entry.
1792 * config/aarch64/aarch64.c (AARCH64_FUSE_ALU_BRANCH): New fusion type.
1793 (thunderx2t99_tunings): Set AARCH64_FUSE_ALU_BRANCH flag.
1794 (aarch_macro_fusion_pair_p): Add support for AARCH64_FUSE_ALU_BRANCH.
1796 2017-06-29 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
1798 * config/aarch64/aarch64.c (aarch_macro_fusion_pair_p): Push the
1799 check for CC usage into AARCH64_FUSE_CMP_BRANCH.
1800 * config/i386/i386.c (ix86_macro_fusion_pair_p): Push the check for
1801 CC usage from generic code to here.
1802 * sched-deps.c (sched_macro_fuse_insns): Move the condition for
1803 CC usage into the target macros.
1805 2017-06-29 Maya Rashish <coypu@sdf.org>
1807 * config/netbsd.h (NETBSD_LIB_SPEC): Add -lc when creating shared
1810 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
1812 * arm/arm-builtins.c: Include profile-count.h
1813 * except.c (sjlj_emit_function_enter): Use
1814 profile_probability::unlikely.
1816 2017-06-29 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
1818 * config/rs6000/rs6000.c (toc_relative_expr_p): Make tocrel_base
1819 and tocrel_offset be pointer args rather than implicitly using
1821 (legitimate_constant_pool_address_p, rs6000_emit_move,
1822 const_load_sequence_p, adjust_vperm): Add local tocrel_base and
1823 tocrel_offset and use in toc_relative_expr_p call.
1824 (print_operand, print_operand_address): Use static tocrel_base_oac
1825 and tocrel_offset_oac.
1826 (rs6000_output_addr_const_extra): Use static tocrel_base_oac and
1829 2017-06-29 Maya Rashish <coypu@sdf.org>
1831 * config/vax/builtins.md (ffssi2_internal): Correct constraint.
1833 2017-06-29 Eric Botcazou <ebotcazou@adacore.com>
1835 * expr.c (expand_expr) <normal_inner_ref>: When testing for unaligned
1836 objects, take into account only the alignment of 'op0' and 'mode1' if
1839 2017-06-29 Steve Ellcey <sellcey@cavium.com>
1841 * ccmp.c (ccmp_tree_comparison_p): New function.
1842 (ccmp_candidate_p): Update to use above function.
1843 (get_compare_parts): New function.
1844 (expand_ccmp_next): Update to use new functions.
1845 (expand_ccmp_expr_1): Take tree arg instead of gimple, update to use
1847 (expand_ccmp_expr): Pass tree instead of gimple to expand_ccmp_expr_1,
1848 take mode as argument.
1849 * ccmp.h (expand_ccmp_expr): Add mode as argument.
1850 * expr.c (expand_expr_real_1): Pass mode as argument.
1852 2017-06-29 Segher Boessenkool <segher@kernel.crashing.org>
1854 * combine.c (combine_instructions): Print insns to dump_file, together
1857 2017-06-29 Jan Hubicka <hubicka@ucw.cz>
1859 * asan.c (asan_emit_stack_protection): Update.
1860 (create_cond_insert_point): Update.
1861 * auto-profile.c (afdo_propagate_circuit): Update.
1862 * basic-block.h (struct edge_def): Turn probability to
1863 profile_probability.
1864 (EDGE_FREQUENCY): Update.
1865 * bb-reorder.c (find_traces_1_round): Update.
1866 (better_edge_p): Update.
1867 (sanitize_hot_paths): Update.
1868 * cfg.c (unchecked_make_edge): Initialize probability to uninitialized.
1869 (make_single_succ_edge): Update.
1870 (check_bb_profile): Update.
1871 (dump_edge_info): Update.
1872 (update_bb_profile_for_threading): Update.
1873 * cfganal.c (connect_infinite_loops_to_exit): Initialize new edge
1874 probabilitycount to 0.
1875 * cfgbuild.c (compute_outgoing_frequencies): Update.
1876 * cfgcleanup.c (try_forward_edges): Update.
1877 (outgoing_edges_match): Update.
1878 (try_crossjump_to_edge): Update.
1879 * cfgexpand.c (expand_gimple_cond): Update make_single_succ_edge.
1880 (expand_gimple_tailcall): Update.
1881 (construct_init_block): Use make_single_succ_edge.
1882 (construct_exit_block): Use make_single_succ_edge.
1883 * cfghooks.c (verify_flow_info): Update.
1884 (redirect_edge_succ_nodup): Update.
1885 (split_edge): Update.
1886 (account_profile_record): Update.
1887 * cfgloopanal.c (single_likely_exit): Update.
1888 * cfgloopmanip.c (scale_loop_profile): Update.
1889 (set_zero_probability): Remove.
1890 (duplicate_loop_to_header_edge): Update.
1891 * cfgloopmanip.h (loop_version): Update prototype.
1892 * cfgrtl.c (try_redirect_by_replacing_jump): Update.
1893 (force_nonfallthru_and_redirect): Update.
1894 (update_br_prob_note): Update.
1895 (rtl_verify_edges): Update.
1896 (purge_dead_edges): Update.
1897 (rtl_lv_add_condition_to_bb): Update.
1898 * cgraph.c: (cgraph_edge::redirect_call_stmt_to_calle): Update.
1899 * cgraphunit.c (init_lowered_empty_function): Update.
1900 (cgraph_node::expand_thunk): Update.
1901 * cilk-common.c: Include profile-count.h
1902 * dojump.c (inv): Remove.
1903 (jumpifnot): Update.
1904 (jumpifnot_1): Update.
1905 (do_jump_1): Update.
1907 (do_jump_by_parts_greater_rtx): Update.
1908 (do_compare_rtx_and_jump): Update.
1909 * dojump.h (jumpifnot, jumpifnot_1, jumpif_1, jumpif, do_jump,
1910 do_jump_1. do_compare_rtx_and_jump): Update prototype.
1911 * dwarf2cfi.c: Include profile-count.h
1912 * except.c (dw2_build_landing_pads): Use make_single_succ_edge.
1913 (sjlj_emit_dispatch_table): Likewise.
1914 * explow.c: Include profile-count.h
1915 * expmed.c (emit_store_flag_force): Update.
1916 (do_cmp_and_jump): Update.
1917 * expr.c (compare_by_pieces_d::generate): Update.
1918 (compare_by_pieces_d::finish_mode): Update.
1919 (emit_block_move_via_loop): Update.
1920 (store_expr_with_bounds): Update.
1921 (store_constructor): Update.
1922 (expand_expr_real_2): Update.
1923 (expand_expr_real_1): Update.
1924 * expr.h (try_casesi, try_tablejump): Update prototypes.
1925 * gimple-pretty-print.c (dump_probability): Update.
1926 (dump_profile): New.
1927 (dump_gimple_label): Update.
1928 (dump_gimple_bb_header): Update.
1929 * graph.c (draw_cfg_node_succ_edges): Update.
1930 * hsa-gen.c (convert_switch_statements): Update.
1931 * ifcvt.c (cheap_bb_rtx_cost_p): Update.
1932 (find_if_case_1): Update.
1933 (find_if_case_2): Update.
1934 * internal-fn.c (expand_arith_overflow_result_store): Update.
1935 (expand_addsub_overflow): Update.
1936 (expand_neg_overflow): Update.
1937 (expand_mul_overflow): Update.
1938 (expand_vector_ubsan_overflow): Update.
1939 * ipa-cp.c (good_cloning_opportunity_p): Update.
1940 * ipa-split.c (split_function): Use make_single_succ_edge.
1941 * ipa-utils.c (ipa_merge_profiles): Update.
1942 * loop-doloop.c (add_test): Update.
1943 (doloop_modify): Update.
1944 * loop-unroll.c (compare_and_jump_seq): Update.
1945 (unroll_loop_runtime_iterations): Update.
1946 * lra-constraints.c (lra_inheritance): Update.
1947 * lto-streamer-in.c (input_cfg): Update.
1948 * lto-streamer-out.c (output_cfg): Update.
1949 * mcf.c (adjust_cfg_counts): Update.
1950 * modulo-sched.c (sms_schedule): Update.
1951 * omp-expand.c (expand_omp_for_init_counts): Update.
1952 (extract_omp_for_update_vars): Update.
1953 (expand_omp_ordered_sink): Update.
1954 (expand_omp_for_ordered_loops): Update.
1955 (expand_omp_for_generic): Update.
1956 (expand_omp_for_static_nochunk): Update.
1957 (expand_omp_for_static_chunk): Update.
1958 (expand_cilk_for): Update.
1959 (expand_omp_simd): Update.
1960 (expand_omp_taskloop_for_outer): Update.
1961 (expand_omp_taskloop_for_inner): Update.
1962 * omp-simd-clone.c (simd_clone_adjust): Update.
1963 * optabs.c (expand_doubleword_shift): Update.
1964 (expand_abs): Update.
1965 (emit_cmp_and_jump_insn_1): Update.
1966 (expand_compare_and_swap_loop): Update.
1967 * optabs.h (emit_cmp_and_jump_insns): Update prototype.
1968 * predict.c (predictable_edge_p): Update.
1969 (edge_probability_reliable_p): Update.
1970 (set_even_probabilities): Update.
1971 (combine_predictions_for_insn): Update.
1972 (combine_predictions_for_bb): Update.
1973 (propagate_freq): Update.
1974 (estimate_bb_frequencies): Update.
1975 (force_edge_cold): Update.
1976 * profile-count.c (profile_count::dump): Add missing space into dump.
1977 (profile_count::debug): Add newline.
1978 (profile_count::differs_from_p): Explicitly convert to unsigned.
1979 (profile_count::stream_in): Update.
1980 (profile_probability::dump): New member function.
1981 (profile_probability::debug): New member function.
1982 (profile_probability::differs_from_p): New member function.
1983 (profile_probability::differs_lot_from_p): New member function.
1984 (profile_probability::stream_in): New member function.
1985 (profile_probability::stream_out): New member function.
1986 * profile-count.h (profile_count_quality): Rename to ...
1987 (profile_quality): ... this one.
1988 (profile_probability): New.
1989 (profile_count): Update.
1990 * profile.c (compute_branch_probabilities): Update.
1991 * recog.c (peep2_attempt): Update.
1992 * sched-ebb.c (schedule_ebbs): Update.
1993 * sched-rgn.c (find_single_block_region): Update.
1994 (compute_dom_prob_ps): Update.
1995 (schedule_region): Update.
1996 * sel-sched-ir.c (compute_succs_info): Update.
1997 * stmt.c (struct case_node): Update.
1998 (do_jump_if_equal): Update.
1999 (get_outgoing_edge_probs): Update.
2000 (conditional_probability): Update.
2001 (emit_case_dispatch_table): Update.
2002 (expand_case): Update.
2003 (expand_sjlj_dispatch_table): Update.
2004 (emit_case_nodes): Update.
2005 * targhooks.c: Update.
2006 * tracer.c (better_p): Update.
2007 (find_best_successor): Update.
2008 * trans-mem.c (expand_transaction): Update.
2009 * tree-call-cdce.c: Update.
2010 * tree-cfg.c (gimple_split_edge): Upate.
2011 (move_sese_region_to_fn): Upate.
2012 * tree-cfgcleanup.c (cleanup_control_expr_graph): Upate.
2013 * tree-eh.c (lower_resx): Upate.
2014 (cleanup_empty_eh_move_lp): Upate.
2015 * tree-if-conv.c (version_loop_for_if_conversion): Update.
2016 * tree-inline.c (copy_edges_for_bb): Update.
2017 (copy_cfg_body): Update.
2018 * tree-parloops.c (gen_parallel_loop): Update.
2019 * tree-profile.c (gimple_gen_ic_func_profiler): Update.
2020 (gimple_gen_time_profiler): Update.
2021 * tree-ssa-dce.c (remove_dead_stmt): Update.
2022 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
2023 * tree-ssa-loop-im.c (execute_sm_if_changed): Update.
2024 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update.
2025 (unloop_loops): Update.
2026 (try_peel_loop): Update.
2027 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
2028 * tree-ssa-loop-split.c (connect_loops): Update.
2029 (split_loop): Update.
2030 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Update.
2031 (hoist_guard): Update.
2032 * tree-ssa-phionlycprop.c (propagate_rhs_into_lhs): Update.
2033 * tree-ssa-phiopt.c (replace_phi_edge_with_variable): Update.
2034 (value_replacement): Update.
2035 * tree-ssa-reassoc.c (branch_fixup): Update.
2036 * tree-ssa-tail-merge.c (replace_block_by): Update.
2037 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges): Update.
2038 (create_edge_and_update_destination_phis): Update.
2039 (compute_path_counts): Update.
2040 (recompute_probabilities): Update.
2041 (update_joiner_offpath_counts): Update.
2042 (freqs_to_counts_path): Update.
2043 (duplicate_thread_path): Update.
2044 * tree-switch-conversion.c (hoist_edge_and_branch_if_true): Update.
2045 (struct switch_conv_info): Update.
2046 (gen_inbound_check): Update.
2047 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
2048 (vect_do_peeling): Update.
2049 (vect_loop_versioning): Update.
2050 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
2051 (optimize_mask_stores): Update.
2052 * ubsan.c (ubsan_expand_null_ifn): Update.
2053 * value-prof.c (gimple_divmod_fixed_value): Update.
2054 (gimple_divmod_fixed_value_transform): Update.
2055 (gimple_mod_pow2): Update.
2056 (gimple_mod_pow2_value_transform): Update.
2057 (gimple_mod_subtract): Update.
2058 (gimple_mod_subtract_transform): Update.
2059 (gimple_ic): Update.
2060 (gimple_stringop_fixed_value): Update.
2061 (gimple_stringops_transform): Update.
2062 * value-prof.h: Update.
2064 2017-06-29 Carl Love <cel@us.ibm.com>
2066 * config/rs6000/rs6000-c.c: Add support for built-in functions
2067 vector signed int vec_signed (vector float);
2068 vector signed long long vec_signed (vector double);
2069 vector signed int vec_signed2 (vector double, vector double);
2070 vector signed int vec_signede (vector double);
2071 vector signed int vec_signedo (vector double);
2072 * config/rs6000/rs6000.c (rs6000_generate_vsigned2_code): Add
2073 instruction generator.
2074 * config/rs6000/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
2075 UNSPEC_VSX_VSIGNED2): Add UNSPECS.
2076 (vsx_xvcvspsxws, vsx_xvcvdpuxds_scale, vsx_xvcvspuxws, vsigned2_v2df):
2078 (vsignedo_v2df, vsignede_v2df, vunsigned2_v2df, vunsignedo_v2df,
2079 vunsignede_v2df): Add define_expands.
2080 * config/rs6000/rs6000-builtin.def (VEC_SIGNED, VEC_UNSIGNED,
2081 VEC_SIGNED2, VEC_UNSIGNED2, VEC_SIGNEDE, VEC_UNSIGNEDE, VEC_SIGNEDO,
2082 VEC_UNSIGNEDO): Add definitions.
2083 * config/vsx.md (UNSPEC_VSX_XVCVSPSXWS, UNSPEC_VSX_XVCVSPSXDS,
2084 UNSPEC_VSX_VSIGNED2): Add UNSPECs.
2085 (vsx_xvcvspsxws, vsx_xvcvspuxws): Add define_insn.
2086 (vsigned2_v2df, vsigendo_v2df, vsignede_v2df,
2087 vunsigned2_v2df, vunsignedo_v2df, vunsignede_v2df): Add define_expands.
2088 * config/rs6000/altivec.h (vec_signed, vec_signed2,
2089 vec_signede and vec_signedo, vec_unsigned, vec_unsigned2,
2090 vec_unsignede, vec_unsignedo): Add builtin defines.
2091 * config/rs6000-protos.h (rs6000_generate_vsigned2_code): Add extern
2093 * doc/extend.texi: Update the built-in documentation file for the
2094 new built-in functions.
2096 2017-06-29 Richard Biener <rguenther@suse.de>
2098 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Do not add
2099 reduction chains to LOOP_VINFO_REDUCTIONS.
2100 * tree-vect-slp.c (vect_analyze_slp): Continue looking for
2101 SLP reductions after processing reduction chains.
2103 2017-06-29 Nathan Sidwell <nathan@acm.org>
2105 * builtins.c (fold_builtin_FUNCTION): Use
2106 lang_hooks.decl_printable_name.
2108 2017-06-29 Peter Bergner <bergner@vnet.ibm.com>
2111 * cfgexpand.c (expand_gimple_stmt_1): Handle switch statements
2112 with only one label.
2113 * stmt.c (expand_case): Assert NCASES is greater than one.
2115 2017-06-29 Richard Biener <rguenther@suse.de>
2117 * tree-cfg.c (group_case_labels_stmt): Return whether we changed
2119 (group_case_labels): Likewise.
2120 (find_taken_edge): Push sanity checking on val to workers...
2121 (find_taken_edge_cond_expr): ... here
2122 (find_taken_edge_switch_expr): ... and here, handle cases
2123 with just a default label.
2124 * tree-cfg.h (group_case_labels_stmt): Adjust prototype.
2125 (group_case_labels): Likewise.
2126 * tree-cfgcleanup.c (execute_cleanup_cfg_post_optimizing): When
2127 group_case_labels does anything cleanup the CFG again.
2129 2017-06-29 Bin Cheng <bin.cheng@arm.com>
2131 PR tree-optimization/81196
2132 * tree-ssa-loop-niter.c (number_of_iterations_cond): Handle loop
2133 exit condition comparing two IVs.
2135 2017-06-29 Richard Earnshaw <rearnsha@arm.com>
2137 * config/arm/parsecpu.awk (gen_comm_data): Add initializer for
2138 profile to the dummy entry at the end of the list of architectures.
2139 * config/arm/arm-cpu-cdata.h: Regenerated.
2141 2017-06-29 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2142 Michael Collison <michael.collison@arm.com>
2145 * config/aarch64/aarch64.md (*aarch64_<optab>_reg_<mode>3_mask1):
2147 (*aarch64_reg_<mode>3_neg_mask2): New pattern.
2148 (*aarch64_reg_<mode>3_minus_mask): New pattern.
2149 (*aarch64_<optab>_reg_di3_mask2): New pattern.
2150 * config/aarch64/aarch64.c (aarch64_rtx_costs): Account for cost
2151 of shift when the shift amount is masked with constant equal to
2152 the size of the mode.
2153 * config/aarch64/predicates.md (subreg_lowpart_operator): New
2156 2017-06-29 Martin Liska <mliska@suse.cz>
2158 * config/i386/i386.opt: Change range from [1,5] to [0,5].
2160 2017-06-29 Yury Gribov <tetra2005@gmail.com>
2163 * ipa-cp.c (allocate_and_init_ipcp_value): Add initialization
2166 (edge_growth_cache_entry::edge_growth_cache_entry): New
2168 (reset_edge_growth_cache): Update to use constructor.
2170 2017-06-28 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
2172 * config/aarch64/aarch64.h (AARCH64_EXPAND_ALIGNMENT): New.
2173 (DATA_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
2174 (LOCAL_ALIGNMENT): Update to use AARCH64_EXPAND_ALIGNMENT.
2176 2017-06-28 Sebastian Peryt <sebastian.peryt@intel.com>
2178 * config/i386/avx512vlintrin.h (_mm256_permutexvar_epi64)
2179 (_mm256_permutexvar_epi32, _mm256_permutex_epi64): New intrinsics.
2181 2017-06-28 Szabolcs Nagy <szabolcs.nagy@arm.com>
2183 * config.gcc (*-linux-musl*): Add t-musl tmake_file.
2184 (*-linux-uclibc*): Add t-uclibc tmake_file.
2185 * config/t-musl: New.
2186 * config/t-uclibc: New.
2188 2017-06-28 Richard Earnshaw <rearnsha@arm.com>
2190 * config/arm/parsecpu.awk (profile): Parse new keyword in an arch
2192 (gen_comm_data): Emit architectural setting of arch_prof.
2193 * config/arm/arm-cpus.in (armv6-m, armv6s-m, armv7-a, armv7ve): Set the
2195 (armv7-r, armv7-m, armv7e-m, armv8-a, armv8.1-a, armv8.2-a): Likewise.
2196 (armv8-m.base, armv8-m.main): Likewise.
2197 * arm-protos.h (arm_build_target): Add profile field.
2198 (arch_option): Likewise.
2199 * config/arm/arm.c (arm_configure_build_target): Copy the profile to
2201 * config/arm/arm.h (TARGET_ARM_ARCH_PROFILE): Use
2202 arm_active_target.profile.
2204 2017-06-28 Richard Biener <rguenther@suse.de>
2207 * fold-const.c (negate_expr_p): Use TYPE_UNSIGNED, not
2208 TYPE_OVERFLOW_WRAPS.
2209 * match.pd (negate_expr_p): Likewise.
2210 * tree-ssa-reassoc.c (optimize_range_tests_diff): Use
2211 fold_build2, not fold_binary.
2213 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
2215 * config/aarch64/aarch64 (aarch64_expand_mov_immediate):
2216 Convert memory address to Pmode.
2217 (aarch64_print_operand): Assert MEM operands are always Pmode.
2219 2017-06-28 Wilco Dijkstra <wdijkstr@arm.com>
2222 * config/arm/aarch-common.c (arm_no_early_alu_shift_dep):
2223 Remove redundant if.
2224 (aarch_forward_to_shift_is_not_shifted_reg): Remove.
2225 * config/arm/aarch-common-protos.h
2226 (aarch_forward_to_shift_is_not_shifted_re): Remove.
2227 * config/arm/cortex-a53.md: Use arm_no_early_alu_shift_dep in bypass.
2229 2017-06-28 Michael Meissner <meissner@linux.vnet.ibm.com>
2232 * multiple_target.c (create_dispatcher_calls): Set the default
2233 clone to be static, not public.
2235 2017-06-28 Richard Biener <rguenther@suse.de>
2237 * tree-vect-loop.c (vectorizable_reduction): Move special
2238 cond reduction IV var creation ...
2239 (vect_create_epilog_for_reduction): ... here. Remove induction_index
2240 parameter. Use STMT_VINFO_VECTYPE.
2241 * tree-vect-slp.c (vect_get_constant_vectors): Properly reset
2244 2017-06-28 Martin Liska <mliska@suse.cz>
2247 * ipa-visibility.c (non_local_p): Handle visibility.
2249 2017-06-28 Martin Liska <mliska@suse.cz>
2252 * common.opt: Add IntegerRange to various options.
2253 * opt-functions.awk (integer_range_info): New function.
2254 * optc-gen.awk: Add integer_range_info to cl_options struct.
2255 * opts-common.c (decode_cmdline_option): Handle
2256 CL_ERR_INT_RANGE_ARG.
2257 (cmdline_handle_error): Likewise.
2258 * opts.c (print_filtered_help): Show valid interval in
2259 when --help is provided.
2260 * opts.h (struct cl_option): Add range_min and range_max fields.
2261 * config/i386/i386.opt: Add IntegerRange for -mbranch-cost.
2263 2017-06-28 Marc Glisse <marc.glisse@inria.fr>
2265 * match.pd ((X & ~Y) | (~X & Y)): Generalize to + and ^.
2266 (x * C EQ/NE y * C): New transformation.
2268 2017-06-28 Christophe Lyon <christophe.lyon@linaro.org>
2270 * genmultilib (combination_space): Accept '+' in option names.
2272 2017-06-28 Martin Liska <mliska@suse.cz>
2275 * asan.c (instrument_derefs): Bail out inner references
2276 that are hard register variables.
2278 2017-06-28 Jakub Jelinek <jakub@redhat.com>
2281 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Use def_builtin
2282 rather than def_builtin_pure for __builtin_ia32_gatherpf*.
2284 2017-06-28 Richard Biener <rguenther@suse.de>
2286 * tree-vectorizer.h (vect_get_vec_defs): Remove.
2287 (vect_get_slp_defs): Adjust.
2288 * tree-vect-loop.c (get_initial_defs_for_reduction): Split
2290 * tree-vect-slp.c (vect_get_constant_vectors): ... here and
2292 * tree-vect-loop.c (vect_create_epilog_for_reduction): Use
2293 get_initial_defs_for_reduction instead of vect_get_vec_defs.
2294 (vectorizable_reduction): Adjust.
2295 * tree-vect-slp.c (vect_get_constant_vectors): Remove reduction
2297 (vect_get_slp_defs): Likewise.
2298 * tree-vect-stmts.c (vect_get_vec_defs): Make static and adjust.
2299 (vectorizable_bswap): Adjust.
2300 (vectorizable_call): Likewise.
2301 (vectorizable_conversion): Likewise.
2302 (vectorizable_assignment): Likewise.
2303 (vectorizable_shift): Likewise.
2304 (vectorizable_operation): Likewise.
2305 (vectorizable_store): Likewise.
2306 (vectorizable_condition): Likewise.
2307 (vectorizable_comparison): Likewise.
2309 2017-06-28 Michael Collison <michael.collison@arm.com>
2312 * config/arm/arm.c (gen_ldm_seq): Remove last unnecessary
2314 (arm_gen_movmemqi): Removed unused variable 'i'.
2315 Convert 'for' loop into 'while' loop.
2316 (arm_expand_prologue): Remove last unnecessary set of insn.
2317 (thumb_pop): Remove unused variable 'pushed_words'.
2318 (thumb_exit): Remove last unnecessary set of regs_to_pop.
2320 2017-06-28 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
2322 * config/s390/predicates.md: Use s390_rel_address_ok_p.
2323 * config/s390/s390-protos.h: Add prototype of
2324 s390_rel_address_ok_p.
2325 * config/s390/s390.c (s390_got_symbol): New function.
2326 (s390_rel_address_ok_p): New function.
2327 (legitimize_pic_address): Use s390_rel_address_ok_p.
2328 (s390_load_got): Use s390_got_symbol.
2329 (s390_option_override): Issue error if
2330 -mno-pic-data-is-text-relative is used without -fpic/-fPIC.
2331 * config/s390/s390.h (TARGET_DEFAULT_PIC_DATA_IS_TEXT_RELATIVE):
2333 * config/s390/s390.opt: New option mpic-data-is-text-relative.
2335 2017-06-27 Andrew Pinski <apinski@cavium.com>
2337 * match.pd (X >/>=/</<= 0 ? 1.0 : -1.0): New patterns.
2338 (X * copysign (1.0, X)): New pattern.
2339 (X * copysign (1.0, -X)): New pattern.
2340 (copysign (-1.0, CST)): New pattern.
2342 2017-06-27 Joseph Myers <joseph@codesourcery.com>
2344 * genmultilib (combination_space): Remove variable.
2345 Validate reuse rules against regular expression for any sequence
2346 of multilib options in any order.
2348 2017-06-27 Michael Collison <michael.collison@arm.com>
2350 * config/aarch64/aarch64-simd.md (aarch64_combine<mode>): Directly
2351 call aarch64_split_simd_combine.
2352 * (aarch64_combine_internal<mode>): Delete pattern.
2353 * config/aarch64/aarch64.c (aarch64_split_simd_combine):
2354 Allow register and subreg operands.
2356 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
2358 * config/i386/vxworks.h (ASM_SPEC): Remove definition. No target
2359 specific need, just fallback on defaults.
2360 (ASM_OUTPUT_ALIGNED_BSS): Add #undef before #define.
2362 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
2363 Olivier Hainque <hainque@adacore.com>
2365 * config/i386/vxworks.h (DBX_REGISTER_NUMBER): Pick distinct
2367 (TARGET_OS_CPP_BUILTINS): builtin_define CPU to X86_64 for 64bit
2368 targets. Pick a default if no particular attempt applied.
2369 (STACK_CHECK_PROTECT): Double for 64bit targets, which have
2372 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
2374 * config.gcc (i*86-wrs-vxworks7): Handle new acceptable triplet.
2375 (x86_64-wrs-vxworks7): Likewise.
2377 2017-06-27 Marek Polacek <polacek@redhat.com>
2380 * ubsan.c (instrument_null): Check get_base_address's result for null.
2382 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
2384 * match.pd ((A+-B)+(C-A), (A+B)-(A-C)): New transformations.
2386 2017-06-27 Marc Glisse <marc.glisse@inria.fr>
2388 * builtin-types.def (BT_FENV_T_PTR, BT_CONST_FENV_T_PTR,
2389 BT_FEXCEPT_T_PTR, BT_CONST_FEXCEPT_T_PTR): New primitive types.
2390 (BT_FN_INT_FENV_T_PTR, BT_FN_INT_CONST_FENV_T_PTR,
2391 BT_FN_INT_FEXCEPT_T_PTR_INT, BT_FN_INT_CONST_FEXCEPT_T_PTR_INT):
2393 * builtins.def (BUILT_IN_FECLEAREXCEPT, BUILT_IN_FEGETENV,
2394 BUILT_IN_FEGETEXCEPTFLAG, BUILT_IN_FEGETROUND,
2395 BUILT_IN_FEHOLDEXCEPT, BUILT_IN_FERAISEEXCEPT,
2396 BUILT_IN_FESETENV, BUILT_IN_FESETEXCEPTFLAG,
2397 BUILT_IN_FESETROUND, BUILT_IN_FETESTEXCEPT,
2398 BUILT_IN_FEUPDATEENV): New builtins.
2399 * tree-core.h (TI_FENV_T_PTR_TYPE, TI_CONST_FENV_T_PTR_TYPE,
2400 TI_FEXCEPT_T_PTR_TYPE, TI_CONST_FEXCEPT_T_PTR_TYPE): New entries.
2401 * tree.h (fenv_t_ptr_type_node, const_fenv_t_ptr_type_node,
2402 fexcept_t_ptr_type_node, const_fexcept_t_ptr_type_node): New
2404 (builtin_structptr_types): Adjust size.
2405 * tree.c (builtin_structptr_types): Add four entries.
2407 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
2408 Olivier Hainque <hainque@adacore.com>
2410 * config/vxworks.h (VXWORKS_LIB_SPEC): Incorporate ...
2411 (TLS_SYM): New local macro, forcing reference to __tls__ on
2412 link command lines for VxWorks 7 RTPs, triggering initialization
2414 (VXWORKS_HAVE_TLS): New macro. State whether the target VxWorks
2415 OS features TLS support, true for RTPs on VxWorks 7.
2416 * config/vxworks.c (vxworks_override_options): Setup emutls
2419 2017-06-27 Jakub Jelinek <jakub@redhat.com>
2421 * predict.c (test_prediction_value_range): Use -1U instead of -1
2422 to avoid narrowing conversion warning.
2423 * dumpfile.c (dump_options): Wrap all value into dump_flags_t cast
2424 to avoid narrowing conversion warning.
2425 * opt-functions.awk (var_ref): Return (unsigned short) -1 instead of
2427 * optc-gen.awk (END): Expect (unsigned short) -1 instead of -1.
2429 2017-06-27 Jerome Lambourg <lambourg@adacore.com>
2431 * config/vxworks.h (VXWORKS_LIBS_RTP): Alternative definition for
2432 64bit configurations.
2433 (PTR_DIFF_TYPE): Alternative definition for TARGET_LP64.
2434 (SIZE_TYPE): Likewise.
2435 * config/vxworks.c (vxworks_emutls_var_fields): Use
2436 long_unsigned_type_node instead of unsigned_type_node as the offset
2437 field type, which is "pointer" mode in emutls.c.
2439 2017-06-27 Jakub Jelinek <jakub@redhat.com>
2442 * ubsan.c (ubsan_encode_value): Initialize DECL_CONTEXT on var.
2445 * gimple-fold.c (replace_call_with_call_and_fold): Handle
2446 gimple_vuse copying separately from gimple_vdef copying.
2448 2017-06-27 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
2450 * value-prof.c (free_hist): Remove call to memset and the enclosing if
2453 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
2454 Olivier Hainque <hainque@adacore.com>
2456 * config.gcc (*-*-vxworks*): Add TARGET_VXWORKS7=1 to tm_defines
2457 for all vxworks7 targets.
2458 * config/vxworks.h (TARGET_VXWORKS7): If not defined, define to 0.
2459 (VXWORKS_ADDITIONAL_CPP_SPEC): Alternative definition for VXWORKS7.
2460 (VXWORKS_LIBS_RTP, VXWORKS_LIBS_RTP_DIR): New macros, allowing
2461 variations for VX6/VX7 and 32/64bits later on in ...
2462 (VXWORKS_LIB_SPEC): Leverage new macros.
2463 (VXWORKS_OS_CPP_BUILTINS): Define _VSB_CONFIG_FILE for VXWORKS7,
2464 as well as _ALLOW_KEYWORD_MACROS when "inline" is not a keyword.
2466 2017-06-26 Jerome Lambourg <lambourg@adacore.com>
2468 * config/vxworks.h (VXWORKS_OS_CPP_BUILTINS): builtin_define
2469 _VX_TOOL_FAMILY and _VX_TOOL to gnu.
2471 2017-06-26 Carl Love <cel@us.ibm.com>
2473 * config/rs6000/rs6000-c.c: Add support for built-in functions
2474 vector bool char vec_reve (vector bool char);
2475 vector signed char vec_reve (vector signed char);
2476 vector unsigned char vec_reve (vector unsigned char);
2477 vector bool int vec_reve (vector bool int);
2478 vector signed int vec_reve (vector signed int);
2479 vector unsigned int vec_reve (vector unsigned int);
2480 vector bool long long vec_reve (vector bool long long);
2481 vector signed long long vec_reve (vector signed long long);
2482 vector unsigned long long vec_reve (vector unsigned long long);
2483 vector bool short vec_reve (vector bool short);
2484 vector signed short vec_reve (vector signed short);
2485 vector double vec_reve (vector double);
2486 vector float vec_reve (vector float);
2487 * config/rs6000/rs6000-builtin.def (VREVE_V2DI, VREVE_V4SI,
2488 VREVE_V8HI, VREVE_V16QI, VREVE_V2DF, VREVE_V4SF, VREVE): New builtin.
2489 * config/rs6000/altivec.md (UNSPEC_VREVEV): New UNSPEC.
2490 (altivec_vreve): New pattern.
2491 * config/rs6000/altivec.h (vec_reve): New define.
2492 * doc/extend.texi (vec_rev): Update the built-in documentation file
2493 for the new built-in functions.
2495 2016-06-26 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
2497 PR tree-optimization/71815
2498 * gimple-ssa-strength-reduction.c (uses_consumed_by_stmt): New
2500 (find_basis_for_candidate): Call uses_consumed_by_stmt rather than
2502 (slsr_process_phi): Likewise.
2503 (replace_uncond_cands_and_profitable_phis): Don't replace a
2504 multiply candidate with a stride of 1 (copy or cast).
2505 (phi_incr_cost): Call uses_consumed_by_stmt rather than
2507 (lowest_cost_path): Likewise.
2508 (total_savings): Likewise.
2510 2017-06-26 Richard Biener <rguenther@suse.de>
2513 * config/i386/i386.c (ix86_init_mmx_sse_builtins):
2514 Use def_builtin_pure for all gather builtins.
2516 2017-06-26 Richard Biener <rguenther@suse.de>
2518 PR tree-optimization/81203
2519 * tree-tailcall.c (find_tail_calls): Do not move stmts into
2522 2017-06-26 Marek Polacek <polacek@redhat.com>
2525 * doc/invoke.texi: Document -Wmultistatement-macros.
2527 2017-06-26 Christophe Lyon <christophe.lyon@linaro.org>
2529 * doc/sourcebuild.texi (ARM-specific attributes): Document new
2530 arm_neon_ok_no_float_abi effective target.
2532 2017-06-26 Richard Biener <rguenther@suse.de>
2534 PR tree-optimization/80928
2535 * cfghooks.c (duplicate_block): Do not copy BB_DUPLICATED flag.
2536 (copy_bbs): Set BB_DUPLICATED flag early.
2537 (execute_on_growing_pred): Do not execute for BB_DUPLICATED
2539 (execute_on_shrinking_pred): Likewise.
2540 * tree-ssa.c (ssa_redirect_edge): Do not look for PHI args in
2541 BB_DUPLICATED blocks.
2542 * tree-ssa-phionlycoprop.c (eliminate_degenerate_phis_1): Properly
2543 iterate over all PHIs considering removal of *gsi.
2545 2017-06-23 Jim Wilson <jim.wilson@linaro.org>
2547 * doc/invoke.texi (AArch64 Options, -mtune): Re-add falkor and
2550 2017-06-23 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
2552 * config/rs6000/rs6000-string.c: (expand_block_clear,
2553 do_load_for_compare, select_block_compare_mode,
2554 compute_current_alignment, expand_block_compare,
2555 expand_strncmp_align_check, expand_strn_compare,
2556 expand_block_move, rs6000_output_load_multiple)
2557 Move functions related to string/block move/compare
2559 * config/rs6000/rs6000.c: Move above functions to rs6000-string.c.
2560 * config/rs6000/rs6000-protos.h (rs6000_emit_dot_insn): Add prototype
2561 for this function which is now used in two files.
2562 * config/rs6000/t-rs6000: Add rule to compile rs6000-string.o.
2563 * config.gcc: Add rs6000-string.o to extra_objs for
2564 targets powerpc*-*-* and rs6000*-*-*.
2566 2017-06-23 Michael Meissner <meissner@linux.vnet.ibm.com>
2569 * config/rs6000/rs6000.md (ALTIVEC_DFORM): Do not allow DImode in
2570 32-bit, since indexed is not valid for DImode.
2571 (mov<mode>_hardfloat32): Reorder ISA 2.07 load/stores before ISA
2572 3.0 d-form load/stores to be the same as mov<mode>_hardfloat64.
2573 (define_peephole2 for Altivec d-form load): Add 32-bit support.
2574 (define_peephole2 for Altivec d-form store): Likewise.
2577 * multiple_target.c (create_dispatcher_calls): Only create the
2578 dispatcher call if the function is the default clone of a
2581 2017-06-23 Segher Boessenkool <segher@kernel.crashing.org>
2584 * builtins.c (expand_builtin_atomic_fetch_op): If emitting code after
2585 a call, force the call to not be a tail call.
2587 2017-06-23 Jeff Law <law@redhat.com>
2589 * doc/contrib.texi: Add entry for Steven Pemberton's work on
2592 2017-06-23 Will Schmidt <will_schmidt@vnet.ibm.com>
2594 * config/rs6000/rs6000.c: Add include of ssa-propagate.h for
2595 update_call_from_tree(). (rs6000_gimple_fold_builtin): Add
2596 handling for early expansion of vector shifts (sl,sr,sra,rl).
2597 (builtin_function_type): Add vector shift right instructions
2598 to the unsigned argument list.
2600 2017-06-23 Bernd Edlinger <bernd.edlinger@hotmail.de>
2602 rtl-optimizatoin/79286
2603 * ira.c (update_equiv_regs): Revert to using may_trap_or_fault_p again.
2604 * rtlanal.c (rtx_addr_can_trap_p_1): SYMBOL_REF_FUNCTION_P can never
2605 trap. PIC register plus a const unspec without offset can never trap.
2607 2017-06-23 Marc Glisse <marc.glisse@inria.fr>
2609 * tree.h (builtin_structptr_type): New type.
2610 (builtin_structptr_types): Declare new array.
2611 * tree.c (builtin_structptr_types): New array.
2612 (free_lang_data, build_common_tree_nodes): Use it.
2614 2017-06-23 Jonathan Wakely <jwakely@redhat.com>
2617 * doc/invoke.texi (-Wnoexcept-type): Fix name of option, from
2620 2017-06-22 Matt Turner <mattst88@gmail.com>
2622 * config/i386/driver-i386.c (host_detect_local_cpu): Add Kaby
2623 Lake models to skylake case. Assume skylake for unknown
2624 models with clflushopt.
2626 2017-06-22 Jeff Law <law@redhat.com>
2628 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Handle
2629 frame sizes that do not satisfy aarch64_uimm12_shift.
2631 2017-06-22 Jan Hubicka <hubicka@ucw.cz>
2633 * profile-count.h (apply_probability,
2634 apply_scale, probability_in): Fix checks for zero.
2636 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2638 * incpath.c (add_sysroot_to_chain): Allow for $SYSROOT prefix.
2639 * doc/cppdiropts.texi (-I @var{dir}): Document it.
2641 2016-06-22 Richard Biener <rguenther@suse.de>
2643 * tree-vect-loop.c (vect_model_reduction_cost): Handle
2644 COND_REDUCTION and INTEGER_INDUC_COND_REDUCTION without
2645 REDUC_MAX_EXPR support.
2646 (vectorizable_reduction): Likewise.
2647 (vect_create_epilog_for_reduction): Likewise.
2649 2017-06-22 James Greenhalgh <james.greenhalgh@arm.com>
2651 * match.pd (A / (1 << B) -> A >> B): New.
2652 * generic-match-head.c: Include optabs-tree.h.
2653 * gimple-match-head.c: Likewise.
2654 * optabs-tree.h (target_supports_op_p): New.
2655 * optabs-tree.c (target_supports_op_p): New.
2657 2017-06-22 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
2659 * configure.ac (gcc_cv_ld_static_dynamic): Also check stderr for
2660 $gcc_cv_ld --help output.
2661 (gcc_cv_ld_demangle): Likewise.
2662 (gcc_cv_ld_eh_frame_hdr): Likewise.
2663 (gcc_cv_ld_pie): Likewise.
2664 (gcc_cv_ld_as_needed): Likewise. Prefer native forms unless $gnu_ld.
2665 (gcc_cv_ld_buildid): Likewise.
2666 (gcc_cv_ld_sysroot): Likewise.
2667 (ld_bndplt_support): Likewise.
2668 (ld_pushpopstate_support): Likewise.
2669 * configure: Regenerate.
2670 * config/sol2.h [!USE_GLD] (SYSROOT_SPEC): Define.
2672 2017-06-21 Jakub Jelinek <jakub@redhat.com>
2675 * config/i386/sse.md (round<mode>2): Renumber match_dup and
2676 operands indexes to avoid gap between operands and match_dups.
2678 2017-06-21 Andrew Pinski <apinski@cavium.com>
2680 * config/aarch64/aarch64-cost-tables.h (thunderx_extra_costs):
2681 Increment Arith_shift and Arith_shift_reg by 1.
2682 * config/aarch64/aarch64-tuning-flags.def (cheap_shift_extend):
2684 * config/aarch64/aarch64.c (thunderx_tunings): Enable
2685 AARCH64_EXTRA_TUNE_CHEAP_SHIFT_EXTEND.
2686 (aarch64_strip_extend): Add new argument and test for it.
2687 (aarch64_cheap_mult_shift_p): New function.
2688 (aarch64_rtx_mult_cost): Call aarch64_cheap_mult_shift_p and don't
2689 add a cost if it is true.
2690 Update calls to aarch64_strip_extend.
2691 (aarch64_rtx_costs): Update calls to aarch64_strip_extend.
2693 2017-06-21 Andrew Pinski <apinski@cavium.com>
2695 * config/aarch64/aarch64-cores.def (thunderxt88p1): Use thunderxt88
2697 (thunderxt88): Likewise.
2698 * config/aarch64/aarch64.c (thunderxt88_prefetch_tune): New variable.
2699 (thunderx_prefetch_tune): New variable.
2700 (thunderx2t99_prefetch_tune): Update for the correct values.
2701 (thunderxt88_tunings): New variable.
2702 (thunderx_tunings): Use thunderx_prefetch_tune instead of
2703 generic_prefetch_tune.
2704 (thunderx2t99_tunings): Use AUTOPREFETCHER_WEAK.
2706 2017-06-21 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
2708 * config/aarch64/atomics.md (aarch64_compare_and_swap<mode>_lse,
2709 SHORT): Relax operand 3 to aarch64_reg_or_zero and constraint to Z.
2710 (aarch64_compare_and_swap<mode>_lse, GPI): Likewise.
2711 (aarch64_atomic_cas<mode>, SHORT): Likewise for operand 2.
2712 (aarch64_atomic_cas<mode>, GPI): Likewise.
2714 2017-06-21 Martin Liska <mliska@suse.cz>
2716 * gimplify.c (gimplify_label_expr): Insert GIMPLE_PREDICT
2717 statements on cold and hot labels.
2718 * predict.c (tree_estimate_probability_bb): Remove the
2719 prediction from this place.
2721 2017-06-21 Martin Liska <mliska@suse.cz>
2723 PR tree-optimization/79489
2724 * gimplify.c (maybe_add_early_return_predict_stmt): New
2726 (gimplify_return_expr): Call the function.
2727 * predict.c (tree_estimate_probability_bb): Remove handling
2729 * predict.def: Update comment about early return predictor.
2730 * gimple-predict.h (is_gimple_predict): New function.
2731 * predict.def: Change default value of early return to 66.
2732 * tree-tailcall.c (find_tail_calls): Skip GIMPLE_PREDICT
2734 * passes.def: Put pass_strip_predict_hints to the beginning of
2737 2017-06-21 Pierre-Marie de Rodat <derodat@adacore.com>
2739 * dwarf2out.c (gen_decl_die): Remove the guard to skip file-scope
2740 FUNCTION_DECL declarations.
2741 (dwarf2out_early_global_decl): Remove the guard to skip FUNCTION_DECL
2743 (dwaf2out_decl): Likewise.
2744 * godump.c (go_early_global_decl): Skip call to the real debug hook
2745 for FUNCTION_DECL declarations.
2746 * passes.c (rest_of_decl_compilation): Skip call to the
2747 early_global_decl debug hook for FUNCTION_DECL declarations, unless
2748 -fdump-go-spec is passed.
2750 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
2752 * config/i386/i386.c (struct builtin_isa): New field pure_p.
2753 Reorder for compactness.
2754 (def_builtin, def_builtin2, ix86_add_new_builtins): Handle pure_p.
2755 (def_builtin_pure, def_builtin_pure2): New functions.
2756 (ix86_init_mmx_sse_builtins) [__builtin_ia32_stmxcsr]: Mark as pure.
2758 2017-06-21 Marc Glisse <marc.glisse@inria.fr>
2760 * match.pd (nop_convert): New predicate.
2761 ((A +- CST1) +- CST2): Allow some NOP conversions.
2763 2017-06-21 Jakub Jelinek <jakub@redhat.com>
2766 * gimplify.c (omp_add_variable): Don't force GOVD_SEEN for types
2767 with ctors/dtors if GOVD_SHARED is set.
2769 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
2771 * config/aarch64/aarch64.md (movti_aarch64):
2772 Emit mov rather than orr.
2773 (movtf_aarch64): Likewise.
2774 * config/aarch64/aarch64-simd.md (aarch64_simd_mov):
2775 Emit mov rather than orr.
2777 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
2779 * config/aarch64/aarch64-simd.md (aarch64_simd_dup):
2780 Swap alternatives, make integer dup more expensive.
2782 2017-06-21 Wilco Dijkstra <wdijkstr@arm.com>
2784 * config/aarch64/aarch64.c (aarch64_legitimate_constant_p):
2785 Return true for non-tls symbols.
2787 2017-06-21 James Greenhalgh <james.greenhalgh@arm.com>
2789 * config/aarch64/aarch64-cores.def (cortex-a55): New.
2790 (cortex-a75): Likewise.
2791 (cortex-a75.cortex-a55): Likewise.
2792 * config/aarch64/aarch64-tune.md: Regenerate.
2793 * doc/invoke.texi (-mtune): Document new values for -mtune.
2795 2017-06-21 Tom de Vries <tom@codesourcery.com>
2797 * doc/sourcebuild.texi (Add Options, Features for dg-add-options): Add
2799 (Effective-Target Keywords, Other attributes): Suggest using
2800 dg-add-options stack_size feature to get stack limit in stack_size
2801 effective target documentation.
2803 2017-06-21 Julian Brown <julian@codesourcery.com>
2804 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
2806 * config/aarch64/aarch64-simd.md (aarch64_crypto_pmulldi)
2807 (aarch64_crypto_pmullv2di): Change type attribute to crypto_pmull.
2808 * config/aarch64/thunderx2t99.md (thunderx2t99_pmull): New
2810 * config/arm/cortex-a53.md (cortex_a53_advsimd_type): Add crypto_pmull to
2811 attribute type list for neon_multiply.
2812 * config/arm/cortex-a57.md (cortex_a57_neon_type): Add crypto_pmull to
2813 attribute type list for neon_multiply.
2814 * config/arm/crypto.md (crypto_vmullp64): Change type to crypto_pmull.
2815 * config/arm/exynos-m1.md (exynos_m1_neon_type): Add crypto_pmull to
2816 attribute type list for neon_multiply.
2817 * config/arm/types.md (crypto_pmull): Add.
2818 * config/arm/xgene1.md (xgene1_neon_pmull): Add crypto_pmull to
2819 attribute type list.
2821 2017-06-20 Andreas Tobler <andreast@gcc.gnu.org>
2823 * config.gcc (armv6*-*-freebsd*): Change the target_cpu_cname to
2826 2017-06-20 Jakub Jelinek <jakub@redhat.com>
2828 * ira-costs.c (find_costs_and_classes): Initialize cost_classes later
2829 to make sure not to dereference a NULL cost_classes_ptr pointer.
2831 2017-06-20 Carl Love <cel@us.ibm.com>
2833 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
2834 ALTIVEC_BUILTIN_VMULESW, ALTIVEC_BUILTIN_VMULEUW,
2835 ALTIVEC_BUILTIN_VMULOSW, ALTIVEC_BUILTIN_VMULOUW entries.
2836 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin,
2837 builtin_function_type): Add ALTIVEC_BUILTIN_* case statements.
2838 * config/rs6000/altivec.md (MVULEUW, VMULESW, VMULOUW,
2839 VMULOSW): New enum "unspec" values.
2840 (vec_widen_umult_even_v4si, vec_widen_smult_even_v4si,
2841 vec_widen_umult_odd_v4si, vec_widen_smult_odd_v4si,
2842 altivec_vmuleuw, altivec_vmulesw, altivec_vmulouw,
2843 altivec_vmulosw): New patterns.
2844 * config/rs6000/rs6000-builtin.def (VMLEUW, VMULESW, VMULOUW,
2845 VMULOSW): Add definitions.
2847 2017-06-20 Julia Koval <julia.koval@intel.com>
2849 * config/i386/i386.c: Fix rounding expand for new pattern.
2850 * config/i386/subst.md: Fix pattern (parallel -> unspec).
2852 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
2854 * config/aarch64/aarch64-option-extensions.def (rcpc): New.
2855 * config/aarch64/aarch64.h (AARCH64_FL_RCPC): New.
2857 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
2859 * config/aarch64/aarch64-option-extensions.def (fp16): Fix expected
2862 2017-06-20 James Greenhalgh <james.greenhalgh@arm.com>
2864 * config/aarch64/aarch64-cores.def: Rearrange to sort by
2865 architecture, then by implementer ID.
2866 * config/aarch64/aarch64-tune.md: Regenerate.
2868 2017-06-20 Richard Biener <rguenther@suse.de>
2871 * fold-const.c (split_tree): Fold to type before negating.
2873 2017-06-20 David Malcolm <dmalcolm@redhat.com>
2875 * diagnostic-show-locus.c
2876 (selftest::test_fixit_deletion_affecting_newline): New function.
2877 (selftest::diagnostic_show_locus_c_tests): Call it.
2879 2017-06-20 Andreas Schwab <schwab@suse.de>
2882 * config/m68k/m68k.md (bsetdreg, bchgdreg, bclrdreg): Use "=d"
2885 2017-06-20 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
2887 * config/arm/arm-c.c (arm_cpu_builtins): New block to define
2888 __ARM_FEATURE_COPROC according to support.
2890 2017-06-20 Jakub Jelinek <jakub@redhat.com>
2892 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
2893 Rewritten to avoid overflow for > 32-bit pointers.
2896 * ubsan.h (ubsan_encode_value): Workaround buggy clang++ parser
2897 by removing enum keyword.
2898 (ubsan_type_descriptor): Likewise. Formatting fix.
2901 * config/i386/i386.md (TARGET_USE_VECTOR_CONVERTS float si->{sf,df}
2902 splitter): Require TARGET_SSE2 in the condition.
2904 2017-06-20 Michael Meissner <meissner@linux.vnet.ibm.com>
2907 * config/rs6000/rs6000.c (rs6000_expand_vector_init): Add support
2908 for doing vector set of SFmode on ISA 3.0.
2909 * config/rs6000/vsx.md (vsx_set_v4sf_p9): Likewise.
2910 (vsx_set_v4sf_p9_zero): Special case setting 0.0f to a V4SF
2912 (vsx_insert_extract_v4sf_p9): Add an optimization for inserting a
2913 SFmode value into a V4SF variable that was extracted from another
2914 V4SF variable without converting the element to double precision
2915 and back to single precision vector format.
2916 (vsx_insert_extract_v4sf_p9_2): Likewise.
2918 2017-06-19 Jakub Jelinek <jakub@redhat.com>
2920 * tree-ssa-structalias.c (get_constraint_for_ptr_offset): Multiply
2921 in UWHI to avoid undefined overflow.
2924 * ubsan.h (enum ubsan_encode_value_phase): New.
2925 (ubsan_encode_value): Change second argument to
2926 enum ubsan_encode_value_phase with default value of
2927 UBSAN_ENCODE_VALUE_GENERIC.
2928 * ubsan.c (ubsan_encode_value): Change second argument to
2929 enum ubsan_encode_value_phase PHASE from bool IN_EXPAND_P,
2930 adjust uses, for UBSAN_ENCODE_VALUE_GENERIC use just
2931 create_tmp_var_raw instead of create_tmp_var and use a
2933 (ubsan_expand_bounds_ifn, ubsan_build_overflow_builtin,
2934 instrument_bool_enum_load, ubsan_instrument_float_cast): Adjust
2935 ubsan_encode_value callers.
2938 * ubsan.c (ubsan_encode_value): If current_function_decl is NULL,
2939 use create_tmp_var_raw instead of create_tmp_var, mark it addressable
2940 just by setting TREE_ADDRESSABLE on the result and use a TARGET_EXPR.
2942 2017-06-19 Richard Biener <rguenther@suse.de>
2945 * tree-cfgcleanup.c (cleanup_tree_cfg_noloop): Clear niter
2946 estimates if we changed anything.
2948 2017-06-19 Richard Biener <rguenther@suse.de>
2950 PR tree-optimization/80887
2951 * tree-ssa-sccvn.c (mprts_hook_cnt): New global.
2952 (vn_lookup_simplify_result): Allow only mprts_hook_cnt succesful
2953 simplified lookups, then reset mprts_hook.
2954 (vn_nary_build_or_lookup_1): Set mprts_hook_cnt to 9 before
2956 (try_to_simplify): Likewise.
2958 2017-06-19 Martin Liska <mliska@suse.cz>
2961 * gimplify.c (gimplify_switch_expr):
2962 Initialize live_switch_vars for SWITCH_BODY == STATEMENT_LIST.
2964 2017-06-19 Martin Liska <mliska@suse.cz>
2966 * doc/install.texi: Document that PGO runs in 4 stages.
2968 2017-06-19 Martin Liska <mliska@suse.cz>
2971 * attribs.c (make_dispatcher_decl): Do not append '.ifunc'
2972 to dispatcher function name.
2973 * multiple_target.c (replace_function_decl): New function.
2974 (create_dispatcher_calls): Redirect both edges and references.
2976 2017-06-19 Jan Hubicka <hubicka@ucw.cz>
2978 * profile-count.c (profile_count::dump): Dump quality.
2979 (profile_count::differs_from_p): Update for unsigned val.
2980 * profile-count.h (profile_count_quality): New enum.
2981 (profile_count): Turn m_val to 62bit unsigned, add quality tracking.
2983 2017-06-19 Richard Biener <rguenther@suse.de>
2985 * tree-ssa-loop-niter.h (estimate_numbers_of_iterations): Take
2986 struct function as arg.
2987 (estimate_numbers_of_iterations): Export overload with loop arg.
2988 (free_numbers_of_iterations_estimates_loop): Use an overload of
2989 free_numbers_of_iterations_estimates instead.
2990 * tree-cfg.c (remove_bb): Adjust.
2991 * tree-cfgcleanup.c (remove_forwarder_block_with_phi): Likewise.
2992 * tree-parloops.c (gen_parallel_loop): Likewise.
2993 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables):
2995 (tree_unroll_loops_completely): Likewise.
2996 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop):
2997 Use an overload instead and export.
2998 (estimated_loop_iterations): Adjust.
2999 (max_loop_iterations): Likewise.
3000 (likely_max_loop_iterations): Likewise.
3001 (estimate_numbers_of_iterations): Take struct function as arg
3003 (loop_exits_before_overflow): Adjust.
3004 (free_numbers_of_iterations_estimates_loop): Use an overload.
3005 * tree-vect-loop.c (vect_analyze_loop_form): Adjust.
3006 * tree-vectorizer.c (vect_free_loop_info_assumptions): Likewise.
3008 2017-06-19 Richard Biener <rguenther@suse.de>
3011 * ipa-prop.c (find_constructor_constant_at_offset): Handle
3012 RANGE_EXPR conservatively.
3014 2017-06-16 Carl Love <cel@us.ibm.com>
3016 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
3017 definitions for vec_float, vec_float2, vec_floato,
3018 vec_floate built-ins.
3019 * config/rs6000/vsx.md (define_c_enum "unspec"): Add RTL code
3020 for instructions vsx_xvcvsxws vsx_xvcvuxwsp, float2, floato and
3022 * config/rs6000/rs6000-builtin.def (FLOAT2_V2DI, FLOATE_V2DF,
3023 FLOATE_2DI, FLOATO_V2DF, FLOATEE_V2DI, XVCVSXWSP_V4SF,
3024 UNS_FLOATO_V2DI, UNS_FLOATE_V2DI): Add definitions.
3025 * config/altivec.md (define_insn "p8_vmrgew_<mode>",
3026 define_mode_attr VF_sxddp): Add V4SF type to p8_vmrgew.
3027 * config/rs6000/altivec.h (vec_float, vec_float2, vec_floate,
3028 vec_floato): Add builtin defines.
3029 * doc/extend.texi (vec_float, vec_float2, vec_floate, vec_floato):
3030 Update the built-in documentation file for the new built-in
3033 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3035 * config/arm/arm.opt (marm): Mark as the negative of of -mthumb.
3036 (mthumb): Mark as the negative of -marm.
3038 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3040 * doc/invoke.texi (ARM Options, -mcpu): Document supported
3042 (ARM Options, -mtune): Document that this accepts the same
3043 extension options as -mcpu.
3044 (ARM Options, -mfpu): Document addition of -mfpu=auto.
3046 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3048 * doc/invoke.texi (ARM Options, -march=): Document new syntax and
3049 permitted extensions.
3051 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3053 * config/arm/arm-cpus.in (armv7): Add extension +nofp.
3054 (armv7-r): Add aliases vfpv3xd and vfpv3-d16.
3055 (armv8-m.main): Add option +nodsp.
3056 * config/arm/arm-cpu-cdata.h: Regenerated.
3058 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3060 * config/arm/t-fuchsia: New file.
3061 * config.gcc (arm*-*-fuchsia*): Use it.
3063 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3065 * config/arm/t-symbian: Rewrite for new option infrastructure.
3067 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3069 * config/arm/t-phoenix (MULTILIB_REUSE): Clear variable.
3070 (MULTILIB_REQUIRED): Likewise.
3072 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3074 * config/arm/t-linux-eabi (MULTILIB_EXCEPTIONS): Set to empty.
3075 (MULTILIB_RESUE): Likewise.
3076 (MULTILIB_MATCHES): Likewise.
3077 (MULTLIB_REQUIRED): Likewise.
3079 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3081 * config/arm/t-rtems: Rewrite for new option framework.
3083 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3085 * config/arm/t-aprofile (v7_a_nosimd_variants, v7_a_simd_variants)
3086 (v7ve_nosimd_variatns, v7ve_vfpv3_simd_variants)
3087 (v7ve_vfpv4_simd_variants, v8_a_nosimd_variants, v8_a_simd_variants)
3088 (v8_1_a_simd_variants, v8_2_a_simd_variants): Move to ...
3089 * config/arm/t-multilib: ... here.
3090 (MULTILIB_OPTIONS): Add armv7 and armv7+fp architectures.
3091 (MULTILIB_MATCHES): Use armv7 libraries for armv7-r. Also use for
3092 armv7-a and armv8*-a when A-profile libraries have not been built.
3093 * config/arm/t-rmprofile: Rewrite.
3095 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3097 * genmultilib (multilib_reuse): Allow an explicit period to be escaped
3098 with a backslash. Remove the backslash after substituting unescaped
3100 * doc/fragments.texi (MULTILIB_REUSE): Document it.
3102 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3104 * config.gcc: (arm*-*-*): When building a-profile libraries, force
3105 the driver to pass through the default setting of -mfloat-abi.
3106 * common/config/arm/arm-common.c (arm_target_thumb_only): Return -marm
3108 * config/arm/t-multilib (MULTILIB_REUSE): Initialize to empty.
3109 (all_feat_combs): New rule.
3110 (MULTILIB_OPTIONS): Use explicit ARM and Thumb directories. Rework
3112 * config/arm/t-aprofile: Rewrite.
3114 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3116 * config/arm/arm.h (FPUTYPE_AUTO): Define.
3117 * config/arm/arm.c (arm_option_override): Use FPUTYPE_AUTO if the
3118 fpu is not specified by the user/command-line.
3119 * config/arm/bpabi.h (FPUTYPE_DEFAULT): Delete.
3120 * config/arm/netbsd-elf.h (FPUTYPE_DEFAULT): Delete.
3121 * config/arm/linux-elf.h (FPUTYPE_DEFAULT): Delete.
3122 * config/arm/vxworks.h (FPUTYPE_DEFAULT): Delete.
3123 * common/config/arm/arm-common.c (arm_canon_arch_option): Use
3124 FPUTYPE_AUTO insted of FPUTYPE_DEFAULT.
3126 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3128 * config/arm/elf.h (MULTILIB_DEFAULTS): Delete.
3129 * config/arm/t-arm-elf: Rewritten.
3131 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3133 * config/arm/arm.h (TARGET_HARD_FLOAT): Also check that we
3134 have some floating-point instructions.
3135 (TARGET_SOFT_FLOAT): Define as inverse of TARGET_HARD_FLOAT.
3136 (TARGET_MAYBE_HARD_FLOAT): New macro.
3137 * config/arm/arm-builtins.c (arm_init_builtins): Use
3138 TARGET_MAYBE_HARD_FLOAT.
3139 * config/arm/arm.c (arm_option_override): Use TARGET_HARD_FLOAT_ABI.
3141 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3143 * common/config/arm/arm-common.c: Define INCLUDE_LIST.
3144 (configargs.h): Include it.
3145 (arm_print_hint_for_fpu_option): New function.
3146 (arm_parse_fpu_option): New function.
3147 (candidate_extension): New class.
3148 (arm_canon_for_multilib): New function.
3149 * config/arm/arm.h (CANON_ARCH_SPEC_FUNCTION): New macro.
3150 (EXTRA_SPEC_FUNCTIONS): Add CANON_ARCH_SPEC_FUNCTION.
3151 (ARCH_CANONICAL_SPECS): New macro.
3152 (DRIVER_SELF_SPECS): Add ARCH_CANONICAL_SPECS.
3154 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3156 * config.gcc (arm*-*-*): Ensure both target_cpu_cname and with_cpu
3157 are set after handling multilib fragments. Set target_cpu_default2
3160 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3162 * config.gcc (arm*-*-fucshia*): Set target_cpu_cname to the real
3164 (arm*-*-*): Set target_cpu_default2 to a quoted string.
3165 * config/arm/parsecpu.awk (check_cpu): Validate any extension
3167 (check_arch): Likewise.
3168 * config/arm/arm.c (arm_configure_build_target): Handle
3169 TARGET_CPU_DEFAULT being a string constant. Scan any feature
3170 options in the default.
3172 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3174 * config/arm/arm-protos.h (cpu_arch_extension): Add field to record
3175 when an option is an alias of another.
3176 * config/arm/parsecpu.awk (optalias): New parser token.
3177 (gen_comm_data): Mark non-alias options as such. Emit entries
3178 for extension aliases.
3179 * config/arm/arm-cpus.in (armv5e): Make vfpv2 an alias.
3180 (armv5te, armv5tej, armv6, armv6j, armv6k, armv6z): Likewise.
3181 (armv6kz, armv6zk, armv6t2): Likewise.
3182 (armv7): Make vfpv3-d16 an alias.
3183 (armv7-a): Make vfpv3-d16, neon and neon-vfpv3 aliases. Sort in
3185 (armv7ve): Make vfpv4-d16, neon-vfpv3 and neon-vfpv4 aliases.
3186 Sort in canonical order.
3187 (armv8-a): Sort in canonical order.
3188 (armv8.1-a, armv8.2-a): Likewise.
3189 (generic-armv7-a): Make neon and neon-vfpv3 aliases. Sort in
3191 (cortex-a9): Sort in canonical order.
3192 * config/arm/arm.c (selftests.h): Include it.
3193 (arm_test_cpu_arch_data): New function.
3194 (arm_run_self_tests): New function.
3195 (TARGET_RUN_TARGET_SELFTESTS): Redefine.
3196 (targetm): Move declaration to the end of the file.
3197 * arm-cpu-cdata.h: Regenerated.
3199 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3201 * config/arm/arm.h (TARGET_MODE_SPECS): Add additional parameter to
3202 call to target_mode_check describing the type of option passed.
3203 * common/config/arm/arm-common.c (arm_arch_core_flag): Delete.
3204 (arm_target_thumb_only): Use arm_parse_arch_option_name or
3205 arm_parse_cpu_option_name to match parameters against list of
3207 * config/arm/parsecpu.awk (gen_comm_data): Don't generate
3208 arm_arch_core_flags data structure.
3209 * config/arm/arm-cpu_cdata.h: Regenerated.
3211 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3213 * common/config/arm/arm-common.c (arm_initialize_isa): Moved here from
3215 (arm_print_hint_for_cpu_option): Likewise.
3216 (arm_print_hint_for_arch_option): Likewise.
3217 (arm_parse_cpu_option_name): Likewise.
3218 (arm_parse_arch_option_name): Likewise.
3219 * config/arm/arm.c (arm_identify_fpu_from_isa): Use the computed number
3220 of entries in the all_fpus list.
3221 * config/arm/arm-protos.h (all_architectures, all_cores): Declare.
3222 (arm_parse_cpu_option_name): Declare.
3223 (arm_parse_arch_option_name): Declare.
3224 (arm_parse_option_features): Declare.
3225 (arm_intialize_isa): Declare.
3226 * config/arm/parsecpu.awk (gen_data): Move CPU and architecture
3228 (gen_comm_data): ... here. Make definitions non-static.
3229 * config/arm/arm-cpu-data.h: Regenerated.
3230 * config/arm/arm-cpu-cdata.h: Regenerated.
3232 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3234 * config/arm/arm-protos.h (arm_build_target): Remove arch_core.
3235 (cpu_arch_extension): New structure.
3236 (cpu_arch_option, arch_option, cpu_option): New structures.
3237 * config/arm/parsecpu.awk (gen_headers): Build an enumeration of
3239 (gen_data): Generate new format data tables.
3240 * config/arm/arm.c (cpu_tune): New structure.
3241 (cpu_option, processors): Delete.
3242 (arm_print_hint_for_core_or_arch): Delete. Replace with ...
3243 (arm_print_hint_for_cpu_option): ... this and ...
3244 (arm_print_hint_for_arch_option): ... this.
3245 (arm_parse_arch_cpu_name): Delete. Replace with ...
3246 (arm_parse_cpu_option_name): ... this and ...
3247 (arm_parse_arch_option_name): ... this.
3248 (arm_unrecognized_feature): Change type of target parameter to
3250 (arm_parse_arch_cpu_features): Delete. Replace with ...
3251 (arm_parse_option_features): ... this.
3252 (arm_configure_build_target): Rework to use new configuration data
3254 (arm_print_tune_info): Rework for new configuration data tables.
3255 * config/arm/arm-cpu-data.h: Regenerated.
3256 * config/arm/arm-cpu.h: Regenerated.
3258 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3260 * Makefile.in (OBJS): Move sbitmap.o from here ...
3261 (OBJS-libcommon): ... to here.
3263 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3265 * config/arm/arm-isa.h (ISA_ALL_FPU_INTERNAL): Renamed from ISA_ALL_FPU.
3266 (ISA_ALL_CRYPTO): New macro.
3267 (ISA_ALL_SIMD): New macro
3268 (ISA_ALL_FP): New macro.
3269 * config/arm/arm.c (fpu_bitlist): Update initializer.
3270 * config/arm/arm-cpus.in: Use new ISA_ALL macros to disable crypto,
3272 (arm9e): Add fpu. Add option for nofp
3273 (arm946e-s, arm966e-s, arm968e-s, arm10e, arm1020e, arm1022e): Likewise.
3274 (arm926ej-s, arm1026ej-s): Likewise.
3275 (generic-armv7-a): Add fpu. Add options for simd, vfpv3, vfpv3-d16,
3276 vfpv3-fp16, vfpv3-d16-fp16, vfpv4, vfpv4-d16, neon, neon-vfp3,
3277 neon-fp16, neon-vfpv4, nofp and nosimd.
3278 (cortex-a5, cortex-a7): Add fpu. Add options for nosimd and nofp.
3279 (cortex-a8): Add fpu. Add option for nofp.
3280 (cortex-a9): Add fpu. Add options for nosimd and nofp.
3281 (cortex-a12, cortex-a15, cortex-a17): Add fpu. Add option for nofp.
3282 (cortex-r4f): Add fpu.
3283 (cortex-r5): Add fpu. Add options for nofp.dp and nofp.
3284 (cortex-r7): Use idiv option from architecture. Add fpu. Add option
3286 (cortex-r8): Likewise.
3287 (cortex-m4): Add fpu. Add option for nofp.
3288 (cortex-a15.cortex-a7): Add fpu. Add option for nofp.
3289 (cortex-a17.cortex-a7): Likewise.
3290 (cortex-a32): Add fpu. Add options for crypto and nofp.
3291 (cortex-a35, cortex-a53): Likewise.
3292 (cortex-a57): Add fpu. Add option for crypto.
3293 (cortex-a72, cortex-a73): Likewise.
3294 (exynos-m1): Likewise.
3295 (cortex-a57.cortex-a53, cortex-a72.cortex-a53): Likewise.
3296 (cortex-a73.cortex-a35, cortex-a73.cortex-a53): Likewise.
3297 (cortex-m33): Add fpu. Add option for nofp.
3298 * config/arm/arm-cpu-cdata.h: Regenerated
3299 * config/arm/arm-cpu-data.h: Regenerated.
3301 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3303 * arm-cpus.in (armv5e): Add options fp, vfpv2 and nofp.
3304 (armv5te, armv5tej): Likewise.
3305 (armv6, armv6j, armv6k, armv6z, armv6kz, armv6zk, armv6t2): Likewise.
3306 (armv7): Add options fp and vfpv3-d16.
3307 (armv7-a): Add options fp, simd, vfpv3, vfpv3-d16, vfpv3-d16-fp16,
3308 vfpv3-fp16, vfpv4, vfpv4-d16, neon, neon-vfpv3, neon-fp16, neon-vfpv4,
3310 (armv7ve): Likewise.
3311 (armv7-r): Add options fp, fp.sp, idiv, nofp and noidiv.
3312 (armv7e-m): Add options fp, fpv5, fp.dp and nofp.
3313 (armv8-a): Add nocrypto option.
3314 (armv8.1-a, armv8.2-a): Likewise.
3315 (armv8-m.main): add options fp, fp.dp and nofp.
3317 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3319 * config/arm/arm-cpus.in (armv8-a): Add options crc, simd crypto and
3321 (armv8-a+crc): Delete.
3322 (armv8.1-a): Add options simd, crypto and nofp.
3323 (armv8.2-a): Add options fp16, simd, crypto and nofp.
3324 (armv8.2-a+fp16): Delete.
3325 (armv8-m.main): Add option dsp.
3326 (armv8-m.main+dsp): Delete.
3327 (cortex-a8): Add fpu. Add nofp option.
3328 (cortex-a9): Add fpu. Add nofp and nosimd options.
3329 * config/arm/parsecpu.awk (gen_data): Generate option tables and
3330 link to main cpu and architecture data structures.
3331 (gen_comm_data): Only put isa attributes from the main architecture
3333 (option): New statement for architecture and CPU entries.
3334 * arm.c (struct cpu_option): New structure.
3335 (struct processors): Add entry for options.
3336 (arm_unrecognized_feature): New function.
3337 (arm_parse_arch_cpu_name): Ignore any characters after the first
3339 (arm_parse_arch_cpu_feature): New function.
3340 (arm_configure_build_target): Separate out any CPU and architecture
3341 features and parse separately. Don't error out if -mfpu=auto is
3342 used with only an architecture string.
3343 (arm_print_asm_arch_directives): New function.
3344 (arm_file_start): Call it.
3345 * config/arm/arm-cpu-cdata.h: Regenerated.
3346 * config/arm/arm-cpu-data.h: Likewise.
3347 * config/arm/arm-tables.opt: Likewise.
3349 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3351 * config/arm/elf.h (ASM_SPEC): Only pass -mfpu through to the
3352 assembler when it is not -mfpu=auto.
3354 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3356 * config/arm/arm.h (BIG_LITTLE_SPEC): Delete macro.
3357 (ASM_REWRITE_SPEC_FUNCTIONS): New macro.
3358 (BIG_LITTLE_CPU_SPEC_FUNCTIONS): Delete macro.
3359 (ASM_CPU_SPEC): Rewrite.
3360 (MCPU_MTUNE_NATIVE_FUNCTIONS): New macro.
3361 (EXTRA_SPEC_FUNCTIONS): Move outside of ifdef. Use
3362 MCPU_MTUNE_NATIVE_FUNCTIONS and ASM_REWRITE_SPEC_FUNCTIONS. Remove
3363 reference to BIG_LITTLE_CPU_SPEC_FUNCTIONS.
3364 * common/config/arm/arm-common.c (arm_rewrite_selected_cpu): Ensure
3365 copied string is NUL-terminated. Also strip any characters prefixed
3367 (arm_rewrite_selected_arch): New function.
3368 (arm_rewrite_march): New function.
3370 2017-06-16 Richard Earnshaw <rearnsha@arm.com>
3372 * config/arm/arm.opt (x_arm_arch_string): New TargetSave option.
3373 (x_arm_cpu_string, x_arm_tune_string): Likewise.
3374 (march, mcpu, mtune): Convert to string-based options.
3375 * config/arm/arm.c (arm_print_hint_for_core_or_arch): New function.
3376 (arm_parse_arch_cpu_name): New function.
3377 (arm_configure_build_target): Use arm_parse_arch_cpu_name to
3378 identify selected architecture or CPU.
3379 (arm_option_save): New function.
3380 (TARGET_OPTION_SAVE): Redefine.
3381 (arm_option_restore): Restore string options.
3382 (arm_option_print): Print string options.
3384 2017-06-16 Martin Sebor <msebor@redhat.com>
3386 PR tree-optimization/80933
3387 PR tree-optimization/80934
3388 * builtins.c (fold_builtin_3): Do not handle bcmp here.
3389 * gimple-fold.c (gimple_fold_builtin_bcmp): New function.
3390 (gimple_fold_builtin_bcopy, gimple_fold_builtin_bzero): Likewise.
3391 (gimple_fold_builtin): Call them.
3393 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
3395 * gimple-ssa-isolate-paths.c (isolate_path): Set edge leading to path
3396 as unlikely; update profile.
3398 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
3400 * predict.c (force_edge_cold): Handle declaring edges impossible
3403 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
3405 * tree-ssa-loop-ivcanon.c (remove_exits_and_undefined_stmts): Update
3407 (try_unroll_loop_completely): Fix reporting.
3409 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
3411 * tree-ssa-tail-merge.c (replace_block_by): Fix profile updating.
3413 2017-06-16 James Greenhalgh <james.greenhalgh@arm.com>
3416 * config/arm/arm-builtins.c (arm_expand_builtin_args): Return TARGET
3417 if given a non-constant argument for an intrinsic which requires a
3420 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
3422 * profile.c (compare_freqs): New function.
3423 (branch_prob): Sort edge list.
3424 (find_spanning_tree): Assume that the list is priority sorted.
3426 2017-06-16 Richard Biener <rguenther@suse.de>
3428 PR tree-optimization/81090
3429 * passes.def (pass_record_bounds): Remove.
3430 * tree-pass.h (make_pass_record_bounds): Likewise.
3431 * tree-ssa-loop.c (pass_data_record_bounds, pass_record_bounds,
3432 make_pass_record_bounds): Likewise.
3433 * tree-ssa-loop-ivcanon.c (canonicalize_induction_variables): Do
3434 not free niter estimates at the beginning but at the end.
3435 * tree-scalar-evolution.c (scev_finalize): Free niter estimates.
3437 2017-06-16 Richard Biener <rguenther@suse.de>
3439 * tree-switch-conversion.c (emit_case_bit_tests): Adjust
3440 initializer to workaround ICE in host GCC 4.8.
3442 2017-06-16 Jan Hubicka <hubicka@ucw.cz>
3444 * ipa-inline-transform.c (update_noncloned_frequencies): Update also
3446 (clone_inlined_nodes): Update.
3448 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
3450 * config/aarch64/aarch64.c (qdf24xx_prefetch_tune): Update
3451 prefetch settings, and enable prefetching by default at -O3.
3453 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
3455 * config/aarch64/aarch64.c (aarch64_override_options_internal):
3456 Set flag_prefetch_loop_arrays according to tuning data.
3458 2017-06-16 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
3460 * config/aarch64/aarch64-protos.h (struct cpu_prefetch_tune):
3462 (struct tune_params): Use cpu_prefetch_tune instead of cache_line_size.
3463 [Unrelated to main purpose of the patch] Place the pointer field last
3464 to enable type checking errors when tune structure are wrongly merged.
3465 * config/aarch64/aarch64.c (generic_prefetch_tune,)
3466 (exynosm1_prefetch_tune, qdf24xx_prefetch_tune,)
3467 (thunderx2t99_prefetch_tune): New tune constants.
3468 (tune_params *_tunings): Update all tunings (no functional change).
3469 (aarch64_override_options_internal): Set PARAM_SIMULTANEOUS_PREFETCHES,
3470 PARAM_L1_CACHE_SIZE, PARAM_L1_CACHE_LINE_SIZE, and PARAM_L2_CACHE_SIZE
3471 from tunings structures.
3473 2017-06-16 Jakub Jelinek <jakub@redhat.com>
3476 * ubsan.c (instrument_null): Add T argument, use it instead
3477 of computing it based on IS_LHS.
3478 (instrument_object_size): Likewise.
3479 (pass_ubsan::execute): Adjust instrument_null and
3480 instrument_object_size callers to pass gimple_get_lhs or
3481 gimple_assign_rhs1 result to it. Use instrument_null instead of
3482 calling get_base_address and instrument_mem_ref. Handle
3483 aggregate call arguments for object-size sanitization.
3485 2017-06-16 Yury Gribov <tetra2005@gmail.com>
3487 PR tree-optimization/81089
3488 * tree-vrp.c (is_masked_range_test): Validate operands of
3491 2017-06-15 Martin Sebor <msebor@redhat.com>
3494 * dumpfile.c (dump_register): Avoid calling memset to initialize
3495 a class with a default ctor.
3496 * gcc.c (struct compiler): Remove const qualification.
3497 * genattrtab.c (gen_insn_reserv): Replace memset with initialization.
3498 * hash-table.h: Ditto.
3499 * ipa-cp.c (allocate_and_init_ipcp_value): Replace memset with
3501 * ipa-prop.c (ipa_free_edge_args_substructures): Ditto.
3502 * omp-low.c (lower_omp_ordered_clauses): Replace memset with
3504 * params.h (struct param_info): Make struct members non-const.
3505 * tree-switch-conversion.c (emit_case_bit_tests): Replace memset
3506 with default initialization.
3507 * vec.h (vec_copy_construct, vec_default_construct): New helper
3509 (vec<T>::copy, vec<T>::splice, vec<T>::reserve): Replace memcpy
3510 with vec_copy_construct.
3511 (vect<T>::quick_grow_cleared): Replace memset with default ctor.
3512 (vect<T>::vec_safe_grow_cleared, vec_safe_grow_cleared): Same.
3513 * doc/invoke.texi (-Wclass-memaccess): Document.
3515 2017-06-15 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
3517 * emit-rtl.h (is_leaf): Update comment about local
3520 2017-06-15 Jozef Lawrynowicz <jozef.l@somniumtech.com>
3523 * config/msp430/msp430.c (msp430_data_attr): Check that it's possible
3524 for a variable to have a section before checking if the section has a
3526 Set section to.persistent if persistent attribute is set.
3527 Warn if .persistent attribute is used on an automatic variable.
3529 2017-06-15 Eric Botcazou <ebotcazou@adacore.com>
3531 PR rtl-optimization/80474
3532 * reorg.c (update_block): Do not ignore instructions in a delay slot.
3534 2017-06-15 Segher Boessenkool <segher@kernel.crashing.org>
3536 * config/rs6000/rs6000.md (add<mode>3): Use reg_or_subregno instead
3539 2017-06-14 Maciej W. Rozycki <macro@imgtec.com>
3541 * config/mips/mips.md (MIPS16_T_REGNUM): Remove constant.
3542 (casesi): Emit bounds checking as RTL.
3543 (casesi_internal_mips16_<mode>): Remove bounds checking.
3545 2017-06-14 Max Filippov <jcmvbkbc@gmail.com>
3547 * config/xtensa/xtensa.c (xtensa_option_override): Append
3548 MASK_CONST16 to target_flags in the absence of TARGET_L32R.
3549 (hwloop_optimize, hwloop_fail, hwloop_pattern_reg,
3550 xtensa_doloop_hooks): Define unconditionally.
3551 (xtensa_reorg_loops): Only call reorg_loops in the presence of
3553 * config/xtensa/xtensa.h (TARGET_L32R): New definition.
3554 (TARGET_DEFAULT): Remove XCHAL_HAVE_L32R condition and account
3555 for it in xtensa_option_override.
3556 (HARD_FRAME_POINTER_IS_FRAME_POINTER,
3557 HARD_FRAME_POINTER_IS_ARG_POINTER): New definitions.
3559 2017-06-14 Boris Kolpackov <boris@codesynthesis.com>
3561 * doc/cppopts.texi: Document '-' special value to -MF.
3563 2017-06-14 Wilco Dijkstra <wdijkstr@arm.com>
3565 * config/arm/cortex-a53.md (cortex_a53_fpalu) Adjust latency.
3566 (cortex_a53_fconst): Likewise.
3567 (cortex_a53_fpmul): Likewise.
3568 (cortex_a53_f_load_64): Likewise.
3569 (cortex_a53_f_load_many): Likewise.
3570 (cortex_a53_advsimd_alu): Likewise.
3571 (cortex_a53_advsimd_alu_q): Likewise.
3572 (cortex_a53_advsimd_mul): Likewise.
3573 (cortex_a53_advsimd_mul_q): Likewise.
3574 (fpmac bypass): Add new bypass for fpmac-fpmac case.
3575 Add missing fmul, r2f_cvt and fconst cases.
3577 2017-06-14 Richard Biener <rguenther@suse.de>
3580 * fold-const.c (split_tree): Drop TREE_OVERFLOW flag from
3582 (fold_binary_loc): When associating do not treat pre-existing
3583 TREE_OVERFLOW on literal constants as a reason to allow
3584 TREE_OVERFLOW on associated literal constants.
3586 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
3588 * config/sparc/sparc.h (MASK_ISA): Add MASK_LEON and MASK_LEON3.
3589 (MASK_FEATURES): New macro.
3590 * config/sparc/sparc.c (sparc_option_override): Remove the special
3591 handling of -mfpu and generalize it to all MASK_FEATURES switches.
3593 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
3595 * simplify-rtx.c (simplify_binary_operation_1) <UDIV>: Do not simplify
3596 a division of 0 if non-call exceptions are enabled.
3598 2017-06-14 Andrew Pinski <apinski@cavium.com>
3599 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
3602 * config/aarch64/aarch64.c (aarch64_expand_vector_init):
3603 Improve vector initialization code gen for only variable case.
3605 2017-06-14 Eric Botcazou <ebotcazou@adacore.com>
3607 * config/sparc/driver-sparc.c (cpu_names): Add SPARC-T5 entry.
3609 2017-06-14 Richard Biener <rguenther@suse.de>
3611 PR tree-optimization/81083
3612 * tree-ssa-sccvn.c (vn_reference_lookup_3): Do not use abnormals
3615 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
3617 * config/rs6000/rs6000.c: Update all comments that mentioned SPE.
3618 (rs6000_expand_builtin): Remove RS6000_BTC_EVSEL.
3619 * config/rs6000/rs6000.h (RS6000_BTC_EVSEL): Delete.
3620 * config/rs6000/vxworks.h (VXCPU_FOR_8548): Delete. Adjust former use.
3621 * config/rs6000/vxworksae.h (VXCPU_FOR_8548): Delete.
3622 * config/rs6000/vxworksmils.h (VXCPU_FOR_8548): Delete.
3624 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
3626 * config/rs6000/rs6000-opts.h (enum rs6000_vector): Delete VECTOR_SPE.
3627 * config/rs6000/rs6000.c (rs6000_debug_vector_unit): Delete VECTOR_SPE.
3629 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
3631 * config/rs6000/rs6000.h (FIXED_SCRATCH): Delete.
3633 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
3635 * config/rs6000/t-rtems: Don't handle SPE.
3637 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
3639 * config/rs6000/t-linux: Don't handle SPE.
3641 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
3643 * config/rs6000/eabispe.h: Delete file.
3645 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
3647 * config/rs6000/t-spe: Delete file.
3649 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
3651 * config/rs6000/rs6000.c (SPE_CONST_OFFSET_OK): Delete.
3652 (rs6000_legitimate_offset_address_p): Return false for anything in
3653 V2SImode or V2SFmode.
3655 2017-06-13 Segher Boessenkool <segher@kernel.crashing.org>
3657 * config/rs6000/rs6000-modes.def: Remove all 8-byte vector modes
3658 except V2SF and V2SI. Rearrange the vector modes, and add comments.
3659 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Remove V8QImode
3661 (reg_offset_addressing_ok_p): Remove V4HImode and V1DImode.
3662 (rs6000_legitimate_offset_address_p): Ditto.
3663 (rs6000_emit_move): Ditto.
3664 (rs6000_init_builtins): Remove V4HI_type_node.
3666 2017-06-13 Martin Liska <mliska@suse.cz>
3669 * asan.c (asan_sanitize_stack_p): Use sanitize_flags_p.
3670 (gate_asan): Likewise.
3671 * asan.h (asan_no_sanitize_address_p): Remove the function.
3672 (sanitize_flags_p): New function.
3673 * builtins.def: Fix coding style.
3674 * common.opt: Use renamed enum value.
3675 * convert.c (convert_to_integer_1): Use sanitize_flags_p.
3676 * doc/extend.texi: Document no_sanitize attribute.
3677 * flag-types.h (enum sanitize_code): Rename SANITIZE_NONDEFAULT
3678 to SANITIZE_UNDEFINED_NONDEFAULT.
3679 * gcc.c (sanitize_spec_function): Use the renamed enum value.
3680 * gimple-fold.c (optimize_atomic_compare_exchange_p):
3681 Use sanitize_flags_p.
3682 * gimplify.c (gimplify_function_tree): Likewise.
3683 * ipa-inline.c (sanitize_attrs_match_for_inline_p): Likewise.
3684 * opts.c (parse_no_sanitize_attribute): New function.
3685 (common_handle_option): Use renamed enum value.
3686 * opts.h (parse_no_sanitize_attribute): Declare.
3687 * tree.c (sanitize_flags_p): New function.
3688 * tree.h: Declared here.
3689 * tsan.c: Use sanitize_flags_p.
3690 * ubsan.c (ubsan_expand_null_ifn): Likewise.
3691 (instrument_mem_ref): Likewise.
3692 (instrument_bool_enum_load): Likewise.
3693 (do_ubsan_in_current_function): Remove the function.
3694 (pass_ubsan::execute): Use sanitize_flags_p.
3695 * ubsan.h: Remove do_ubsan_in_current_function
3696 * tree-cfg.c (print_no_sanitize_attr_value): New function.
3697 (dump_function_to_file): Use it here.
3699 2017-06-13 Martin Jambor <mjambor@suse.cz>
3701 PR tree-optimization/80803
3702 PR tree-optimization/81063
3703 * tree-sra.c (subtree_mark_written_and_enqueue): Move up in the file.
3704 (propagate_subaccesses_across_link): Enqueue subtree whenever
3705 necessary instead of relying on the caller.
3707 2017-06-13 Martin Jambor <mjambor@suse.cz>
3709 * tree-sra.c (add_access_to_work_queue): Only enqueue accesses
3710 that have a first_link.
3711 (sort_and_splice_var_accesses): Do not check first_link before
3713 (subtree_mark_written_and_enqueue): Likewise.
3714 (propagate_all_subaccesses): Likewise and do not stop at first
3715 parent with a first_link.
3717 2017-06-13 Martin Jambor <mjambor@suse.cz>
3719 * tree-sra.c (dump_access_tree_1): Fix accidental dumping to stderr
3722 2017-06-13 Yury Gribov <tetra2005@gmail.com>
3724 * match.pd: New pattern.
3726 2017-06-13 Yury Gribov <tetra2005@gmail.com>
3728 * tree-vrp.c (is_masked_range_test): New function.
3729 (register_edge_assert_for): Determine ranges for
3732 2017-06-13 Yury Gribov <tetra2005@gmail.com>
3734 PR tree-optimization/67328
3735 * fold-const.c (maskable_range_p): New function.
3736 (build_range_check): Generate bittests if possible.
3738 2017-06-13 Martin Liska <mliska@suse.cz>
3740 * gimple-pretty-print.c (dump_probability): Add new argument.
3741 (dump_edge_probability): Dump both probability and count.
3742 (dump_gimple_label): Likewise.
3743 (dump_gimple_bb_header): Likewise.
3745 2017-06-13 Georg-Johann Lay <avr@gjlay.de>
3748 * config/avr/avr-devices.c: Fix indentation.
3749 * config/avr/gen-avr-mmcu-specs.c: Dito.
3751 2017-06-13 Richard Biener <rguenther@suse.de>
3753 * tree-vect-loop.c (vect_model_reduction_cost): Do not fail,
3754 instead get vector type from stmt_info.
3755 (vectorizable_reduction): Adjust. Remove dead code.
3757 2017-06-13 Richard Biener <rguenther@suse.de>
3760 * fold-const.c (extract_muldiv_1): Remove bogus distribution
3761 case of C * (x * C2 + C3).
3762 (fold_addr_of_array_ref_difference): Properly fold index difference.
3764 2017-06-12 David S. Miller <davem@davemloft.net>
3767 * config/sparc/sparc.md (return expander): Emit frame blockage if
3768 function uses alloca.
3770 2017-06-12 Richard Sandiford <richard.sandiford@linaro.org>
3772 * combine.c (make_field_assignment): Check len rather than the mode
3773 precision when calling force_to_mode.
3775 2017-06-12 Georg-Johann Lay <avr@gjlay.de>
3777 Support multilibs and devices that see flash in RAM address range.
3780 * config/avr/avr-arch.h (avr_arch_id) <ARCH_AVRXMEGA3>: New enum.
3781 (avr_mcu_t) <flash_pm_offset>: New field.
3782 (avr_device_specific_features) <AVR_ISA_RCALL>: New enum.
3783 * config/avr/avr.h (AVR_SHORT_CALLS): New define.
3784 (AVR_HAVE_JMP_CALL): Don't set if AVR_SHORT_CALLS.
3785 (AVR_TINY_PM_OFFSET): Remove macro.
3786 * config/avr/avr.opt (-mshort-calls): New option.
3787 * config/avr/gen-avr-mmcu-specs.c (print_mcu)
3788 [*self_spec]: Add / remove -mshort-calls depending on AVR_ISA_RCALL.
3789 * config/avr/avr-c.c (avr_cpu_cpp_builtins)
3790 <__AVR_SHORT_CALLS__>: Built-in define if AVR_SHORT_CALLS.
3791 <__AVR_HAVE_JMP_CALL__>: Use AVR_HAVE_JMP_CALL as condition
3792 instead of avr_arch->have_jmp_call.
3793 <__AVR_PM_BASE_ADDRESS__>: Built-in define if avr_arch->flash_pm_offset.
3794 [AVR_TINY] <__AVR_TINY_PM_BASE_ADDRESS__>: Use
3795 avr_arch->flash_pm_offset to define.
3796 * config/avr/avr-devices.c (avr_arch_types): Add initializers for
3797 new field flash_pm_offset. Add entry for avrxmega3.
3798 (avr_texinfo): Add entry for avrxmega3.
3799 * config/avr/avr-mcus.def: Add entries for: avrxmega3,
3800 attiny212, attiny214,
3801 attiny412, attiny414, attiny416, attiny417,
3802 attiny814, attiny816, attiny817,
3803 attiny1614, attiny1616, attiny1617,
3804 attiny3214, attiny3216, attiny3217.
3805 * config/avr/avr.c (avr_assemble_integer)[AVR_TINY]: Use
3806 avr_arch->flash_pm_offset instead of AVR_TINY_PM_OFFSET.
3807 (avr_print_operand_address) [AVR_TINY]: Same.
3808 (avr_asm_init_sections) <readonly_data_section>: Only patch
3809 callback if avr_arch->flash_pm_offset = 0.
3810 (avr_asm_named_section) <avr_need_copy_data_p>: Skip setting it
3811 for rodata if avr_arch->flash_pm_offset != 0.
3812 (avr_encode_section_info) [AVR_TINY]: Adjust comment.
3813 * config/avr/genmultilib.awk (dir_rcall, opt_rcall): New vars.
3814 (opts) [AVR_ISA_RCALL]: Append opt_rcall.
3815 (m_options): Append opt_rcall.
3816 (m_dirnames): Append dir_rcall.
3817 * config/avr/t-multilib: Regenerate.
3819 * configure.ac [target=avr]: Check whether avrxmega3 default
3820 linker description file works as needed.
3821 * configure: Regenerate.
3822 * doc/avr-mmcu.texi: Regenerate.
3823 * doc/invoke.texi (AVR Options) <-mshort-calls>: Document it.
3824 <__AVR_ARCH__>: Document avrxmega3 and 103.
3825 <__AVR_HAVE_JMP_CALL__>: Adjust documentation.
3826 <__AVR_SHORT_CALLS__>: Document it.
3827 <__AVR_PM_BASE_ADDRESS__>: Document it.
3828 * doc/extend.texi (AVR Options) <-mshort-calls>: Document it.
3829 (AVR Variable Attributes) <progmem>: Document this is
3830 not needed for avrxmega3.
3831 (AVR Named Address Spaces) <__flash>: Dito.
3833 2017-06-12 Jan Hubicka <hubicka@ucw.cz>
3835 * cgraph.c (cgraph_node::dump): Complain about profile insanities.
3837 2017-06-12 Doug Rupp <rupp@adacore.com>
3839 * config.gcc (*-*-vxworks*): Set use_gcc_stdint to "provide".
3840 Append vxworks-stdint.h to the tm_file list.
3841 * config/vxworks-stdint.h: New file.
3843 2017-06-12 Martin Liska <mliska@suse.cz>
3845 PR tree-optimization/81041
3846 * tree-profile.c (gimple_gen_ic_func_profiler):
3847 Create an extra BB in profile-generate
3848 (gimple_gen_time_profiler): Likewise.
3850 2017-06-12 Jakub Jelinek <jakub@redhat.com>
3852 PR tree-optimization/81003
3853 * tree-ssa-reassoc.c (force_into_ssa_name): New function.
3854 (update_range_test): Use it instead of force_gimple_operand_gsi.
3856 2017-06-12 Richard Biener <rguenther@suse.de>
3858 PR tree-optimization/81053
3859 * tree-vect-loop.c (vect_is_simple_reduction): Handle PHI
3860 with backedge value not defined in loop. Simplify def stmt
3863 2017-06-11 Tom de Vries <tom@codesourcery.com>
3866 * config/nvptx/nvptx.c (nvptx_cannot_force_const_mem): New function.
3868 (TARGET_CANNOT_FORCE_CONST_MEM): Redefine to
3869 nvptx_cannot_force_const_mem.
3871 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
3873 * opts.c (finish_options): Move test for flag_split_stack after
3874 it has been initialized.
3876 2017-06-11 Jason Merrill <jason@redhat.com>
3878 * tree.h (id_equal): New.
3879 * dwarf2out.c, hsa-gen.c, ipa-devirt.c, omp-expand.c,
3880 omp-simd-clone.c, read-rtl-function.c, tree-chkp.c, tree.c: Use it
3881 instead of strcmp of IDENTIFIER_POINTER.
3883 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
3885 * ipa-inline-transform.c: Include function.h, cfg.h and basic-block.h
3886 (mark_all_inlined_calls_cdtor): Fix formating.
3887 (inline_transform): Rescale profile before inlining.
3889 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
3891 * cgraph.h (cgraph_edge::clone): Update prototype.
3892 * cgraphclones.c (cgraph_edge::clone): Update profile scaling.
3893 (cgraph_node::create_clone): Update.
3894 (cgraph_node::create_version_clone): Update.
3895 * tree-inline.c (copy_bb): Update.
3896 (expand_call_inline): Update.
3898 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
3900 * config/rs6000/rs6000.c (emit_vrsave_prologue): New function,
3901 factored out from ...
3902 (rs6000_emit_prologue): ... here.
3904 2017-06-10 Segher Boessenkool <segher@kernel.crashing.org>
3906 * config/rs6000/rs6000.c (emit_split_stack_prologue): New function,
3907 factored out from ...
3908 (rs6000_emit_prologue): ... here.
3910 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
3912 * predict.c (drop_profile): Also drop individual bb/edge and cgraph
3914 (handle_missing_profiles): Fix computation of tp_first_run.
3915 (counts_to_freqs): Do not touch freqs when count is 0.
3917 2017-06-10 Jan Hubicka <hubicka@ucw.cz>
3919 * cgraphbuild.c (cgraph_edge::rebuild_references): Do not touch
3922 2017-06-10 Tom de Vries <tom@codesourcery.com>
3924 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
3925 attributes): Document signal effective target.
3927 2017-06-10 Tom de Vries <tom@codesourcery.com>
3929 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
3930 Document effective target stack_size.
3932 2017-06-09 David Malcolm <dmalcolm@redhat.com>
3934 * diagnostic.c (diagnostic_report_diagnostic): Only add fixits
3935 to the edit_context if they can be auto-applied.
3937 2017-06-9 Ian Lance Taylor <iant@golang.org>
3939 * opts.c (finish_options): If -fsplit-stack, disable implicit
3940 -forder-blocks-and-partition.
3941 * doc/invoke.texi (Optimize Options): Document that when using
3942 -fsplit-stack -forder-blocks-and-partition is not implicitly
3945 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
3947 * builtin-attrs.def (ATTR_NORETURN_NOTHROW_LEAF_COLD_LIST,
3948 ATTR_CONST_NORETURN_NOTHROW_LEAF_COLD_LIST,
3949 ATTR_TMPURE_NORETURN_NOTHROW_LEAF_COLD_LIST): New.
3950 * builtins.def (abort, trap, unreachable): Declare cold.
3951 * calls.c (flags_from_decl_or_type): Lookup ECF_COLD.
3952 * tree-core.h (ECF_COLD): New.
3953 * tree.c (set_call_expr_flags): Handle ECF_COLD.
3954 (build_common_builtin_nodes): Mark unreachable and abort as cold.
3956 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
3958 * predict.c (unlikely_executed_stmt_p): Cleanup.
3960 2017-06-09 Richard Biener <rguenther@suse.de>
3962 * tree-ssa-loop-im.c (execute_sm): Do not force multi-threaded
3963 model if the ref is always written to.
3965 2017-06-09 Tamar Christina <tamar.christina@arm.com>
3967 * config/aarch64/aarch64.md (lrint<GPF:mode><GPI:mode>2): New.
3969 2017-06-09 Tamar Christina <tamar.christina@arm.com>
3971 * config/arm/arm.c (arm_rtx_costs_internal): Make sdiv more expensive
3974 2017-06-09 Tom de Vries <tom@codesourcery.com>
3977 * config/nvptx/nvptx.md (define_expand "mov<QHSDISDFM>"): Error out with
3978 "target cannot support label values" when encountering LABEL_REF.
3980 2017-06-09 Martin Liska <mliska@suse.cz>
3982 * tree-profile.c (gimple_gen_ic_profiler): Update comment.
3983 (gimple_gen_ic_func_profiler): Emit direct comparison
3984 of __gcov_indirect_call_callee with NULL.
3985 (gimple_gen_time_profiler): Change probability from
3986 PROB_VERY_UNLIKELY to PROB_UNLIKELY.
3988 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
3990 * profile.c (edge_gcov_counts): Turn to pointer.
3991 (compute_branch_probabilities, compute_branch_probabilities): Update.
3992 (branch_prob): Do not clear edge_gcov_count.
3993 * profile.h (edge_gcov_counts): Turn to pointer.
3994 (edge_gcov_count): Update.
3996 2017-06-09 Jan Hubicka <hubicka@ucw.cz>
3998 * gimple.h (gimple_check_failed): Mark cold.
4000 2017-06-09 Richard Biener <rguenther@suse.de>
4002 PR tree-optimization/66623
4003 * tree-vect-loop.c (vect_is_simple_reduction): Cleanup,
4004 refactor check_reduction into two parts, properly computing
4005 whether we have to check reduction validity for outer loop
4008 2017-06-09 Richard Biener <rguenther@suse.de>
4010 PR tree-optimization/79483
4011 * graphite-scop-detection.c (order): New global.
4012 (get_order): Compute bb to order mapping that satisfies code
4013 generation constraints.
4014 (cmp_pbbs): New helper.
4015 (build_scops): Start domwalk at entry block, sort generated
4018 2017-06-09 Richard Biener <rguenther@suse.de>
4021 * ipa-polymorphic-call.c
4022 (ipa_polymorphic_call_context::restrict_to_inner_class):
4023 Skip FIELD_DECLs with error_mark_node type.
4024 * passes.def (all_lowering_passes): Run pass_build_cgraph_edges
4027 2017-06-09 Martin Liska <mliska@suse.cz>
4029 * predict.c (struct branch_predictor): New struct.
4030 (test_prediction_value_range): New test.
4031 (predict_c_tests): New function.
4032 * selftest-run-tests.c (selftest::run_tests): Run the function.
4033 * selftest.h: Declare new tests.
4035 2017-06-09 Segher Boessenkool <segher@kernel.crashing.org>
4038 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Assert that
4039 gen_add3_insn did not fail.
4040 * config/rs6000/rs6000.md (add<mode>3): If asked to add a constant to
4041 r0, construct that number in a temporary reg and add that reg to r0.
4042 If asked to put the result in r0 as well, fail.
4044 2017-06-08 Will Schmidt <will_schmidt@vnet.ibm.com>
4046 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
4047 for early expansion of vec_eqv.
4049 2017-06-08 Jakub Jelinek <jakub@redhat.com>
4052 * ubsan.c (instrument_null): Avoid pointless code temporary.
4053 (pass_ubsan::execute): Instrument aggregate arguments of calls.
4055 2017-06-08 Uros Bizjak <ubizjak@gmail.com>
4059 2016-12-14 Uros Bizjak <ubizjak@gmail.com>
4062 * config/i386/i386.md (*ctzhi2): New insn_and_split pattern.
4065 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
4067 * predict.c (unlikely_executed_edge_p): Move ahead.
4068 (probably_never_executed_edge_p): Use it.
4070 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
4073 * tree-chkp.c (chkp_gimple_call_builtin_p): Remove
4074 gimple_call_builtin_p call.
4076 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
4078 * system.h (fancy_abort): Annotate by ATTRIBUTE_COLD.
4079 * rtl.h (rtl_check_failed_bounds, rtl_check_failed_type1,
4080 rtl_check_failed_type2, rtl_check_failed_code1,
4081 rtl_check_failed_code2, rtl_check_failed_code_mode,
4082 rtl_check_failed_block_symbol, cwi_check_failed_bounds,
4083 rtvec_check_failed_bounds, rtl_check_failed_flag,
4084 _fatal_insn_not_found, _fatal_insn): Likewise.
4085 * tree.h (tree_contains_struct_check_failed,
4086 tree_check_failed, tree_not_check_failed,
4087 tree_class_check_failed, tree_range_check_failed,
4088 tree_not_class_check_failed, tree_int_cst_elt_check_failed,
4089 tree_vec_elt_check_failed, phi_node_elt_check_failed,
4090 tree_operand_check_failed, omp_clause_check_failed,
4091 omp_clause_operand_check_failed, omp_clause_range_check_failed):
4094 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
4096 * cgraph.c (cgraph_edge::maybe_hot_p): Do not check
4097 flag_branch_probabilities.
4098 * ipa-inline.c (edge_badness): Likewise.
4099 * ipa-profile.c (ipa_propagate_frequency_1): Likewise.
4100 * postreload-gcse.c (eliminate_partially_redundant_load): Likewise.
4101 * predict.c (maybe_hot_frequency_p): Likewise.
4102 (probably_never_executed): Likewise.
4103 * sched-ebb.c (schedule_ebbs): Likewise.
4104 * sched-rgn.c (find_single_block_region): Likewise.
4105 * tracer.c (tail_duplicate): Likewise.
4107 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
4109 * opts.c (finish_options): x_flag_reorder_blocks_and_partition no
4110 longer requires x_flag_profile_use.
4112 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
4114 * cfgrtl.c (cfg_layout_initialize): Check crtl->has_bb_partition
4115 instead of flag_reorder_blocks_and_partition.
4116 * dbxout.c (dbxout_function_end): Likewise.
4117 * dwarf2out.c (gen_subprogram_die): Likewise.
4118 * haifa-sched.c (sched_create_recovery_edges): Likewise.
4119 * hw-doloop.c (reorg_loops): Likewise.
4120 * varasm.c (assemble_start_function,
4121 assemble_end_function): Likewise.
4122 (decide_function_section): Do not check for
4123 flag_reorder_blocks_and_partition.
4125 2017-06-08 Alexander Ivchenko <alexander.ivchenko@intel.com>
4127 * tree-chkp.c (chkp_get_hard_register_var_fake_base_address):
4129 (chkp_get_hard_register_fake_addr_expr): Ditto.
4130 (chkp_build_addr_expr): Add check for hard reg case.
4131 (chkp_parse_array_and_component_ref): Ditto.
4132 (chkp_find_bounds_1): Ditto.
4133 (chkp_process_stmt): Don't generate bounds store for
4136 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
4138 * predict.c (maybe_hot_bb_p): Do not check profile status.
4139 (maybe_hot_edge_p): Likewise.
4140 (probably_never_executed): Check for zero counts even if profile
4142 (unlikely_executed_edge_p): New function.
4143 (unlikely_executed_stmt_p): New function.
4144 (unlikely_executed_bb_p): New function.
4145 (set_even_probabilities): Use unlikely predicates.
4146 (combine_predictions_for_bb): Likewise.
4147 (predict_paths_for_bb): Likewise.
4148 (predict_paths_leading_to_edge): Likewise.
4149 (determine_unlikely_bbs): New function.
4150 (estimate_bb_frequencies): Use it.
4151 (compute_function_frequency): Use zero counts even if profile is
4153 * profile-count.h: Fix typo.
4155 2017-08-08 Julia Koval <julia.koval@intel.com>
4157 * config/i386/avx512bwintrin.h (_mm512_mask_cvtepi16_storeu_epi8,
4158 _mm512_mask_cvtsepi16_storeu_epi8,
4159 _mm512_mask_cvtusepi16_storeu_epi8): New intrinsics.
4160 * config/i386/avx512vlbwintrin.h (_mm256_mask_cvtepi16_storeu_epi8,
4161 _mm_mask_cvtsepi16_storeu_epi8, _mm256_mask_cvtsepi16_storeu_epi8,
4162 _mm_mask_cvtusepi16_storeu_epi8, _mm256_mask_cvtusepi16_storeu_epi8,
4163 _mm_mask_cvtepi16_storeu_epi8): New intrinsics.
4164 * config/i386/i386-builtin-types.def (PV8Q, V8QI): New pointer type.
4165 (VOID_FTYPE_PV32QI_V32HI_USI, VOID_FTYPE_PV8QI_V8HI_UQI,
4166 VOID_FTYPE_PV16QI_V16HI_UHI): New function types.
4167 * config/i386/i386-builtin.def (__builtin_ia32_pmovwb128mem_mask,
4168 __builtin_ia32_pmovwb256mem_mask, __builtin_ia32_pmovswb128mem_mask,
4169 __builtin_ia32_pmovswb256mem_mask, __builtin_ia32_pmovuswb128mem_mask,
4170 __builtin_ia32_pmovuswb256mem_mask,
4171 __builtin_ia32_pmovuswb512mem_mask, __builtin_ia32_pmovswb512mem_mask)
4172 __builtin_ia32_pmovwb512mem_mask): New builtins.
4174 2017-08-08 Julia Koval <julia.koval@intel.com>
4176 PR target/73350,80862
4177 * config/i386/subst.md (round): Fix round pattern.
4178 * config/i386/i386.c (ix86_erase_embedded_rounding):
4179 Fix erasing rounding for the fixed pattern.
4181 2017-06-08 Jan Hubicka <hubicka@ucw.cz>
4183 * cfgbuild.c (find_many_sub_basic_blocks): Fix thinko.
4185 2017-06-08 Martin Liska <mliska@suse.cz>
4187 PR gcov-profile/80911
4188 * gcov.c (block_info::block_info): New constructor.
4190 2017-06-07 Carl Love <cel@us.ibm.com>
4192 * config/rs6000/rs6000-c: The return type of the following
4193 built-in functions was implemented as int not long long. Fix sign
4194 of return value for the unsigned version of vec_mulo and vec_mule.
4195 vector unsigned long long vec_bperm (vector unsigned long long,
4196 vector unsigned char)
4197 vector signed long long vec_mule (vector signed int,
4199 vector unsigned long long vec_mule (vector unsigned int,
4200 vector unsigned int)
4201 vector signed long long vec_mulo (vector signed int,
4203 vector unsigned long long vec_mulo (vector unsigned int,
4204 vector unsigned int)
4205 * doc/extend.texi: Fix the documentation for the built-in
4208 2017-06-07 Carl Love <cel@us.ibm.com>
4211 * config/rs6000/altivec.md (double<mode>2): Fix the implementation of
4214 2017-06-07 Carl Love <cel@us.ibm.com>
4216 * config/rs6000/altivec.md: Fix argument swizzle in vec_doublel
4217 support, Generate doublehv for signed int/float for BE case only.
4219 2017-06-07 Alexander Monakov <amonakov@ispras.ru>
4221 * doc/invoke.texi (mcx16): Rewrite.
4223 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4225 * config/rs6000/predicates.md (rs6000_nonimmediate_operand): Delete.
4226 * config/rs6000/rs6000.md (*movsi_internal1, movsi_from_sf,
4227 *mov<mode>_softfloat, and an anonymous splitter): Use
4228 nonimmediate_operand instead of rs6000_nonimmediate_operand.
4230 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4232 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE_ACC and
4234 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
4235 (enum rs6000_reg_type): Delete SPE_ACC_TYPE and SPEFSCR_REG_TYPE.
4236 (rs6000_debug_reg_global): Adjust.
4237 (rs6000_init_hard_regno_mode_ok): Adjust.
4238 (rs6000_dbx_register_number): Adjust.
4239 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change to 115.
4240 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
4241 Remove SPE_ACC and SPEFSCR.
4242 (REG_ALLOC_ORDER): Ditto.
4243 (FRAME_POINTER_REGNUM): Change to 111.
4244 (enum reg_class): Remove the SPE_ACC and SPEFSCR registers.
4245 (REG_CLASS_NAMES): Ditto.
4246 (REG_CLASS_CONTENTS): Delete the SPE_ACC and SPEFSCR registers.
4247 (REGISTER_NAMES): Ditto.
4248 (ADDITIONAL_REG_NAMES): Ditto.
4249 (rs6000_reg_names): Ditto.
4250 * config/rs6000/rs6000.md: Renumber some register number
4253 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4255 * config/rs6000/darwin.h (REGISTER_NAMES): Delete the SPE high
4257 * config/rs6000/rs6000.c (rs6000_reg_names, alt_reg_names): Ditto.
4258 * config/rs6000/rs6000.h (FIRST_PSEUDO_REGISTER): Change from 149
4260 (DWARF_REG_TO_UNWIND_COLUMN): Do not define.
4261 (FIXED_REGISTERS, CALL_USED_REGISTERS, CALL_REALLY_USED_REGISTERS):
4262 Delete the SPE high registers.
4263 (REG_ALLOC_ORDER): Ditto.
4264 (enum reg_class): Remove SPE_HIGH_REGS.
4265 (REG_CLASS_NAMES): Ditto.
4266 (REG_CLASS_CONTENTS): Delete the SPE high registers.
4267 (REGISTER_NAMES): Ditto.
4268 (rs6000_reg_names): Ditto.
4269 * doc/tm.texi.in: Remove SPE as example.
4270 * doc/tm.texi: Regenerate.
4272 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4274 * config/rs6000/8540.md (ppc8540_brinc): Delete.
4275 * config/rs6000/e500mc.md (e500mc_brinc): Delete.
4276 * config/rs6000/e500mc64.md (e500mc64_brinc): Delete.
4277 * config/rs6000/rs6000.md (type): Remove "brinc".
4279 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4281 * config.gcc (powerpc*-*-*): Don't add spe.h to extra_headers.
4282 (powerpc*-linux*spe*): Use ${cpu_type} instead of rs6000.
4283 * config/rs6000/linuxspe.h: Delete file.
4284 * config/rs6000/rs6000.md: Don't include spe.md.
4285 * config/rs6000/spe.h: Delete file.
4286 * config/rs6000/spe.md: Delete file.
4287 * config/rs6000/t-rs6000: Remove spe.md.
4289 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4291 * config/rs6000/predicates.md (reg_or_mem_operand): Reformat.
4292 (reg_or_none500mem_operand): Delete.
4293 * config/rs6000/rs6000.md (extendsfdf2): Use reg_or_mem_operand
4294 instead of reg_or_none500mem_operand.
4296 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4298 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
4299 handling of SPE flags.
4300 * config/rs6000/rs6000.opt (-mspe, -mspe=no, -mspe=yes): Delete.
4302 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4304 * config/rs6000/rs6000-common.c (rs6000_handle_option): Remove
4306 * config/rs6000/paired.md (paired_negv2sf2): Rename to negv2sf2.
4307 (paired_absv2sf2, paired_addv2sf3, paired_subv2sf3, paired_mulv2sf3,
4308 paired_divv2sf3): Similar.
4309 * config/rs6000/predicates.md: Replace TARGET_SPE, TARGET_SPE_ABI,
4310 SPE_VECTOR_MODE and SPE_HIGH_REGNO_P by 0; simplify.
4311 * config/rs6000/rs6000-builtin.def: Delete RS6000_BUILTIN_E and
4313 Delete BU_SPE_1, BU_SPE_2, BU_SPE_3, BU_SPE_E, BU_SPE_P, and BU_SPE_X.
4314 Rename the paired_* instruction patterns.
4315 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Do not
4317 * config/rs6000/rs6000-protos.h (invalid_e500_subreg): Delete.
4318 * config/rs6000/rs6000.c: Delete RS6000_BUILTIN_E and RS6000_BUILTIN_S.
4319 (struct rs6000_stack): Delete fields spe_gp_save_offset, spe_gp_size,
4320 spe_padding_size, and spe_64bit_regs_used. Replace TARGET_SPE and
4321 TARGET_SPE_ABI with 0, simplify. Replace SPE_VECTOR_MODE with
4323 (struct machine_function): Delete field spe_insn_chain_scanned_p.
4324 (spe_func_has_64bit_regs_p): Delete.
4325 (spe_expand_predicate_builtin): Delete.
4326 (spe_expand_evsel_builtin): Delete.
4327 (TARGET_DWARF_REGISTER_SPAN): Do not define.
4328 (TARGET_MEMBER_TYPE_FORCES_BLK): Do not define.
4329 (invalid_e500_subreg): Delete.
4330 (rs6000_legitimize_address): Always force_reg op2 as well, for
4331 paired single memory accesses.
4332 (rs6000_member_type_forces_blk): Delete.
4333 (rs6000_spe_function_arg): Delete.
4334 (rs6000_expand_unop_builtin): Delete SPE handling.
4335 (rs6000_expand_binop_builtin): Ditto.
4336 (spe_expand_stv_builtin): Delete.
4337 (bdesc_2arg_spe): Delete.
4338 (spe_expand_builtin): Delete.
4339 (spe_expand_predicate_builtin): Delete.
4340 (spe_expand_evsel_builtin): Delete.
4341 (rs6000_invalid_builtin): Remove RS6000_BTM_SPE handling.
4342 (spe_init_builtins): Delete.
4343 (spe_func_has_64bit_regs_p): Delete.
4344 (savres_routine_name): Delete "info" parameter. Adjust callers.
4345 (rs6000_emit_stack_reset): Ditto.
4346 (rs6000_dwarf_register_span): Delete.
4347 * config/rs6000/rs6000.h (TARGET_SPE_ABI, TARGET_SPE,
4348 UNITS_PER_SPE_WORD, SPE_HIGH_REGNO_P, SPE_SIMD_REGNO_P,
4349 SPE_VECTOR_MODE, RS6000_BTM_SPE, RS6000_BUILTIN_E, RS6000_BUILTIN_S):
4351 * config/rs6000/rs6000.md (FIRST_SPE_HIGH_REGNO, LAST_SPE_HIGH_REGNO):
4353 * config/rs6000/rs6000.opt (-mabi=spe, -mabi=no-spe): Delete.
4354 * config/rs6000/spe.md: Delete every pattern that uses TARGET_SPE.
4355 * config/rs6000/vector.md (absv2sf2, negv2sf2, addv2sf3, subv2sf3,
4356 mulv2sf3, divv2sf3): Delete expanders.
4358 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4360 config/rs6000/rs6000.md (UNSPEC_MV_CR_GT): Delete.
4362 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4364 * config/rs6000/rs6000-protos.h (output_e500_flip_gt_bit): Delete.
4365 * config/rs6000/rs6000.c: Ditto.
4367 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4369 * config/rs6000/predicated.md (rs6000_cbranch_operator): Delete.
4370 * config/rs6000/rs6000.md: Replace rs6000_cbranch_operator by
4371 comparison_operator.
4373 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4375 * config/rs6000/rs6000.c: Remove everything related to -mfloat-gprs.
4376 * config/rs6000/rs6000.opt: Ditto.
4377 * config/rs6000/t-rtems: Ditto.
4379 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4381 * config/rs6000/predicates.md: Replace TARGET_E500_DOUBLE and
4382 TARGET_E500_SINGLE by 0, simplify.
4383 * config/rs6000/rs6000.c: Ditto.
4384 (rs6000_option_override_internal): Delete CHECK_E500_OPTIONS.
4385 (spe_build_register_parallel): Delete.
4386 * config/rs6000/rs6000.h: Delete TARGET_E500_SINGLE,
4387 TARGET_E500_DOUBLE, and CHECK_E500_OPTIONS.
4388 * config/rs6000/rs6000.md: Replace TARGET_E500_DOUBLE,
4389 TARGET_E500_SINGLE, and <E500_CONVERT> by 0, simplify.
4390 (E500_CONVERT): Delete.
4391 * config/rs6000/spe.md: Remove many patterns and all define_constants.
4393 2017-06-07 Segher Boessenkool <segher@kernel.crashing.org>
4395 * config/rs6000/darwin.md: Replace TARGET_FPRS by 1 and simplify.
4396 * config/rs6000/dfp.md: Ditto.
4397 (negdd2, *negdd2_fpr): Merge.
4398 (absdd2, *absdd2_fpr): Merge.
4399 (negtd2, *negtd2_fpr): Merge.
4400 (abstd2, *abstd2_fpr): Merge.
4401 * config/rs6000/e500.h: Delete file.
4402 * config/rs6000/predicates.md (rs6000_cbranch_operator): Replace
4403 TARGET_FPRS by 1 and simplify.
4404 * config/rs6000/rs6000-c.c: Ditto.
4405 * config/rs6000/rs6000.c: Ditto. Also replace TARGET_SF_SPE and
4407 * config/rs6000/rs6000.h: Ditto. Delete TARGET_SF_SPE and
4409 * config/rs6000/rs6000.md: Ditto.
4410 (floatdidf2, *floatdidf2_fpr): Merge.
4411 (move_from_CR_gt_bit): Delete.
4412 * config/rs6000/spe.md: Replace TARGET_FPRS by 1 and simplify.
4413 (E500_CR_IOR_COMPARE): Delete.
4414 (All patterns that require !TARGET_FPRS): Delete.
4415 * config/rs6000/vsx.md: Replace TARGET_FPRS by 1 and simplify.
4417 2017-06-07 Bin Cheng <bin.cheng@arm.com>
4419 * passes.def (pass_iv_canon): Move before pass_loop_distribution.
4421 2017-06-07 Bin Cheng <bin.cheng@arm.com>
4423 * graphds.c (add_edge): Intitialize edge's attached data.
4424 (foll_in_subgraph, dfs_fst_edge, dfs_next_edge): New function
4425 pointer parameter. Call pointed function on each edge during
4426 graph traversing. Skip traversing the edge when the function
4428 (graphds_dfs, graphds_scc): Ditto.
4429 (for_each_edge): New parameter. Pass the new parameter to callback
4431 * graphds.h (skip_edge_callback): New function pointer type.
4432 (graphds_dfs, graphds_scc): New function pointer parameter.
4433 (graphds_edge_callback, for_each_edge): New parameter.
4435 2017-06-07 Bin Cheng <bin.cheng@arm.com>
4437 * tree-vect-data-refs.c (vect_mark_for_runtime_alias_test): Factor
4438 out code checking if runtime alias check is possible to below ...
4439 Call the new function.
4440 * tree-data-ref.c (runtime_alias_check_p): ... to new function.
4441 * tree-data-ref.h (runtime_alias_check_p): New decalaration.
4443 2017-06-07 Marek Polacek <polacek@redhat.com>
4446 * fold-const.c (extract_muldiv_1) <case MINUS_EXPR>: Add
4447 TYPE_OVERFLOW_WRAPS check.
4449 2017-06-07 Bin Cheng <bin.cheng@arm.com>
4451 * tree-vect-loop-manip.c (vect_do_peeling): Don't skip vector loop
4452 if versioning is required.
4453 * tree-vect-loop.c (vect_analyze_loop_2): Merge niter check for loop
4454 peeling with the check for versioning.
4456 2017-06-07 Bin Cheng <bin.cheng@arm.com>
4458 * tree-vectorizer.h (vect_build_loop_niters): New parameter.
4459 * tree-vect-loop-manip.c (vect_build_loop_niters): New parameter.
4460 Set true to new parameter if new ssa variable is defined.
4461 (vect_gen_vector_loop_niters): Refactor. Set range information
4462 for the new vector loop bound variable.
4463 (vect_do_peeling): Ditto.
4465 2017-06-07 Bin Cheng <bin.cheng@arm.com>
4467 * tree-affine.c (ssa.h): Include header file.
4468 (tree_to_aff_combination): Handle (T1)(X - CST) when inner type
4469 has wrapping overflow behavior.
4471 2017-06-07 Bin Cheng <bin.cheng@arm.com>
4473 * tree-affine.c (tree_to_aff_combination): Handle (T1)(X + X).
4475 2017-06-07 Bin Cheng <bin.cheng@arm.com>
4477 (aff_combination_expand): Move (T1)(X *+- CST) simplification to ...
4478 (tree_to_aff_combination): ... here.
4480 2017-06-07 Bin Cheng <bin.cheng@arm.com>
4482 * tree-ssa-loop-ivopts.c (ivopts_estimate_reg_pressure): New
4483 reg_pressure model function.
4484 (ivopts_global_cost_for_size): Delete.
4485 (determine_set_costs, iv_ca_recount_cost): Call new model function
4486 ivopts_estimate_reg_pressure.
4488 2017-06-07 Tamar Christina <tamar.christina@arm.com>
4490 * config/aarch64/aarch64.c (aarch64_rtx_costs): Make sdiv more
4491 expensive than udiv. Remove floating point cases from mod.
4493 2017-06-07 Tamar Christina <tamar.christina@arm.com>
4495 * config/arm/aarch-cost-tables.h (cortexa53_extra_cost):
4498 2017-06-07 Tamar Christina <tamar.christina@arm.com>
4500 * config/aarch64/aarch64.md
4501 (copysignsf3): Fix mask generation.
4503 2017-06-07 Jakub Jelinek <jakub@redhat.com>
4505 * dumpfile.h (enum tree_dump_index): Rename TDI_generic to
4507 (class dump_manager): Add register_dumps method.
4508 * dumpfile.c: Include langhooks.h.
4509 (dump_files): Use 0 instead of 3/4/5 for TDI_{original,gimple,nested}.
4510 (FIRST_AUTO_NUMBERED_DUMP): Decrease to 1.
4511 (FIRST_ME_AUTO_NUMBERED_DUMP): Define.
4512 (dump_manager::dump_register): Start with 512 entries instead of 32.
4513 (dump_manager::register_dumps): New method.
4514 * toplev.c (general_init): Instead of invoking register_dumps
4515 langhook, invoke register_dumps method on the dump manager.
4516 * gimplify.c (gimplify_function_tree): Use TDI_gimple instead of
4519 2017-06-07 Richard Sandiford <richard.sandiford@linaro.org>
4521 * doc/md.texi: Clarify the restrictions on a define_insn condition.
4522 Say that # requires an associated define_split to exist, and that
4523 the define_split must be suitable for use after register allocation.
4525 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
4527 * cfgbuild.c (find_bb_boundaries): Initialize profile of split blocks.
4528 (compute_outgoing_frequencies): Also initialize zero counts.
4529 (find_many_sub_basic_blocks): Do not produce uninitialized profile
4530 around loops; preserve more of profile when nothing changes.
4532 2017-06-06 Jim Wilson <jim.wilson@linaro.org>
4534 * config/aarch64/aarch64-cost-tables.h (qdf24xx_extra_costs): Move to
4536 * config/arm/aarch-cost-tables.h (qdf24xx_extra_costs): From here.
4537 * config/arm/arm-cpu-cdata.h: Regenerate.
4538 * config/arm/arm-cpu-data.h, config/arm/arm-cpu.h: Likewise.
4539 * config/arm/arm-tables.opt, config/arm/arm-tune.md: Likewise.
4540 * config/arm/arm-cpus.in: Delete falkor and qdf24xx entries.
4541 * config/arm/arm.c (arm_qdf24xx_tune): Delete.
4542 * config/arm/bpabi.h (BE8_LINK_SPEC): Delete falkor and qdf24xx
4544 * config/arm/t-aprofile (MULTILIB_MATCHES): Delete falkor and qdf24xx
4546 * config/arm/t-rmprofile: Likewise.
4547 * doc/invoke.texi (ARM Options): Drop falkor and qdf24xx support.
4549 2017-06-06 David S. Miller <davem@davemloft.net>
4552 * config/sparc/sparc.c (sparc_expand_prologue): Emit frame
4553 blockage if function uses alloca.
4555 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4557 * tree-ssa-loop-prefetch.c (struct mem_ref_group, struct mem_ref):
4558 New "uid" fields to hold pretty-print IDs of group and ref.
4559 Memory references are now identified as <group_id>:<ref_id>
4560 instead of using [random] addresses.
4561 (dump_mem_details): Simplify, no functional change.
4562 (dump_mem_ref): Simplify and make output more concise.
4563 Replace couple of fprintf's throughout code with calls to dump_mem_ref.
4564 (find_or_create_group): Initialize group uid.
4565 (record_ref): Initialize ref uid. Improve debug output.
4566 (prune_group_by_reuse, should_issue_prefetch_p,)
4567 (should_issue_prefetch_p, schedule_prefetches, issue_prefetch_ref,)
4568 (mark_nontemporal_store, determine_loop_nest_reuse):
4569 Improve debug output.
4571 2017-06-06 Maxim Kuvyrkov <maxim.kuvyrkov@linaro.org>
4573 * dbgcnt.def (prefetch): New debug counter.
4574 * tree-ssa-loop-prefetch.c (dbgcnt.h): New include.
4575 (schedule_prefetches): Stop issueing prefetches if debug counter
4578 2017-06-06 Tom de Vries <tom@codesourcery.com>
4580 * doc/sourcebuild.texi (Testsuites, C Language Testsuites,
4581 gcc.c-torture/compile): Remove mention of NO_LABEL_VALUES in fixme.
4583 2017-06-06 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4585 * config/aarch64/atomics.md (atomic_compare_and_swap<mode> expander):
4586 Use aarch64_reg_or_zero predicate for operand 4.
4587 (aarch64_compare_and_swap<mode> define_insn_and_split):
4588 Use aarch64_reg_or_zero predicate for operand 3. Add 'Z' constraint.
4589 (aarch64_store_exclusive<mode>): Likewise for operand 2.
4591 2017-06-06 Thomas Preud'homme <thomas.preudhomme@arm.com>
4593 * config/arm/arm.c (arm_compute_save_reg_mask): Rename into ...
4594 (arm_compute_save_core_reg_mask): This.
4595 (thumb1_compute_save_reg_mask): Rename into ...
4596 (thumb1_compute_save_core_reg_mask): This.
4597 (arm_compute_save_reg0_reg12_mask): Adapt comment.
4598 (arm_compute_frame_layout): Likewise.
4600 2017-06-06 Richard Biener <rguenther@suse.de>
4602 PR tree-optimization/80974
4603 * tree-ssa-sccvn.c (set_ssa_val_to): Do not change but only
4604 keep or clear leaders SSA info.
4606 2017-06-06 Tom de Vries <tom@codesourcery.com>
4608 * config/nvptx/nvptx.c (split_mode_p): New function.
4609 (nvptx_declare_function_name, nvptx_print_operand): Use split_mode_p.
4611 2017-06-06 Tom de Vries <tom@codesourcery.com>
4613 * config/nvptx/nvptx.c (nvptx_print_operand): Use maybe_split_mode.
4615 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
4618 * tree-cfg.c (execute_fixup_cfg): Fix condition on when to rescale
4621 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
4623 * shrink-wrap.c (handle_simple_exit): Update profile.
4624 (try_shrink_wrapping): Upate profile.
4626 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
4628 * predict.c (tree_estimate_probability_bb): Add LOCAL_ONLY.
4629 (tree_guess_outgoing_edge_probabilities): New.
4630 * predict.h (tree_guess_outgoing_edge_probabilities): Declare.
4631 * tree-cfg.c (gimple_find_sub_bbs): Propagate profile.
4633 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
4635 * ipa-split.c (split_function): Initialize return bb profile.
4637 2017-06-06 Jan Hubicka <hubicka@ucw.cz>
4639 * profile.c (compute_branch_probabilities): Also initialize
4642 2017-06-06 Richard Biener <rguenther@suse.de>
4644 PR tree-optimization/80928
4645 * tree-vect-loop.c (vect_update_vf_for_slp): Amend dumps.
4646 (vect_analyze_loop_operations): Properly guard analysis for
4648 (vect_transform_loop): Likewise.
4649 (vect_analyze_loop_2): Also reset SLP type on PHIs.
4650 (vect_model_induction_cost): Do not cost for pure SLP.
4651 (vectorizable_induction): Pass in SLP node, implement SLP vectorization
4652 of induction in inner loop vectorization.
4653 * tree-vect-slp.c (vect_create_new_slp_node): Handle PHIs.
4654 (vect_get_and_check_slp_defs): Handle vect_induction_def.
4655 (vect_build_slp_tree): Likewise. Handle PHIs as terminating the
4657 (vect_analyze_slp_cost_1): Cost induction.
4658 (vect_detect_hybrid_slp_stmts): Handle PHIs.
4659 (vect_get_slp_vect_defs): Likewise.
4660 * tree-vect-stmts.c (vect_analyze_stmt): Handle induction.
4661 (vect_transform_stmt): Handle SLP reductions.
4662 * tree-vectorizer.h (vectorizable_induction): Adjust.
4664 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
4666 * config/rs6000/rs6000.c (make_resolver_func): Update
4667 init_lowered_empty_function call.
4669 2017-06-05 Bernd Edlinger <bernd.edlinger@hotmail.de>
4671 * doc/invoke.texi: Document the -fprofile-abs-path option.
4672 * common.opt (fprofile-abs-path): New option.
4673 * gcov-io.h (gcov_write_filename): Declare.
4674 * gcov-io.c (gcov_write_filename): New function.
4675 * coverage.c (coverage_begin_function): Use gcov_write_filename.
4676 * profile.c (output_location): Likewise.
4678 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
4680 * shring-wrap.c: Revert accidental commit.
4682 2017-06-05 Volker Reichelt <v.reichelt@netcologne.de>
4684 * doc/invoke.texi (-Wduplicated-branches): Add to warning list.
4686 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
4688 * cfgexpand.c (expand_gimple_tailcall): Initialize profile of
4690 * ipa-inline.c (want_inline_self_recursive_call_p): Watch for missing
4691 profile in callgraph edge.
4692 * profile-count.h (apply_probability): If THIS is 0, then result is 0
4693 (apply_scale): Likewise.
4694 * tree-inline.c (copy_bb, copy_edges_for_bb, copy_cfg_body):
4695 Also scale profile when inlining function with zero profile.
4696 (initialize_cfun): Update exit block profile even when it is zero.
4697 * tree-ssa-threadupdate.c (clear_counts_path): Handle correctly case
4698 when profile is read.
4700 2017-06-05 Michael Meissner <meissner@linux.vnet.ibm.com>
4702 * config/rs6000/rs6000.c (toplevel): Include attribs.h.
4703 (CLONE_*): New constants to define the processors we can generate
4704 code for with the target_clone attribute.
4705 (rs6000_clone_map): New array to identify which clone processors
4706 the current program is running on.
4707 (TARGET_COMPARE_VERSION_PRIORITY): Define to enable the
4708 target_clone attribute.
4709 (TARGET_GENERATE_VERSION_DISPATCHER_BODY): Likewise.
4710 (TARGET_GET_FUNCTION_VERSIONS_DISPATCHER): Likewise.
4711 (TARGET_OPTION_FUNCTION_VERSIONS): Likewise.
4712 (cpu_expand_builtin): Add support for target_clone attribute.
4713 (rs6000_valid_attribute_p): Allow "default" attribute.
4714 (get_decl_name): New debug function to simplify printing the
4715 current function name in debugging statements.
4716 (rs6000_clone_priority): New functions to support the target_clone
4717 attribute, and be able to generate code to switch between ISA 2.05
4718 through ISA 3.0 (power6 through power9).
4719 (rs6000_compare_version_priority): Likewise.
4720 (rs6000_get_function_versions_dispatcher): Likewise.
4721 (make_resolver_func): Likewise.
4722 (add_condition_to_bb): Likewise.
4723 (dispatch_function_versions): Likewise.
4724 (rs6000_generate_version_dispatcher_body): Likewise.
4725 (rs6000_can_inline_p): Call get_decl_name for debugging usage.
4726 (fusion_gpr_load_p): Fix a spacing issue.
4727 * doc/extend.texi (Common Function Attributes): Document that the
4728 PowerPC supports the target_clone attribute.
4730 2017-06-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
4732 * config/arm/arm.h: explain F symbol found in description of ARM
4733 register allocation in its legend.
4735 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
4737 * config/mips/frame-header-opt.c: Include profile-count.h.
4738 * config/riscv/riscv.c: Include profile-count.h
4740 2017-06-05 Jan Hubicka <hubicka@ucw.cz>
4742 * tree-ssa-loop-im.c (execute_sm_if_changed): Add FLAG_BBS parameter;
4744 (sm_set_flag_if_changed): Add bbs field.
4745 (execute_sm_if_changed_flag_set): Pass BBS.
4746 (execute_sm): Update.
4748 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4750 * config/aarch64/aarch64-simd.md (aarch64_store_lane0<mode>):
4753 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4755 * config/aarch64/aarch64.md (sub<mode>3_compare1_imm): New define_insn.
4756 (peephole2): New peephole2 to emit the above.
4757 * config/aarch64/predicates.md (aarch64_sub_immediate): New predicate.
4759 2017-06-05 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
4761 * config/aarch64/aarch64.c (define_peephole2 above
4762 *sub_<shift>_<mode>): New peephole.
4764 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
4766 * config/i386/i386.c (make_resolver_func): Update.
4767 * Makefile.in: Add profile-count.h and profile-count.o
4768 * auto-profile.c (afdo_indirect_call): Update to new API.
4769 (afdo_set_bb_count): Update.
4770 (afdo_propagate_edge): Update.
4771 (afdo_propagate_circuit): Update.
4772 (afdo_calculate_branch_prob): Update.
4773 (afdo_annotate_cfg): Update.
4774 * basic-block.h: Include profile-count.h
4775 (struct edge_def): Turn count to profile_count.
4776 (struct basic_block_def): Likewie.
4777 (REG_BR_PROB_BASE): Move to profile-count.h
4778 (RDIV): Move to profile-count.h
4779 * bb-reorder.c (max_entry_count): Turn to profile_count.
4780 (find_traces): Update.
4781 (rotate_loop):Update.
4782 (connect_traces):Update.
4783 (sanitize_hot_paths):Update.
4784 * bt-load.c (migrate_btr_defs): Update.
4785 * cfg.c (RDIV): Remove.
4786 (init_flow): Use alloc_block.
4787 (alloc_block): Uninitialize count.
4788 (unchecked_make_edge): Uninitialize count.
4789 (check_bb_profile): Update.
4790 (dump_edge_info): Update.
4791 (dump_bb_info): Update.
4792 (update_bb_profile_for_threading): Update.
4793 (scale_bbs_frequencies_int): Update.
4794 (scale_bbs_frequencies_gcov_type): Update.
4795 (scale_bbs_frequencies_profile_count): New.
4796 * cfg.h (update_bb_profile_for_threading): Update.
4797 (scale_bbs_frequencies_profile_count): Declare.
4798 * cfgbuild.c (compute_outgoing_frequencies): Update.
4799 (find_many_sub_basic_blocks): Update.
4800 * cfgcleanup.c (try_forward_edges): Update.
4801 (try_crossjump_to_edge): Update.
4802 * cfgexpand.c (expand_gimple_tailcall): Update.
4803 (construct_exit_block): Update.
4804 * cfghooks.c (verify_flow_info): Update.
4805 (dump_bb_for_graph): Update.
4806 (split_edge): Update.
4807 (make_forwarder_block): Update.
4808 (duplicate_block): Update.
4809 (account_profile_record): Update.
4810 * cfgloop.c (find_subloop_latch_edge_by_profile): Update.
4811 (get_estimated_loop_iterations): Update.
4812 * cfgloopanal.c (expected_loop_iterations_unbounded): Update.
4813 (single_likely_exit): Update.
4814 * cfgloopmanip.c (scale_loop_profile): Update.
4816 (set_zero_probability): Update.
4817 (lv_adjust_loop_entry_edge): Update.
4818 * cfgrtl.c (force_nonfallthru_and_redirect): Update.
4819 (purge_dead_edges): Update.
4820 (rtl_account_profile_record): Update.
4821 * cgraph.c (cgraph_node::create): Uninitialize count.
4822 (symbol_table::create_edge): Uninitialize count.
4823 (cgraph_update_edges_for_call_stmt_node): Update.
4824 (cgraph_edge::dump_edge_flags): Update.
4825 (cgraph_node::dump): Update.
4826 (cgraph_edge::maybe_hot_p): Update.
4827 * cgraph.h: Include profile-count.h
4828 (create_clone), create_edge, create_indirect_edge): Update.
4829 (cgraph_node): Turn count to profile_count.
4830 (cgraph_edge0: Likewise.
4831 (make_speculative, clone): Update.
4832 (create_edge): Update.
4833 (init_lowered_empty_function): Update.
4834 * cgraphclones.c (cgraph_edge::clone): Update.
4835 (duplicate_thunk_for_node): Update.
4836 (cgraph_node::create_clone): Update.
4837 * cgraphunit.c (cgraph_node::analyze): Update.
4838 (cgraph_node::expand_thunk): Update.
4839 * final.c (dump_basic_block_info): Update.
4840 * gimple-streamer-in.c (input_bb): Update.
4841 * gimple-streamer-out.c (output_bb): Update.
4842 * graphite.c (print_global_statistics): Update.
4843 (print_graphite_scop_statistics): Update.
4844 * hsa-brig.c: Include basic-block.h.
4845 * hsa-dump.c: Include basic-block.h.
4846 * hsa-gen.c (T sum_slice): Update.
4847 (convert_switch_statements):Update.
4848 * hsa-regalloc.c: Include basic-block.h.
4849 * ipa-chkp.c (chkp_produce_thunks): Update.
4850 * ipa-cp.c (struct caller_statistics): Update.
4851 (init_caller_stats): Update.
4852 (gather_caller_stats): Update.
4853 (ipcp_cloning_candidate_p): Update.
4854 (good_cloning_opportunity_p): Update.
4855 (get_info_about_necessary_edges): Update.
4856 (dump_profile_updates): Update.
4857 (update_profiling_info): Update.
4858 (update_specialized_profile): Update.
4859 (perhaps_add_new_callers): Update.
4860 (decide_about_value): Update.
4861 (ipa_cp_c_finalize): Update.
4862 * ipa-devirt.c (struct odr_type_warn_count): Update.
4863 (struct decl_warn_count): Update.
4864 (struct final_warning_record): Update.
4865 (possible_polymorphic_call_targets): Update.
4866 (ipa_devirt): Update.
4867 * ipa-fnsummary.c (redirect_to_unreachable): Update.
4868 * ipa-icf.c (sem_function::merge): Update.
4869 * ipa-inline-analysis.c (do_estimate_edge_time): Update.
4870 * ipa-inline.c (compute_uninlined_call_time): Update.
4871 (compute_inlined_call_time): Update.
4872 (want_inline_small_function_p): Update.
4873 (want_inline_self_recursive_call_p): Update.
4874 (edge_badness): Update.
4875 (lookup_recursive_calls): Update.
4876 (recursive_inlining): Update.
4877 (inline_small_functions): Update.
4878 (dump_overall_stats): Update.
4879 (dump_inline_stats): Update.
4880 * ipa-profile.c (ipa_profile_generate_summary): Update.
4881 (ipa_propagate_frequency): Update.
4882 (ipa_profile): Update.
4883 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
4884 * ipa-utils.c (ipa_merge_profiles): Update.
4885 * loop-doloop.c (doloop_modify): Update.
4886 * loop-unroll.c (report_unroll): Update.
4887 (unroll_loop_runtime_iterations): Update.
4888 * lto-cgraph.c (lto_output_edge): Update.
4889 (lto_output_node): Update.
4890 (input_node): Update.
4891 (input_edge): Update.
4892 (merge_profile_summaries): Update.
4893 * lto-streamer-in.c (input_cfg): Update.
4894 * lto-streamer-out.c (output_cfg): Update.
4895 * mcf.c (create_fixup_graph): Update.
4896 (adjust_cfg_counts): Update.
4897 (sum_edge_counts): Update.
4898 * modulo-sched.c (sms_schedule): Update.
4899 * postreload-gcse.c (eliminate_partially_redundant_load): Update.
4900 * predict.c (maybe_hot_count_p): Update.
4901 (probably_never_executed): Update.
4902 (dump_prediction): Update.
4903 (combine_predictions_for_bb): Update.
4904 (propagate_freq): Update.
4905 (handle_missing_profiles): Update.
4906 (counts_to_freqs): Update.
4907 (rebuild_frequencies): Update.
4908 (force_edge_cold): Update.
4909 * predict.h: Include profile-count.h
4910 (maybe_hot_count_p, counts_to_freqs): UPdate.
4911 * print-rtl-function.c: Do not include cfg.h
4912 * print-rtl.c: Include basic-block.h
4913 * profile-count.c: New file.
4914 * profile-count.h: New file.
4915 * profile.c (is_edge_inconsistent): Update.
4916 (correct_negative_edge_counts): Update.
4917 (is_inconsistent): Update.
4918 (set_bb_counts): Update.
4919 (read_profile_edge_counts): Update.
4920 (compute_frequency_overlap): Update.
4921 (compute_branch_probabilities): Update; Initialize and deinitialize
4923 (branch_prob): Update.
4924 * profile.h (bb_gcov_counts, edge_gcov_counts): New.
4925 (edge_gcov_count): New.
4926 (bb_gcov_count): New.
4927 * shrink-wrap.c (try_shrink_wrapping): Update.
4928 * tracer.c (better_p): Update.
4929 * trans-mem.c (expand_transaction): Update.
4930 (ipa_tm_insert_irr_call): Update.
4931 (ipa_tm_insert_gettmclone_call): Update.
4932 * tree-call-cdce.c: Update.
4933 * tree-cfg.c (gimple_duplicate_sese_region): Update.
4934 (gimple_duplicate_sese_tail): Update.
4935 (gimple_account_profile_record): Update.
4936 (execute_fixup_cfg): Update.
4937 * tree-inline.c (copy_bb): Update.
4938 (copy_edges_for_bb): Update.
4939 (initialize_cfun): Update.
4940 (freqs_to_counts): Update.
4941 (copy_cfg_body): Update.
4942 (expand_call_inline): Update.
4943 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): Update.
4944 * tree-ssa-loop-ivcanon.c (unloop_loops): Update.
4945 (try_unroll_loop_completely): Update.
4946 (try_peel_loop): Update.
4947 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Update.
4948 * tree-ssa-loop-niter.c (estimate_numbers_of_iterations_loop): Update.
4949 * tree-ssa-loop-split.c (connect_loops): Update.
4950 * tree-ssa-loop-unswitch.c (hoist_guard): Update.
4951 * tree-ssa-reassoc.c (branch_fixup): Update.
4952 * tree-ssa-tail-merge.c (replace_block_by): Update.
4953 * tree-ssa-threadupdate.c (create_block_for_threading): Update.
4954 (compute_path_counts): Update.
4955 (update_profile): Update.
4956 (recompute_probabilities): Update.
4957 (update_joiner_offpath_counts): Update.
4958 (estimated_freqs_path): Update.
4959 (freqs_to_counts_path): Update.
4960 (clear_counts_path): Update.
4961 (ssa_fix_duplicate_block_edges): Update.
4962 (duplicate_thread_path): Update.
4963 * tree-switch-conversion.c (case_bit_test_cmp): Update.
4964 (struct switch_conv_info): Update.
4965 * tree-tailcall.c (decrease_profile): Update.
4966 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Update.
4967 * tree-vect-loop.c (scale_profile_for_vect_loop): Update.
4968 * value-prof.c (check_counter): Update.
4969 (gimple_divmod_fixed_value): Update.
4970 (gimple_mod_pow2): Update.
4971 (gimple_mod_subtract): Update.
4972 (gimple_ic_transform): Update.
4973 (gimple_stringop_fixed_value): Update.
4974 * value-prof.h (gimple_ic): Update.
4976 2017-06-02 Carl Love <cel@us.ibm.com>
4978 * config/rs6000/rs6000-c: Add support for built-in functions
4979 vector double vec_doublee (vector signed int);
4980 vector double vec_doublee (vector unsigned int);
4981 vector double vec_doublee (vector float);
4982 vector double vec_doubleh (vector signed int);
4983 vector double vec_doubleh (vector unsigned int);
4984 vector double vec_doubleh (vector float);
4985 vector double vec_doublel (vector signed int);
4986 vector double vec_doublel (vector unsigned int);
4987 vector double vec_doublel (vector float);
4988 vector double vec_doubleo (vector signed int);
4989 vector double vec_doubleo (vector unsigned int);
4990 vector double vec_doubleo (vector float);.
4991 * config/rs6000/rs6000-builtin.def: Add definitions for DOUBLEE,
4992 DOUBLEO, DOUBLEH, DOUBLEL, UNS_DOUBLEO, UNS_DOUBLEE, UNS_DOUBLEH,
4994 * config/rs6000/altivec.md: Add code generator for doublee<mode>2,
4995 unsdoubleev4si2, doubleo<mode>2, unsdoubleov4si2, doubleh<mode>2,
4996 unsdoublehv4si2, doublel<mode>2, unsdoublelv4si2, add mode attribute
4998 * config/rs6000/altivec.h: Add define for vec_doublee, vec_doubleo,
4999 vec_doublel, vec_doubleh.
5000 * doc/extend.texi: Update the built-in documentation file for the
5001 new built-in functions.
5003 2017-06-02 David Malcolm <dmalcolm@redhat.com>
5006 * ipa-inline-analysis.c (free_growth_caches): Set
5007 edge_removal_hook_holder to NULL after removing it.
5009 2017-06-02 Sudakshina Das <sudi.das@arm.com>
5011 * config/aarch64/aarch64.c (aarch64_select_cc_mode): Return CC_SWP for
5012 comparision with zero.
5014 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
5015 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5016 for early expansion of vec_min and vec_max builtins.
5017 (builtin_function_type): Add min/max unsigned variants to those
5018 identified as having unsigned arguments.
5020 2017-06-02 Olivier Hainque <hainque@adacore.com>
5022 * config/vx-common.h (DWARF_UNWIND_INFO): Switch #define to 1.
5024 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5026 * config/aarch64/aarch64-simd.md (*aarch64_simd_vec_copy_lane<mode>):
5027 Use VALL_F16 iterator rather than VALL.
5029 2017-06-02 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
5031 * config/aarch64/aarch64.c (aarch64_split_compare_and_swap):
5032 Emit CBNZ inside loop when doing a strong exchange and comparing
5033 against zero. Generate the CC flags after the loop.
5035 2017-06-02 David Edelsohn <dje.gcc@gmail.com>
5037 * dwarf2out.c (DWARF_INITIAL_LENGTH_SIZE_STR): New.
5038 (dl_section_ref): New.
5039 (dwarf2out_finish): Copy debug_line_section_label to dl_section_ref.
5040 On AIX, append an expression to subtract the size of the
5041 section length to dl_section_ref.
5043 2017-06-02 Will Schmidt <will_schmidt@vnet.ibm.com>
5045 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling
5046 for early expansion of vector absolute builtins.
5048 2017-06-02 Richard Biener <rguenther@suse.de>
5050 * tree-vect-slp.c (vect_detect_hybrid_slp_2): Match up
5051 what we consider a relevant use stmt with vect_detect_hybrid_slp_stmts.
5053 2017-06-02 Richard Biener <rguenther@suse.de>
5055 PR tree-optimization/80948
5056 * tree-tailcall.c (find_tail_calls): Track stmts to move in
5059 2017-06-02 Richard Biener <rguenther@suse.de>
5061 * tree-vect-loop.c (vect_analyze_loop_operations): Not relevant
5063 * tree-vect-stmts.c (process_use): Do not mark backedge defs
5064 for inductions as relevant.
5066 2017-06-02 Richard Biener <rguenther@suse.de>
5068 * tree-vect-loop.c (get_initial_def_for_induction): Inline into ...
5069 (vectorizable_induction): ... this. Remove dead code.
5071 2017-06-02 Eric Botcazou <ebotcazou@adacore.com>
5073 * builtins. (expand_builtin_alloca): Remove second parameter and
5074 infer its value from the first parameter instead.
5075 (expand_builtin) <BUILT_IN_ALLOCA>: Adjust call to above.
5077 2017-06-02 Jakub Jelinek <jakub@redhat.com>
5079 PR rtl-optimization/80903
5080 * loop-doloop.c (add_test): Unshare sequence.
5082 2017-06-02 Bernd Edlinger <bernd.edlinger@hotmail.de>
5084 * doc/invoke.texi: Document the -Wsizeof-pointer-div warning.
5086 2017-06-01 Bernd Edlinger <bernd.edlinger@hotmail.de>
5088 * config/i386/i386.c (x86_64_ms_sysv_extra_clobbered_registers): Make
5090 (xlogue_layout::get_stack_space_used, xlogue_layout::s_instances,
5091 xlogue_layout::get_instance, logue_layout::xlogue_layout,
5092 sp_valid_at, fp_valid_at, choose_basereg): Formatting.
5093 (xlogue_layout::get_stub_rtx): Make static.
5094 (xlogue_layout::get_stub_name): Avoid const-cast, make static.
5095 (xlogue_layout::compute_stub_managed_regs): Rename to...
5096 (xlogue_layout::count_stub_managed_regs): ...this.
5097 (xlogue_layout::is_stub_managed_reg): New function.
5098 (xlogue_layout::m_stub_names): Rename to...
5099 (xlogue_layout::s_stub_names): ...this, make static.
5100 (xlogue_layout::STUB_INDEX_OFFSET, xlogue_layout::MIN_REGS,
5101 xlogue_layout::MAX_REGS, xlogue_layout::MAX_EXTRA_REGS,
5102 xlogue_layout::VARIANT_COUNT, xlogue_layout::STUB_NAME_MAX_LEN,
5103 xlogue_layout::s_stub_names): Instantiate statics.
5104 (stub_managed_regs): Remove.
5105 (ix86_save_reg): Use xlogue_layout::compute_stub_managed_regs.
5106 (disable_call_ms2sysv_xlogues): Rename to...
5107 (warn_once_call_ms2sysv_xlogues): ...this, and warn only once.
5108 (ix86_initial_elimination_offset, ix86_expand_call): Fix call_ms2sysv
5110 (ix86_static_chain): Make sure that ix86_static_chain_on_stack can't
5111 change after reload_completed.
5112 (ix86_can_use_return_insn_p): Use the ix86_frame data structure
5114 (ix86_expand_prologue): Likewise.
5115 (ix86_expand_epilogue): Likewise.
5116 (ix86_expand_split_stack_prologue): Likewise.
5117 (ix86_compute_frame_layout): Remove frame parameter ...
5118 (TARGET_COMPUTE_FRAME_LAYOUT): ... and export it as a target hook.
5119 (ix86_finalize_stack_realign_flags): Call ix86_compute_frame_layout
5121 (ix86_init_machine_status): Don't set use_fast_prologue_epilogue_nregs.
5122 (ix86_frame): Move from here ...
5123 * config/i386/i386.h (ix86_frame): ... to here.
5124 (machine_function): Remove use_fast_prologue_epilogue_nregs, cache the
5125 complete ix86_frame data structure instead. Remove some_ld_name.
5127 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
5129 * dwarf2out.c (dwarf2out_late_global_decl): Add locations for
5130 symbols that hold a DECL_VALUE_EXPR.
5132 2017-06-01 Martin Jambor <mjambor@suse.cz>
5134 PR tree-optimization/80898
5135 * tree-sra.c (process_subtree_disqualification): Removed.
5136 (disqualify_candidate): Do not acll
5137 process_subtree_disqualification.
5138 (subtree_mark_written_and_enqueue): New function.
5139 (propagate_all_subaccesses): Set grp_write of LHS subtree if the
5140 RHS has been disqualified and re-queue LHS if necessary. Apart
5141 from that, ignore disqualified RHS.
5143 2017-06-01 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5145 * config/s390/s390.c (s390_emit_epilogue): Disable early return
5146 address fetch for z10 or later.
5148 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
5150 * config/arc/arc.md (tst_movb): Add guard when splitting.
5152 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
5154 * config/arc/arc.c (arc_can_eliminate): Test against
5155 arc_frame_pointer_needed.
5157 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
5159 * config/arc/arc.c (arc_expand_prologue): Emit a special barrier
5160 to prevent store reordering.
5161 * config/arc/arc.md (UNSPEC_ARC_STKTIE): Define.
5162 (type): Add block type.
5163 (stack_tie): Define special instruction to be used in
5166 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
5168 * config/arc/arc.md (commutative_binary_comparison): Remove 'I'
5169 constraint. It is not valid for the pattern.
5170 (noncommutative_binary_comparison): Likewise.
5172 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
5174 * config/arc/simdext.md (movv2hi_insn): Change predicate to avoid
5177 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
5179 * config/arc/arc.c (arc_conditional_register_usage): Allow r30 to
5180 be used by the reg-alloc.
5182 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
5184 * config/arc/arc.md (mulsi3): Avoid use of hard registers before
5185 reg-alloc when having mul64 or mul32x16 instructions.
5186 (mulsidi3): Likewise.
5187 (umulsidi3): Likewise.
5188 (mulsi32x16): New pattern.
5189 (mulsi64): Likewise.
5190 (mulsidi64): Likewise.
5191 (umulsidi64): Likewise.
5192 (MUL32x16_REG): Define.
5193 (mul64_600): Use MUL32x16_REG.
5194 (mac64_600): Likewise.
5195 (umul64_600): Likewise.
5196 (umac64_600): Likewise.
5198 2017-06-01 Claudiu Zissulescu <claziss@synopsys.com>
5200 * config/arc/arc.md (mulsi3_700): Make it commutative.
5202 2017-06-01 Jose E. Marchesi <jose.marchesi@oracle.com>
5204 * config/sparc/sparc.md (*zero_extendsidi2_insn_sp64): Set insn
5206 (*sign_extendsidi2_insn): Likewise for movstosw.
5208 2017-06-01 Pierre-Marie de Rodat <derodat@adacore.com>
5210 * dwarf2out.c (get_discr_value): Call the get_debug_type hook on
5211 the type of the input discriminant value. Convert the
5212 discriminant value of signedness vary.
5214 2017-06-01 Volker Reichelt <v.reichelt@netcologne.de>
5216 * doc/invoke.texi (-Wcatch-value): Document new shortcut.
5217 Add to -Wall section.
5219 2017-06-01 Richard Biener <rguenther@suse.de>
5222 * fold-const.c (fold_plusminus_mult_expr): If the factored
5223 factor may be zero use a wrapping type for the inner operation.
5224 * tree-tailcall.c (independent_of_stmt_p): Pass in to_move bitmap
5225 and handle moved defs.
5226 (process_assignment): Properly guard the unary op case. Return a
5227 tri-state indicating that moving the stmt before the call may allow
5228 to continue. Pass through to_move.
5229 (find_tail_calls): Handle moving unrelated defs before
5232 2017-05-31 Segher Boessenkool <segher@kernel.crashing.org>
5235 * config/rs6000/vector.md (*vector_uneq<mode>): Write the nor in the
5236 splitter result in the canonical way.
5238 2017-05-31 Uros Bizjak <ubizjak@gmail.com>
5240 * config/i386/i386.md (*zero_extendsidi2): Enable alternative (?r, *Yj)
5241 also for 32bit target. Update insn attributes.
5242 (zero-extendsidi2 splitter): Allow all registers for operand 1.
5244 2017-05-31 Sebastian Peryt <sebastian.peryt@intel.com>
5246 * config/i386/avx512fintrin.h (_mm_mask_max_sd)
5247 (_mm_maskz_max_sd, _mm_mask_max_ss, _mm_maskz_max_ss)
5248 (_mm_mask_min_sd, _mm_maskz_min_sd, _mm_mask_min_ss)
5249 (_mm_maskz_min_ss): New intrinsics.
5251 2017-05-31 Martin Liska <mliska@suse.cz>
5253 * tree-vect-loop.c (vect_create_epilog_for_reduction):
5254 Change comment style to one we normally use.
5255 (vectorizable_reduction): Likewise.
5256 (vectorizable_induction): Likewise.
5257 * tree-vect-stmts.c (vectorizable_mask_load_store): Likewise.
5258 (vectorizable_call): Likewise.
5259 (vectorizable_simd_clone_call): Likewise.
5260 (vectorizable_conversion): Likewise.
5261 (vectorizable_assignment): Likewise.
5262 (vectorizable_shift): Likewise.
5263 (vectorizable_operation): Likewise.
5264 (vectorizable_store): Likewise.
5265 (vectorizable_load): Likewise.
5266 * tree-vectorizer.h: Likewise.
5268 2017-05-31 Alexander Monakov <amonakov@ispras.ru>
5270 * passes.c (emergency_dump_function): New.
5271 * tree-pass.h (emergency_dump_function): Declare.
5272 * plugin.c (plugins_internal_error_function): Remove.
5273 * plugin.h (plugins_internal_error_function): Remove declaration.
5274 * toplev.c (internal_error_function): New static function. Use it...
5275 (general_init): ...here.
5277 2017-05-31 Graham Markall <graham.markall@embecosm.com>
5279 * config/arc/arc.c (arc_print_operand): Handle constant operands.
5280 (arc_rtx_costs): Add costs for new patterns.
5281 * config/arc/arc.md: Additional *add_n and *sub_n patterns.
5282 * config/arc/predicates.md: Add _1_2_3_operand predicate.
5284 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
5286 * tree-ssa-strlen.c (get_next_strinfo): New function.
5287 (get_stridx_plus_constant): Use it.
5288 (zero_length_string): Likewise.
5289 (adjust_related_strinfos): Likewise.
5290 (adjust_last_stmt): Likewise.
5292 2017-05-31 Richard Biener <rguenther@suse.de>
5295 * config/i386/i386.c (ix86_expand_builtin): Remove assert
5296 for arg being an SSA name when expanding IX86_BUILTIN_BNDRET.
5298 2017-05-31 Richard Sandiford <richard.sandiford@linaro.org>
5300 * tree-vect-data-refs.c (vect_find_same_alignment_drs): Remove
5301 loop_vinfo argument and use of dependence distance vectors.
5302 Check instead whether the two references differ only in their
5303 initial value and assume that they have the same alignment if the
5304 difference is a multiple of the vector alignment.
5305 (vect_analyze_data_refs_alignment): Update call accordingly.
5307 2017-05-31 Martin Liska <mliska@suse.cz>
5310 * config/i386/cpuid.h: Fix typo in a comment in cpuid.h.
5312 2017-05-31 Bin Cheng <bin.cheng@arm.com>
5314 * tree-vect-loop-manip.c (create_intersect_range_checks_index)
5315 (create_intersect_range_checks): Move from ...
5316 * tree-data-ref.c (create_intersect_range_checks_index)
5317 (create_intersect_range_checks): ... to here.
5318 (create_runtime_alias_checks): New function factored from ...
5319 * tree-vect-loop-manip.c (vect_create_cond_for_alias_checks): ...
5320 here. Call above function.
5321 * tree-data-ref.h (create_runtime_alias_checks): New function.
5323 2017-05-31 Bin Cheng <bin.cheng@arm.com>
5325 * tree-data-ref.c (prune_runtime_alias_test_list): Relax minimal
5326 segment length for dr_b and compute it in wide_int.
5328 2017-05-31 Richard Biener <rguenther@suse.de>
5330 PR tree-optimization/80906
5331 * graphite-isl-ast-to-gimple.c (copy_loop_close_phi_nodes): Get
5332 and pass through iv_map.
5333 (copy_bb_and_scalar_dependences): Adjust.
5334 (translate_pending_phi_nodes): Likewise.
5335 (copy_loop_close_phi_args): Handle code-generating IVs instead
5338 2017-05-30 David Malcolm <dmalcolm@redhat.com>
5340 * diagnostic-color.c (color_dict): Add "type-diff".
5341 (parse_gcc_colors): Update comment.
5342 * doc/invoke.texi (Diagnostic Message Formatting Options): Add
5343 -fdiagnostics-show-template-tree and -fno-elide-type.
5344 (GCC_COLORS): Add type-diff to example.
5346 (-fdiagnostics-show-template-tree): New.
5347 (-fno-elide-type): New.
5348 * pretty-print.c (pp_format): Pass quote and formatters[argno] to
5349 the pp_format_decoder callback. Call any m_format_postprocessor's
5351 (pretty_printer::pretty_printer): Initialize
5352 m_format_postprocessor.
5353 (pretty_printer::~pretty_printer): Delete any
5354 m_format_postprocessor.
5355 * pretty-print.h (printer_fn): Add bool and const char ** parameters.
5356 (class format_postprocessor): New class.
5357 (struct pretty_printer::format_decoder): Document the new parameters.
5358 (struct pretty_printer::m_format_postprocessor): New field.
5359 * tree-diagnostic.c (default_tree_printer): Update for new
5360 bool and const char ** params.
5361 * tree-diagnostic.h (default_tree_printer): Likewise.
5363 2017-05-30 Segher Boessenkool <segher@kernel.crashing.org>
5365 * config/rs6000/predicates.md (cc_reg_not_micro_cr0_operand): Delete.
5366 (lwa_operand): Delete rs6000_gen_cell_microcode test.
5367 * config/rs6000/rs6000.c (rs6000_option_override_internal): Delete
5368 rs6000_gen_cell_microcode code.
5369 (rs6000_final_prescan_insn): Delete.
5370 (rs6000_opt_vars): Delete the "gen-cell-microcode" and
5371 "warn-cell-microcode" entries.
5372 * config/rs6000/rs6000.h (FINAL_PRESCAN_INSN): Delete.
5373 * config/rs6000/rs6000.md: Delete rs6000_gen_cell_microcode tests
5374 throughout. Change cc_reg_not_micro_cr0_operand to
5375 cc_reg_not_cr0_operand throughout.
5376 (*extendhi<mode>2_noload): Delete.
5377 * config/rs6000/rs6000.opt (mgen-cell-microcode): Replace by stub.
5378 (mwarn-cell-microcode): Delete.
5379 * doc/invoke.texi (RS/6000 and PowerPC Options): Delete
5380 -mgen-cell-microcode and -mwarn-cell-microcode.
5382 2017-05-30 Uros Bizjak <ubizjak@gmail.com>
5385 * config/i386/constraints.md (Yd): New constraint.
5387 * config/i386/i386.md (*movti_internal): Add (?r, Ye)
5388 and (?Yd, r) alternatives. Update insn attributes.
5389 * config/i386/i386.md (*movti_internal): Add (?r, *Ye)
5390 and (?*Yd, r) alternatives. Update insn attributes.
5391 (double-mode inter-unit splitters): Add new GR<->XMM splitters.
5393 2017-05-30 Pierre-Marie de Rodat <derodat@adacore.com>
5395 * gimplify.c (gimplify_modify_expr): Don't create a
5396 DECL_DEBUG_EXPR link if *FROM_P does not belong to the current
5399 2017-05-30 Wilco Dijkstra <wdijkstr@arm.com>
5401 * config/arm/arm-builtins.c (arm_expand_builtin): Remove const.
5403 2017-05-30 Richard Biener <rguenther@suse.de>
5405 * tree-vectorizer.h (struct _stmt_vec_info): Add reduc_type
5406 and reduc_def fields.
5407 (STMT_VINFO_REDUC_TYPE): New define.
5408 (STMT_VINFO_REDUC_DEF): Likewise.
5409 (vect_force_simple_reduction): Adjust prototype.
5410 * tree-vect-loop.c (vect_analyze_scalar_cycles_1): Adjust.
5411 (vect_is_simple_reduction): Remove check_reduction argument.
5412 (vect_force_simple_reduction): Adjust and set
5413 STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
5414 (vectorizable_reduction): Do not re-do reduction analysis
5415 but use STMT_VINFO_REDUC_TYPE and STMT_VINFO_REDUC_DEF.
5416 * tree-parloops.c (gather_scalar_reductions): Adjust.
5418 2017-05-30 Richard Biener <rguenther@suse.de>
5421 * cfgexpand.c (expand_gimple_cond): Match up loop fixup with
5424 2017-05-24 Robin Dapp <rdapp@linux.vnet.ibm.com>
5426 * tree-vect-data-refs.c (vect_get_peeling_costs_all_drs):
5427 Introduce unknown_misalignment parameter and remove vf.
5428 (vect_peeling_hash_get_lowest_cost):
5429 Pass unknown_misalignment parameter.
5430 (vect_enhance_data_refs_alignment):
5431 Fix unsupportable data ref treatment.
5433 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
5435 * tree-vect-data-refs.c (vect_get_data_access_cost):
5436 Workaround for SLP handling.
5437 (vect_enhance_data_refs_alignment):
5438 Compute costs for doing no peeling at all, compare to the best
5439 peeling costs so far and avoid peeling if cheaper.
5441 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
5443 * tree-vect-data-refs.c (vect_peeling_hash_choose_best_peeling):
5444 Return peeling info and set costs to zero for unlimited cost
5446 (vect_enhance_data_refs_alignment): Also inspect all datarefs
5447 with unknown misalignment. Compute and costs for unknown
5448 misalignment, compare them to the costs for known misalignment
5449 and choose the cheapest for peeling.
5451 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
5453 * tree-vect-data-refs.c (vect_update_misalignment_for_peel): Rename.
5454 (vect_get_peeling_costs_all_drs): Create function.
5455 (vect_peeling_hash_get_lowest_cost):
5456 Use vect_get_peeling_costs_all_drs.
5457 (vect_peeling_supportable): Create function.
5458 (vect_enhance_data_refs_alignment): Use vect_peeling_supportable.
5460 2017-05-30 Robin Dapp <rdapp@linux.vnet.ibm.com>
5462 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Create
5463 DR_HAS_NEGATIVE_STEP.
5464 (vect_update_misalignment_for_peel): Define DR_MISALIGNMENT.
5465 (vect_enhance_data_refs_alignment): Use.
5466 (vect_duplicate_ssa_name_ptr_info): Use.
5467 * tree-vectorizer.h (dr_misalignment): Use.
5468 (known_alignment_for_access_p): Use.
5470 2017-05-30 Jozef Lawrynowicz <jozef.l@somniumtech.com>
5473 * config/msp430/msp430.c (gen_prefix): Return NULL when section name is
5475 (has_section_name): New function.
5477 2017-05-30 Martin Liska <mliska@suse.cz>
5480 * auto-profile.c (get_function_decl_from_block): Fix
5483 2017-05-30 Richard Biener <rguenther@suse.de>
5486 * cfgexpand.c (expand_gimple_cond): Fixup preserving loops again.
5488 2017-05-30 Martin Liska <mliska@suse.cz>
5490 * dumpfile.c: Use newly added macro DUMP_FILE_INFO.
5491 * dumpfile.h (struct dump_file_info): Remove ctors.
5493 2017-05-30 Martin Liska <mliska@suse.cz>
5495 * predict.def: Fix GNU coding style.
5497 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
5499 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
5500 Mark 'to' argument with ATTRIBUTE_UNUSED.
5502 2017-05-29 Max Filippov <jcmvbkbc@gmail.com>
5504 * config/xtensa/xtensa.c (xtensa_emit_call): Use
5505 HOST_WIDE_INT_PRINT_HEX instead of 0x%lx format string.
5506 (print_operand): Use HOST_WIDE_INT_PRINT_DEC instead of %ld
5509 2017-05-29 Eric Botcazou <ebotcazou@adacore.com>
5511 * doc/install.texi (Options specification): Restore entry of
5512 --enable-sjlj-exceptions.
5514 2017-05-27 Michael Eager <eager@eagercon.com>
5517 2016-01-21 Ajit Agarwal <ajitkum@xilinx.com>
5519 See https://gcc.gnu.org/ml/gcc/2017-05/msg00221.html.
5521 * config/microblaze/microblaze.h
5522 (FIXED_REGISTERS): Update in macro.
5523 (CALL_USED_REGISTERS): Update in macro.
5525 2017-05-27 François-Xavier Coudett <fxcoudert@gcc.gnu.org>
5527 * doc/install.texi: Add links to macOS binary distributions.
5529 2017-05-27 Jakub Jelinek <jakub@redhat.com>
5533 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
5535 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
5537 2017-05-26 Martin Liska <mliska@suse.cz>
5539 * dumpfile.h (enum dump_kind): Renumber TDF_* flags to be contiguous.
5541 2017-05-26 Martin Liska <mliska@suse.cz>
5543 * cfg.c (check_bb_profile): Do not use TDF_COMMENT and print
5544 always leading ';; '.
5545 (dump_bb_info): Likewise.
5546 (brief_dump_cfg): Likewise.
5547 * cfgrtl.c (print_rtl_with_bb): Do not use TDF_COMMENT.
5548 * dumpfile.c: Remove usage of TDF_VERBOSE.
5549 * dumpfile.h (enum dump_kind): Likewise.
5550 (dump_gimple_bb_header): Do not use TDF_COMMENT.
5551 * print-tree.c (debug_verbose): Remove.
5552 * tree-cfg.c (gimple_dump_cfg): Do not use TDF_COMMENT.
5553 (dump_function_to_file): Remove dumps guarded with TDF_VERBOSE.
5554 * tree-diagnostic.c (default_tree_printer): Replace
5555 TDF_DIAGNOSTIC with TDF_SLIM.
5557 2017-05-26 Bin Cheng <bin.cheng@arm.com>
5559 * tree-vect-loop-manip.c (create_intersect_range_checks_index): Pass
5560 in parameter loop, rather than loop_vinfo.
5561 (create_intersect_range_checks): Ditto.
5562 (vect_create_cond_for_alias_checks): Update call to above functions.
5564 2017-05-26 Bin Cheng <bin.cheng@arm.com>
5566 PR tree-optimization/80815
5567 * tree-data-ref.c (prune_runtime_alias_test_list): Simplify condition
5568 for merging runtime alias checks. Handle negative DR_STEPs.
5570 2017-05-26 Bin Cheng <bin.cheng@arm.com>
5572 * tree-vect-data-refs.c (Operator==, comp_dr_with_seg_len_pair):
5574 * tree-data-ref.c (Operator==, comp_dr_with_seg_len_pair): To here.
5575 * tree-vect-data-refs.c (vect_prune_runtime_alias_test_list): Factor
5576 out code pruning runtime alias checks.
5577 * tree-data-ref.c (prune_runtime_alias_test_list): New function
5578 factored out from above.
5579 * tree-vectorizer.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
5581 * tree-data-ref.h (struct dr_with_seg_len, dr_with_seg_len_pair_t):
5583 (prune_runtime_alias_test_list): New decalaration.
5585 2017-05-26 Bin Cheng <bin.cheng@arm.com>
5587 * tree-vect-data-refs.c (compare_tree): Rename and move ...
5588 * tree-data-ref.c (data_ref_compare_tree): ... to here.
5589 * tree-data-ref.h (data_ref_compare_tree): New decalaration.
5590 * tree-vect-data-refs.c (dr_group_sort_cmp): Update uses.
5591 (operator==, comp_dr_with_seg_len_pair): Ditto.
5592 (vect_prune_runtime_alias_test_list): Ditto.
5594 2017-05-26 Martin Liska <mliska@suse.cz>
5597 * params.def: Bound partial-inlining-entry-probability param.
5599 2017-05-26 Marek Polacek <polacek@redhat.com>
5602 * fold-const.c (fold_binary_loc) <case MULT_EXPR>: Check if OP1
5605 2017-05-26 Richard Biener <rguenther@suse.de>
5607 PR tree-optimization/80842
5608 * tree-ssa-ccp.c (set_lattice_value): Always meet with the old
5611 2017-05-26 Richard Biener <rguenther@suse.de>
5613 PR tree-optimization/80844
5614 * tree-vectorizer.c (adjust_simduid_builtins): Propagate results.
5616 2017-05-25 Sebastian Peryt <sebastian.peryt@intel.com>
5618 * doc/md.texi (Machine Constraints): Update x86 family
5619 machine constraints section to match 'config/i386/constraints.md'.
5621 2017-05-25 Volker Reichelt <v.reichelt@netcologne.de>
5623 * doc/invoke.texi (-Wcatch-value=): Document new warning option.
5625 2017-05-25 Nathan Sidwell <nathan@acm.org>
5627 * doc/invoke.texi (--enable-languages): Update documentation.
5629 2017-05-25 Martin Liska <mliska@suse.cz>
5631 * dumpfile.c: Add TDF_FOLDING.
5632 * dumpfile.h (enum dump_kind): Likewise.
5633 * genmatch.c (dt_simplify::gen_1): Use it.
5635 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
5637 * match.pd (view_convert (convert@0 @1)): Handle zero-extension.
5639 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
5641 * match.pd ((A +- CST1) +- CST2): Allow some conversions.
5642 * tree.c (drop_tree_overflow): Handle COMPLEX_CST and VECTOR_CST.
5644 2017-05-25 Marc Glisse <marc.glisse@inria.fr>
5646 * fold-const.c (fold_binary_loc) [(A & C) == D]: Remove transformation.
5647 * match.pd (X == C): Rewrite it here.
5648 (with_possible_nonzero_bits, with_possible_nonzero_bits2,
5649 with_certain_nonzero_bits2): New predicates.
5650 * tree-ssanames.c (get_nonzero_bits): Handle INTEGER_CST.
5652 2017-05-24 Nathan Sidwell <nathan@acm.org>
5654 * lto-streamer-in.c (lto_input_data_block): Adjust T const cast to
5657 * auto-profile.c (afdo_propagate): Adjust T const cast to avoid
5660 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
5662 * config/powerpcspe: New port. Files are copied from the rs6000
5663 port, with "rs6000" in filenames replaced by "powerpcspe".
5665 2017-05-24 Wilco Dijkstra <wdijkstr@arm.com>
5667 PR rtl-optimization/80754
5668 * lra-remat.c (do_remat): Add overlap checks for dst_regno.
5670 2017-05-24 Sheldon Lobo <smlobo@sheldon.us.oracle.com>
5672 * config/sparc/sparc.md (length): Return the correct value for -mflat
5673 sibcalls to match output_sibcall.
5675 2017-05-24 Segher Boessenkool <segher@kernel.crashing.org>
5679 * config/rs6000/rs6000.c (struct machine_function): Add new field
5681 (rs6000_get_separate_components): Init that field, use it.
5682 (rs6000_components_for_bb): Use the field.
5684 2017-05-24 Thomas Preud'homme <thomas.preudhomme@arm.com>
5686 * config/arm/arm.c (arm_expand_prologue): Fix typo in comment.
5688 2017-05-24 Peter Bergner <bergner@vnet.ibm.com>
5691 * tree-cfg.c (group_case_labels_stmt): Delete increment of "i";
5693 2017-05-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
5696 * config/s390/s390.c (s390_check_qrst_address): Check incoming
5697 address against address_operand predicate.
5698 * config/s390/s390.md ("*indirect_jump"): Swap alternatives.
5700 2017-05-24 Eric Botcazou <ebotcazou@adacore.com>
5702 * var-tracking.c (track_expr_p): Do not return 0 for tracked record
5703 parameters passed indirectly.
5705 2017-05-23 Uros Bizjak <ubizjak@gmail.com>
5707 * config/i386/i386.md (*movdi_internal): Remove SSE4
5708 alternative 18 (?r, *v). Update insn attributes.
5709 (*movsi_internal): Remove SSE4 alternative 13 (?r, *v).
5710 Update insn attributes.
5711 (*zero_extendsidi2): Remove SSE4 alternative (?r, *x).
5712 Update insn attributes.
5713 * config/i386/sse.md (vec_extract<ssevecmodelower>_0): Remove SSE4
5714 alternative 1 (r, v). Remove isa attribute.
5715 * config/i386/i386.c (dimode_scalar_chain::make_vector_copies):
5716 Always move value through stack for !TARGET_INTER_UNIT_MOVES_TO_VEC
5717 and !TARGET_INTER_UNIT_MOVES_TO_VEC targets.
5719 2017-05-23 Tom de Vries <tom@codesourcery.com>
5721 * doc/sourcebuild.texi (Directives, Verify compiler message): Document
5724 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
5726 * cgraphunit.c (symbol_table::process_new_functions): Update.
5727 * ipa-fnsummary.c (pass_data_inline_parameters): Remove.
5728 (inline_generate_summary): Rename to ...
5729 (ipa_fn_summary_generate): ... this one.
5730 (inline_read_summary): Rename to ...
5731 (ipa_fn_summary_read): ... this one.
5732 (inline_write_summary): Rename to ...
5733 (ipa_fn_summary_write): ... this one.
5734 (inline_free_summary): Rename to ...
5735 (ipa_free_fn_summary): ... this one.
5736 (pass_data_local_fn_summary, pass_local_fn_summary,
5737 make_pass_local_fn_summary, pass_data_ipa_free_fn_summary,
5738 pass_ipa_free_fn_summary, make_pass_ipa_free_fn_summary,
5739 pass_data_ipa_fn_summary, pass_ipa_fn_summary,
5740 make_pass_ipa_fn_summary): New.
5741 * ipa-fnsummary.h (inline_generate_summary, inline_read_summary,
5742 inline_write_summary, inline_free_summary): Remove.
5743 (ipa_free_fn_summary) : New.
5744 * ipa-inline.c (ipa_inline): Update.
5745 (pass_ipa_inline): Do not generate summaries.
5746 * ipa.c (pass_data_ipa_free_fn_summary, pass_ipa_free_fn_summary):
5748 * passes.def: Replace pass_inline_parameters by pass_local_fn_summary
5749 and add pass_ipa_fn_summary.
5750 * tree-pass.h (make_pass_ipa_fn_summary, make_pass_local_fn_summary):
5752 (make_pass_inline_parameters): Remove.
5754 2017-05-23 Thomas Schwinge <thomas@codesourcery.com>
5756 * omp-low.c (struct omp_context): Remove "default_kind" member.
5759 * omp-offload.c (execute_oacc_device_lower): Remove the
5760 parallelism dimensions function attributes for unparallelized
5761 OpenACC kernels constructs.
5763 2017-05-23 Martin Liska <mliska@suse.cz>
5765 * cgraph.c (cgraph_node::get_create): Use symtab_node::dump_{asm_,}name
5767 (cgraph_edge::make_speculative): Likewise.
5768 (cgraph_edge::resolve_speculation): Likewise.
5769 (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
5770 (cgraph_node::dump): Likewise.
5771 * cgraph.h: Likewise.
5772 * cgraphunit.c (analyze_functions): Likewise.
5773 (symbol_table::compile): Likewise.
5774 * ipa-cp.c (print_all_lattices): Likewise.
5775 (determine_versionability): Likewise.
5776 (initialize_node_lattices): Likewise.
5777 (ipcp_verify_propagated_values): Likewise.
5778 (estimate_local_effects): Likewise.
5779 (update_profiling_info): Likewise.
5780 (create_specialized_node): Likewise.
5781 (perhaps_add_new_callers): Likewise.
5782 (decide_about_value): Likewise.
5783 (decide_whether_version_node): Likewise.
5784 (identify_dead_nodes): Likewise.
5785 (ipcp_store_bits_results): Likewise.
5786 * ipa-devirt.c (dump_targets): Likewise.
5787 (ipa_devirt): Likewise.
5788 * ipa-icf.c (sem_item::dump): Likewise.
5789 (sem_function::equals): Likewise.
5790 (sem_variable::equals): Likewise.
5791 (sem_item_optimizer::read_section): Likewise.
5792 (sem_item_optimizer::execute): Likewise.
5793 (congruence_class::dump): Likewise.
5794 * ipa-inline-analysis.c (dump_ipa_call_summary): Likewise.
5795 (dump_inline_summary): Likewise.
5796 (estimate_node_size_and_time): Likewise.
5797 (inline_analyze_function): Likewise.
5798 * ipa-inline-transform.c (inline_call): Likewise.
5799 * ipa-inline.c (report_inline_failed_reason): Likewise.
5800 (want_early_inline_function_p): Likewise.
5801 (edge_badness): Likewise.
5802 (update_edge_key): Likewise.
5803 (inline_small_functions): Likewise.
5804 * ipa-profile.c (ipa_profile): Likewise.
5805 * ipa-prop.c (ipa_print_node_jump_functions): Likewise.
5806 (ipa_make_edge_direct_to_target): Likewise.
5807 (remove_described_reference): Likewise.
5808 (ipa_impossible_devirt_target): Likewise.
5809 (propagate_controlled_uses): Likewise.
5810 (ipa_print_node_params): Likewise.
5811 (ipcp_transform_function): Likewise.
5812 * ipa-pure-const.c (pure_const_read_summary): Likewise.
5813 (propagate_pure_const): Likewise.
5814 * ipa-reference.c (generate_summary): Likewise.
5815 (read_write_all_from_decl): Likewise.
5816 (propagate): Likewise.
5817 (ipa_reference_read_optimization_summary): Likewise.
5818 * ipa-utils.c (ipa_merge_profiles): Likewise.
5819 * ipa.c (walk_polymorphic_call_targets): Likewise.
5820 (symbol_table::remove_unreachable_nodes): Likewise.
5821 (ipa_single_use): Likewise.
5822 * passes.c (execute_todo): Likewise.
5823 * predict.c (drop_profile): Likewise.
5824 * symtab.c (symtab_node::get_dump_name): New function.
5825 (symtab_node::dump_name): Likewise.
5826 (symtab_node::dump_asm_name): Likewise.
5827 (symtab_node::dump_references): Likewise.
5828 (symtab_node::dump_referring): Likewise.
5829 (symtab_node::dump_base): Likewise.
5830 (symtab_node::debug_symtab): Likewise.
5831 * tree-sra.c (convert_callers_for_node): Likewise.
5832 * tree-ssa-structalias.c (ipa_pta_execute): Likewise.
5833 * value-prof.c (init_node_map): Likewise.
5835 2017-05-23 Martin Liska <mliska@suse.cz>
5837 * cgraph.h: Move symtab_node::dump_table to symbol_table::dump
5838 and symtab_node::debug_symtab to symbol_table::debug.
5839 * cgraphunit.c (analyze_functions): Use the renamed function.
5840 (symbol_table::compile): Likewise.
5841 * ipa-cp.c (ipcp_verify_propagated_values): Likewise.
5842 * ipa-icf.c (sem_item_optimizer::execute): Likewise.
5843 * passes.c (execute_todo): Likewise.
5844 * symtab.c (symbol_table::dump): New function.
5845 * tree-ssa-structalias.c (ipa_pta_execute): Use the renamed function.
5847 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
5849 * ipa-fnsummary.c (estimate_node_size_and_time): Do not sanity check
5850 that nonconst implies exec.
5852 2017-05-23 Jan Hubicka <hubicka@ucw.cz>
5854 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
5855 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
5856 (inline_edge_summary_vec): Turn into ...
5857 (ipa_call_summaries): ... this one.
5858 (redirect_to_unreachable, edge_set_predicate,
5859 evaluate_properties_for_edge, inline_summary_alloc,
5860 reset_ipa_call_summary, reset_inline_summary,
5861 inline_summary_t::duplicate): Update.
5862 (inline_edge_duplication_hook): Turn to ...
5863 (ipa_call_summary_t::duplicate): ... this one.
5864 (inline_edge_removal_hook): Turn to ...
5865 (ipa_call_summary_t::remove): ... this one.
5866 (dump_inline_edge_summary): Turn to ...
5867 (dump_ipa_call_summary): ... this one.
5868 (estimate_function_body_sizes): Update.
5869 (inline_update_callee_summaries): Update.
5870 (remap_edge_change_prob): Update.
5871 (remap_edge_summaries): Update.
5872 (inline_merge_summary): Update.
5873 (do_estimate_edge_time): Update.
5874 (inline_generate_summary): Update.
5875 (inline_read_section): Update.
5876 (inline_read_summary): Update.
5877 (inline_free_summary): Update.
5878 * ipa-inline.c (can_inline_edge_p): Update.
5879 (compute_inlined_call_time): Update.
5880 (want_inline_small_function_p): Update.
5881 (edge_badness): Update.
5882 (early_inliner): Update.
5883 * ipa-inline.h (inline_edge_summary): Turn to ...
5884 (ipa_call_summary): ... this one.
5885 (ipa_call_summary_t): New class.
5886 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
5887 (ipa_call_summaries): New.
5888 (inline_edge_summary): Remove.
5889 (estimate_edge_growth): Update.
5890 * ipa-profile.c (ipa_propagate_frequency_1): Update.
5891 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
5892 * ipa-split.c (execute_split_functions): Update.
5893 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
5895 2017-05-23 Tom de Vries <tom@codesourcery.com>
5897 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
5898 attributes): Document rdrand effective target.
5900 2017-05-23 Tom de Vries <tom@codesourcery.com>
5902 * doc/sourcebuild.texi (Effective-Target Keywords, Other hardware
5903 attributes): Sort alphabetically.
5905 2017-05-23 Georg-Johann Lay <avr@gjlay.de>
5907 * config/avr/genmultilib.awk: Use gsub instead of gensub.
5909 2017-05-22 Michael Meissner <meissner@linux.vnet.ibm.com>
5912 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Split
5913 V2DF/V2DI splat into two separate patterns, one that handles
5914 registers, and the other that only handles memory. Drop support
5915 for splatting from a GPR on ISA 2.07 and then splitting the
5916 splat into direct move and splat.
5917 (vsx_splat_<mode>_reg): Likewise.
5918 (vsx_splat_<mode>_mem): Likewise.
5920 2017-05-22 Segher Boessenkool <segher@kernel.crashing.org>
5922 * cfgcleanup.c (bb_is_just_return): Allow CLOBBERs.
5924 2017-05-22 Jakub Jelinek <jakub@redhat.com>
5927 * omp-low.c (finish_taskreg_remap): New function.
5928 (finish_taskreg_scan): If unit size of ctx->record_type
5929 is non-constant, unshare the size expression and replace
5930 decls in it with possible outer var refs.
5933 * gimplify.c (omp_add_variable): For GOVD_DEBUG_PRIVATE use
5934 GOVD_SHARED rather than GOVD_PRIVATE with it.
5935 (gimplify_adjust_omp_clauses_1, gimplify_adjust_omp_clauses): Expect
5936 GOVD_SHARED rather than GOVD_PRIVATE with GOVD_DEBUG_PRIVATE.
5939 * omp-low.c (lower_reduction_clauses): Pass OMP_CLAUSE_PRIVATE
5940 as last argument to build_outer_var_ref for pointer bases of array
5943 2017-05-19 Martin Sebor <msebor@redhat.com>
5945 * print-tree.c (print_node): Print DECL_READ_P flag.
5947 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
5949 * Makefile.in: Add ipa-fnsummary.o and ipa-fnsummary.h
5950 * auto-profile.c: Replace ipa-inline.h by ipa-fnsummary.h
5951 * cgraph.c: Likewise.
5952 * cgraphunit.c: Likewise.
5953 * gengtype.c: Likewise.
5954 * ipa-cp.c: Likewise.
5955 * ipa-devirt.c: Likewise.
5956 * ipa-icf.c: Likewise.
5957 * ipa-predicate.c: Likewise.
5958 * ipa-profile.c: Likewise.
5959 * ipa-prop.c: Likewise.
5960 * ipa-split.c: Likewise.
5962 * ipa-inline-analysis.c (inline_summaries, ipa_call_summaries,
5963 edge_predicate_pool, dump_inline_hints,
5964 inline_summary::account_size_time, redirect_to_unreachable,
5965 edge_set_predicate, set_hint_predicate,
5966 evaluate_conditions_for_known_args, evaluate_properties_for_edge,
5967 inline_summary_alloc, ipa_call_summary::reset, inline_summary::reset,
5968 inline_summary_t::remove, remap_hint_predicate_after_duplication,
5969 inline_summary_t::duplicate, ipa_call_summary_t::duplicate,
5970 ipa_call_summary_t::remove, initialize_growth_caches,
5971 free_growth_caches, dump_ipa_call_summary, dump_inline_summary,
5972 debug_inline_summary, dump_inline_summaries, initialize_inline_failed,
5973 mark_modified, unmodified_parm_1, unmodified_parm,
5974 unmodified_parm_or_parm_agg_item, eliminated_by_inlining_prob,
5975 set_cond_stmt_execution_predicate, set_switch_stmt_execution_predicate,
5976 compute_bb_predicates, will_be_nonconstant_expr_predicate,
5977 will_be_nonconstant_predicate, record_modified_bb_info,
5978 get_minimal_bb, record_modified, param_change_prob,
5979 phi_result_unknown_predicate, predicate_for_phi_result,
5980 array_index_predicate, clobber_only_eh_bb_p, fp_expression_p,
5981 estimate_function_body_sizes, compute_inline_parameters,
5982 compute_inline_parameters_for_curren, pass_data_inline_parameters,
5983 estimate_node_size_and_time, estimate_ipcp_clone_size_and_time,
5984 inline_update_callee_summaries, remap_edge_change_prob,
5985 remap_edge_summaries, remap_hint_predicate, inline_merge_summary,
5986 inline_update_overall_summary, inline_indirect_intraprocedural_analysis,
5987 inline_analyze_function, inline_summary_t::insert,
5988 inline_generate_summary, read_ipa_call_summary, inline_read_section,
5989 inline_read_summary, write_ipa_call_summary, inline_write_summary,
5990 inline_free_summary): Move to ipa-fnsummary.h
5991 (predicate_t): Remove.
5992 * ipa-fnsummary.c: New file.
5993 * ipa-inline.h: Do not include sreal.h and ipa-predicate.h
5994 (enum inline_hints_vals, inline_hints, agg_position_info,
5995 INLINE_SIZE_SCALE, size_time_entry, inline_summary, inline_summary_t,
5996 inline_summaries, ipa_call_summary, ipa_call_summary_t,
5997 ipa_call_summaries, debug_inline_summary, dump_inline_summaries,
5998 dump_inline_summary, dump_inline_hints, inline_generate_summary,
5999 inline_read_summary, inline_write_summary, inline_free_summary,
6000 inline_analyze_function, initialize_inline_failed,
6001 inline_merge_summary, inline_update_overall_summary,
6002 compute_inline_parameters): Move to ipa-fnsummary.h
6003 * ipa-fnsummary.h: New file.
6004 * ipa-inline-transform.h: Include ipa-inline.h.
6005 * ipa-inline.c: LIkewise.
6007 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
6009 * ipa-inline.c (edge_badness): Use inlined_time instead of
6010 inline_summaries->get.
6012 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
6014 * ipa-inline.c (edge_badness): Use estimate_size_after_inlining.
6016 2017-05-22 Nathan Sidwell <nathan@acm.org>
6018 * doc/invoke.texi (fdump-translation-unit): Delete documentation.
6019 (fdump-lang): Document 'raw' option.
6020 * dumpfile.h (TDI_tu): Delete.
6021 * dumpfile.c (dump_files): Remove translation-unit.
6022 (FIRST_AUTO_NUMBERED_DUMP): Decrement.
6024 2017-05-22 Georg-Johann Lay <avr@gjlay.de>
6026 * config/avr/t-avr (AWK) [t-multilib]: Remove "-v FORMAT=Makefile"
6027 command option from $(AWK) call.
6028 * config/avr/genmultilib.awk: Simplify and rewrite so that it
6029 generates MULTILIB_REQUIRED instead of MULTILIB_EXCEPTIONS.
6030 [FORMAT]: Remove handling of variable.
6031 * config/avr/t-multilib: Regenerate.
6033 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
6035 * ipa-inline-analysis.c (inline_summary::reset): Do not reset
6037 (dump_inline_summary): Do not print self_time.
6038 (estimate_function_body_sizes): Do not set self_time.
6039 (compute_inline_parameters): Likewise.
6040 (inline_read_section, inline_write_summary): Do not stream self_time.
6041 * ipa-inline.h (inline_summary): Drop self_time.
6043 2017-05-22 Jan Hubicka <hubicka@ucw.cz>
6045 * ipa-inline-analysis.c (account_size_time): Rename to ...
6046 (inline_summary::account_size_time): ... this one.
6047 (reset_ipa_call_summary): Turn to ...
6048 (ipa_call_summary::reset): ... this one.
6049 (reset_inline_summary): Turn to ...
6050 (inline_summary::reset): ... this one.
6051 (inline_summary_t::remove): Update.
6052 (inline_summary_t::duplicate): Update.
6053 (ipa_call_summary_t::remove): Update.
6054 (dump_inline_summary): Update.
6055 (estimate_function_body_sizes): Update.
6056 (compute_inline_parameters): Update.
6057 (estimate_node_size_and_time): Update.
6058 (inline_merge_summary): Update.
6059 (inline_update_overall_summary): Update.
6060 (inline_read_section): Update.
6061 (inline_write_summary): Update.
6062 * ipa-inline.h (inline_summary): Rename entry to size_time_table;
6063 add account_size_time and reset member functions.
6064 (ipa_call_summary): Add reset function.
6065 * ipa-predicate.h (predicate::operator &): Constify.
6067 2017-05-22 Richard Biener <rguenther@suse.de>
6069 * df-scan.c (df_insn_refs_verify): Speedup when not verifying.
6071 2017-05-19 Jason Merrill <jason@redhat.com>
6073 * tree.c (make_tree_vec_stat, grow_tree_vec_stat): Use size_t.
6075 2017-05-19 Marek Polacek <polacek@redhat.com>
6078 * fold-const.c (extract_muldiv_1) <case TRUNC_DIV_EXPR>: Add
6079 TYPE_OVERFLOW_WRAPS checks.
6081 2017-05-19 Thomas Schwinge <thomas@codesourcery.com>
6083 * tree-core.h (enum omp_clause_default_kind): Add
6084 "OMP_CLAUSE_DEFAULT_PRESENT".
6085 * tree-pretty-print.c (dump_omp_clause): Handle it.
6086 * gimplify.c (enum gimplify_omp_var_data): Add
6087 "GOVD_MAP_FORCE_PRESENT".
6088 (gimplify_adjust_omp_clauses_1): Map it to
6089 "GOMP_MAP_FORCE_PRESENT".
6090 (oacc_default_clause): Handle "OMP_CLAUSE_DEFAULT_PRESENT".
6092 * gimplify.c (oacc_default_clause): Clarify.
6094 2017-05-19 Nathan Sidwell <nathan@acm.org>
6096 LANG_HOOK_REGISTER_DUMPS
6097 * toplev.c (general_init): Call register dump lang hook.
6098 * doc/invoke.texi: Document -fdump-lang option family.
6099 * dumpfile.c (dump_files): Remove class dump here.
6100 (FIRST_AUTO_NUMBERED_DUMP): Adjust.
6101 * dumpfile.h (tree_dump_index): Remove TDI_class.
6102 * langhooks-def.h (lhd_register_dumps): Declare.
6103 (LANG_HOOKS_REGISTER_DUMPS): Define.
6104 (LANG_HOOKS_INITIALIZER): Add it.
6105 * langhooks.c (lhd_register_dumps): Define.
6106 * langhooks.h (struct lang_hooks): Add register_dumps.
6108 2017-05-19 Nathan Sidwell <nathan@acm.org>
6110 * context.h (context::set_passes): New.
6111 * context.c (context::context): Do not create pass manager.
6112 * toplev.c (general_init): Create pass manager here.
6114 2017-05-19 Segher Boessenkool <segher@kernel.crashing.org>
6116 * config/rs6000/rs6000.md (splitter to load of -1 and mask): Don't
6117 use this splitter if two add or or instructions would also work for
6118 the constant we want to generate.
6120 2017-05-19 Richard Biener <rguenther@suse.de>
6123 * genmatch.c (dt_node::gen_kids_1): Add missing scope around
6124 predicate evaluation.
6126 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
6128 * ipa-inline.h (ipa_call_summary): Turn sizes into signed;
6130 * ipa-inline.c (want_inline_small_function_p): Do not cast to
6133 2017-05-19 Jan Hubicka <hubicka@ucw.cz>
6135 * ipa-inline-analysis.c (cgraph_2edge_hook_list, cgraph_edge_hook_list,
6136 inline_edge_removal_hook, inline_edge_duplication_hook): Remove.
6137 (inline_edge_summary_vec): Turn into ...
6138 (ipa_call_summaries): ... this one.
6139 (redirect_to_unreachable, edge_set_predicate,
6140 evaluate_properties_for_edge, inline_summary_alloc,
6141 reset_ipa_call_summary, reset_inline_summary,
6142 inline_summary_t::duplicate): Update.
6143 (inline_edge_duplication_hook): Turn to ...
6144 (ipa_call_summary_t::duplicate): ... this one.
6145 (inline_edge_removal_hook): Turn to ...
6146 (ipa_call_summary_t::remove): ... this one.
6147 (dump_inline_edge_summary): Turn to ...
6148 (dump_ipa_call_summary): ... this one.
6149 (estimate_function_body_sizes): Update.
6150 (inline_update_callee_summaries): Update.
6151 (remap_edge_change_prob): Update.
6152 (remap_edge_summaries): Update.
6153 (inline_merge_summary): Update.
6154 (do_estimate_edge_time): Update.
6155 (inline_generate_summary): Update.
6156 (inline_read_section): Update.
6157 (inline_read_summary): Update.
6158 (inline_free_summary): Update.
6159 * ipa-inline.c (can_inline_edge_p): Update.
6160 (compute_inlined_call_time): Update.
6161 (want_inline_small_function_p): Update.
6162 (edge_badness): Update.
6163 (early_inliner): Update.
6164 * ipa-inline.h (inline_edge_summary): Turn to ...
6165 (ipa_call_summary): ... this one.
6166 (ipa_call_summary_t): New class.
6167 (inline_edge_summary_t, inline_edge_summary_vec): Remove.
6168 (ipa_call_summaries): New.
6169 (inline_edge_summary): Remove.
6170 (estimate_edge_growth): Update.
6171 * ipa-profile.c (ipa_propagate_frequency_1): Update.
6172 * ipa-prop.c (ipa_make_edge_direct_to_target): Update.
6173 * ipa-split.c (execute_split_functions): Update.
6174 * ipa.c (symbol_table::remove_unreachable_nodes): Update.
6176 2017-05-19 Richard Biener <rguenther@suse.de>
6179 * cfgexpand.c (expand_gimple_cond): Fix loop fixup.
6181 2017-05-18 Segher Boessenkool <segher@kernel.crashing.org>
6183 * config/rs6000/rs6000.c (struct machine_function): Add field
6184 fpr_is_wrapped_separately.
6185 (rs6000_get_separate_components): Use 64 components. Handle the
6187 (rs6000_components_for_bb): Handle the FPR components.
6188 (rs6000_emit_prologue_components): Handle the FPR components.
6189 (rs6000_emit_epilogue_components): Handle the FPR components.
6190 (rs6000_set_handled_components): Handle the FPR components.
6191 (rs6000_emit_prologue): Don't output prologue code for those FPRs
6192 that are already separately shrink-wrapped.
6193 (rs6000_emit_epilogue): Don't output epilogue code for those FPRs
6194 that are already separately shrink-wrapped.
6196 2017-05-18 Michael Meissner <meissner@linux.vnet.ibm.com>
6199 * config/rs6000/predicates.md (simple_offsettable_mem_operand):
6202 * config/rs6000/rs6000.md (ALTIVEC_DFORM): New iterator.
6203 (define_peephole2 for Altivec d-form load): Add peepholes to catch
6204 cases where the register allocator uses a move and an offsettable
6205 memory operation to/from a FPR register on ISA 2.06/2.07.
6206 (define_peephole2 for Altivec d-form store): Likewise.
6208 2017-05-18 Uros Bizjak <ubizjak@gmail.com>
6211 * config/i386/mmx.md (*mov<mode>_internal): Enable
6212 alternatives 11, 12, 13 and 14 also for 32bit targets.
6213 Remove alternatives 15, 16, 17 and 18.
6214 * config/i386/sse.md (vec_concatv2di): Change
6215 alternative (!x, *y) to (x, ?!*Yn).
6217 2017-05-18 Paolo Carlini <paolo.carlini@oracle.com>
6219 * dumpfile.h (enum dump_kind): Remove stray comma.
6221 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
6223 * Makefile.in: Add ipa-predicate.o and ipa-predicate.h
6224 * ipa-inline-analysis.c (NUM_CONDITIONS): turn into
6225 predicate::num_conditions
6226 (IS_NOT_CONSTANT): turn into predicate::is_not_constant.
6227 (CHANGED): turn into predicate::changed.
6228 (agg_position_info): Move to ipa-predicate.h
6229 (add_condition, predicate::add_clause, predicate::operator &=,
6230 predicate::or_with, predicate::evaluate, predicate::probability,
6231 dump_condition, dump_clause, predicate::dump,
6232 predicate::remap_after_duplication, predicate::remap_after_inlining,
6233 predicate::stream_in, predicate::stream_out): Move to ipa-predicate.c
6234 (evaluate_conditions_for_known_args): Update.
6235 (set_cond_stmt_execution_predicate): Update.
6236 * ipa-inline.h: Include ipa-predicate.h
6237 (condition, inline_param_summary, conditions, agg_position_info,
6238 predicate): Move to ipa-predicate.h
6239 * ipa-predicate.c: New file.
6240 * ipa-predicate.h: New file.
6242 2017-05-18 Wilco Dijkstra <wdijkstr@arm.com>
6244 * final.c (leaf_function_p): Check we are not in a sequence.
6246 2017-05-18 Martin Liska <mliska@suse.cz>
6248 * cfgrtl.c (rtl_verify_edges): Remove usage of TDF_RTL.
6249 * dumpfile.c (dump_register): Use new enum dump_kind.
6250 (get_dump_file_name): Likewise.
6251 (dump_enable_all): Likewise.
6252 (dump_switch_p_1): Likewise.
6253 (enable_rtl_dump_file): Remove usage of TDF_RTL.
6254 * dumpfile.h (enum dump_kind): New enum type.
6255 (struct dump_file_info): Create constructor and
6256 format fields and comments.
6257 * passes.c (pass_manager::register_one_dump_file):
6259 * statistics.c (statistics_early_init): Likewise.
6260 * tree-ssa-loop-prefetch.c (dump_mem_details): Replace
6261 TDF_TREE with TDF_SLIM.
6262 (gather_memory_references_ref): Likewise.
6264 2017-05-18 Martin Liska <mliska@suse.cz>
6266 * vec.h (struct vnull): Use it.
6268 2017-05-18 Jan Hubicka <hubicka@ucw.cz>
6270 * ipa-inline-analysis.c (predicate_conditions): Move to ipa-inline.h
6271 (true_predicate, false_predicate, true_predicate_p,
6272 false_predicate_p): Remove.
6273 (single_cond_predicate, not_inlined_predicate): Turn to member function
6275 (add_condition): Update.
6276 (add_clause): Turn to...
6277 (predicate::add_clause): ... this one; update; allow passing NULL
6279 (and_predicates): Turn to ...
6280 (predicate::operator &=): ... this one.
6281 (predicates_equal_p): Move to predicate::operator == in ipa-inline.h
6282 (or_predicates): Turn to ...
6283 (predicate::or_with): ... this one.
6284 (evaluate_predicate): Turn to ...
6285 (predicate::evaluate): ... this one.
6286 (predicate_probability): Turn to ...
6287 (predicate::probability): ... this one.
6288 (dump_condition): Update.
6289 (dump_predicate): Turn to ...
6290 (predicate::dump): ... this one.
6291 (account_size_time): Update.
6292 (edge_set_predicate): Update.
6293 (set_hint_predicate): UPdate.
6294 (evaluate_conditions_for_known_args): Update.
6295 (evaluate_properties_for_edge): Update.
6296 (remap_predicate_after_duplication): Turn to...
6297 (predicate::remap_after_duplication): ... this one.
6298 (remap_hint_predicate_after_duplication): Update.
6299 (inline_summary_t::duplicate): UPdate.
6300 (dump_inline_edge_summary): Update.
6301 (dump_inline_summary): Update.
6302 (set_cond_stmt_execution_predicate): Update.
6303 (set_switch_stmt_execution_predicate): Update.
6304 (compute_bb_predicates): Update.
6305 (will_be_nonconstant_expr_predicate): Update.
6306 (will_be_nonconstant_predicate): Update.
6307 (phi_result_unknown_predicate): Update.
6308 (predicate_for_phi_result): Update.
6309 (array_index_predicate): Update.
6310 (estimate_function_body_sizes): Update.
6311 (estimate_node_size_and_time): Update.
6312 (estimate_ipcp_clone_size_and_time): Update.
6313 (remap_predicate): Rename to ...
6314 (predicate::remap_after_inlining): ... this one.
6315 (remap_hint_predicate): Update.
6316 (inline_merge_summary): Update.
6317 (inline_update_overall_summary): Update.
6318 (estimate_size_after_inlining): Update.
6319 (read_predicate): Rename to ...
6320 (predicate::stream_in): ... this one.
6321 (read_inline_edge_summary): Update.
6322 (write_predicate): Rename to ...
6323 (predicate::stream_out): ... this one.
6324 (write_inline_edge_summary): Update.
6325 * ipa-inline.h (MAX_CLAUSES): Turn to predicate::max_clauses.
6326 (clause_t): Turn to uint32_t
6327 (predicate): Turn to class; implement constructor and operators
6329 (size_time_entry): Update.
6330 (inline_summary): Update.
6331 (inline_edge_summary): Update.
6333 2017-05-18 Marc Glisse <marc.glisse@inria.fr>
6335 * fold-const.c (fold_binary_loc): Move transformation...
6336 * match.pd (C - X CMP X): ... here.
6338 2017-05-18 Sheldon Lobo <sheldon.lobo@oracle.com>
6340 * config/sparc/sparc.c (sparc_option_override): Set function
6341 alignment for -mcpu=niagara7 to 64 to match the I$ line.
6342 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC M7 branch
6344 * config/sparc/sparc.h (BRANCH_COST): Set the SPARC T4 branch
6346 * config/sparc/sol2.h: Fix a ASM_CPU32_DEFAULT_SPEC typo.
6348 2017-05-18 Marek Polacek <polacek@redhat.com>
6351 * ubsan.c (instrument_null): Unwrap ADDR_EXPRs.
6352 (pass_ubsan::execute): Call gimple_assign_single_p instead of
6353 gimple_assign_load_p.
6355 2017-05-17 Segher Boessenkool <segher@kernel.crashing.org>
6358 * real.c (do_compare): Give decimal_do_compare preference over
6359 comparing just the signs.
6361 2017-05-17 Uros Bizjak <ubizjak@gmail.com>
6363 * doc/md.texi (Canonicalization of Instructions): Describe the
6364 canonical form of instructions that inherently set a condition
6367 2017-05-17 Peter Bergner <bergner@vnet.ibm.com>
6370 * tree-cfg.c: Move deletion of unreachable case statements to after
6371 the merging of consecutive case labels.
6373 2017-05-17 Thomas Preud'homme <thomas.preudhomme@arm.com>
6375 * config/arm/arm.c (cmse_nonsecure_call_clear_caller_saved): Refer
6376 readers to __gnu_cmse_nonsecure_call libcall for saving, clearing and
6377 restoring of callee-saved registers.
6379 2017-05-17 Eric Botcazou <ebotcazou@adacore.com>
6381 * compare-elim.c (equivalent_reg_at_start): Adjust outdated comment.
6382 * config/visium/visium.c (single_set_and_flags): Likewise.
6383 * config/visium/visium.md (Substitutions): Likewise.
6385 2017-05-17 Martin Liska <mliska@suse.cz>
6387 * cfg.c: Introduce dump_flags_t type and
6388 use it instead of int type.
6390 * cfghooks.c: Likewise.
6391 * cfghooks.h (struct cfg_hooks): Likewise.
6392 * cfgrtl.c: Likewise.
6393 * cfgrtl.h: Likewise.
6394 * cgraph.c (cgraph_node::get_body): Likewise.
6395 * coretypes.h: Likewise.
6396 * domwalk.c: Likewise.
6397 * domwalk.h: Likewise.
6398 * dumpfile.c (struct dump_option_value_info): Likewise.
6399 (dump_enable_all): Likewise.
6400 (dump_switch_p_1): Likewise.
6401 (opt_info_switch_p): Likewise.
6402 * dumpfile.h (enum tree_dump_index): Likewise.
6403 (struct dump_file_info): Likewise.
6404 * genemit.c: Likewise.
6405 * generic-match-head.c: Likewise.
6406 * gengtype.c (open_base_files): Likewise.
6407 * gimple-pretty-print.c: Likewise.
6408 * gimple-pretty-print.h: Likewise.
6409 * graph.c (print_graph_cfg): Likewise.
6410 * graphite-scop-detection.c (dot_all_sese): Likewise.
6411 * ipa-devirt.c (build_type_inheritance_graph): Likewise.
6412 * loop-unroll.c (report_unroll): Likewise.
6413 * passes.c (pass_manager::register_one_dump_file): Likewise.
6414 * print-tree.c: Likewise.
6415 * statistics.c: Likewise.
6416 * tree-cfg.c: Likewise.
6417 * tree-cfg.h: Likewise.
6418 * tree-dfa.c: Likewise.
6419 * tree-dfa.h: Likewise.
6420 * tree-dump.c (dump_function): Likewise.
6421 * tree-dump.h (struct dump_info): Likewise.
6422 * tree-pretty-print.c: Likewise.
6423 * tree-pretty-print.h: Likewise.
6424 * tree-ssa-live.c: Likewise.
6425 * tree-ssa-live.h: Likewise.
6426 * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Likewise.
6427 * tree-vect-loop.c: Likewise.
6428 * tree-vect-slp.c: Likewise.
6430 2017-05-16 James Greenhalgh <james.greenhalgh@arm.com>
6431 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
6433 PR tree-optimization/80457
6434 * tree-vect-stmts.c (vect_model_simple_cost): Model the cost
6435 of all arguments to a statement as scalar_to_vec operations.
6436 (vectorizable_call): Adjust call to vect_model_simple_cost for
6438 (vectorizable_conversion): Likewise.
6439 (vectorizable_assignment): Likewise.
6440 (vectorizable_shift): Likewise.
6441 (vectorizable_operation): Likewise.
6442 (vectorizable_comparison): Likewise.
6443 (vect_is_simple_cond): Record the def types for operands.
6444 (vectorizable_condition): Likewise, call vect_model_simple_cost.
6445 * tree-vectorizer.h (vect_model_simple_cost): Add new parameter
6446 for statement argument count.
6448 2017-05-16 Carl Love <cel@us.ibm.com>
6450 * config/rs6000/rs6000-c: Add support for built-in functions
6451 vector unsigned long long vec_bperm (vector unsigned long long,
6452 vector unsigned char)
6453 vector signed long long vec_mule (vector signed int,
6455 vector unsigned long long vec_mule (vector unsigned int,
6456 vector unsigned int)
6457 vector signed long long vec_mulo (vector signed int,
6459 vector unsigned long long vec_mulo (vector unsigned int,
6460 vector unsigned int)
6461 vector signed char vec_sldw (vector signed char,
6464 vector unsigned char vec_sldw (vector unsigned char,
6465 vector unsigned char,
6467 vector signed short vec_sldw (vector signed short,
6468 vector signed short,
6470 vector unsigned short vec_sldw (vector unsigned short,
6471 vector unsigned short,
6473 vector signed int vec_sldw (vector signed int,
6476 vector unsigned int vec_sldw (vector unsigned int,
6477 vector unsigned int,
6479 vector signed long long vec_sldw (vector signed long long,
6480 vector signed long long,
6482 vector unsigned long long vec_sldw (vector unsigned long long,
6483 vector unsigned long long,
6485 * config/rs6000/rs6000-c: Add support for built-in functions
6486 * config/rs6000/rs6000-builtin.def: Add definition for SLDW.
6487 * config/rs6000/altivec.h: Add defintion for vec_sldw.
6488 * doc/extend.texi: Update the built-in documentation for the
6489 new built-in functions.
6491 2017-05-16 Marek Polacek <polacek@redhat.com>
6495 * tree.c (save_expr): Don't fold the expression.
6497 2017-05-16 Uros Bizjak <ubizjak@gmail.com>
6499 * config/i386/i386.md (*movsi_internal): Split (?rm,*y) alternative
6500 to (?r,*Yn) and (?m,*y) alternatives, and (?*y,rm) to (?*Ym,r)
6501 and (?*y,m). Update insn attributes.
6503 2017-05-16 Martin Liska <mliska@suse.cz>
6505 * cgraph.c (cgraph_edge::resolve_speculation): Add default value for
6506 flags argument of print_gimple_stmt, print_gimple_expr,
6507 print_generic_stmt and print_generic_expr.
6508 * cgraphclones.c (symbol_table::materialize_all_clones): Likewise.
6509 * coretypes.h: Likewise.
6510 * except.c (dump_eh_tree): Likewise.
6511 * gimple-fold.c (gimple_fold_stmt_to_constant_1): Likewise.
6512 * gimple-pretty-print.h: Likewise.
6513 * gimple-ssa-backprop.c (dump_usage_prefix): Likewise.
6514 (backprop::push_to_worklist): Likewise.
6515 (backprop::pop_from_worklist): Likewise.
6516 (backprop::process_use): Likewise.
6517 (backprop::intersect_uses): Likewise.
6518 (note_replacement): Likewise.
6519 * gimple-ssa-store-merging.c
6520 (pass_store_merging::terminate_all_aliasing_chains): Likewise.
6521 (imm_store_chain_info::coalesce_immediate_stores): Likewise.
6522 (pass_store_merging::execute): Likewise.
6523 * gimple-ssa-strength-reduction.c (dump_candidate): Likewise.
6524 (ssa_base_cand_dump_callback): Likewise.
6525 (dump_incr_vec): Likewise.
6526 (replace_refs): Likewise.
6527 (replace_mult_candidate): Likewise.
6528 (create_add_on_incoming_edge): Likewise.
6529 (create_phi_basis): Likewise.
6530 (insert_initializers): Likewise.
6531 (all_phi_incrs_profitable): Likewise.
6532 (introduce_cast_before_cand): Likewise.
6533 (replace_one_candidate): Likewise.
6534 * gimplify.c (gimplify_expr): Likewise.
6535 * graphite-isl-ast-to-gimple.c (is_valid_rename): Likewise.
6536 (set_rename): Likewise.
6537 (rename_uses): Likewise.
6538 (copy_loop_phi_nodes): Likewise.
6539 (add_close_phis_to_merge_points): Likewise.
6540 (copy_loop_close_phi_args): Likewise.
6541 (copy_cond_phi_args): Likewise.
6542 (graphite_copy_stmts_from_block): Likewise.
6543 (translate_pending_phi_nodes): Likewise.
6544 * graphite-poly.c (print_pdr): Likewise.
6545 (dump_gbb_cases): Likewise.
6546 (dump_gbb_conditions): Likewise.
6547 (print_scop_params): Likewise.
6548 * graphite-scop-detection.c (build_cross_bb_scalars_def): Likewise.
6549 (build_cross_bb_scalars_use): Likewise.
6550 (gather_bbs::before_dom_children): Likewise.
6551 * hsa-dump.c (dump_hsa_immed): Likewise.
6552 * ipa-cp.c (print_ipcp_constant_value): Likewise.
6553 (get_replacement_map): Likewise.
6554 * ipa-inline-analysis.c (dump_condition): Likewise.
6555 (estimate_function_body_sizes): Likewise.
6556 * ipa-polymorphic-call.c (check_stmt_for_type_change): Likewise.
6557 (ipa_polymorphic_call_context::get_dynamic_type): Likewise.
6558 * ipa-prop.c (ipa_dump_param): Likewise.
6559 (ipa_print_node_jump_functions_for_edge): Likewise.
6560 (ipa_modify_call_arguments): Likewise.
6561 (ipa_modify_expr): Likewise.
6562 (ipa_dump_param_adjustments): Likewise.
6563 (ipa_dump_agg_replacement_values): Likewise.
6564 (ipcp_modif_dom_walker::before_dom_children): Likewise.
6565 * ipa-pure-const.c (check_stmt): Likewise.
6566 (pass_nothrow::execute): Likewise.
6567 * ipa-split.c (execute_split_functions): Likewise.
6568 * omp-offload.c (dump_oacc_loop_part): Likewise.
6569 (dump_oacc_loop): Likewise.
6570 * trans-mem.c (tm_log_emit): Likewise.
6571 (tm_memopt_accumulate_memops): Likewise.
6572 (dump_tm_memopt_set): Likewise.
6573 (dump_tm_memopt_transform): Likewise.
6574 * tree-cfg.c (gimple_verify_flow_info): Likewise.
6575 (print_loop): Likewise.
6576 * tree-chkp-opt.c (chkp_print_addr): Likewise.
6577 (chkp_gather_checks_info): Likewise.
6578 (chkp_get_check_result): Likewise.
6579 (chkp_remove_check_if_pass): Likewise.
6580 (chkp_use_outer_bounds_if_possible): Likewise.
6581 (chkp_reduce_bounds_lifetime): Likewise.
6582 * tree-chkp.c (chkp_register_addr_bounds): Likewise.
6583 (chkp_mark_completed_bounds): Likewise.
6584 (chkp_register_incomplete_bounds): Likewise.
6585 (chkp_mark_invalid_bounds): Likewise.
6586 (chkp_maybe_copy_and_register_bounds): Likewise.
6587 (chkp_build_returned_bound): Likewise.
6588 (chkp_get_bound_for_parm): Likewise.
6589 (chkp_build_bndldx): Likewise.
6590 (chkp_get_bounds_by_definition): Likewise.
6591 (chkp_generate_extern_var_bounds): Likewise.
6592 (chkp_get_bounds_for_decl_addr): Likewise.
6593 * tree-chrec.c (chrec_apply): Likewise.
6594 * tree-data-ref.c (dump_data_reference): Likewise.
6595 (dump_subscript): Likewise.
6596 (dump_data_dependence_relation): Likewise.
6597 (analyze_overlapping_iterations): Likewise.
6598 * tree-inline.c (expand_call_inline): Likewise.
6599 (tree_function_versioning): Likewise.
6600 * tree-into-ssa.c (dump_defs_stack): Likewise.
6601 (dump_currdefs): Likewise.
6602 (dump_names_replaced_by): Likewise.
6603 (dump_update_ssa): Likewise.
6604 (update_ssa): Likewise.
6605 * tree-object-size.c (pass_object_sizes::execute): Likewise.
6606 * tree-parloops.c (build_new_reduction): Likewise.
6607 (try_create_reduction_list): Likewise.
6608 (ref_conflicts_with_region): Likewise.
6609 (oacc_entry_exit_ok_1): Likewise.
6610 (oacc_entry_exit_single_gang): Likewise.
6611 * tree-pretty-print.h: Likewise.
6612 * tree-scalar-evolution.c (set_scalar_evolution): Likewise.
6613 (get_scalar_evolution): Likewise.
6614 (add_to_evolution): Likewise.
6615 (get_loop_exit_condition): Likewise.
6616 (analyze_evolution_in_loop): Likewise.
6617 (analyze_initial_condition): Likewise.
6618 (analyze_scalar_evolution): Likewise.
6619 (instantiate_scev): Likewise.
6620 (number_of_latch_executions): Likewise.
6621 (gather_chrec_stats): Likewise.
6622 (final_value_replacement_loop): Likewise.
6623 (scev_const_prop): Likewise.
6624 * tree-sra.c (dump_access): Likewise.
6625 (disqualify_candidate): Likewise.
6626 (create_access): Likewise.
6628 (maybe_add_sra_candidate): Likewise.
6629 (create_access_replacement): Likewise.
6630 (analyze_access_subtree): Likewise.
6631 (analyze_all_variable_accesses): Likewise.
6632 (sra_modify_assign): Likewise.
6633 (initialize_constant_pool_replacements): Likewise.
6634 (find_param_candidates): Likewise.
6635 (decide_one_param_reduction): Likewise.
6636 (replace_removed_params_ssa_names): Likewise.
6637 * tree-ssa-ccp.c (ccp_fold_stmt): Likewise.
6638 * tree-ssa-copy.c (dump_copy_of): Likewise.
6639 (copy_prop_visit_cond_stmt): Likewise.
6640 * tree-ssa-dce.c (mark_operand_necessary): Likewise.
6641 * tree-ssa-dom.c (pass_dominator::execute): Likewise.
6642 (record_equivalences_from_stmt): Likewise.
6643 * tree-ssa-dse.c (compute_trims): Likewise.
6644 (delete_dead_call): Likewise.
6645 (delete_dead_assignment): Likewise.
6646 * tree-ssa-forwprop.c (forward_propagate_into_gimple_cond): Likewise.
6647 (forward_propagate_into_cond): Likewise.
6648 (pass_forwprop::execute): Likewise.
6649 * tree-ssa-ifcombine.c (ifcombine_ifandif): Likewise.
6650 * tree-ssa-loop-im.c (invariantness_dom_walker::before_dom_children):
6652 (move_computations_worker): Likewise.
6653 (execute_sm): Likewise.
6654 * tree-ssa-loop-ivcanon.c (tree_estimate_loop_size): Likewise.
6655 (remove_exits_and_undefined_stmts): Likewise.
6656 (remove_redundant_iv_tests): Likewise.
6657 * tree-ssa-loop-ivopts.c (dump_use): Likewise.
6658 (adjust_iv_update_pos): Likewise.
6659 * tree-ssa-math-opts.c (bswap_replace): Likewise.
6660 * tree-ssa-phiopt.c (factor_out_conditional_conversion): Likewise.
6661 (value_replacement): Likewise.
6662 * tree-ssa-phiprop.c (phiprop_insert_phi): Likewise.
6663 * tree-ssa-pre.c (print_pre_expr): Likewise.
6664 (get_representative_for): Likewise.
6665 (create_expression_by_pieces): Likewise.
6666 (insert_into_preds_of_block): Likewise.
6667 (eliminate_insert): Likewise.
6668 (eliminate_dom_walker::before_dom_children): Likewise.
6669 (eliminate): Likewise.
6670 (remove_dead_inserted_code): Likewise.
6671 * tree-ssa-propagate.c (substitute_and_fold): Likewise.
6672 * tree-ssa-reassoc.c (get_rank): Likewise.
6673 (eliminate_duplicate_pair): Likewise.
6674 (eliminate_plus_minus_pair): Likewise.
6675 (eliminate_not_pairs): Likewise.
6676 (undistribute_ops_list): Likewise.
6677 (eliminate_redundant_comparison): Likewise.
6678 (update_range_test): Likewise.
6679 (optimize_range_tests_var_bound): Likewise.
6680 (optimize_vec_cond_expr): Likewise.
6681 (rewrite_expr_tree): Likewise.
6682 (rewrite_expr_tree_parallel): Likewise.
6683 (linearize_expr): Likewise.
6684 (break_up_subtract): Likewise.
6685 (linearize_expr_tree): Likewise.
6686 (attempt_builtin_powi): Likewise.
6687 (attempt_builtin_copysign): Likewise.
6688 (transform_stmt_to_copy): Likewise.
6689 (transform_stmt_to_multiply): Likewise.
6690 (dump_ops_vector): Likewise.
6691 * tree-ssa-sccvn.c (vn_nary_build_or_lookup_1): Likewise.
6692 (print_scc): Likewise.
6693 (set_ssa_val_to): Likewise.
6694 (visit_reference_op_store): Likewise.
6695 (visit_use): Likewise.
6696 (sccvn_dom_walker::before_dom_children): Likewise.
6697 (run_scc_vn): Likewise.
6698 * tree-ssa-scopedtables.c (avail_exprs_stack::lookup_avail_expr):
6700 (expr_hash_elt::print): Likewise.
6701 (const_and_copies::pop_to_marker): Likewise.
6702 (const_and_copies::record_const_or_copy_raw): Likewise.
6703 * tree-ssa-structalias.c (compute_dependence_clique): Likewise.
6704 * tree-ssa-uninit.c (collect_phi_def_edges): Likewise.
6705 (dump_predicates): Likewise.
6706 (find_uninit_use): Likewise.
6707 (warn_uninitialized_phi): Likewise.
6708 (pass_late_warn_uninitialized::execute): Likewise.
6709 * tree-ssa.c (verify_vssa): Likewise.
6710 (verify_ssa): Likewise.
6711 (maybe_optimize_var): Likewise.
6712 * tree-vrp.c (dump_value_range): Likewise.
6713 (dump_all_value_ranges): Likewise.
6714 (dump_asserts_for): Likewise.
6715 (register_edge_assert_for_2): Likewise.
6716 (vrp_visit_cond_stmt): Likewise.
6717 (vrp_visit_switch_stmt): Likewise.
6718 (vrp_visit_stmt): Likewise.
6719 (vrp_visit_phi_node): Likewise.
6720 (simplify_cond_using_ranges_1): Likewise.
6721 (fold_predicate_in): Likewise.
6722 (evrp_dom_walker::before_dom_children): Likewise.
6723 (evrp_dom_walker::push_value_range): Likewise.
6724 (evrp_dom_walker::pop_value_range): Likewise.
6725 (execute_early_vrp): Likewise.
6727 2017-05-16 Richard Biener <rguenther@suse.de>
6729 * dwarf2out.c (loc_list_from_tree_1): Do not create
6730 DW_OP_GNU_variable_value for DECL_IGNORED_P decls.
6732 2017-05-16 Richard Biener <rguenther@suse.de>
6734 * dwarf2out.c (resolve_variable_value_in_expr): Lookup DIE
6736 (note_variable_value_in_expr): If we resolved the decl ref
6737 do not push to the stack.
6739 2017-05-16 Matthew Wahab <matthew.wahab@arm.com>
6741 * config/arm/arm_neon.h (vadd_f16): Use standard arithmetic
6742 operations in fast-math mode.
6743 (vaddq_f16): Likewise.
6744 (vmul_f16): Likewise.
6745 (vmulq_f16): Likewise.
6746 (vsub_f16): Likewise.
6747 (vsubq_f16): Likewise.
6748 * config/arm/neon.md (add<mode>3): New.
6750 (fma:<VH:mode>3): New. Also remove outdated comment.
6753 2017-05-16 Martin Liska <mliska@suse.cz>
6757 * ipa-devirt.c (warn_types_mismatch): Fix typo.
6758 (odr_types_equivalent_p): Likewise.
6760 2017-05-15 Sylvestre Ledru <sylvestre@debian.org>
6762 * plugin.c (try_init_one_plugin): Fix ressource leaks (CID 726637).
6764 2017-05-15 Uros Bizjak <ubizjak@gmail.com>
6767 * config/i386.i386.md (*zero_extendsidi2): Do not penalize
6768 non-interunit SSE move alternatives with '?'.
6769 (zero-extendsidi peephole2): New peephole to skip intermediate
6770 general register in SSE zero-extend sequence.
6772 2017-05-15 Jeff Law <law@redhat.com>
6774 * reorg.c (relax_delay_slots): Create a new variable to hold
6775 the temporary target rather than clobbering TARGET_LABEL.
6777 * config/tilegx/tilegx.c (tilegx_expand_unaligned_load): Add
6778 missing argument to extract_bit_field call.
6779 * config/tilepro/tilepro.c (tilepro_expand_unaligned_load): Likewise.
6781 2017-05-15 Martin Liska <mliska@suse.cz>
6784 * gcc.c (process_command): Do not allow empty argument of -o option.
6786 2017-05-15 Renlin Li <renlin.li@arm.com>
6788 * config/aarch64/aarch64-protos.h (aarch64_expand_call): Declare.
6789 * config/aarch64/aarch64.c (aarch64_expand_call): Define.
6790 * config/aarch64/constraints.md (Usf): Add long call check.
6791 * config/aarch64/aarch64.md (call): Use aarch64_expand_call.
6792 (call_value): Likewise.
6793 (sibcall): Likewise.
6794 (sibcall_value): Likewise.
6796 (call_value_insn): New.
6797 (sibcall_insn): Update rtx pattern.
6798 (sibcall_value_insn): Likewise.
6799 (call_internal): Remove.
6800 (call_value_internal): Likewise.
6801 (sibcall_internal): Likewise.
6802 (sibcall_value_internal): Likewise.
6803 (call_reg): Likewise.
6804 (call_symbol): Likewise.
6805 (call_value_reg): Likewise.
6806 (call_value_symbol): Likewise.
6808 2017-05-14 Krister Walfridsson <krister.walfridsson@gmail.com>
6811 * config/netbsd.h (NETBSD_LIBGCC_SPEC): Always add -lgcc.
6813 2017-05-14 Uros Bizjak <ubizjak@gmail.com>
6815 * config/i386.i386.c (ix86_cc_modes_compatible): CCNOmode is
6816 compatible with CCGOCmode and with CCZmode.
6818 2017-05-14 Martin Sebor <msebor@redhat.com>
6821 * gimple-fold.c (gimple_fold_builtin_sprintf): Make extern.
6822 (gimple_fold_builtin_snprintf): Same.
6823 * gimple-fold.h (gimple_fold_builtin_sprintf): Declare.
6824 (gimple_fold_builtin_snprintf): Same.
6825 * gimple-ssa-sprintf.c (get_format_string): Correct the detection
6827 (is_call_safe): New function.
6828 (try_substitute_return_value): Call it.
6829 (try_simplify_call): New function.
6830 (pass_sprintf_length::handle_gimple_call): Call it.
6832 2017-05-14 Martin Sebor <msebor@redhat.com>
6835 * builtins.c (expand_builtin_stpncpy): Simplify.
6837 2017-05-14 Daniel Santos <daniel.santos@pobox.com>
6839 * config/i386/i386.opt: Add option -mcall-ms2sysv-xlogues.
6840 * config/i386/i386.h
6841 (x86_64_ms_sysv_extra_clobbered_registers): Change type to unsigned.
6842 (NUM_X86_64_MS_CLOBBERED_REGS): New macro.
6843 (struct machine_function): Add new members call_ms2sysv,
6844 call_ms2sysv_pad_in, call_ms2sysv_pad_out and call_ms2sysv_extra_regs.
6845 (struct machine_frame_state): New fields sp_realigned and
6846 sp_realigned_offset.
6847 * config/i386/i386.c
6848 (enum xlogue_stub): New enum.
6849 (enum xlogue_stub_sets): New enum.
6850 (class xlogue_layout): New class.
6851 (struct ix86_frame): New fields stack_realign_allocate_offset,
6852 stack_realign_offset and outlined_save_offset. Modify comments to
6853 detail stack layout when using out-of-line stubs.
6854 (ix86_target_string): Add -mcall-ms2sysv-xlogues option.
6855 (ix86_option_override_internal): Add sorry() for TARGET_SEH and
6856 -mcall-ms2sysv-xlogues.
6857 (stub_managed_regs): New static variable.
6858 (ix86_save_reg): Add new parameter ignore_outlined to optionally omit
6859 registers managed by out-of-line stub.
6860 (disable_call_ms2sysv_xlogues): New function.
6861 (ix86_compute_frame_layout): Modify re-alignment calculations, disable
6862 m->call_ms2sysv when appropriate and compute frame layout for
6864 (sp_valid_at, fp_valid_at): New inline functions.
6865 (choose_basereg): New function.
6866 (choose_baseaddr): Add align parameter, use choose_basereg and modify
6868 (ix86_emit_save_reg_using_mov, ix86_emit_restore_sse_regs_using_mov):
6869 Use align parameter of choose_baseaddr to generated aligned SSE movs
6871 (pro_epilogue_adjust_stack): Modify to track
6872 machine_frame_state::sp_realigned.
6873 (ix86_nsaved_regs): Modify to accommodate changes to ix86_save_reg.
6874 (ix86_nsaved_sseregs): Likewise.
6875 (ix86_emit_save_regs): Likewise.
6876 (ix86_emit_save_regs_using_mov): Likewise.
6877 (ix86_emit_save_sse_regs_using_mov): Likewise.
6878 (get_scratch_register_on_entry): Likewise.
6879 (gen_frame_set): New function.
6880 (gen_frame_load): Likewise.
6881 (gen_frame_store): Likewise.
6882 (emit_outlined_ms2sysv_save): Likewise.
6883 (emit_outlined_ms2sysv_restore): Likewise.
6884 (ix86_expand_prologue): Modify stack re-alignment code and call
6885 emit_outlined_ms2sysv_save when appropriate.
6886 (ix86_emit_leave): Clear machine_frame_state::sp_realigned. Add
6887 parameter rtx_insn *insn, which allows the function to be used to only
6889 (ix86_expand_epilogue): Modify validity checks of frame and stack
6890 pointers, and call emit_outlined_ms2sysv_restore when appropriate.
6891 (ix86_expand_call): Modify to enable m->call_ms2sysv when appropriate.
6892 * config/i386/predicates.md
6893 (save_multiple): New predicate.
6894 (restore_multiple): Likewise.
6895 * config/i386/sse.md
6896 (save_multiple<mode>): New pattern.
6897 (save_multiple_realign<mode>): Likewise.
6898 (restore_multiple<mode>): Likewise.
6899 (restore_multiple_and_return<mode>): Likewise.
6900 (restore_multiple_leave_return<mode>): Likewise.
6901 * Makefile.in: Export HOSTCXX and HOSTCXXFLAGS to site.exp
6903 2017-05-14 Julia Koval <julia.koval@intel.com>
6905 * config/i386/i386-builtin-types.def (VOID_FTYPE_INT_INT64): New type.
6906 * config/i386/i386-builtin.def (__builtin_ia32_xgetbv)
6907 (__builtin_ia32_xsetbv): New builtins.
6908 * config/i386/i386.c (ix86_expand_special_args_builtin):
6910 (ix86_expand_builtin): Special expand for new intrinsics.
6911 * config/i386/i386.md (UNSPECV_XGETBV, UNSPECV_XSETBV): New.
6912 (xsetbv, xsetbv_rex64, xgetbv, xgetbv_rex64): New insn patterns.
6913 * config/i386/xsaveintrin.h (_xsetbv, _getbv): New intrinsics.
6915 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6917 * cfganal.c (inverted_post_order_compute): Change argument type
6919 * cfganal.h (inverted_post_order_compute): Adjust prototype.
6920 * df-core.c (rest_of_handle_df_initialize): Adjust.
6921 (rest_of_handle_df_finish): Likewise.
6922 (df_analyze_1): Likewise.
6923 (df_analyze): Likewise.
6924 (loop_inverted_post_order_compute): Change argument to be a vec *.
6925 (df_analyze_loop): Adjust.
6926 (df_get_n_blocks): Likewise.
6927 (df_get_postorder): Likewise.
6928 * df.h (struct df_d): Change field to be a vec.
6929 * lcm.c (compute_laterin): Adjust.
6930 (compute_available): Likewise.
6931 * lra-lives.c (lra_create_live_ranges_1): Likewise.
6932 * tree-ssa-dce.c (remove_dead_stmt): Likewise.
6933 * tree-ssa-pre.c (compute_antic): Likewise.
6935 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6937 * cfganal.c (connect_infinite_loops_to_exit): Adjust.
6938 (depth_first_search::depth_first_search): Change structure init
6939 function to this constructor.
6940 (depth_first_search::add_bb): Rename function to this member.
6941 (depth_first_search::execute): Likewise.
6942 (flow_dfs_compute_reverse_finish): Adjust.
6944 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6946 * ddg.c (find_nodes_on_paths): Use auto_sbitmap.
6947 (longest_simple_path): Likewise.
6948 * shrink-wrap.c (spread_components): Likewise.
6949 (disqualify_problematic_components): Likewise.
6950 (emit_common_heads_for_components): Likewise.
6951 (emit_common_tails_for_components): Likewise.
6952 (insert_prologue_epilogue_for_components): Likewise.
6954 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6956 * tree-ssa-dse.c (dse_dom_walker): Make m_live_byes a
6959 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6961 * df-core.c (df_set_blocks): Start using auto_bitmap.
6962 (df_compact_blocks): Likewise.
6963 * df-problems.c (df_rd_confluence_n): Likewise.
6964 * df-scan.c (df_insn_rescan_all): Likewise.
6965 (df_process_deferred_rescans): Likewise.
6966 (df_update_entry_block_defs): Likewise.
6967 (df_update_exit_block_uses): Likewise.
6968 (df_entry_block_bitmap_verify): Likewise.
6969 (df_exit_block_bitmap_verify): Likewise.
6970 (df_scan_verify): Likewise.
6971 * lra-constraints.c (lra_constraints): Likewise.
6972 (undo_optional_reloads): Likewise.
6973 (lra_undo_inheritance): Likewise.
6974 * lra-remat.c (calculate_gen_cands): Likewise.
6975 (do_remat): Likewise.
6976 * lra-spills.c (assign_spill_hard_regs): Likewise.
6977 (spill_pseudos): Likewise.
6978 * tree-ssa-pre.c (bitmap_set_and): Likewise.
6979 (bitmap_set_subtract_values): Likewise.
6981 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6983 * haifa-sched.c (estimate_shadow_tick): Replace manual bitmap
6984 management with auto_bitmap.
6985 (fix_inter_tick): Likewise.
6986 (fix_recovery_deps): Likewise.
6987 * ira.c (add_store_equivs): Likewise.
6988 (find_moveable_pseudos): Likewise.
6989 (split_live_ranges_for_shrink_wrap): Likewise.
6990 * print-rtl.c (rtx_reuse_manager::rtx_reuse_manager): Likewise.
6991 (rtx_reuse_manager::seen_def_p): Likewise.
6992 (rtx_reuse_manager::set_seen_def): Likewise.
6993 * print-rtl.h (class rtx_reuse_manager): Likewise.
6995 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
6997 * bt-load.c (combine_btr_defs): Use auto_bitmap to manage bitmap
6999 (migrate_btr_def): Likewise.
7000 * cfgloop.c (get_loop_body_in_bfs_order): Likewise.
7001 * df-core.c (loop_post_order_compute): Likewise.
7002 (loop_inverted_post_order_compute): Likewise.
7003 * hsa-common.h: Likewise.
7004 * hsa-gen.c (hsa_bb::~hsa_bb): Likewise.
7005 * init-regs.c (initialize_uninitialized_regs): Likewise.
7006 * ipa-inline.c (resolve_noninline_speculation): Likewise.
7007 (inline_small_functions): Likewise.
7008 * ipa-reference.c (ipa_reference_write_optimization_summary): Likewise.
7009 * ira.c (combine_and_move_insns): Likewise.
7010 (build_insn_chain): Likewise.
7011 * loop-invariant.c (find_invariants): Likewise.
7012 * lower-subreg.c (propagate_pseudo_copies): Likewise.
7013 * predict.c (tree_predict_by_opcode): Likewise.
7014 (predict_paths_leading_to): Likewise.
7015 (predict_paths_leading_to_edge): Likewise.
7016 (estimate_loops_at_level): Likewise.
7017 (estimate_loops): Likewise.
7018 * shrink-wrap.c (try_shrink_wrapping): Likewise.
7019 (spread_components): Likewise.
7020 * tree-cfg.c (remove_edge_and_dominated_blocks): Likewise.
7021 * tree-loop-distribution.c (rdg_build_partitions): Likewise.
7022 * tree-predcom.c (tree_predictive_commoning_loop): Likewise.
7023 * tree-ssa-coalesce.c (coalesce_ssa_name): Likewise.
7024 * tree-ssa-phionlycprop.c (pass_phi_only_cprop::execute): Likewise.
7025 * tree-ssa-pre.c (remove_dead_inserted_code): Likewise.
7026 * tree-ssa-sink.c (nearest_common_dominator_of_uses): Likewise.
7027 * tree-ssa-threadupdate.c (compute_path_counts): Likewise.
7028 (mark_threaded_blocks): Likewise.
7029 (thread_through_all_blocks): Likewise.
7030 * tree-ssa.c (verify_ssa): Likewise.
7031 (execute_update_addresses_taken): Likewise.
7032 * tree-ssanames.c (verify_ssaname_freelists): Likewise.
7034 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7036 * cfganal.c (mark_dfs_back_edges): Replace manual stack with
7038 (post_order_compute): Likewise.
7039 (inverted_post_order_compute): Likewise.
7040 (pre_and_rev_post_order_compute_fn): Likewise.
7042 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7044 * genrecog.c (int_set::int_set): Explicitly construct our
7045 auto_vec base class.
7046 * vec.h (auto_vec::auto_vec): New constructor.
7048 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7050 * bitmap.h (class auto_bitmap): New constructor taking
7051 bitmap_obstack * argument.
7053 2017-05-13 Trevor Saunders <tbsaunde+gcc@tbsaunde.org>
7055 * bitmap.h (class auto_bitmap): Change type of m_bits to
7056 bitmap_head, and adjust ctor / dtor and member operators.
7058 2017-05-13 Uros Bizjak <ubizjak@gmail.com>
7060 * compare-elim.c (equivalent_reg_at_start): Return NULL_RTX
7061 when returned register mode doesn't match original mode.
7063 2017-05-12 Jeff Law <law@redhat.com>
7064 Jakub Jelinek <jakub@redhat.com>
7066 * config/mn10300/mn10300.c (mn10300_match_ccmode): Fix where
7067 we look for cc setter after the compare-elim changes.
7068 * config/mn10300/mn10300.md (addsi3_flags): Fix order of patterns
7069 within the vector to match what compare-elim now expects.
7070 (subsi3_flags, andsi3_flags, iorsi3_flags): Likewise.
7071 (xorsi3_flags, one_cmplsi2_flags): Likewise.
7073 * config/rx/rx.c (rx_match_ccmode): Fix where we look cc setter
7074 after the compare-elim changes.
7075 * config/rx/rx.md (abssi2_flags): Fix order of patterns within
7076 the vector to match what compare-elim now expects.
7077 (addsi3_flags, adc_flags, addsi3_flags peepholes): Likewise.
7078 (andsi3_flags, negsi2_flags, one_cmplsi2_flags): Likewise.
7079 (iorsi3_flags, rotlsi3_flags, rotrsi3_flags): Likewise.
7080 (ashrsi3_flags, lshrsi3_flags, ashlsi3_flags): Likewise.
7081 (ssaddsi3, subsi3_flags, sbb_flags, xorsi3_flags): Likewise.
7083 * config/visium/visium.c (single_set_and_flags): Fix where
7084 we look for cc setter after the compare-elim changes.
7085 * config/visium/visium.md (flags_subst_logic): Fix order of patterns
7086 with the vector to match what compare-elim now expects.
7087 (flags_subst_arith, add<mode>3_insn_set_carry): Likewise.
7088 (add<mode>3_insn_set_overflow, addsi3_insn_set_carry): Likewise.
7089 (addsi3_insn_set_overflow, sub<mode>3_insn_set_carry): Likewise.
7090 (sub<mode>3_insn_set_overflow, subsi3_insn_set_carry): Likewise.
7091 (subsi3_insn_set_overflow, negsi2_insn_set_carry): Likewise.
7092 (neg<mode>2_insn_set_overflow): Likewise.
7094 2017-05-12 Jim Wilson <jim.wilson@linaro.org>
7097 * expmed.c (extract_bit_field_1): Add alt_rtl argument. Before
7098 maybe_expand_insn call, set ops[0].target. If still set after call,
7099 set alt_rtl. Add extra arg to recursive calls.
7100 (extract_bit_field): Add alt_rtl argument. Pass to
7102 * expmed.h (extract_bit_field): Fix prototype.
7103 * expr.c (emit_group_load_1, copy_blkmode_from_reg)
7104 (copy_blkmode_to_reg, read_complex_part, store_field): Pass extra NULL
7105 to extract_bit_field_calls.
7106 (expand_expr_real_1): Pass alt_rtl to expand_expr_real instead of 0.
7107 Pass alt_rtl to extract_bit_field calls.
7108 * calls.c (store_unaligned_arguments_into_psuedos)
7109 load_register_parameters): Pass extra NULL to extract_bit_field calls.
7110 * optabs.c (maybe_legitimize_operand): Clear op->target when call
7112 * optabs.h (struct expand_operand): Add target bitfield.
7114 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
7116 * compare-elim.c (try_eliminate_compare): Canonicalize
7117 operation with embedded compare to
7118 [(set (reg:CCM) (compare:CCM (operation) (immediate)))
7119 (set (reg) (operation)].
7121 * config/i386/i386.c (TARGET_FLAGS_REGNUM): New define.
7123 2017-05-12 Uros Bizjak <ubizjak@gmail.com>
7126 * config/i386/i386.c (ix86_rtx_cost) [case PLUS]: Ignore the
7127 cost of adding a carry flag for ADC instruction.
7128 [case MINUS]: Ignore the cost of subtracting a carry flag
7129 for SBB instruction.
7131 2017-05-12 Steven Munroe <munroesj@gcc.gnu.org>
7133 * config.gcc (powerpc*-*-*): Add bmi2intrin.h, bmiintrin.h,
7135 * config/rs6000/bmiintrin.h: New file.
7136 * config/rs6000/bmi2intrin.h: New file.
7137 * config/rs6000/x86intrin.h: New file.
7139 2017-05-12 Jeff Law <law@redhat.com>
7141 * tree-vrp.c (vrp_dom_walker::before_dom_children): Push unwinding
7144 2017-05-12 Peter Bergner <bergner@vnet.ibm.com>
7147 * tree-cfg.c: Remove cfg edges of unreachable case statements.
7149 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
7151 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
7152 early expansion of vector divide builtins.
7153 (builtin_function_type): Add VSX_BUILTIN_UDIV_V2DI to the list of
7154 builtins identified as having unsigned arguments.
7156 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
7158 * config/rs6000/rs6000.c (gimple-fold.h): New #include.
7159 (rs6000_gimple_fold_builtin): Add handling for early GIMPLE
7160 expansion of vector logical operations (and, andc, or, xor,
7163 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
7165 * gimple-fold.c (create_tmp_reg_or_ssa_name): Remove static declaration.
7166 * gimple-fold.h (create_tmp_reg_or_ssa_name): New prototype.
7168 2017-05-12 Will Schmidt <will_schmidt@vnet.ibm.com>
7170 * config/rs6000/rs6000.c (rs6000_gimple_fold_builtin): Add handling for
7171 early GIMPLE expansion of vector multiplies.
7173 2017-05-12 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
7175 * config/arm/arm.md (movsi): Add TARGET_32BIT in addition to the
7176 TARGET_HAVE_MOVT conditional.
7177 (movt splitter): Likewise.
7179 2017-05-12 Richard Biener <rguenther@suse.de>
7181 * tree-ssa-sccvn.h (has_VN_INFO): Declare.
7182 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
7183 Fold all stmts not inplace.
7185 2017-05-12 Richard Biener <rguenther@suse.de>
7187 PR tree-optimization/80713
7188 * tree-ssa-pre.c (remove_dead_inserted_code): Clear
7189 inserted_exprs bit for not removed stmts.
7191 2017-05-12 Thomas Schwinge <thomas@codesourcery.com>
7194 * tree-parloops.c (create_parallel_loop): Set "oacc kernels
7195 parallelized" attribute for parallelized OpenACC kernels.
7196 * omp-offload.c (execute_oacc_device_lower): Use it.
7198 * omp-expand.c (expand_omp_target) <GF_OMP_TARGET_KIND_OACC_KERNELS>:
7199 Set "oacc kernels" attribute.
7200 * omp-general.c (oacc_set_fn_attrib): Remove is_kernel formal
7201 parameter. Adjust all users.
7202 (oacc_fn_attrib_kernels_p): Remove function.
7203 * omp-offload.c (execute_oacc_device_lower): Look for "oacc
7204 kernels" attribute instead of calling oacc_fn_attrib_kernels_p.
7205 * tree-ssa-loop.c (gate_oacc_kernels): Likewise.
7206 * tree-parloops.c (create_parallel_loop): If oacc_kernels_p,
7207 assert "oacc kernels" attribute is set.
7209 2017-05-11 Carl Love <cel@us.ibm.com>
7211 * config/rs6000/rs6000-c: Add support for built-in functions
7212 vector unsigned char vec_popcnt (vector signed char)
7213 vector unsigned char vec_popcnt (vector unsigned char)
7214 vector unsigned short vec_popcnt (vector signed short)
7215 vector unsigned short vec_popcnt (vector unsigned short)
7216 vector unsigned int vec_popcnt (vector signed int)
7217 vector unsigned int vec_popcnt (vector unsigned int)
7218 vector unsigned long long vec_popcnt (vector signed long long)
7219 vector unsigned long long vec_popcnt (vector unsigned long long)
7220 vector signed long long vec_slo (vector signed long long,
7222 vector signed long long vec_slo (vector signed long long,
7223 vector unsigned char)
7224 vector unsigned long long vec_slo (vector unsigned long long,
7226 vector unsigned long long vec_slo (vector unsigned long long,
7227 vector unsigned char)
7228 * config/rs6000/rs6000-builtin.def: Add definitions for VPOPCNTUB,
7229 VPOPCNTUH, VPOPCNTUW, and VPOPCNTUD overloads.
7230 * config/rs6000/altivec.h: Add define for vec_popcnt, vec_popcntb,
7231 vec_popcnth, vec_popcntw and vec_popcntd built-in functions.
7232 * doc/extend.texi: Update the built-in documentation file for the
7233 new built-in functions.
7235 2017-05-11 Michael Meissner <meissner@linux.vnet.ibm.com>
7237 * attribs.h (sorted_attr_string): Move machine independent
7238 functions for target clone support from the i386 port to common
7239 code. Rename ix86_function_versions to common_function_versions.
7240 Rename make_name to make_unique_name.
7241 (common_function_versions): Likewise.
7242 (make_unique_name): Likewise.
7243 (make_dispatcher_decl): Likewise.
7244 (is_function_default_version): Likewise.
7245 * attribs.c (attr_strcmp): Likewise.
7246 (sorted_attr_string): Likewise.
7247 (common_function_versions): Likewise.
7248 (make_unique_name): Likewise.
7249 (make_dispatcher_decl): Likewise.
7250 (is_function_default_version): Likewise.
7251 * config/i386/i386.c (attr_strcmp): Likewise.
7252 (sorted_attr_string): Likewise.
7253 (ix86_function_versions): Likewise.
7254 (make_name): Likewise.
7255 (make_dispatcher_decl): Likewise.
7256 (is_function_default_version): Likewise.
7257 (TARGET_OPTION_FUNCTION_VERSIONS): Update target function hook.
7259 2017-05-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7262 * config/rs6000/rs6000.c (rs6000_builtin_vectorization_cost):
7263 Account for direct move costs for vec_construct of integer
7266 2017-05-11 Uros Bizjak <ubizjak@gmail.com>
7269 * config/i386/sync.md (UNSPEC_LDX_ATOMIC): New unspec.
7270 (UNSPEC_STX_ATOMIC): Ditto.
7271 (loaddi_via_sse): New insn.
7272 (storedi_via_sse): Ditto.
7273 (atomic_loaddi_fpu): Emit loaddi_via_sse and storedi_via_sse.
7274 Update corresponding peephole2 patterns.
7275 (atomic_storedi_fpu): Ditto.
7277 2017-05-11 Julia Koval <julia.koval@intel.com>
7279 * config/i386/avx512fintrin.h (_mm_mask_rsqrt14_sd)
7280 (_mm_maskz_rsqrt14_sd, _mm_mask_rsqrt14_ss, _mm_maskz_rsqrt14_ss):
7282 * config/i386/i386-builtin.def (__builtin_ia32_rsqrt14sd_mask)
7283 (__builtin_ia32_rsqrt14ss_mask): New builtins.
7284 * config/i386/sse.md (rsqrt14_<mode>_mask): New pattern.
7286 2017-05-11 Nathan Sidwell <nathan@acm.org>
7288 * graphite-poly.c: Include dumpfile.h.
7290 * dumpfle.h (dump_function): Declare here ...
7291 * tree-dump.h (dump_function): ... not here.
7292 * dumpfile.c: #include tree-cfg.h.
7293 (dump_function): Move here from ...
7294 * tree-dump.c (dump_function): ... here.
7295 * gimplify.c: #include splay-tree.h, not tree-dump.h.
7296 * graphite-poly.c: Don't include tree-dump.h.
7297 * cgraphclones.c: Include dumpfile.h not tree-dump.h.
7298 * print-tree.c: Likewise.
7299 * stor-layout.c: Likewise.
7300 * tree-nested.c: Likewise.
7302 * dumpfile.c (dump_start): Use TDF_FLAGS.
7303 (dump_enable_all): Fix TDF_KIND check thinko.
7305 2017-05-11 Kelvin Nilsen <kelvin@gcc.gnu.org>
7307 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
7308 array entries to represent two legal parameterizations of the
7309 overloaded __builtin_cmpb function, as represented by the
7310 P6_OV_BUILTIN_CMPB constant.
7311 (altivec_resolve_overloaded_builtin): Add special case handling
7312 for the __builtin_cmpb function, as represented by the
7313 P6_OV_BUILTIN_CMPB constant.
7314 * config/rs6000/rs6000-builtin.def (BU_P6_2): New macro.
7315 (BU_P6_64BIT_2): New macro.
7316 (BU_P6_OVERLOAD_2): New macro
7317 (CMPB_32): Add 32-bit compare-bytes support for 32-bit only targets.
7318 (CMPB): Add 64-bit compare-bytes support for 32-bit and 64-bit targets.
7319 (CMPB): Add overload support to represent both 32-bit and 64-bit
7320 compare-bytes function.
7321 * config/rs6000/rs6000.c (rs6000_builtin_mask_calculate): Add
7322 support for TARGET_CMPB.
7323 * config/rs6000/rs6000.h: Add support for RS6000_BTM_CMPB.
7324 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
7325 documentation of the __builtin_cmpb overloaded built-in function.
7327 2017-05-11 Richard Biener <rguenther@suse.de>
7329 PR tree-optimization/80705
7330 * tree-vect-data-refs.c (vect_analyze_data_refs): DECL_NONALIASED
7331 bases are not vectorizable.
7333 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7335 * tree-ssa-loop-ivopts.c (determine_set_costs): Skip non-interger
7336 when counting register pressure.
7338 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7340 * tree-ssa-loop-ivopts.c (compare_cost_pair): New.
7341 (iv_ca_more_deps): Renamed to ...
7342 (iv_ca_compare_deps): ... this.
7343 (iv_ca_extend): Extend iv_ca if NEW_CP is cheaper than OLD_CP.
7345 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7347 * tree-ssa-loop-ivopts.c (find_interesting_uses): Move inv vars dump
7349 (determine_group_iv_costs): ... here.
7350 (find_inv_vars_cb): Record inv var if it's not recorded before.
7352 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7354 * tree-ssa-loop-ivopts.c (allow_ip_end_pos_p): Refine comments.
7355 (get_shiftadd_cost): Ditto.
7357 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7359 * tree-ssa-address.c: Include header file.
7360 (move_hint_to_base): Return TRUE if BASE_HINT is moved to memory
7362 (add_to_parts): Refactor.
7363 (addr_to_parts): New parameter. Update use of move_hint_to_base.
7364 (create_mem_ref): Update use of addr_to_parts. Re-associate addr
7367 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7369 PR tree-optimization/53090
7370 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New enum value
7372 (extract_cond_operands): Detect condition with IV on both sides
7373 and return COMP_IV_EXPR_2.
7374 (find_interesting_uses_cond): Add iv_use for both IVs in condition.
7375 (rewrite_use_compare): Simplify by removing call to function
7376 extract_cond_operands.
7378 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7380 * tree-ssa-loop-ivopts.c (enum comp_iv_rewrite): New.
7381 (extract_cond_operands): Detect condition comparing against non-
7382 invariant bound and return appropriate enum value.
7383 (find_interesting_uses_cond): Update use of extract_cond_operands.
7384 Handle its return value accordingly.
7385 (determine_group_iv_cost_cond, rewrite_use_compare): Ditto.
7387 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7389 * tree-ssa-loop-ivopts.c (rewrite_use_nonlinear_expr): Re-associate
7390 nonlinear iv_use computation in loop invariant sensitive way.
7392 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7394 * tree-ssa-loop-ivopts.c (relate_compare_use_with_all_cands): New.
7395 (find_iv_candidates): Call relate_compare_use_with_all_cands.
7397 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7399 * tree-ssa-loop-ivopts.c (struct iv_cand): New field inv_exprs.
7400 (dump_cand): Support iv_cand.inv_exprs.
7401 (add_candidate_1): Record invariant exprs in iv_cand.inv_exprs
7403 (iv_ca_set_no_cp, iv_ca_set_cp, free_loop_data): Support
7406 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7408 * tree-ssa-loop-ivopts.c (multiplier_allowed_in_address_p): Move
7410 * tree-ssa-address.c (multiplier_allowed_in_address_p): ... to here
7411 as local function. Include necessary header files.
7412 * tree-ssa-loop-ivopts.h (multiplier_allowed_in_address_p): Delete.
7414 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7416 * tree-ssa-loop-ivopts.c (autoinc_possible_for_pair): Simplify.
7418 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7420 * tree-ssa-loop-ivopts.c (force_expr_to_var_cost): Handle more
7421 operators: TRUNC_DIV_EXPR, BIT_AND_EXPR, BIT_IOR_EXPR, LSHIFT_EXPR,
7422 RSHIFT_EXPR and BIT_NOT_EXPR.
7424 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7426 * tree-ssa-loop-ivopts.c (get_loop_invariant_expr): Simplify.
7427 (adjust_setup_cost): New parameter supporting round up adjustment.
7428 (struct address_cost_data): Delete.
7429 (force_expr_to_var_cost): Don't bound cost with spill_cost.
7430 (split_address_cost, ptr_difference_cost): Delete.
7431 (difference_cost, compare_aff_trees, record_inv_expr): Delete.
7432 (struct ainc_cost_data): New struct.
7433 (get_address_cost_ainc): New function.
7434 (get_address_cost, get_computation_cost): Reimplement.
7435 (determine_group_iv_cost_address): Record inv_expr for all uses of
7437 (determine_group_iv_cost_cond): Call get_loop_invariant_expr.
7438 (iv_ca_has_deps): Reimplemented to ...
7439 (iv_ca_more_deps): ... this. Check if NEW_CP introduces more deps
7441 (iv_ca_extend): Call iv_ca_more_deps.
7443 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7445 * tree-ssa-address.c (struct mem_address): Move to header file.
7446 (valid_mem_ref_p, move_fixed_address_to_symbol): Make it global.
7447 * tree-ssa-address.h (struct mem_address): Move from C file.
7448 (valid_mem_ref_p, move_fixed_address_to_symbol): Declare.
7450 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7452 * tree-affine.h (aff_combination_type): New interface.
7453 (aff_combination_zero_p): Remove static.
7454 (aff_combination_const_p): New interface.
7455 (aff_combination_singleton_var_p): New interfaces.
7457 2017-05-11 Richard Biener <rguenther@suse.de>
7459 * tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
7460 Skip unreachable blocks and destinations.
7461 (eliminate): Move stmt removal and fixup ...
7462 (fini_eliminate): ... here. Skip inserted exprs.
7463 (pass_pre::execute): Move fini_pre after fini_eliminate.
7464 * tree-ssa-tailmerge.c: Include tree-cfgcleanup.h.
7465 (tail_merge_optimize): Run cleanup_tree_cfg if requested by
7466 PRE to get rid of dead code that has invalid SSA form and
7467 split critical edges again.
7469 2017-05-11 Bin Cheng <bin.cheng@arm.com>
7471 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
7473 2017-05-11 Richard Biener <rguenther@suse.de>
7475 * passes.c (execute_function_todo): Verify loops if they are
7476 said to be up-to-date.
7477 * cfgexpand.c (pass_expand::execute): Discard loops for -dx.
7478 * trans-mem.c (pass_tm_edges::execute): Mark loops for fixup.
7480 2017-05-10 John David Anglin <danglin@gcc.gnu.org>
7483 * config/pa/pa.c (pa_assemble_integer): When outputting a SYMBOL_REF,
7484 handle calling assemble_external ourself.
7487 * config/pa/pa.c (pa_cannot_change_mode_class): Reject changes to/from
7488 modes with zero size. Enhance comment.
7490 2017-05-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7492 * config/rs6000/rs6000.c (altivec_init_builtins): Define POWER8
7493 built-ins for vec_xl and vec_xst with short and char pointer
7496 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
7498 * config/i386/avx512fintrin.h (_mm_mask_max_round_sd)
7499 (_mm_maskz_max_round_sd, _mm_mask_max_round_ss)
7500 (_mm_maskz_max_round_ss, _mm_mask_min_round_sd)
7501 (_mm_maskz_min_round_sd, _mm_mask_min_round_ss)
7502 (_mm_maskz_min_round_ss): New intrinsics.
7503 * config/i386/i386-builtin-types.def (V2DF, V2DF, V2DF, V2DF, UQI, INT)
7504 (V4SF, V4SF, V4SF, V4SF, UQI, INT): New function type aliases.
7505 * config/i386/i386-builtin.def (__builtin_ia32_maxsd_mask_round)
7506 (__builtin_ia32_maxss_mask_round, __builtin_ia32_minsd_mask_round)
7507 (__builtin_ia32_minss_mask_round): New builtins.
7508 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
7509 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
7510 * config/i386/sse.md (<sse>_vm<code><mode>3<round_saeonly_name>):
7512 (<sse>_vm<code><mode>3<mask_name><round_saeonly_name>): ... this.
7513 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_op3>%2, %1, %0|%0, %1, %<iptr>2<round_saeonly_op3>}):
7515 (v<maxmin_float><ssescalarmodesuffix>\t{<round_saeonly_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_saeonly_mask_op3>}):
7518 2017-05-10 Sebastian Peryt <sebastian.peryt@intel.com>
7520 * config/i386/avx512fintrin.h (_mm_mask_mul_round_sd)
7521 (_mm_maskz_mul_round_sd, _mm_mask_mul_round_ss)
7522 (_mm_maskz_mul_round_ss, _mm_mask_div_round_sd)
7523 (_mm_maskz_div_round_sd, _mm_mask_div_round_ss)
7524 (_mm_maskz_div_round_ss, _mm_mask_mul_sd, _mm_maskz_mul_sd)
7525 (_mm_mask_mul_ss, _mm_maskz_mul_ss, _mm_mask_div_sd)
7526 (_mm_maskz_div_sd, _mm_mask_div_ss, _mm_maskz_div_ss): New intrinsics.
7527 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
7528 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
7529 * config/i386/i386-builtin.def (__builtin_ia32_divsd_mask_round)
7530 (__builtin_ia32_divss_mask_round, __builtin_ia32_mulsd_mask_round)
7531 (__builtin_ia32_mulss_mask_round): New builtins.
7532 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
7533 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
7534 * config/i386/sse.md (<sse>_vm<multdiv_mnemonic><mode>3<round_name>):
7536 (<sse>_vm<multdiv_mnemonic><mode>3<mask_name><round_name>): ... this.
7537 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
7539 (v<multdiv_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
7542 2017-05-10 Julia Koval <julia.koval@intel.com>
7544 * config/i386/avxintrin.h (_mm256_set_m128, _mm256_set_m128d)
7545 (_mm256_set_m128i, _mm256_setr_m128, _mm256_setr_m128d)
7546 (_mm256_setr_m128i): New intrinsics.
7548 2017-05-10 Julia Koval <julia.koval@intel.com>
7550 * config/i386/avx512fintrin.h (_mm_mask_rcp14_sd)
7551 (_mm_maskz_rcp14_sd, _mm_mask_rcp14_ss)
7552 (_mm_maskz_rcp14_ss): New intrinsics.
7553 * config/i386/i386-builtin.def (__builtin_ia32_rcp14sd_mask)
7554 (__builtin_ia32_rcp14ss_mask): New builtins.
7555 * config/i386/sse.md (srcp14<mode>_mask): New pattern.
7557 2017-05-10 Peter Bergner <bergner@vnet.ibm.com>
7559 PR tree-optimization/51513
7560 * tree-cfg.c (gimple_seq_unreachable_p): New function.
7561 (assert_unreachable_fallthru_edge_p): Use it.
7562 (group_case_labels_stmt): Likewise.
7563 * tree-cfg.h: Prototype it.
7564 * stmt.c: Include cfghooks.h and tree-cfg.h.
7565 (emit_case_dispatch_table) <gap_label>: New local variable.
7566 Use it to fill dispatch table gaps.
7567 Test for default_label before updating probabilities.
7568 (expand_case) <default_label>: Remove unneeded initialization.
7569 Test for unreachable default case statement and remove its edge.
7570 Set default_label accordingly.
7571 * tree-ssa-ccp.c (optimize_unreachable): Update comment.
7573 2017-05-10 Carl Love <cel@us.ibm.com>
7575 * config/rs6000/rs6000-c: Add support for built-in functions
7576 vector signed char vec_neg (vector signed char)
7577 vector signed short int vec_neg (vector short int)
7578 vector signed int vec_neg (vector signed int)
7579 vector signed long long vec_neg (vector signed long long)
7580 vector float vec_neg (vector float)
7581 vector double vec_neg (vector double)
7582 * config/rs6000/rs6000-builtin.def: Add definitions for NEG function
7584 * config/rs6000/altivec.h: Add define for vec_neg
7585 * doc/extend.texi: Update the built-in documentation for the
7586 new built-in functions.
7588 2017-05-10 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
7590 PR tree-optimization/77644
7591 * match.pd (sqrt(x) cmp sqrt(y) -> x cmp y): New pattern.
7593 2017-05-10 Nathan Sidwell <nathan@acm.org>
7595 * dumpfile.h (TDI_lang_all): New.
7596 (TDF_KIND): New. Renumber others
7597 (TDF_LANG, TDF_TREE, TDF_RTL, TDF_IPA): Enumerate value, rather
7599 * dumpfile.c (dump_files): Mark language dumps as TDF_LANG. add
7601 (get_dump_file_name): Adjust suffix generation.
7602 (dump_enable_all): Use TDF_KIND.
7603 * doc/invoke.texi (-fdump-lang-all): Document.
7605 * dumpfile.h: Tabify.
7607 2017-05-10 Wilco Dijkstra <wdijkstr@arm.com>
7610 * config/aarch64/cortex-a57-fma-steering.c (merge_forest):
7611 Move member access before delete.
7613 2017-05-10 Alexandre Oliva <aoliva@redhat.com>
7615 * tree-inline.c (expand_call_inline): Split block at stmt
7618 2017-05-09 Michael Meissner <meissner@linux.vnet.ibm.com>
7621 * config/rs6000/rs6000.md (f32_lr): Delete mode attributes that
7622 are now unused after splitting mov{sf,sd}_hardfloat.
7623 (f32_lr2): Likewise.
7625 (f32_lm2): Likewise.
7627 (f32_li2): Likewise.
7630 (f32_sr2): Likewise.
7632 (f32_sm2): Likewise.
7634 (f32_si2): Likewise.
7637 (f32_vsx): Likewise.
7639 (mov<mode>_hardfloat): Split into separate movsf and movsd pieces.
7640 For movsf, order stores so the VSX stores occur before the GPR
7641 store which encourages the register allocator to use a traditional
7642 FPR instead of a GPR. For movsd, order the stores so that the GPR
7643 store comes before the VSX stores to allow the power6 to work.
7644 This is due to the power6 not having a 32-bit integer store
7645 instruction from a FPR.
7646 (movsf_hardfloat): Likewise.
7647 (movsd_hardfloat): Likewise.
7649 2017-05-09 Martin Sebor <msebor@redhat.com>
7651 PR translation/80280
7652 * config/sol2-c.c (cmn_err_flag_specs): Initialize new data member
7655 PR translation/80280
7656 * config/i386/msformat-c.c (ms_printf_flag_specs): Initialize new
7657 data member added in r247778.
7658 (ms_scanf_flag_specs, ms_strftime_flag_specs): Same.
7660 2017-05-09 Nathan Sidwell <nathan@acm.org>
7662 * tree.h (tree_fits_shwi_p, tree_fits_uhwi_p): Unconditionally pure.
7664 * ipa-devirt.c (default_hash_traits<type_pair>): Skip struct-scope
7667 2017-05-09 Marek Polacek <polacek@redhat.com>
7669 * doc/invoke.texi: Fix typo.
7671 2017-05-09 Richard Biener <rguenther@suse.de>
7673 * tree-vrp.c (vrp_val_is_max): Adjust comment.
7674 (vrp_val_is_min): Likewise.
7675 (set_value_range_to_value): Likewise.
7676 (set_value_range_to_nonnegative): Likewise.
7677 (gimple_assign_nonzero_p): Likewise.
7678 (gimple_stmt_nonzero_p): Likewise.
7679 (vrp_int_const_binop): Likewise. Remove unreachable case.
7680 (adjust_range_with_scev): Adjust comments.
7681 (compare_range_with_value): Likewise.
7682 (extract_range_from_phi_node): Likewise.
7683 (test_for_singularity): Likewise.
7685 2017-05-09 Richard Biener <rguenther@suse.de>
7687 * tree-vrp.c (get_single_symbol): Add assert that we don't
7688 get overflowed constants as invariant part.
7689 (compare_values_warnv): Add comment before the TREE_NO_WARNING
7690 checks. Use wi::cmp instead of recursing for integer constants.
7691 (compare_values): Just ignore whether we assumed undefined
7692 overflow instead of failing the compare.
7693 (extract_range_for_var_from_comparison_expr): Add comment before the
7694 TREE_NO_WARNING sets.
7695 (test_for_singularity): Likewise.
7696 (extract_range_from_comparison): Do not disable optimization
7697 when we assumed undefined overflow.
7698 (extract_range_basic): Remove init of unused var.
7700 2017-05-09 Richard Biener <rguenther@suse.de>
7702 * tree-vrp.c (vrp_int_const_binop): Use wide-ints and simplify.
7703 (extract_range_from_multiplicative_op_1): Adjust.
7704 (extract_range_from_binary_expr_1): Use int_const_binop.
7706 2017-05-08 Kelvin Nilsen <kelvin@gcc.gnu.org>
7709 * config/rs6000/power6.md: Replace store_data_bypass_p calls with
7710 rs6000_store_data_bypass_p in seven define_bypass directives and
7711 in several comments.
7712 * config/rs6000/rs6000-protos.h: Add prototype for
7713 rs6000_store_data_bypass_p function.
7714 * config/rs6000/rs6000.c (rs6000_store_data_bypass_p): New
7715 function implements slightly different (rs6000-specific) semantics
7716 than store_data_bypass_p, returning false rather than aborting
7717 with assertion error when arguments do not satisfy the
7718 requirements of store data bypass.
7719 (rs6000_adjust_cost): Replace six calls of store_data_bypass_p with
7720 rs6000_store_data_bypass_p.
7722 2017-05-08 Max Filippov <jcmvbkbc@gmail.com>
7724 * config/xtensa/xtensa-protos.h
7725 (xtensa_initial_elimination_offset): New declaration.
7726 * config/xtensa/xtensa.c (xtensa_initial_elimination_offset):
7727 New function. Move its body from the INITIAL_ELIMINATION_OFFSET
7728 macro definition, add case for FRAME_POINTER_REGNUM when
7729 FRAME_GROWS_DOWNWARD.
7730 * config/xtensa/xtensa.h (FRAME_GROWS_DOWNWARD): New macro definition.
7731 (INITIAL_ELIMINATION_OFFSET): Replace body with call to
7732 xtensa_initial_elimination_offset.
7734 2017-05-08 Nathan Sidwell <nathan@acm.org>
7736 * doc/invoke.texi: Alphabetize -fdump options.
7738 2017-05-08 Martin Sebor <msebor@redhat.com>
7740 PR translation/80280
7741 * config/sol2-c.c (solaris_pragma_align): Correct quoting.
7743 2017-05-08 Bernd Edlinger <bernd.edlinger@hotmail.de>
7745 * target.def (compute_frame_layout): New optional target hook.
7746 * doc/tm.texi.in (TARGET_COMPUTE_FRAME_LAYOUT): Add hook.
7747 * doc/tm.texi (TARGET_COMPUTE_FRAME_LAYOUT): Add documentation.
7748 * lra-eliminations.c (update_reg_eliminate): Call compute_frame_layout
7750 * reload1.c (verify_initial_elim_offsets): Likewise.
7751 * config/arm/arm.c (TARGET_COMPUTE_FRAME_LAYOUT): Define.
7752 (use_simple_return_p): Call arm_compute_frame_layout if needed.
7753 (arm_get_frame_offsets): Split up into this ...
7754 (arm_compute_frame_layout): ... and this function.
7756 2017-05-08 Richard Sandiford <richard.sandiford@arm.com>
7758 * config/aarch64/constraints.md (Usa): New constraint.
7759 * config/aarch64/aarch64.md (*movsi_aarch64, *movdi_aarch64): Use it.
7761 2017-05-08 Thomas Preud'homme <thomas.preudhomme@arm.com>
7763 * config.gcc (arm*-*-*): Set TM_MULTILIB_CONFIG from
7764 with_multilib_list after it has been checked.
7766 2017-05-08 Richard Biener <rguenther@suse.de>
7768 * tree-ssa-pre.c (bitmap_set_and): Avoid bitmap copy.
7769 (bitmap_set_subtract_values): Likewise.
7771 2017-05-08 Richard Biener <rguenther@suse.de>
7773 * tree-vrp.c (gimple_assign_nonzero_warnv_p): Rename to ...
7774 (gimple_assign_nonzero): ... this and remove strict_overflow_p
7776 (gimple_stmt_nonzero_warnv_p): Rename to ...
7777 (gimple_stmt_nonzero_p): ... this and remove strict_overflow_p
7779 (vrp_stmt_computes_nonzero): Remove strict_overflow_p argument.
7780 (extract_range_basic): Adjust, do not disable propagation on
7781 strict overflow sensitive simplification.
7782 (vrp_visit_cond_stmt): Likewise.
7784 2017-05-05 Jan Hubicka <hubicka@ucw.cz>
7786 * ipa-inline-analysis.c (estimate_function_body_sizes): Recompute
7787 body size unconditionally.
7789 2017-05-07 Jeff Law <law@redhat.com>
7792 2017-05-06 Jeff Law <law@redhat.com>
7793 PR tree-optimization/78496
7794 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
7797 PR tree-optimization/78496
7798 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
7799 (simplify_stmt_using_ranges): Call it.
7800 (vrp_dom_walker::before_dom_children): Extract equivalences
7801 from an ASSERT_EXPR with an equality comparison against a
7804 2017-05-06 Jeff Law <law@redhat.com>
7806 PR tree-optimization/78496
7807 * tree-vrp.c (simplify_assert_expr_using_ranges): Remove debugging
7810 PR tree-optimization/78496
7811 * tree-vrp.c (simplify_assert_expr_using_ranges): New function.
7812 (simplify_stmt_using_ranges): Call it.
7813 (vrp_dom_walker::before_dom_children): Extract equivalences
7814 from an ASSERT_EXPR with an equality comparison against a
7817 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
7819 * lra-constraints.c (lra_copy_reg_equiv): New function.
7820 (split_reg): Use it to copy equivalence information from the
7821 original register to the spill register.
7823 2017-05-06 Richard Sandiford <richard.sandiford@linaro.org>
7825 PR rtl-optimization/75964
7826 * simplify-rtx.c (simplify_const_relational_operation): Remove
7827 invalid handling of comparisons of integer ABS.
7829 2017-05-06 Uros Bizjak <ubizjak@gmail.com>
7831 * config/i386/i386.c (ext_80387_constant_init): Do not explicitly
7833 (init_regs): Remove declaration.
7834 (function_arg_advance_32): Initialize error_p as boolean variable.
7836 2017-05-05 Nathan Sidwell <nathan@acm.org>
7838 * store-motion.c (remove_reachable_equiv_notes): Reformat long
7839 lines. Use for (;;).
7841 2017-05-05 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
7843 * config/rs6000/rs6000.c (rs6000_vect_nonmem): New static var.
7844 (rs6000_init_cost): Initialize rs6000_vect_nonmem.
7845 (rs6000_add_stmt_cost): Update rs6000_vect_nonmem.
7846 (rs6000_finish_cost): Avoid vectorizing simple copy loops with
7847 VF=2 that require versioning.
7849 2017-05-05 David Malcolm <dmalcolm@redhat.com>
7851 * diagnostic.h (CARET_LINE_MARGIN): Convert from macro to const
7854 2017-05-05 David Malcolm <dmalcolm@redhat.com>
7856 * diagnostic.h (diagnostic_override_option_index): Convert from
7857 macro to inline function.
7859 2017-05-05 David Malcolm <dmalcolm@redhat.com>
7861 * diagnostic.c (last_module_changed_p): New function.
7862 (set_last_module): New function.
7863 (diagnostic_report_current_module): Convert macro usage to
7864 the above functions.
7865 * diagnostic.h (diagnostic_context::last_module): Strengthen
7866 from const line_map * to const line_map_ordinary *.
7867 (diagnostic_last_module_changed): Delete macro.
7868 (diagnostic_set_last_module): Delete macro.
7870 2017-05-05 David Malcolm <dmalcolm@redhat.com>
7872 * diagnostic.c (diagnostic_impl): Replace report_diagnostic
7873 with diagnostic_report_diagnostic.
7874 (diagnostic_n_impl_richloc): Likewise.
7875 * diagnostic.h (report_diagnostic): Delete macro.
7876 * rtl-error.c (diagnostic_for_asm): Replace report_diagnostic
7877 with diagnostic_report_diagnostic.
7878 * substring-locations.c (format_warning_va): Likewise.
7880 2017-05-05 David Malcolm <dmalcolm@redhat.com>
7882 * diagnostic.c (diagnostic_report_diagnostic): Eliminate
7883 save/restor of format_spec. Move option-printing code to...
7884 (print_option_information): ...this new function, and
7885 reimplement by simply printing to the pretty_printer,
7886 rather than appending to the format string.
7888 2017-05-05 David Malcolm <dmalcolm@redhat.com>
7890 * diagnostic.c (diagnostic_report_diagnostic): Split out pragma
7891 handling logic into...
7892 (update_effective_level_from_pragmas): ...this new function.
7894 2017-05-04 Andrew Waterman <andrew@sifive.com>
7896 * config/riscv/riscv.opt (mstrict-align): New option.
7897 * config/riscv/riscv.h (STRICT_ALIGNMENT): Use it. Update comment.
7898 (SLOW_UNALIGNED_ACCESS): Define.
7899 (riscv_slow_unaligned_access): Declare.
7900 * config/riscv/riscv.c (riscv_tune_info): Add slow_unaligned_access
7902 (riscv_slow_unaligned_access): New variable.
7903 (rocket_tune_info): Set slow_unaligned_access to true.
7904 (optimize_size_tune_info): Set slow_unaligned_access to false.
7905 (riscv_cpu_info_table): Add entry for optimize_size_tune_info.
7906 (riscv_valid_lo_sum_p): Use TARGET_STRICT_ALIGN.
7907 (riscv_option_override): Set riscv_slow_unaligned_access.
7908 * doc/invoke.texi: Add -mstrict-align to RISC-V.
7910 2017-05-04 Kito Cheng <kito.cheng@gmail.com>
7912 * config/riscv/riscv.md: Unify indentation.
7914 2017-05-05 Michael Meissner <meissner@linux.vnet.ibm.com>
7919 * config/rs6000/rs6000.md (u code attribute): Add FIX and
7921 (extendsi<mode>2): Add support for doing sign extension via
7922 VUPKHSW and XXPERMDI if the value is in Altivec registers and we
7923 don't have ISA 3.0 instructions.
7924 (extendsi<mode>2 splitter): Likewise.
7925 (fix_trunc<mode>si2): If we are at ISA 2.07 (VSX small integer),
7926 generate the normal insns since SImode can now go in vector
7927 registers. Disallow the special UNSPECs needed for previous
7928 machines to hide SImode being used. Add new insns
7929 fctiw{,w}_<mode>_smallint if SImode can go in vector registers.
7930 (fix_trunc<mode>si2_stfiwx): Likewise.
7931 (fix_trunc<mode>si2_internal): Likewise.
7932 (fixuns_trunc<mode>si2): Likewise.
7933 (fixuns_trunc<mode>si2_stfiwx): Likewise.
7934 (fctiw<u>z_<mode>_smallint): Likewise.
7935 (fctiw<u>z_<mode>_mem): New combiner pattern to prevent conversion
7936 of floating point to 32-bit integer from doing a direct move to
7937 the GPR registers to do a store.
7938 (fctiwz_<mode>): Break long line.
7940 2017-05-05 Bin Cheng <bin.cheng@arm.com>
7942 * Makefile.in (GTFILES): Add tree-ssa-loop-ivopts.c.
7943 * tree-ssa-loop-ivopts.c (compute_max_addr_offset): Delete.
7944 (addr_list, addr_offset_valid_p): New.
7945 (split_address_groups): Check offset validity with above function.
7946 (gt-tree-ssa-loop-ivopts.h): Include header file.
7948 2017-05-05 Nathan Sidwell <nathan@acm.org>
7950 * config.gcc (arm*-*-*): Add missing 'fi'.
7952 2017-05-05 Steve Ellcey <sellcey@cavium.com>
7954 * doc/invoke.texi (-fopt-info): Explicitly say order of options
7955 included in -fopt-info does not matter.
7956 * doc/optinfo.texi (-fopt-info): Fix description of default
7957 behavour. Explicitly say order of options included in -fopt-info
7960 2017-05-05 Thomas Preud'homme <thomas.preudhomme@arm.com>
7962 * config.gcc: Allow combinations of aprofile and rmprofile values for
7963 --with-multilib-list.
7964 * config/arm/t-multilib: New file.
7965 * config/arm/t-aprofile: Remove initialization of MULTILIB_*
7966 variables. Remove setting of ISA and floating-point ABI in
7967 MULTILIB_OPTIONS and MULTILIB_DIRNAMES. Set architecture and FPU in
7968 MULTI_ARCH_OPTS_A and MULTI_ARCH_DIRS_A rather than MULTILIB_OPTIONS
7969 and MULTILIB_DIRNAMES respectively. Add comment to introduce all
7970 matches. Add architecture matches for marvel-pj4 and generic-armv7-a
7972 * config/arm/t-rmprofile: Likewise except for the matches changes.
7973 * doc/install.texi (--with-multilib-list): Document the combination of
7974 aprofile and rmprofile values and warn about pitfalls in doing that.
7976 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
7978 * config/aarch64/aarch64.md (movsi_aarch64): Remove '*' from r=w.
7979 (movdi_aarch64): Likewise.
7981 2017-05-05 Jakub Jelinek <jakub@redhat.com>
7983 PR tree-optimization/80632
7984 * tree-switch-conversion.c (struct switch_conv_info): Add target_vop
7986 (build_arrays): Initialize it for virtual phis.
7987 (fix_phi_nodes): Use it for virtual phis.
7989 PR tree-optimization/80558
7990 * tree-vrp.c (extract_range_from_binary_expr_1): Optimize
7991 [x, y] op z into [x op, y op z] for op & or | if conditions
7994 2017-05-05 Andre Vieira <andre.simoesdiasvieira@arm.com>
7995 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
7998 * config/arm/arm.md (use_literal_pool): Remove.
7999 (64-bit immediate split): No longer takes cost into consideration
8000 if arm_disable_literal_pool is enabled.
8001 * config/arm/arm.c (arm_tls_referenced_p): Add diagnostic if TLS is
8002 used when arm_disable_literal_pool is enabled.
8003 (arm_max_const_double_inline_cost): Remove use of
8004 arm_disable_literal_pool.
8005 (push_minipool_fix): Add assert.
8006 (arm_reorg): Add return if arm_disable_literal_pool is enabled.
8007 * config/arm/vfp.md (no_literal_pool_df_immediate): New.
8008 (no_literal_pool_sf_immediate): New.
8010 2017-05-05 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8012 PR tree-optimization/80613
8013 * tree-ssa-dce.c (propagate_necessity): Remove cases for
8014 BUILT_IN_STRDUP and BUILT_IN_STRNDUP.
8016 2017-05-05 Richard Biener <rguenther@suse.de>
8018 * tree-ssa-pre.c (get_or_alloc_expr_for): Simplify.
8020 2017-05-05 Georg-Johann Lay <avr@gjlay.de>
8022 * config/avr/avr.md [flag_strict_overflow]: Remove any occurence
8023 of this flag from insn conditions due to removal from r247495.
8025 2017-05-05 Wilco Dijkstra <wdijkstr@arm.com>
8027 * config/arm/aarch-common.c (arm_early_load_addr_dep_ptr):
8029 (arm_early_store_addr_dep_ptr): Likewise.
8030 * config/arm/aarch-common-protos.h
8031 (arm_early_load_addr_dep_ptr): Add prototype.
8032 (arm_early_store_addr_dep_ptr): Likewise.
8033 * config/arm/cortex-a53.md: Add new bypasses.
8035 2017-05-05 Jakub Jelinek <jakub@redhat.com>
8037 * tree.c (next_type_uid): Change type to unsigned.
8038 (type_hash_canon): Decrement back next_type_uid if
8039 freeing a type node with the highest TYPE_UID. For INTEGER_TYPEs
8040 also ggc_free TYPE_MIN_VALUE, TYPE_MAX_VALUE and TYPE_CACHED_VALUES
8043 2017-05-04 Martin Sebor <msebor@redhat.com>
8045 * builtins.c: Fix a trivial typo in a comment.
8048 * builtins.c (check_sizes): Adjust to handle reading past the end.
8049 Avoid printing excessive upper bound of ranges. Use %E to print
8050 tree nodes instead of converting them to %wu.
8051 (expand_builtin_memchr): New function.
8052 (compute_dest_size): Rename...
8053 (compute_objsize): ...to this.
8054 (expand_builtin_memcpy): Adjust.
8055 (expand_builtin_mempcpy): Adjust.
8056 (expand_builtin_strcat): Adjust.
8057 (expand_builtin_strcpy): Adjust.
8058 (check_strncat_sizes): Adjust.
8059 (expand_builtin_strncat): Adjust.
8060 (expand_builtin_strncpy): Adjust and simplify.
8061 (expand_builtin_memset): Adjust.
8062 (expand_builtin_bzero): Adjust.
8063 (expand_builtin_memcmp): Adjust.
8064 (expand_builtin): Handle memcmp.
8065 (maybe_emit_chk_warning): Check strncat just once.
8067 2017-05-04 Martin Sebor <msebor@redhat.com>
8069 PR preprocessor/79214
8072 * builtins.c (check_sizes): Add inlining context and issue
8073 warnings even when -Wno-system-headers is set.
8074 (check_strncat_sizes): Same.
8075 (expand_builtin_strncat): Same.
8076 (expand_builtin_memmove): New function.
8077 (expand_builtin_stpncpy): Same.
8078 (expand_builtin): Handle memmove and stpncpy.
8080 2017-05-04 Bin Cheng <bin.cheng@arm.com>
8082 * tree-ssa-loop-ivopts.c (struct cost_pair): Remove field inv_expr
8083 which is not used any more.
8085 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
8087 * config/aarch64/aarch64.c (generic_tunings): Update prefetch model.
8089 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
8091 * config/aarch64/aarch64.c (cortexa35_tunings): Set jump alignment to 4.
8092 (cortexa53_tunings): Likewise.
8093 (cortexa57_tunings): Likewise.
8094 (cortexa72_tunings): Likewise.
8095 (cortexa73_tunings): Likewise.
8097 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
8099 * config/aarch64/aarch64.c (generic_tunings): Set jump alignment to 4.
8100 Set loop alignment to 8.
8102 2017-05-04 Martin Sebor <msebor@redhat.com>
8104 PR translation/80280
8105 * builtins.c (expand_builtin_object_size): Add missing quoting to
8106 %D and like directives.
8107 * hsa-gen.c (hsa_type_for_scalar_tree_type): Same.
8108 (hsa_type_for_tree_type): Same.
8109 (verify_function_arguments): Same.
8110 * symtab.c (symbol_table::change_decl_assembler_name): Same.
8111 * varasm.c (get_section): Same.
8114 2017-05-04 Martin Sebor <msebor@redhat.com>
8116 PR translation/80280
8117 * config/i386/i386.c (ix86_function_versions): Quote a %D directive.
8119 2017-05-04 Wilco Dijkstra <wdijkstr@arm.com>
8121 * config/aarch64/aarch64.c (generic_addrcost_table):
8122 Change HI/TI mode setting.
8124 2017-05-04 Martin Jambor <mjambor@suse.cz>
8126 PR tree-optimization/80622
8127 * tree-sra.c (comes_initialized_p): New function.
8128 (build_accesses_from_assign): Only set write lazily when
8129 comes_initialized_p is false.
8130 (analyze_access_subtree): Use comes_initialized_p.
8131 (propagate_subaccesses_across_link): Assert !comes_initialized_p
8132 instead of testing for PARM_DECL.
8134 2017-05-04 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
8136 * config/aarch64/aarch64.md (prefetch); Adjust predicate and
8137 constraint on operand 0 to allow more general addressing modes.
8138 Adjust output template.
8139 * config/aarch64/aarch64.c (aarch64_address_valid_for_prefetch_p):
8141 * config/aarch64/aarch64-protos.h
8142 (aarch64_address_valid_for_prefetch_p): Declare prototype.
8143 * config/aarch64/constraints.md (Dp): New address constraint.
8144 * config/aarch64/predicates.md (aarch64_prefetch_operand): New
8147 2017-05-04 Jan Hubicka <hubicka@ucw.cz>
8149 * ipa-cp.c (perform_estimation_of_a_value): Drop base_time parameter;
8150 update use of estimate_ipcp_clone_size_and_time.
8151 (estimate_local_effects): Update use of
8152 estimate_ipcp_clone_size_and_time and perform_estimation_of_a_value.
8153 * ipa-inline.h (estimate_ipcp_clone_size_and_time): Update prototype.
8154 * ipa-inline-analysis.c (estimate_ipcp_clone_size_and_time):
8155 Return nonspecialized time.
8157 2017-05-04 Richard Biener <rguenther@suse.de>
8159 * tree-ssa-alias.c (get_continuation_for_phi): Improve looking
8160 for the last VUSE which def dominates the PHI. Directly call
8162 (get_continuation_for_phi_1): Remove.
8164 2017-05-04 Richard Sandiford <richard.sandiford@linaro.org>
8166 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): Add commentary
8167 to explain the use of truncating division. Cap the number of
8168 iterations to the maximum given by nb_iterations_upper_bound,
8171 2017-05-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
8173 * configure.ac (--enable-mingw-wildcard): Add new configurable feature.
8174 * configure: Regenerate.
8175 * config.in: Regenerate.
8176 * config/i386/driver-mingw32.c: new file.
8177 * config/i386/x-mingw32: Add rule to build driver-mingw32.o.
8178 * config.host: Link driver-mingw32.o on MinGW host.
8179 * doc/install.texi: Document new --enable-mingw-wildcard configure
8182 2017-05-04 Marek Polacek <polacek@redhat.com>
8184 PR tree-optimization/80612
8185 * calls.c (get_size_range): Check for INTEGRAL_TYPE_P.
8187 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8188 Andre Simoes Dias Vieira <andre.simoesdiasvieira@arm.com>
8190 * config/arm/arm.md (movsi): Change TARGET_32BIT to TARGET_HAVE_MOVT.
8191 (movt splitter): Likewise.
8192 * config/arm/arm.c (arm_option_check_internal): Change arm_arch_thumb2
8193 to TARGET_HAVE_MOVT, and merge with -mslow-flash-data check.
8194 (const_ok_for_arm): Change else to else if (TARGET_THUMB2) and add else
8195 block for Thumb-1 with MOVT.
8196 (thumb2_legitimate_address_p): Move code block ...
8197 (can_avoid_literal_pool_for_label_p): ... into this new function.
8198 (thumb1_legitimate_address_p): Add check for TARGET_HAVE_MOVT and
8200 (thumb_legitimate_constant_p): Add conditional on TARGET_HAVE_MOVT
8201 * doc/invoke.texi (-mpure-code): Change "ARMv7-M targets" for
8202 "M-profile targets with the MOVT instruction".
8204 2017-05-04 Prakhar Bahuguna <prakhar.bahuguna@arm.com>
8206 * config/arm/arm-builtins.c (arm_init_builtins): Rename
8207 __builtin_arm_ldfscr to __builtin_arm_get_fpscr, and rename
8208 __builtin_arm_stfscr to __builtin_arm_set_fpscr.
8210 2017-05-04 Martin Liska <mliska@suse.cz>
8212 * tree-vrp.c (simplify_cond_using_ranges_2): Remove unused
8215 2017-05-04 Richard Biener <rguenther@suse.de>
8217 * tree.c (array_at_struct_end_p): Handle arrays at struct
8218 end with flexarrays more conservatively. Refactor and treat
8219 arrays of arrays or aggregates more strict. Fix
8220 VIEW_CONVERT_EXPR handling. Remove allow_compref argument.
8221 * tree.c (array_at_struct_end_p): Adjust prototype.
8222 * emit-rtl.c (set_mem_attributes_minus_bitpos): Adjust.
8223 * gimple-fold.c (get_range_strlen): Likewise.
8224 * tree-chkp.c (chkp_may_narrow_to_field): Likewise.
8226 2017-05-04 Richard Biener <rguenther@suse.de>
8228 PR tree-optimization/31130
8229 * tree-vrp.c (needs_overflow_infinity): Remove as always returning
8231 (supports_overflow_infinity): Likewise.
8232 (is_negative_overflow_infinity): Likewise.
8233 (is_positive_overflow_infinity): Likewise.
8234 (is_overflow_infinity): Likewise.
8235 (stmt_overflow_infinity): Likewise.
8236 (overflow_infinity_range_p): Likewise.
8237 (usable_range_p): Remove as always returning true.
8238 (make_overflow_infinity): Remove.
8239 (negative_overflow_infinity): Likewise.
8240 (positive_overflow_infinity): Likewise.
8241 (avoid_overflow_infinity): Likewise.
8242 (set_value_range): Adjust accordingly.
8243 (set_value_range_to_nonnegative): Likewise, remove now unused
8244 overflow_infinity arg.
8245 (vrp_operand_equal_p): Adjust.
8246 (update_value_range): Likewise.
8247 (range_int_cst_singleton_p): Likewise.
8248 (operand_less_p): Likewise.
8249 (compare_values_warnv): Likewise.
8250 (extract_range_for_var_from_comparison_expr): Likewise.
8251 (vrp_int_const_binop): Likewise.
8252 (zero_nonzero_bits_from_vr): Likewise.
8253 (extract_range_from_multiplicative_op_1): Likewise.
8254 (extract_range_from_binary_expr_1): Likewise.
8255 (extract_range_from_unary_expr): Likewise.
8256 (extract_range_from_comparison): Likewise.
8257 (extract_range_basic): Likewise.
8258 (adjust_range_with_scev): Likewise.
8259 (compare_ranges): Likewise.
8260 (compare_range_with_value): Likewise.
8261 (dump_value_range): Likewise.
8262 (test_for_singularity): Likewise, remove strict_overflow_p parameter
8264 (simplify_cond_using_ranges): Adjust.
8266 2017-05-04 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
8268 * brig-builtins.def: Added a builtin for class_f64.
8269 * builtin-types.def: Added a builtin type needed by class_f64.
8271 2017-05-03 Jason Merrill <jason@redhat.com>
8273 * timevar.def: Add TV_CONSTEXPR.
8275 2017-05-03 David Malcolm <dmalcolm@redhat.com>
8277 * common.opt (fdiagnostics-parseable-fixits): Fix typo.
8279 2017-05-03 Martin Jambor <mjambor@suse.cz>
8281 * ipa-prop.c (ipa_update_after_lto_read): Removed.
8282 * ipa-prop.h (ipa_update_after_lto_read): Remove declaration.
8283 * ipa-cp.c (ipcp_propagate_stage): Do not call
8284 ipa_update_after_lto_read.
8285 * ipa-inline.c (ipa_inline): Likewise.
8287 2017-05-03 Martin Jambor <mjambor@suse.cz>
8289 * ipa-prop.h (ipa_edge_args): Make a class. Mark with for_user GTY
8290 tag. Added a default constructor and a destructor.
8291 (ipa_edge_args_sum_t): New class;
8292 (ipa_edge_args_sum): Declare.
8293 (ipa_edge_args_vector): Remove declaration.
8294 (IPA_EDGE_REF): Use ipa_edge_args_sum.
8295 (ipa_free_edge_args_substructures): Remove declaration.
8296 (ipa_check_create_edge_args): Use ipa_edge_args_sum.
8297 (ipa_edge_args_info_available_for_edge_p): Likewise.
8298 * ipa-prop.c (ipa_edge_args_vector): Removed.
8299 (edge_removal_hook_holder): Likewise.
8300 (edge_duplication_hook_holder): Likewise.
8301 (ipa_edge_args_sum): New variable.
8302 (ipa_propagate_indirect_call_infos): Test ipa_edge_args_sum instead of
8303 ipa_edge_args_vector.
8304 (ipa_free_edge_args_substructures): Likewise.
8305 (ipa_free_all_edge_args): Free ipa_edge_args_sum instead of
8306 ipa_edge_args_vector.
8307 (ipa_edge_removal_hook): Turned into method
8308 ipa_edge_args_sum_t::remove.
8309 (ipa_edge_duplication_hook): Turned into method
8310 ipa_edge_args_sum_t::duplicate.
8311 (ipa_register_cgraph_hooks): Create ipa_edge_args_sum instead of
8312 registering edge hooks.
8313 (ipa_unregister_cgraph_hooks): Do not unregister edge hooks.
8314 * ipa-inline-analysis.c (estimate_function_body_sizes): Test
8315 ipa_edge_args_sum instead of ipa_edge_args_vector.
8316 * ipa-profile.c (ipa_profile): Likewise.
8318 2017-05-03 Martin Jambor <mjambor@suse.cz>
8320 * symbol-summary.h (function_summary): New method exists.
8321 (function_summary::symtab_removal): Deallocate through release.
8322 (call_summary): New class.
8323 (gt_ggc_mx): New overload.
8324 (gt_pch_nx): Likewise.
8325 (gt_pch_nx): Likewise.
8327 2017-05-03 Jeff Law <law@redhat.com>
8329 PR tree-optimization/78496
8330 * tree-vrp.c (simplify_cond_using_ranges_1): Renamed
8331 from simplify_cond_using_ranges. Split off code to walk
8332 backwards through casts into ...
8333 (simplify_cond_using_ranges_2): New function.
8334 (simplify_stmt_using_ranges): Call simplify_cond_using_ranges_1.
8335 (execute_vrp): After identifying jump threads, call
8336 simplify_cond_using_ranges_2.
8338 2017-05-03 Jan Hubicka <hubicka@ucw.cz>
8341 * ipa-inline.h (inline_summary): Add ctor.
8342 (create_ggc): Do not use ggc_cleared_alloc.
8344 2017-05-03 Jeff Downs <heydowns@somuchpressure.net>
8345 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
8347 * gcc.c (handle_braces): Support escaping in switch matching
8349 * doc/invoke.texi (Spec Files): Document it.
8350 Remove superfluous @code markup in items.
8352 2017-05-03 David Malcolm <dmalcolm@redhat.com>
8354 * diagnostic-show-locus.c (struct column_range): New struct.
8355 (get_affected_columns): New function.
8356 (get_printed_columns): New function.
8357 (struct correction): New struct.
8358 (correction::ensure_capacity): New function.
8359 (correction::ensure_terminated): New function.
8360 (struct line_corrections): New struct.
8361 (line_corrections::~line_corrections): New dtor.
8362 (line_corrections::add_hint): New function.
8363 (layout::print_trailing_fixits): Reimplement in terms of the new
8365 (selftest::test_overlapped_fixit_printing): New function.
8366 (selftest::diagnostic_show_locus_c_tests): Call it.
8368 2017-05-03 Nathan Sidwell <nathan@acm.org>
8370 Canonicalize canonical type hashing
8371 * tree.h (type_hash_canon_hash): Declare.
8372 * tree.c (type_hash_list, attribute_hash_list): Move into
8373 type_hash_canon_hash.
8374 (build_type_attribute_qual_variant): Break out hash code calc into
8375 type_hash_canon_hash.
8376 (type_hash_canon_hash): New. Generic type hash computation.
8377 (build_range_type_1, build_array_type_1, build_function_type,
8378 build_method_type_directly, build_offset_type, build_complex_type,
8379 make_vector_type): Call it.
8381 2017-05-03 Richard Biener <rguenther@suse.de>
8383 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment):
8384 When all DRs have unknown misaligned do not always peel
8385 when there is a store but apply the same costing model as if
8386 there were only loads.
8388 2017-05-03 Richard Biener <rguenther@suse.de>
8391 PR tree-optimization/80492
8392 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
8393 compare_base_decls returning dont-know properly.
8395 2017-05-03 Thomas Preud'homme <thomas.preudhomme@arm.com>
8397 * config/arm/iterators.md (CCSI): New mode iterator.
8398 (arch): New mode attribute.
8399 * config/arm/sync.md (atomic_compare_and_swap<mode>_1): Rename into ...
8400 (atomic_compare_and_swap<CCSI:arch><NARROW:mode>_1): This and ...
8401 (atomic_compare_and_swap<CCSI:arch><SIDI:mode>_1): This. Use CCSI
8402 code iterator for success result mode.
8403 * config/arm/arm.c (arm_expand_compare_and_swap): Adapt code to use
8404 the corresponding new insn generators.
8406 2017-05-03 Bin Cheng <bin.cheng@arm.com>
8409 2017-05-02 Bin Cheng <bin.cheng@arm.com>
8410 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
8412 2017-05-03 Richard Sandiford <richard.sandiford@linaro.org>
8414 * tree-data-ref.h (SUB_CONFLICTS_IN_A): Wrap SUB argument in brackets.
8415 (SUB_CONFLICTS_IN_B, SUB_LAST_CONFLICT, SUB_DISTANCE): Likewise.
8416 (DDR_A): Wrap DDR argument in brackets.
8417 (DDR_B, DDR_AFFINE_P, DDR_ARE_DEPENDENT, DDR_SUBSCRIPTS): Likewise.
8418 (DDR_LOOP_NEST, DDR_INNER_LOOP, DDR_SELF_REFERENCE): Likewise.
8419 (DDR_REVERSED_P): Likewise.
8421 2017-05-03 Jakub Jelinek <jakub@redhat.com>
8423 PR tree-optimization/79472
8424 * tree-switch-conversion.c (struct switch_conv_info): Add
8425 contiguous_range and default_case_nonstandard fields.
8426 (collect_switch_conv_info): Compute contiguous_range and
8427 default_case_nonstandard fields, don't clear final_bb if
8428 contiguous_range and only the default case doesn't have the required
8430 (check_all_empty_except_final): Set default_case_nonstandard instead
8431 of failing if contiguous_range and the default case doesn't have empty
8433 (check_final_bb): Add SWTCH argument, don't fail if contiguous_range
8434 and only the default case doesn't have the required constants. Skip
8436 (gather_default_values): Skip virtual phis. Allow non-NULL CASE_LOW
8437 if default_case_nonstandard.
8438 (build_constructors): Build constant 1 just once. Assert that default
8439 values aren't inserted in between cases if contiguous_range. Skip
8441 (build_arrays): Skip virtual phis.
8442 (prune_bbs): Add DEFAULT_BB argument, don't remove that bb.
8443 (fix_phi_nodes): Don't add e2f phi arg if default_case_nonstandard.
8444 Handle virtual phis.
8445 (gen_inbound_check): Handle default_case_nonstandard case.
8446 (process_switch): Adjust check_final_bb caller. Call
8447 gather_default_values with the first non-default case instead of
8448 default case if default_case_nonstandard.
8450 2017-05-02 Nathan Sidwell <nathan@acm.org>
8452 * ggc-page.c (move_ptes_to_front): Replace unsigned >0 with i--
8453 check. Fix formatting.
8455 2017-05-02 Jan Hubicka <hubicka@ucw.cz>
8457 * ipa-inline-analysis.c (estimate_node_size_and_time): Allow roundoff
8458 errors when comparing specialized and unspecialized times.
8460 2017-05-02 David Malcolm <dmalcolm@redhat.com>
8462 * diagnostic-show-locus.c
8463 (layout::should_print_annotation_line_p): Make private.
8464 (layout::print_annotation_line): Make private.
8465 (layout::annotation_line_showed_range_p): Make private.
8466 (layout::show_ruler): Make private.
8467 (layout::print_source_line): Make private. Pass in line and
8468 line_width, rather than calling location_get_source_line. Drop
8470 (layout::print_leading_fixits): New method.
8471 (layout::print_any_fixits): Rename to...
8472 (layout::print_trailing_fixits): ...this, and make private.
8473 Don't print newline fixits.
8474 (diagnostic_show_locus): Move logic for printing one row into...
8475 (layout::print_line): ...this new function. Move the
8476 location_get_source_line call and error-handling from
8477 print_source_line to here. Call print_leading_fixits, and rename
8478 print_any_fixits to print_trailing_fixits.
8479 (selftest::test_fixit_insert_containing_newline): Update now that
8480 newlines are partially supported.
8481 (selftest::test_fixit_insert_containing_newline_2): New test.
8482 (selftest::test_fixit_replace_containing_newline): Update comments.
8483 (selftest::diagnostic_show_locus_c_tests): Call the new test.
8484 * edit-context.c (class added_line): New class.
8485 (class edited_line): Describe newline handling in comment.
8486 (edited_line::actually_edited_p): New method.
8487 (edited_line::print_content): Delete redundant decl.
8488 (edited_line::m_predecessors): New field.
8489 (edited_file::print_content): Call edited_line::print_content.
8490 (edited_file::print_diff): Update to support newlines.
8491 (edited_file::print_diff_hunk): Likewise.
8492 (edited_file::print_run_of_changed_lines): New function.
8493 (edited_file::print_diff_line): Convert to...
8494 (print_diff_line): ...this.
8495 (edited_file::get_effective_line_count): New function.
8496 (edited_line::edited_line): Initialize new field m_predecessors.
8497 (edited_line::~edited_line): Clean up m_predecessors.
8498 (edited_line::apply_fixit): Handle newlines.
8499 (edited_line::get_effective_line_count): New function.
8500 (edited_line::print_content): New function.
8501 (edited_line::print_diff_lines): New function.
8502 (selftest::test_applying_fixits_insert_containing_newline): New
8504 (selftest::test_applying_fixits_replace_containing_newline): New
8506 (selftest::insert_line): New function.
8507 (selftest::test_applying_fixits_multiple_lines): Add example of
8509 (selftest::edit_context_c_tests): Call the new tests.
8511 2017-05-02 Bin Cheng <bin.cheng@arm.com>
8513 * tree-ssa-loop-ivopts.c (get_scaled_computation_cost_at): Delete
8514 parameter cand. Update dump information.
8515 (get_computation_cost): Update uses.
8517 2017-05-02 Bin Cheng <bin.cheng@arm.com>
8519 * tree-ssa-loop-ivopts.c (get_computation_aff_1): New.
8520 (get_computation_aff): Reorder parameters. Use get_computation_aff_1.
8521 (get_computation_at, rewrite_use_address): Update use of
8522 get_computation_aff.
8524 2017-05-02 Bin Cheng <bin.cheng@arm.com>
8526 * tree-ssa-loop-ivopts.c (get_computation_at): Reorder parameters.
8527 (get_computation): Delete.
8528 (get_computation_cost): Implement like get_computation_cost_at.
8529 Use get_computation_at.
8530 (get_computation_cost_at): Delete.
8531 (rewrite_use_nonlinear_expr): Use get_computation_at.
8532 (rewrite_use_compare, remove_unused_ivs): Ditto.
8534 2017-05-02 Bin Cheng <bin.cheng@arm.com>
8536 * tree-ssa-loop-ivopts.c (rewrite_use_address): Simple refactor.
8538 2017-05-02 Bin Cheng <bin.cheng@arm.com>
8540 * tree-ssa-loop-ivopts.c (struct iv_ca): Rename n_regs to n_invs.
8541 (ivopts_global_cost_for_size): Rename parameter and update uses.
8542 (iv_ca_recount_cost): Update uses.
8543 (iv_ca_set_remove_invs, iv_ca_set_no_cp): Record invariants and
8544 candidates seperately in n_invs and n_cands.
8545 (iv_ca_set_add_invs, iv_ca_set_cp, iv_ca_new): Ditto.
8547 2017-05-02 Bin Cheng <bin.cheng@arm.com>
8549 * tree-ssa-loop-ivopts.c (struct walk_tree_data): New.
8550 (find_inv_vars_cb): New.
8551 (find_depends): Renamed to ...
8552 (find_inv_vars): ... this.
8553 (add_candidate_1, force_var_cost): Call find_inv_vars.
8554 (split_address_cost, determine_group_iv_cost_cond): Ditto.
8556 2017-05-02 Bin Cheng <bin.cheng@arm.com>
8558 * tree-ssa-loop-ivopts.c (struct cost_pair): Rename depends_on to
8559 inv_vars. Add inv_exprs.
8560 (struct iv_cand): Rename depends_on to inv_vars.
8561 (struct ivopts_data): Rename max_inv_id/n_invariant_uses to
8562 max_inv_var_id/n_inv_var_uses. Move max_inv_expr_id around.
8563 Refactor field used_inv_exprs from has_map to array n_inv_expr_uses.
8564 (dump_cand): Dump inv_vars.
8565 (tree_ssa_iv_optimize_init): Support inv_vars and inv_exprs.
8566 (record_invariant, find_depends, add_candidate_1): Ditto.
8567 (set_group_iv_cost, force_var_cost): Ditto.
8568 (split_address_cost, ptr_difference_cost, difference_cost): Ditto.
8569 (get_computation_cost_at, get_computation_cost): Ditto.
8570 (determine_group_iv_cost_generic): Ditto.
8571 (determine_group_iv_cost_address): Ditto.
8572 (determine_group_iv_cost_cond, autoinc_possible_for_pair): Ditto.
8573 (determine_group_iv_costs): Ditto.
8574 (iv_ca_recount_cost): Update call to ivopts_global_cost_for_size.
8575 (iv_ca_set_remove_invariants): Renamed to ...
8576 (iv_ca_set_remove_invs): ... this. Support inv_vars and inv_exprs.
8577 (iv_ca_set_no_cp): Use iv_ca_set_remove_invs.
8578 (iv_ca_set_add_invariants): Renamed to ...
8579 (iv_ca_set_add_invs): ... this. Support inv_vars and inv_exprs.
8580 (iv_ca_set_cp): Use iv_ca_set_add_invs.
8581 (iv_ca_has_deps): Support inv_vars and inv_exprs.
8582 (iv_ca_new, iv_ca_free, iv_ca_dump, free_loop_data): Ditto.
8583 (create_new_ivs): Remove useless dump.
8585 2017-05-02 Bin Cheng <bin.cheng@arm.com>
8587 * tree-ssa-loop-ivopts.c (get_computation_cost_at): Remove pseudo
8589 (determine_group_iv_cost_cond, determine_iv_cost): Ditto.
8590 (iv_ca_set_no_cp, create_new_iv): Ditto.
8592 2017-05-02 Bin Cheng <bin.cheng@arm.com>
8594 * rtlanal.c (rtx_cost): Handle TRUNCATE between tieable modes.
8596 2017-05-02 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
8598 * tree.h (EXPR_CILK_SPAWN): Use macro TREE_CHECK2 instead of
8599 function tree_check2.
8601 2017-05-02 Martin Liska <mliska@suse.cz>
8603 * doc/gcov.texi: Add missing preposition.
8604 * gcov.c (function_info::function_info): Properly fill up
8605 all member variables.
8607 2017-05-02 Tamar Christina <tamar.christina@arm.com>
8609 * expr.c (expand_expr_real_2): Re-cost if previous costs are the same.
8611 2017-05-02 Tamar Christina <tamar.christina@arm.com>
8613 * simplify-rtx.c (simplify_binary_operation_1): Add LSHIFTRT case.
8615 2017-05-02 Martin Liska <mliska@suse.cz>
8618 * lto-streamer-in.c (lto_read_tree_1): Remove
8620 * lto-streamer.c (struct tree_hash_entry): Likewise.
8621 (struct tree_entry_hasher): Likewise.
8622 (tree_entry_hasher::hash): Likewise.
8623 (tree_entry_hasher::equal): Likewise.
8624 (lto_streamer_init): Likewise.
8625 (lto_orig_address_map): Likewise.
8626 (lto_orig_address_get): Likewise.
8627 (lto_orig_address_remove): Likewise.
8628 * lto-streamer.h: Likewise.
8629 * tree-streamer-in.c (streamer_alloc_tree): Likewise.
8630 * tree-streamer-out.c (streamer_write_tree_header): Likewise.
8632 2017-05-02 Sebastian Peryt <sebastian.peryt@intel.com>
8634 * config/i386/avx512fintrin.h (_mm_mask_add_round_sd)
8635 (_mm_maskz_add_round_sd, _mm_mask_add_round_ss)
8636 (mm_maskz_add_round_ss, _mm_mask_sub_round_sd)
8637 (mm_maskz_sub_round_sd, _mm_mask_sub_round_ss)
8638 (mm_maskz_sub_round_ss, _mm_mask_add_sd)
8639 (mm_maskz_add_sd, _mm_mask_add_ss, _mm_maskz_add_ss)
8640 (mm_mask_sub_sd, _mm_maskz_sub_sd, _mm_mask_sub_ss)
8641 (mm_maskz_sub_ss): New intrinsics.
8642 * config/i386/i386-builtin-types.def (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8643 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): New function type aliases.
8644 * config/i386/i386-builtin.def (__builtin_ia32_addsd_mask_round)
8645 (__builtin_ia32_addss_mask_round, __builtin_ia32_subsd_mask_round)
8646 (__builtin_ia32_subss_mask_round): New builtins.
8647 * config/i386/i386.c (V2DF_FTYPE_V2DF_V2DF_V2DF_UQI_INT)
8648 (V4SF_FTYPE_V4SF_V4SF_V4SF_UQI_INT): Handle new types.
8649 * config/i386/sse.md (<sse>_vm<plusminus_insn><mode>3<round_name>):
8651 (<sse>_vm<plusminus_insn><mode>3<mask_name><round_name>): ... this.
8652 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_op3>%2, %1, %0|%0, %1, %<iptr>2<round_op3>}):
8654 (v<plusminus_mnemonic><ssescalarmodesuffix>\t{<round_mask_op3>%2, %1, %0<mask_operand3>|%0<mask_operand3>, %1, %<iptr>2<round_mask_op3>}):
8657 2017-05-02 Martin Jambor <mjambor@suse.cz>
8659 PR tree-optimization/78687
8660 * tree-sra.c (access): New field parent.
8661 (process_subtree_disqualification): New function.
8662 (disqualify_candidate): Call it.
8663 (build_accesses_from_assign): Reset write flag if creating an
8665 (build_access_subtree): Fill in parent field and also prpagate
8666 down grp_write flag.
8667 (create_artificial_child_access): New parameter set_grp_write, set
8668 grp_write to its value.
8669 (propagate_subaccesses_across_link): Also propagate grp_write flag
8671 (propagate_all_subaccesses): Push the closest parent back to work
8672 queue if add_access_to_work_queue returned true.
8674 2017-05-02 Richard Biener <rguenther@suse.de>
8676 * common.opt (fstrict-overflow): Alias negative to fwrapv.
8677 * doc/invoke.texi (fstrict-overflow): Remove all traces of
8678 -fstrict-overflow documentation.
8679 * tree.h (TYPE_OVERFLOW_UNDEFINED): Do not test flag_strict_overflow.
8680 (POINTER_TYPE_OVERFLOW_UNDEFINED): Test !flag_wrapv instead of
8681 flag_strict_overflow.
8682 * ipa-inline.c (can_inline_edge_p): Do not test flag_strict_overflow.
8683 * lto-opts.c (lto_write_options): Do not stream it.
8684 * lto-wrapper.c (merge_and_complain): Do not handle it.
8685 * opts.c (default_options_table): Do not set -fstrict-overflow.
8686 (finish_options): Likewise do not clear it when sanitizing.
8687 * simplify-rtx.c (simplify_const_relational_operation): Do not
8688 test flag_strict_overflow.
8690 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
8692 * config/alpha/alpha.md (*add<mode>3_ieee): Merge to add<mode>3
8693 using enabled attribute.
8694 (*sub<mode>3_ieee): Merge to sub<mode>3 using enabled attribute.
8695 (*mul<mode>3_ieee): Merge to mul<mode>3 using enabled attribute.
8696 (*div<mode>3_ieee): Merge to div<mode>3 using enabled attribute.
8697 (*sqrt<mode>2_ieee): Merge to sqrt<mode>2 using enabled attribute.
8698 (*fix_truncdfdi_ieee): Merge to *fix_truncdfdi2 using enabled attribute.
8699 (*fix_truncsfdi_ieee): Merge to *fix_truncsfdi2 using enabled attribute.
8700 (*floatdisf_ieee): Merge to floatdisf2 using enabled attribute.
8701 (*floatdidf_ieee): Merge to floatdidf2 using enabled attribute.
8702 (*truncdfsf2_ieee): Merge to truncdfsf2 using enabled attribute.
8703 (*cmpdf_ieee): Merge to *cmpdf_internal using enabled attribute.
8705 2017-05-02 Uros Bizjak <ubizjak@gmail.com>
8707 * config/i386/i386.c (ix86_code_end): Use {FIRST,LAST}_INT_REG.
8709 2017-05-02 Richard Biener <rguenther@suse.de>
8711 PR tree-optimization/80591
8713 2017-04-10 Richard Biener <rguenther@suse.de>
8715 * tree-ssa-structalias.c (find_func_aliases): Properly handle
8718 2017-05-02 Richard Biener <rguenther@suse.de>
8720 PR tree-optimization/80549
8721 * tree-cfgcleanup.c (mfb_keep_latches): New helper.
8722 (cleanup_tree_cfg_noloop): Create forwarders to known loop
8723 headers if they do not have a preheader.
8725 2017-05-02 Martin Liska <mliska@suse.cz>
8728 * common.opt: Fix typo.
8729 * doc/invoke.texi: Likewise.
8731 2017-05-01 Jan Beulich <jbeulich@suse.com>
8733 * config/i386/sse.md (xop_vpermil2<mode>3): Do not allow operand
8734 swapping, add (x,x,m,x,n) alternative.
8736 2017-05-01 Nathan Sidwell <nathan@acm.org>
8738 * calls.c (combine_pending_stack_adjustment_and_call): Remove
8739 unnecessary unadjusted_alignment check.
8741 2017-05-01 Xi Ruoyao <ryxi@stu.xidian.edu.cn>
8744 * cilk_common.c (expand_builtin_cilk_detach): Move pedigree
8746 * gimplify.c (gimplify_cilk_detach): New function.
8747 (gimplify_call_expr, gimplify_modify_expr): Call it as needed.
8748 * tree-core.h: Document EXPR_CILK_SPAWN.
8749 * tree.h (EXPR_CILK_SPAWN): Define.
8751 2017-05-01 David Malcolm <dmalcolm@redhat.com>
8753 * diagnostic-show-locus.c (layout::get_expanded_location): Rewrite
8754 to use new fixit_hint representation, using the "replace" logic.
8755 (get_line_span_for_fixit_hint): Likewise.
8756 (layout::print_any_fixits): Likewise.
8757 (selftest::test_one_liner_many_fixits): Rename to...
8758 (selftest::test_one_liner_many_fixits_1): ...this, and update
8759 comment and expected output to reflect that the multiple fix-it
8760 hints are now consolidated into one insertion.
8761 (selftest::test_one_liner_many_fixits_2): New test.
8762 (selftest::test_diagnostic_show_locus_one_liner): Update for
8764 (selftest::test_fixit_consolidation): Update for fix-it API
8766 * diagnostic.c (print_parseable_fixits): Likewise.
8767 * edit-context.c (edited_line::m_line_events): Convert from
8768 auto_vec <line_event *> to auto_vec <line_event>.
8769 (class line_event): Convert from abstract base class to a concrete
8770 class, taking over the role of replace_event.
8771 (class insert_event): Delete.
8772 (class replace_event): Rename to class line_event. Convert to
8774 (edit_context::add_fixits): Reimplement.
8775 (edit_context::apply_insert): Delete.
8776 (edit_context::apply_replace): Rename to...
8777 (edit_context::apply_fixit): ...this. Convert to half-open range.
8778 (edited_file::apply_insert): Delete.
8779 (edited_file::apply_replace): Rename to...
8780 (edited_file::apply_fixit): ...this.
8781 (edited_line::~edited_line): Drop deletion of events.
8782 (edited_line::apply_insert): Delete.
8783 (edited_line::apply_replace): Rename to...
8784 (edited_line::apply_fixit): ...this. Convert to half-open range.
8785 Update for change to type of m_line_events.
8786 * edit-context.h (edit_context::apply_insert): Delete.
8787 (edit_context::apply_replace): Rename to...
8788 (edit_context::apply_fixit): ...this.
8790 2017-05-01 Martin Sebor <msebor@redhat.com>
8792 * gimple-ssa-sprintf.c (format_integer): Set knownrange when it's
8795 2017-05-01 Uros Bizjak <ubizjak@gmail.com>
8798 * config/i386/cpuid.h (__get_cpuid): Always return 0 when
8799 __get_cpuid_max returns 0.
8800 (__get_cpuid_count): Ditto.
8802 2017-05-01 Eric Botcazou <ebotcazou@adacore.com>
8804 * tree.c (substitute_in_expr) <tcc_vl_exp>: Also inline a call if the
8805 replacement expression is another instance of one of its arguments.
8807 2017-05-01 Jakub Jelinek <jakub@redhat.com>
8810 * rtlanal.c (reg_set_p): If reg is a stack_pointer_rtx, also
8811 check for stack push/pop autoinc.
8812 * config/i386/i386.c (ix86_agi_dependent): Return false
8813 if the only reason why modified_in_p returned true is that
8814 addr is SP based and set_insn is a push or pop.
8816 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
8818 * ipa-inline.c (compute_inlined_call_time): Remove now unnecesary
8821 2017-04-29 Jan Hubicka <hubicka@ucw.cz>
8824 * ipa-inline-analysis.c (dump_predicate): Add optional parameter NL.
8825 (account_size_time): Use two predicates - exec_pred and
8827 (evaluate_conditions_for_known_args): Compute both clause and
8829 (evaluate_properties_for_edge): Evaulate both clause and nonspec_clause.
8830 (inline_summary_t::duplicate): Update.
8831 (estimate_function_body_sizes): Caluculate exec and nonconst predicates
8833 (compute_inline_parameters): Likewise.
8834 (estimate_edge_size_and_time): Update caluclation of time.
8835 (estimate_node_size_and_time): Compute both time and nonspecialized
8837 (estimate_ipcp_clone_size_and_time): Update.
8838 (inline_merge_summary): Update.
8839 (do_estimate_edge_time): Update.
8840 (do_estimate_edge_size): Update.
8841 (do_estimate_edge_hints): Update.
8842 (inline_read_section, inline_write_summary): Stream both new predicates.
8843 * ipa-inline.c (compute_uninlined_call_time): Take uninlined_call_time
8845 (compute_inlined_call_time): Cleanup.
8846 (big_speedup_p): Update.
8847 (edge_badness): Update.
8848 * ipa-inline.h (INLINE_TIME_SCALE): Remove.
8849 (size_time_entry): Replace predicate by exec_predicate and
8851 (edge_growth_cache_entry): Cache both time nad nonspecialized time.
8852 (estimate_edge_time): Return also nonspec_time.
8853 (reset_edge_growth_cache): Update.
8855 2017-04-29 Jakub Jelinek <jakub@redhat.com>
8857 PR rtl-optimization/80491
8858 * ifcvt.c (noce_process_if_block): When looking for x setter
8859 with missing else_bb, don't check only the insn right before
8860 cond_earliest, but look for the last insn that x is modified in
8863 PR rtl-optimization/80491
8864 * alias.c (memory_modified_in_insn_p): Return true for CALL_INSNs.
8866 2017-04-29 Marc Glisse <marc.glisse@inria.fr>
8868 PR tree-optimization/80487
8869 * tree-ssa-alias.c (stmt_kills_ref_p): Handle stpncpy and strncpy.
8871 2017-04-29 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
8873 PR tree-optimization/79697
8874 * tree-ssa-dce.c (mark_stmt_if_obviously_necessary): Check if callee
8875 is BUILT_IN_STRDUP, BUILT_IN_STRNDUP, BUILT_IN_REALLOC.
8876 (propagate_necessity): Check if def_callee is BUILT_IN_STRDUP or
8878 * gimple-fold.c (gimple_fold_builtin_realloc): New function.
8879 (gimple_fold_builtin): Call gimple_fold_builtin_realloc.
8881 2017-04-28 Martin Sebor <msebor@redhat.com>
8883 PR tree-optimization/80523
8884 * gimple-ssa-sprintf.c (target_to_host_charmap): New global variable.
8885 (init_target_to_host_charmap, target_to_host, target_strtol10): New
8887 (maybe_warn, format_directive, parse_directive): Use new functions.
8888 (pass_sprintf_length::execute): Call init_target_to_host_charmap.
8890 2017-04-28 Marc Glisse <marc.glisse@inria.fr>
8892 * match.pd (X+Z OP Y+Z, X-Z OP Y-Z, Z-X OP Z-Y): New transformations.
8894 2017-04-28 Bernd Edlinger <bernd.edlinger@hotmail.de>
8896 * configure.ac (SYSTEM_HEADER_DIR, BUILD_SYSTEM_HEADER_DIR,
8897 target_header_dir): Set correctly.
8898 * configure: Regenerated.
8899 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
8900 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
8901 instead of SYSTEM_HEADER_DIR.
8903 2017-04-28 Jan Hubicka <hubicka@ucw.cz>
8905 * ipa-cp.c (perform_estimation_of_a_value): Turn time to sreal.
8906 (estimate_local_effects): Likewise.
8907 * ipa-inline.c (compute_inlined_call_time, want_inline_small_function_p,
8908 edge_badness, inline_small_functions, dump_overall_stats): LIkewise.
8909 * ipa-inline.h (edge_growth_cache_entry, estimate_time_after_inlining,
8910 estimate_ipcp_clone_size_and_time, do_estimate_edge_time,
8911 do_estimate_edge_time, estimate_edge_time): Likewise.
8912 * ipa-inline-analysis.c (estimate_node_size_and_time,
8913 estimate_ipcp_clone_size_and_time, do_estimate_edge_time): Likewise.
8914 (estimate_time_after_inlining): Remove.
8916 2017-04-28 Martin Liska <mliska@suse.cz>
8918 * doc/gcov.texi: Enhance documentation of gcov.
8920 2017-04-28 Martin Liska <mliska@suse.cz>
8922 * doc/gcov.texi: Sort options in alphabetic order.
8923 * doc/gcov-dump.texi: Likewise.
8924 * doc/gcov-tool.texi: Likewise.
8925 * gcov.c (print_usage): Likewise.
8926 * gcov-dump.c (print_usage): Likewise.
8927 * gcov-tool.c (print_merge_usage_message): Likewise.
8928 (print_rewrite_usage_message): Likewise.
8929 (print_overlap_usage_message): Likewise.
8931 2017-04-28 Martin Liska <mliska@suse.cz>
8933 PR gcov-profile/53915
8934 * gcov.c (format_gcov): Print 'NAN %' when top > bottom.
8936 2017-04-28 Martin Liska <mliska@suse.cz>
8938 PR gcov-profile/79891
8939 * gcov.c (add_line_counts): Assign BBs to lines just if the BB
8940 is marked by compiler as living on a line.
8941 (get_cycles_count): Remove usage of the union.
8942 (output_intermediate_file): Likewise.
8943 (find_source): Fix GNU coding style.
8944 (accumulate_line_counts): Remove old non-all block mode.
8945 (output_lines): Remove usage of the union.
8946 * profile.c (output_location): Include all BBs, even if
8947 belonging to a same line (and file) as a previous BB.
8949 2017-04-28 Martin Liska <mliska@suse.cz>
8951 * gcov.c (process_args): Handle new argument 'w'.
8952 (read_graph_file): Assign ID to BBs.
8953 (output_branch_count): Display BB # if verbose flag is set.
8954 (output_lines): Likewise for arcs.
8955 (print_usage): Add '--verbose' option help.
8956 * doc/gcov.texi: Document --verbose (-w) option.
8958 2017-04-28 Martin Liska <mliska@suse.cz>
8960 * gcov.c (struct block_location_info): New struct.
8961 (process_file): Fill up the new structure.
8962 (read_graph_file): Replace usage of encoding by the newly added
8964 (add_line_counts): Likewise.
8965 (accumulate_line_counts): Remove usage of the union.
8966 (function_info::function_info): New function.
8967 (function_info::~function_info): Likewise.
8968 (process_file): Call delete instead of release_function.
8969 (release_function): Release the function.
8970 (release_structures): Call delete instead of release_function.
8971 (solve_flow_graph): Replace usage of num_blocks.
8972 (find_exception_blocks): Likewise.
8973 (output_lines): Fix GNU coding style.
8975 2017-04-28 Martin Liska <mliska@suse.cz>
8978 * coverage.c (coverage_remove_note_file): New function.
8979 * coverage.h: Declare the function.
8980 * toplev.c (finalize): Clean if an error has been seen.
8982 2017-04-28 Martin Liska <mliska@suse.cz>
8984 PR gcov-profile/80031
8985 * gcov-dump.c (tag_blocks): Just print number of basic blocks.
8986 * gcov-io.h (GCOV_TAG_BLOCKS_NUM): Remove unused macro.
8987 * gcov.c (read_graph_file): Read just number of blocks.
8988 * profile.c (branch_prob): Do not stream 0 flags per a basic
8991 2017-04-28 Martin Liska <mliska@suse.cz>
8993 * gcov-dump.c (tag_*): Add new argument to declarations.
8994 (dump_gcov_file): Likewise.
8995 (tag_blocks): Add and use new argument depth.
8996 (tag_arcs): Likewise.
8997 (tag_lines): Likewise.
8998 (tag_counters): Likewise.
8999 (tag_summary): Likewise.
9000 (dump_working_sets): Use depth to do a proper indentation.
9002 2017-04-28 Jakub Jelinek <jakub@redhat.com>
9005 * cgraph.h (symtab_node::debug_symtab): No longer inline.
9006 * symtab.c (symtab_node::debug_symtab): Move definition here.
9008 2017-04-28 Richard Biener <rguenther@suse.de>
9010 * lto-streamer.h (LTO_major_version): Bump to 7.
9012 2017-04-28 Richard Biener <rguenther@suse.de>
9014 * tree-vrp.c (assert_info): New struct.
9015 (add_assert_info): New helper.
9016 (register_edge_assert_for_2): Refactor to add asserts to a vector
9018 (register_edge_assert_for_1): Likewise.
9019 (register_edge_assert_for): Likewise.
9020 (finish_register_edge_assert_for): New helper actually registering
9021 asserts where live on edge.
9022 (find_conditional_asserts): Adjust.
9023 (find_switch_asserts): Likewise.
9024 (evrp_dom_walker::try_find_new_range): Generalize.
9025 (evrp_dom_walker::before_dom_children): Use register_edge_assert_for.
9027 2017-04-27 Marek Polacek <polacek@redhat.com>
9030 * fold-const.c (fold_binary_loc) <case EQ_EXPR, NE_EXPR>: Convert
9031 arg10 and arg11 to itype.
9033 2017-04-27 Jonathan Wakely <jwakely@redhat.com>
9035 * doc/extend.texi (Object Size Checking): Improve grammar.
9037 2017-04-27 Richard Earnshaw <rearnsha@arm.com>
9040 * config/aarch64/aarch64.c (aarch64_emit_approx_sqrt): Ensure
9041 that the logic for permitting reciprocal estimates matches that
9044 2017-04-27 Jakub Jelinek <jakub@redhat.com>
9047 * tree.c (type_cache_hasher::equal): Only compare
9048 TYPE_TYPELESS_STORAGE flag on non-aggregate element types.
9049 (build_array_type_1): Only hash TYPE_TYPELESS_STORAGE flag on
9050 non-aggregate element types.
9051 * tree.h (TYPE_TYPELESS_STORAGE): Fix comment typo, add more details
9052 about the flag on ARRAY_TYPEs in the comment, formatting fix.
9054 2017-04-27 Richard Biener <rguenther@suse.de>
9057 * emit-rtl.c (set_mem_attributes_minus_bitpos): When
9058 stripping ARRAY_REFs from MEM_EXPR make sure we're not
9059 keeping a reference to a trailing array.
9061 2017-04-27 Richard Biener <rguenther@suse.de>
9064 * tree-chrec.c (chrec_fold_plus_poly_poly): Deal with not
9065 being in loop-closed SSA form conservatively.
9066 (chrec_fold_multiply_poly_poly): Likewise.
9068 2017-04-27 Tamar Christina <tamar.christina@arm.com>
9071 * expr.c (expand_expr_real_2): Move TRUNC_MOD_EXPR, FLOOR_MOD_EXPR,
9072 CEIL_MOD_EXPR, ROUND_MOD_EXPR cases.
9074 2017-04-27 Jakub Jelinek <jakub@redhat.com>
9077 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): Remove.
9078 (aarch64_function_arg_alignment): Return unsigned int again, but still
9079 ignore TYPE_FIELDS chain decls other than FIELD_DECLs.
9080 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
9081 Don't emit -Wpsabi note.
9082 (aarch64_function_arg_boundary): Likewise.
9083 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
9086 2017-04-26 Nathan Sidwell <nathan@acm.org>
9088 * tree.h (crc32_unsigned_n): Declare.
9089 (crc32_unsigned, crc32_unsigned): Make inline.
9090 * tree.c (crc32_unsigned_bits): Replace with ...
9091 (crc32_unsigned_n): ... this.
9092 (crc32_unsigned, crc32_byte): Remove.
9093 (crc32_string): Remove unnecessary braces.
9095 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
9097 * ipa-cp.c (estimate_local_effects): Convert sreal to int.
9098 * ipa-inline-analysis.c (MAX_TIME): Remove.
9099 (account_size_time): Use sreal for time.
9100 (dump_inline_summary): Update.
9101 (estimate_function_body_sizes): Update.
9102 (estimate_edge_size_and_time): Update.
9103 (estimate_calls_size_and_time): Update.
9104 (estimate_node_size_and_time): Update.
9105 (inline_merge_summary): Update.
9106 (inline_update_overall_summary): Update.
9107 (estimate_time_after_inlining): Update.
9108 (inline_read_section): Update.
9109 (inline_write_summary): Update.
9110 * ipa-inline.c (compute_uninlined_call_time): Update.
9111 (compute_inlined_call_time): Update.
9112 (recursive_inlining): Update.
9113 (inline_small_functions): Update.
9114 (dump_overall_stats): Update.
9115 * ipa-inline.h: Include sreal.h.
9116 (size_time_entry): Turn time to sreal.
9117 (inline_summary): Turn self_time nad time to sreal.
9119 2017-04-25 Jan Hubicka <hubicka@ucw.cz>
9121 * sreal.c: Include backend.h, tree.h, gimple.h, cgraph.h and
9123 (sreal::stream_out, sreal::stream_in): New.
9124 * sreal.h (sreal::stream_out, sreal::stream_in): Declare.
9126 2017-04-25 Jakub Jelinek <jakub@redhat.com>
9128 * Makefile.in (s-options): Invoke opt-gather.awk with LC_ALL=C in the
9131 2017-04-25 Uros Bizjak <ubizjak@gmail.com>
9134 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p):
9136 (dimode_scalar_chain::compute_convert_gain): Ditto.
9137 (dimode_scalar_chain::make_vector_copies): Ditto.
9138 (dimode_scalar_chain::convert_reg): Ditto.
9139 (dimode_scalar_chain::convert_insn): Ditto.
9140 * config/i386/sse.md (VI24_AVX512BW_1): Remove mode iterator.
9141 (VI248_AVX512BW_1): New mode iterator.
9142 (<mask_codefor>ashr<mode>3<mask_name>): Merge insn pattern with
9143 <mask_codefor>ashrv2di3<mask_name> insn using VI248_AVX512BW_1
9146 2017-04-25 Martin Sebor <msebor@redhat.com>
9148 PR tree-optimization/80497
9149 * gimple-ssa-sprintf.c (get_int_range): Avoid assuming all integer
9150 constants are representable in HOST_WIDE_INT.
9151 (parse_directive): Ditto.
9153 2017-04-25 Martin Sebor <msebor@redhat.com>
9156 * dominance.c (dom_info::m_n_basic_blocks): Change type to unsigned.
9157 (new_zero_array): Adjust signature.
9158 (dom_info::dom_init): Used unsigned rather that size_t.
9159 (dom_info::dom_info): Same.
9161 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9162 Jakub Jelinek <jakub@redhat.com>
9165 * config/arm/arm.c: Include gimple.h.
9166 (aapcs_layout_arg): Emit -Wpsabi note if arm_needs_doubleword_align
9167 returns negative, increment ncrn only if it returned positive.
9168 (arm_needs_doubleword_align): Return int instead of bool,
9169 ignore DECL_ALIGN of non-FIELD_DECL TYPE_FIELDS chain
9170 members, but if there is any such non-FIELD_DECL
9171 > PARM_BOUNDARY aligned decl, return -1 instead of false.
9172 (arm_function_arg): Emit -Wpsabi note if arm_needs_doubleword_align
9173 returns negative, increment nregs only if it returned positive.
9174 (arm_setup_incoming_varargs): Likewise.
9175 (arm_function_arg_boundary): Emit -Wpsabi note if
9176 arm_needs_doubleword_align returns negative, return
9177 DOUBLEWORD_ALIGNMENT only if it returned positive.
9179 2017-04-25 Marek Polacek <polacek@redhat.com>
9182 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
9183 first argument to type.
9185 2017-04-25 Bill Seurer <seurer@linux.vnet.ibm.com>
9188 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Change
9189 type checks to test for compatibility instead of equality.
9191 2017-04-25 Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>
9192 Jakub Jelinek <jakub@redhat.com>
9195 * config/aarch64/aarch64.c (struct aarch64_fn_arg_alignment): New
9197 (aarch64_function_arg_alignment): Return aarch64_fn_arg_alignment
9198 struct. Ignore DECL_ALIGN of decls other than FIELD_DECL for
9199 the alignment computation, but return their maximum in warn_alignment.
9200 (aarch64_layout_arg): Adjust aarch64_function_arg_alignment caller.
9201 Emit a -Wpsabi note if warn_alignment is 16 bytes, but alignment
9203 (aarch64_function_arg_boundary): Likewise. Simplify using MIN/MAX.
9204 (aarch64_gimplify_va_arg_expr): Adjust aarch64_function_arg_alignment
9207 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
9209 * config/arc/simdext.md (dmpyh): Fix typo.
9211 2017-04-25 Richard Biener <rguenther@suse.de>
9213 PR tree-optimization/80492
9214 * alias.c (compare_base_decls): Handle registers with asm
9215 specification conservatively.
9216 * tree-ssa-alias.c (decl_refs_may_alias_p): Handle
9217 compare_base_decls returning dont-know properly.
9219 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
9221 * config/arc/arc.c (LEGITIMATE_OFFSET_ADDRESS_P): Delete macro.
9222 (legitimate_offset_address_p): New function.
9223 (arc_legitimate_address_p): Use above function.
9225 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
9227 * config/arc/arc.c (arc_output_mi_thunk): Emit PIC calls.
9229 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
9231 * config/arc/arc.c (arc_conditional_register_usage): Use ACCL,
9232 ACCH registers whenever they are available.
9234 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
9236 * config/arc/arc.c (arc_conditional_register_usage): Make D0, D1
9237 double regs fix when not used.
9239 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
9241 * config/arc/arc.h (REGNO_OK_FOR_BASE_P): Consider also extension
9243 (REG_OK_FOR_INDEX_P_NONSTRICT): Likewise.
9244 (REG_OK_FOR_BASE_P_NONSTRICT): Likewise.
9246 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
9248 * config/arc/arc.c (arc_output_addsi): Check for h-register class
9249 when emitting short ADD instructions.
9251 2017-04-25 Claudiu Zissulescu <claziss@synopsys.com>
9253 * config/arc/arc.md (cmpsi_cc_insn_mixed): Use 'h' register
9255 (cmpsi_cc_c_insn): Likewise.
9256 (cbranchsi4_scratch): Compute proper instruction length using
9257 compact_hreg_operand.
9258 * config/arc/predicates.md (compact_hreg_operand): New predicate.
9260 2017-04-25 Richard Biener <rguenther@suse.de>
9263 * passes.c (pass_manager::pass_manager): Initialize
9266 2017-04-25 Richard Biener <rguenther@suse.de>
9268 PR tree-optimization/79201
9269 * tree-ssa-sink.c (statement_sink_location): Handle calls.
9271 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9274 * config/s390/vector.md: Split MEM->GPR vector moves for
9275 non-s_operand addresses.
9277 2017-04-25 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
9280 * config/s390/predicates.md (reload_const_wide_int_operand): New
9282 * config/s390/s390.md ("movti"): Remove d/P alternative.
9283 ("movti_bigconst"): New pattern definition.
9285 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
9288 * s390-protos.h (s390_expand_cs_hqi): Removed.
9289 (s390_expand_cs, s390_expand_atomic_exchange_tdsi): New prototypes.
9290 * config/s390/s390.c (s390_emit_compare_and_swap): Handle all integer
9291 modes as well as CCZ1mode and CCZmode.
9292 (s390_expand_atomic_exchange_tdsi, s390_expand_atomic): Adapt to new
9293 signature of s390_emit_compare_and_swap.
9294 (s390_expand_cs_hqi): Likewise, make static.
9295 (s390_expand_cs_tdsi): Generate an explicit compare before trying
9296 compare-and-swap, in some cases.
9297 (s390_expand_cs): Wrapper function.
9298 (s390_expand_atomic_exchange_tdsi): New backend specific expander for
9300 (s390_match_ccmode_set): Allow CCZmode <-> CCZ1 mode.
9301 * config/s390/s390.md ("atomic_compare_and_swap<mode>"): Merge the
9302 patterns for small and large integers. Forbid symref memory operands.
9303 Move expander to s390.c. Require cc register.
9304 ("atomic_compare_and_swap<DGPR:mode><CCZZ1:mode>_internal")
9305 ("*atomic_compare_and_swap<TDI:mode><CCZZ1:mode>_1")
9306 ("*atomic_compare_and_swapdi<CCZZ1:mode>_2")
9307 ("*atomic_compare_and_swapsi<CCZZ1:mode>_3"): Use s_operand to forbid
9308 symref memory operands. Remove CC mode and call s390_match_ccmode
9310 ("atomic_exchange<mode>"): Allow and implement all integer modes.
9312 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
9314 * config/s390/s390.md (define_peephole2): New peephole to help
9315 combining the load-and-test pattern with volatile memory.
9317 2017-04-25 Dominik Vogt <vogt@linux.vnet.ibm.com>
9319 * config/s390/s390.md ("cstorecc4"): Use load-on-condition and deal
9320 with CCZmode for TARGET_Z196.
9322 2017-04-25 Jakub Jelinek <jakub@redhat.com>
9324 PR rtl-optimization/80501
9325 * combine.c (make_compound_operation_int): Set subreg_code to SET
9326 even for AND with mask of the sign bit of mode.
9328 PR rtl-optimization/80500
9329 * loop-unroll.c (combine_var_copies_in_loop_exit): Call copy_rtx on
9330 sum's initial value.
9332 2017-04-25 Julian Brown <julian@codesourcery.com>
9333 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9335 * config/aarch64/thunderx2t99.md (thunderx2t99_crc): New Reservation.
9337 2017-04-25 Marc Glisse <marc.glisse@inria.fr>
9339 * fold-const.c (tree_single_nonzero_warnv_p): Handle SSA_NAME.
9341 2017-04-25 Julian Brown <julian@codesourcery.com>
9342 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9344 * config/aarch64/thunderx2t99.md (thunderx2t99_aes): New Reservation.
9345 (thunderx2t99_sha): New Reservation.
9347 2017-04-25 Julian Brown <julian@codesourcery.com>
9348 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
9350 * config/aarch64/aarch64-simd.md (aarch64_simd_vec_set<mode>): Fix
9351 type for 1-element load.
9353 2017-04-24 Marc Glisse <marc.glisse@inria.fr>
9355 * match.pd (X/[ex]C CMP Y/[ex]C): New transformation.
9357 2017-04-24 Martin Jambor <mjambor@suse.cz>
9359 PR tree-optimization/80293
9360 * tree-sra.c (scalarizable_type_p): New parameter const_decl, make
9361 char arrays not totally scalarizable if it is false.
9362 (analyze_all_variable_accesses): Pass correct value in the new
9363 parameter. Add a statistics counter.
9365 2017-04-24 Jan Hubicka <hubicka@ucw.cz>
9368 * ipa-devirt.c (dump_possible_polymorphic_call_targets): Fix ICE.
9370 2017-04-24 Richard Biener <rguenther@suse.de>
9372 PR tree-optimization/80494
9373 * tree-scalar-evolution.c (analyze_scalar_evolution_1): Bail
9374 out for complex types.
9376 2017-04-24 Richard Biener <rguenther@suse.de>
9378 * tree-ssa-sccvn.h (run_scc_vn): Adjust prototype.
9379 * tree-ssa-sccvn.c (print_scc): Print SCC size.
9380 (extract_and_process_scc_for_name): Never fail but drop SCC to varying.
9381 (DFS): Adjust and never fail.
9382 (sccvn_dom_walker::fail): Remove.
9383 (sccvn_dom_walker::before_dom_children): Adjust.
9384 (run_scc_vn): Likewise and never fail.
9385 * tree-ssa-pre.c (pass_pre::execute): Adjust.
9386 (pass_fre::execute): Likewise.
9388 2017-04-24 Richard Biener <rguenther@suse.de>
9390 PR tree-optimization/79725
9391 * tree-ssa-sink.c (statement_sink_location): Return whether
9392 failure reason was zero uses. Move that check later.
9393 (sink_code_in_bb): Deal with zero uses by removing the stmt
9396 2017-04-24 Richard Biener <rguenther@suse.de>
9399 * tree-ssa-uninit.c (warn_uninitialized_vars): Handle some
9400 pointer-based references.
9402 2017-04-24 Richard Biener <rguenther@suse.de>
9405 * pass_manager.h (pass_manager::operator new): Remove.
9406 (pass_manager::operator delete): Likewise.
9407 * passes.c (pass_manager::operator new): Remove.
9408 (pass_manager::operator delete): Likewise.
9409 (pass_manager::pass_manager): Zero individual pass members.
9411 2017-04-23 Uros Bizjak <ubizjak@gmail.com>
9414 * config/i386/i386.c (dimode_scalar_to_vector_candidate_p)
9415 <case ASHIFT, case LSHIFTRT>: Also consider variable shifts.
9416 Check "XEXP (src, 1)" operand here.
9417 <case PLUS, case MINUS, case IOR, case XOR, case AND>:
9418 Check "XEXP (src, 1)" operand here.
9419 (dimode_scalar_chain::make_vector_copies): Detect count register
9420 of a shift instruction. Zero extend count register from QImode
9421 to DImode to satisfy vector shift pattern count operand predicate.
9422 Substitute vector shift count operand with a DImode copy.
9423 (dimode_scalar_chain::convert_reg): Ditto, zero-extend from
9426 2017-04-21 Uros Bizjak <ubizjak@gmail.com>
9428 * config/i386/i386.md (*extzvqi_mem_rex64): Move above *extzv<mode>.
9429 Remove UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
9430 (*insvqi_1_mem_rex64): Move above insv<mode>_1. Remove
9431 UNSPEC_NOREX_MEM tag. Update corresponding peephole2 pattern.
9432 (UNSPEC_NOREX_MEM): Remove definition.
9434 2017-04-21 Richard Biener <rguenther@suse.de>
9436 PR tree-optimization/79547
9437 * tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
9438 Handle strlen, strcmp, strncmp, strcasecmp, strncasecmp, memcmp,
9439 bcmp, strspn, strcspn, __builtin_object_size and __builtin_constant_p
9440 without any constraints.
9442 2017-04-21 Richard Biener <rguenther@suse.de>
9444 PR tree-optimization/78847
9445 * fold-const.c (split_tree): Handle POINTER_PLUS_EXPR.
9447 2017-04-21 Richard Biener <rguenther@suse.de>
9449 * tree.h (build_qualified_type): Annotate with CXX_MEM_STAT_INFO.
9450 (build_distinct_type_copy): Likewise.
9451 (build_variant_type_copy): Likewise.
9452 * tree.c (build_qualified_type): Pass down mem-stat info.
9453 (build_distinct_type_copy): Likewise.
9454 (build_variant_type_copy): Likewise.
9456 2017-04-21 Richard Biener <rguenther@suse.de>
9458 PR tree-optimization/80237
9459 * tree-ssa-pre.c (find_leader_in_sets): Add third set argument,
9461 (phi_translate_1): Also allow a leader in AVAIL_OUT of pred
9462 for a simplified result.
9464 2016-04-21 Richard Biener <rguenther@suse.de>
9466 * tree-ssa-loop-ivcanon.c (constant_after_peeling): Do not require
9467 sth as strict as a simple_iv but a chrec without symbols and an
9468 operand defined in the loop we are peeling (and not some subloop).
9469 (propagate_constants_for_unrolling): Propagate all constants.
9471 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
9474 * config/i386/i386.c (print_reg): Remove assert for disalowed
9475 regno values, call output_operand_lossage instead.
9477 2017-04-20 Uros Bizjak <ubizjak@gmail.com>
9480 * config/i386/constraints.md (Yc): New register constraint.
9481 * config/i386/i386.md (*float<SWI48:mode><MODEF:mode>2_mixed):
9482 Use Yc constraint for alternative 2 of operand 0. Remove
9483 preferred_for_speed attribute.
9485 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
9487 * omp-low.c (lower_lastprivate_clauses): Correct handling of linear and
9488 lastprivate clauses in SIMT case.
9490 2017-04-20 Volker Reichelt <v.reichelt@netcologne.de>
9492 * doc/invoke.texi (-Wextra-semi): Document new warning option.
9494 2017-04-20 Richard Biener <rguenther@suse.de>
9496 PR tree-optimization/57796
9497 * tree-vect-stmts.c (vect_model_store_cost): Cost scatters
9499 (vect_model_load_cost): Cost gathers as N scalar loads.
9501 2017-04-20 Richard Biener <rguenther@suse.de>
9503 * ggc-page.c (ggc_allocated_p): Rename to ...
9504 (safe_lookup_page_table_entry): ... this and return the lookup
9506 (gt_ggc_m_S): Use safe_lookup_page_table_entry.
9508 2017-04-20 Richard Biener <rguenther@suse.de>
9510 PR tree-optimization/80453
9511 * tree-ssa-sccvn.h (struct vn_phi_s): Add cclhs and ccrhs members.
9512 * tree-ssa-sccvn.c (cond_stmts_equal_p): Use recorded lhs and rhs
9513 from the conditions.
9514 (vn_phi_eq): Pass them down.
9515 (vn_phi_lookup): Record them.
9516 (vn_phi_insert): Likewise.
9518 2017-04-20 Matthew Fortune <matthew.fortune@imgtec.com>
9520 * config/mips/mips.c (mips_expand_vec_perm_const): Re-fix
9521 uninitialized variable warning to avoid buffer overrun.
9523 2017-04-20 Alexander Monakov <amonakov@ispras.ru>
9526 * doc/invoke.texi (-Wmissing-field-initializers): Mention that warning
9527 is suppressed for '{ 0 }' in C.
9529 2017-04-20 Jakub Jelinek <jakub@redhat.com>
9531 * BASE-VER: Set to 8.0.0.
9533 2017-04-20 Thomas Preud'homme <thomas.preudhomme@arm.com>
9535 * config/arm/arm.c (arm_elf_asm_cdtor): Create non-default
9536 priority .init_array and .fini_array section with SECTION_NOTYPE
9539 2017-04-20 Jakub Jelinek <jakub@redhat.com>
9542 * tree.h (build_array_type): Add typeless_storage default argument.
9543 * tree.c (type_cache_hasher::equal): Also compare
9544 TYPE_TYPELESS_STORAGE flag for ARRAY_TYPEs.
9545 (build_array_type): Add typeless_storage argument, set
9546 TYPE_TYPELESS_STORAGE to it, if shared also hash it, and pass to
9548 (build_nonshared_array_type): Adjust build_array_type_1 caller.
9549 (build_array_type): Likewise. Add typeless_storage argument.
9551 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
9552 Jakub Jelinek <jakub@redhat.com>
9554 PR tree-optimization/80426
9555 * tree-vrp.c (extract_range_from_binary_expr_1): For an additive
9556 operation on symbolic operands, also compute the overflow for the
9557 invariant part when the operation degenerates into a negation.
9559 2017-04-19 Jakub Jelinek <jakub@redhat.com>
9562 * dwarf2out.c (modified_type_die, gen_type_die_with_usage):
9563 Check for t with zero TYPE_QUALS_NO_ADDR_SPACE.
9566 * tree-ssa-loop-manip.c (find_uses_to_rename_def): Ignore debug uses.
9568 2017-04-19 Georg-Johann Lay <avr@gjlay.de>
9571 * config/avr/avr.c (tree.h): Include it.
9572 (cgraph.h): Include it.
9573 (avr_encode_section_info): Don't warn for uninitialized progmem
9574 variable if it's just an alias.
9576 2017-04-19 Richard Biener <rguenther@suse.de>
9579 * auto-profile.c (afdo_vpt_for_early_inline): Update SSA
9580 when needed by AutoPGO.
9582 2017-04-19 Paulo J. Matos <paulo@matos-sorge.com>
9585 * doc/lto.texi: Remove an extra 'that'.
9587 2017-04-19 Segher Boessenkool <segher@kernel.crashing.org>
9589 PR rtl-optimization/80429
9590 * ira.c (split_live_ranges_for_shrink_wrap): Don't split regs that
9591 are only used in debug insns.
9593 2017-04-19 Eric Botcazou <ebotcazou@adacore.com>
9594 Vladimir Makarov <vmakarov@redhat.com>
9596 * config/sparc/predicates.md (input_operand): Add comment. Return
9597 true for any memory operand when LRA is in progress.
9598 * config/sparc/sparc.c (sparc_expand_move): Minor formatting fix.
9600 2017-04-18 Jeff Law <law@redhat.com>
9603 * mips.md ({return,simple_return}_internal): Do not overwrite
9606 2017-04-18 Jakub Jelinek <jakub@redhat.com>
9608 PR tree-optimization/80443
9609 * tree-vrp.c (intersect_ranges): For signed 1-bit precision type,
9610 instead of adding 1, subtract -1 and similarly instead of subtracting
9613 2017-04-18 Richard Sandiford <richard.sandiford@arm.com>
9615 PR rtl-optimization/80357
9616 * haifa-sched.c (tmp_bitmap): New variable.
9617 (model_recompute): Handle duplicate use records.
9618 (alloc_global_sched_pressure_data): Initialize tmp_bitmap.
9619 (free_global_sched_pressure_data): Free it.
9621 2017-04-18 Bernd Edlinger <bernd.edlinger@hotmail.de>
9624 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
9625 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
9626 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
9627 instead of SYSTEM_HEADER_DIR.
9629 2017-04-18 Jeff Law <law@redhat.com>
9632 * cfgcleanup.c (try_crossjump_to_edge): Verify SRC1 and SRC2 have
9633 predecessors after walking up the insn chain.
9635 2017-04-18 Jakub Jelinek <jakub@redhat.com>
9638 * dwarf2out.c (modified_type_die): Try harder not to emit internal
9639 sizetype type into debug info.
9641 2017-04-18 Michael Meissner <meissner@linux.vnet.ibm.com>
9644 * config/rs6000/rs6000.c (rs6000_expand_vector_extract): Eliminate
9645 unneeded test for TARGET_UPPER_REGS_SF.
9646 * config/rs6000/vsx.md (vsx_extract_v4sf_var): Likewise.
9648 2017-04-18 Jakub Jelinek <jakub@redhat.com>
9651 * sancov.c (sancov_pass): Use gsi_start_nondebug_after_labels_bb
9652 instead of gsi_after_labels.
9654 2017-04-18 Jeff Law <law@redhat.com>
9656 * regcprop.c (maybe_mode_change): Avoid creating copies of the
9660 2017-04-13 Jeff Law <law@redhat.com>
9661 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
9662 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
9664 2017-04-18 Georg-Johann Lay <avr@gjlay.de>
9667 * config/avr/avr.c (intl.h): Include it.
9668 (avr_pgm_check_var_decl) [reason]: Wrap diagnostic snippets into _().
9670 2017-04-18 Martin Liska <mliska@suse.cz>
9672 PR gcov-profile/78783
9673 * gcov-tool.c (gcov_output_files): Validate that destination
9674 file is either removed by the tool or by a user.
9676 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
9677 Guy Benyei <guybe@mellanox.com>
9679 * config/arc/arc.c (arc_reorg): Move loop_end_id into a more local
9680 block, and do not negate it, the stored id is already negative.
9682 2017-04-14 Andrew Burgess <andrew.burgess@embecosm.com>
9684 * config/arc/arc.md (doloop_begin_i): Use @pcl assembler syntax.
9686 2017-04-14 Michael Meissner <meissner@linux.vnet.ibm.com>
9689 * config/rs6000/rs6000-cpus.def (OTHER_P9_VECTOR_MASKS): Define
9690 masks of options that should be turned off if the VSX vector
9691 options are turned off.
9692 (OTHER_P8_VECTOR_MASKS): Likewise.
9693 (OTHER_VSX_VECTOR_MASKS): Likewise.
9694 * config/rs6000/rs6000.c (rs6000_option_override_internal): Call
9695 rs6000_disable_incompatible_switches to validate no type switches
9697 (rs6000_incompatible_switch): New function to disallow turning on
9698 other vector options if -mno-vsx, -mno-power8-vector, or
9699 -mno-power9-vector are specified.
9701 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
9703 * config/arc/arc.h (CRT_CALL_STATIC_FUNCTION): Use long calls.
9705 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
9707 * config/arc/arc-protos.h (arc_decl_pretend_args): Remove.
9708 * config/arc/arc.c (arc_decl_pretend_args): Likewise.
9709 * config/arc/arc.h (CFA_FRAME_BASE_OFFSET): Likewise.
9710 (ARG_POINTER_CFA_OFFSET): Likewise.
9712 2017-04-14 Claudiu Zissulescu <claziss@synopsys.com>
9714 * config/arc/arc.c (arc_mode_dependent_address_p): Relax
9715 conditions to take advantage of various optimizations.
9717 2017-04-13 Jeff Law <law@redhat.com>
9719 * config/mips.mips.md (zero_extendsidi2): Do not allow SP to appear
9720 in operands[1] if it is a MEM and TARGET_MIPS16 is active.
9721 (zero_extendsidi2_dext): Likewise.
9723 2017-04-13 Jakub Jelinek <jakub@redhat.com>
9726 * fold-const.c (fold_ternary_loc): Revert
9727 use op0 instead of fold_convert_loc (loc, type, arg0) part of
9730 2017-04-13 Vladimir Makarov <vmakarov@redhat.com>
9732 PR rtl-optimization/80343
9733 * lra-remat.c (update_scratch_ops): Assign original hard reg to
9736 2017-04-13 Denis Khalikov <d.khalikov@partner.samsung.com>
9739 * ubsan.c (ubsan_expand_bounds_ifn): Pass original index
9740 to ubsan_encode_value.
9742 2017-04-13 Jeff Law <law@redhat.com>
9744 * reload1.c (eliminate_regs_1): Call gen_rtx_raw_SUBREG for SUBREGs
9745 appearing in DEBUG_INSNs.
9747 2017-04-13 Martin Liska <mliska@suse.cz>
9749 PR gcov-profile/80413
9750 * gcov-io.c (gcov_write_string): Copy to buffer just when
9751 allocated size is greater than zero.
9753 2017-04-13 Jakub Jelinek <jakub@redhat.com>
9756 * dwarf2out.c (decls_for_scope): Ignore declarations of
9757 current_function_decl in BLOCK_NONLOCALIZED_VARS.
9759 2017-04-12 Jan Hubicka <hubicka@ucw.cz>
9762 * ipa-visibility.c (non_local_p): Fix typos.
9763 (localize_node): When localizing symbol in same comdat group,
9764 dissolve the group only when we know external symbols are going
9766 (function_and_variable_visibility): Do not localize DECL_EXTERNAL.
9768 2017-04-12 Jakub Jelinek <jakub@redhat.com>
9770 PR tree-optimization/79390
9771 * optabs.c (emit_conditional_move): If the preferred op2/op3 operand
9772 order does not result in usable sequence, retry with reversed operand
9778 * fold-const.c (fold_ternary_loc): Use op1 instead of arg1 as argument
9779 to fold_build2_loc. Convert TREE_OPERAND (tem, 0) to type. Use
9780 op0 instead of fold_convert_loc (loc, type, arg0).
9782 2017-04-12 Jeff Law <law@redhat.com>
9784 * genattrtab.c (write_eligible_delay): Verify DELAY_INSN still
9785 has a delay slot in the generated code.
9787 * config/cris/cris.md (cris_preferred_reload_class): Return
9788 GENNONACR_REGS rather than GENERAL_REGS.
9790 2017-04-12 Jakub Jelinek <jakub@redhat.com>
9793 * expr.c <CASE_CONVERT>: For EXPAND_INITIALIZER determine SIGN_EXTEND
9794 vs. ZERO_EXTEND based on signedness of treeop0's type rather than
9795 signedness of the result type.
9797 2017-04-12 Richard Biener <rguenther@suse.de>
9798 Jeff Law <law@redhat.com>
9800 PR tree-optimization/80359
9801 * tree-ssa-dse.c (maybe_trim_partially_dead_store): Do not
9802 trim stores to TARGET_MEM_REFs.
9804 2017-04-12 Richard Biener <rguenther@suse.de>
9806 PR tree-optimization/79390
9807 * gimple-ssa-split-paths.c (is_feasible_trace): Restrict
9808 threading case even more.
9810 2017-04-12 Segher Boessenkool <segher@kernel.crashing.org>
9813 * config/rs6000/sync.md (atomic_load<mode>, atomic_store<mode): Test
9814 for quad_address_p for TImode, instead of just not indexed_address.
9816 2017-04-12 Richard Biener <rguenther@suse.de>
9817 Bernd Edlinger <bernd.edlinger@hotmail.de>
9820 * alias.c (component_uses_parent_alias_set_from): Handle
9821 TYPE_TYPELESS_STORAGE.
9822 (get_alias_set): Likewise.
9823 * tree-core.h (tree_type_common): Add typeless_storage flag.
9824 * tree.h (TYPE_TYPELESS_STORAGE): New macro.
9825 * stor-layout.c (place_union_field): Set TYPE_TYPELESS_STORAGE
9826 for types containing members with TYPE_TYPELESS_STORAGE.
9827 (place_field): Likewise.
9828 (layout_type): Likewise for ARRAY_TYPE.
9829 * lto-streamer-out.c (hash_tree): Hash TYPE_TYPELESS_STORAGE.
9830 * tree-streamer-in.c (unpack_ts_type_common_value_fields): Stream
9831 TYPE_TYPELESS_STORAGE.
9832 * tree-streamer-out.c (pack_ts_type_common_value_fields): Likewise.
9834 2017-04-12 Jakub Jelinek <jakub@redhat.com>
9837 * fold-const.c (fold_binary_loc) <case BIT_IOR_EXPR>: Convert arg0's
9838 first argument to type.
9840 2017-04-11 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
9844 * config/rs6000/rs6000.c (rs6000_expand_unop_builtin): Return
9845 CONST0_RTX (mode) rather than const0_rtx where appropriate.
9846 (rs6000_expand_binop_builtin): Likewise.
9847 (rs6000_expand_ternop_builtin): Likewise; also add missing
9848 vsx_xxpermdi_* variants; also fix typo (arg1 => arg2) for
9849 vshasigma built-ins.
9850 * doc/extend.texi: Document that vec_xxpermdi's third argument
9853 2017-04-11 Uros Bizjak <ubizjak@gmail.com>
9855 * config/i386/i386.c (dimode_scalar_chain::compute_convert_gain):
9856 Use shift_const cost parameter when calculating gain of STV shifts.
9858 2017-04-11 Vladimir Makarov <vmakarov@redhat.com>
9860 PR rtl-optimization/70478
9861 * lra-constraints.c (process_alt_operands): Check memory for
9862 disfavoring memory insn operand.
9864 2017-04-11 Jakub Jelinek <jakub@redhat.com>
9867 * simplify-rtx.c (simplify_binary_operation_1) <case IOR>: Perform
9868 left shift in unsigned HOST_WIDE_INT type.
9870 PR rtl-optimization/80385
9871 * simplify-rtx.c (simplify_unary_operation_1): Don't transform
9872 (not (neg X)) into (plus X -1) for complex or non-integral modes.
9875 * omp-low.c (scan_omp_task): Don't optimize away empty tasks
9876 if they have any depend clauses.
9878 2017-04-11 Martin Liska <mliska@suse.cz>
9881 * cgraph.c (cgraph_node::dump): Dump calls_comdat_local.
9882 * ipa-split.c (split_function): Create a local comdat symbol
9883 if caller is in a comdat group.
9885 2017-04-11 Martin Liska <mliska@suse.cz>
9888 * ipa-cp.c (determine_versionability): Handle calls_comdat_local
9891 2017-04-11 Martin Sebor <msebor@redhat.com>
9894 * gimple-ssa-sprintf.c (get_int_range): Remove second argument and
9895 always use the int type. Use INTEGRAL_TYPE_P() rather than testing
9897 (directive::set_width, directive::set_precision, format_character):
9899 (parse_directive): Use INTEGRAL_TYPE_P() rather than testing for
9902 2017-04-11 Richard Earnshaw <rearnsha@arm.com>
9905 * config/arm/arm.c (arm_configure_build_target): When -mcpu and -arch
9906 conflict, set target->arch_name instead of target->cpu_name.
9908 2017-04-11 Richard Biener <rguenther@suse.de>
9910 PR tree-optimization/80374
9911 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Use
9912 build_zero_cst, remove fold_convertible_p check again.
9914 2017-04-11 Martin Liska <mliska@suse.cz>
9917 * ubsan.c (instrument_object_size): Do not instrument register
9920 2017-04-11 Jakub Jelinek <jakub@redhat.com>
9923 * config/i386/i386-builtin-types.def
9924 (V16HI_FTYPE_V16HI_INT_V16HI_UHI_COUNT,
9925 V16HI_FTYPE_V16HI_V8HI_V16HI_UHI_COUNT,
9926 V16SI_FTYPE_V16SI_INT_V16SI_UHI_COUNT,
9927 V16SI_FTYPE_V16SI_V4SI_V16SI_UHI_COUNT,
9928 V2DI_FTYPE_V2DI_INT_V2DI_UQI_COUNT,
9929 V2DI_FTYPE_V2DI_V2DI_V2DI_UQI_COUNT,
9930 V32HI_FTYPE_V32HI_INT_V32HI_USI_COUNT,
9931 V32HI_FTYPE_V32HI_V8HI_V32HI_USI_COUNT,
9932 V4DI_FTYPE_V4DI_INT_V4DI_UQI_COUNT,
9933 V4DI_FTYPE_V4DI_V2DI_V4DI_UQI_COUNT,
9934 V4SI_FTYPE_V4SI_INT_V4SI_UQI_COUNT,
9935 V4SI_FTYPE_V4SI_V4SI_V4SI_UQI_COUNT,
9936 V8DI_FTYPE_V8DI_INT_V8DI_UQI_COUNT,
9937 V8DI_FTYPE_V8DI_V2DI_V8DI_UQI_COUNT,
9938 V8HI_FTYPE_V8HI_INT_V8HI_UQI_COUNT,
9939 V8HI_FTYPE_V8HI_V8HI_V8HI_UQI_COUNT,
9940 V8SI_FTYPE_V8SI_INT_V8SI_UQI_COUNT,
9941 V8SI_FTYPE_V8SI_V4SI_V8SI_UQI_COUNT): New function type aliases.
9942 * config/i386/i386-builtin.def (__builtin_ia32_pslld512_mask,
9943 __builtin_ia32_pslldi512_mask, __builtin_ia32_psllq512_mask,
9944 __builtin_ia32_psllqi512_mask, __builtin_ia32_psrad512_mask,
9945 __builtin_ia32_psradi512_mask, __builtin_ia32_psraq512_mask,
9946 __builtin_ia32_psraqi512_mask, __builtin_ia32_psrld512_mask,
9947 __builtin_ia32_psrldi512_mask, __builtin_ia32_psrlq512_mask,
9948 __builtin_ia32_psrlqi512_mask, __builtin_ia32_psllwi128_mask,
9949 __builtin_ia32_pslldi128_mask, __builtin_ia32_psllqi128_mask,
9950 __builtin_ia32_psllw128_mask, __builtin_ia32_pslld128_mask,
9951 __builtin_ia32_psllq128_mask, __builtin_ia32_psllwi256_mask,
9952 __builtin_ia32_psllw256_mask, __builtin_ia32_pslldi256_mask,
9953 __builtin_ia32_pslld256_mask, __builtin_ia32_psllqi256_mask,
9954 __builtin_ia32_psllq256_mask, __builtin_ia32_psradi128_mask,
9955 __builtin_ia32_psrad128_mask, __builtin_ia32_psradi256_mask,
9956 __builtin_ia32_psrad256_mask, __builtin_ia32_psraqi128_mask,
9957 __builtin_ia32_psraq128_mask, __builtin_ia32_psraqi256_mask,
9958 __builtin_ia32_psraq256_mask, __builtin_ia32_psrldi128_mask,
9959 __builtin_ia32_psrld128_mask, __builtin_ia32_psrldi256_mask,
9960 __builtin_ia32_psrld256_mask, __builtin_ia32_psrlqi128_mask,
9961 __builtin_ia32_psrlq128_mask, __builtin_ia32_psrlqi256_mask,
9962 __builtin_ia32_psrlq256_mask, __builtin_ia32_psrawi256_mask,
9963 __builtin_ia32_psraw256_mask, __builtin_ia32_psrawi128_mask,
9964 __builtin_ia32_psraw128_mask, __builtin_ia32_psrlwi256_mask,
9965 __builtin_ia32_psrlw256_mask, __builtin_ia32_psrlwi128_mask,
9966 __builtin_ia32_psrlw128_mask, __builtin_ia32_psllwi512_mask,
9967 __builtin_ia32_psllw512_mask, __builtin_ia32_psrawi512_mask,
9968 __builtin_ia32_psraw512_mask, __builtin_ia32_psrlwi512_mask,
9969 __builtin_ia32_psrlw512_mask): Use _COUNT suffixed function type
9971 * config/i386/i386.c (ix86_expand_args_builtin): Rename last_arg_count
9972 flag to second_arg_count, handle 4 argument function type _COUNT
9973 aliases, handle second_arg_count on second argument rather than last.
9975 2017-04-10 Jeff Law <law@redhat.com>
9977 PR tree-optimization/80374
9978 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Do not try to
9979 record anything if we can not convert integer_zero_node to the
9982 2017-04-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
9985 * config/rs6000/rs6000.c (rs6000_option_override_internal):
9986 Enhance special handling given to the TARGET_P9_MINMAX option in
9987 relation to certain other options.
9989 2017-04-10 Bin Cheng <bin.cheng@arm.com>
9991 PR tree-optimization/80153
9992 * tree-ssa-loop-ivopts.c (add_iv_candidate_for_use): Check and
9993 remove POINTER_PLUS_EXPR's base part directly, rather than through
9996 2017-04-10 Richard Biener <rguenther@suse.de>
9997 Bin Cheng <bin.cheng@arm.com>
9999 PR tree-optimization/80153
10000 * tree-affine.c (aff_combination_to_tree): Get base pointer from
10001 the first element of pointer type aff_tree. Build result expr in
10003 (add_elt_to_tree): Convert to type unconditionally. Remove other
10004 fold_convert calls.
10005 * tree-ssa-loop-ivopts.c (alloc_iv): Pass in consistent types.
10006 (rewrite_use_nonlinear_expr): Check invariant using iv information.
10008 2017-04-10 Richard Biener <rguenther@suse.de>
10010 * tree-ssa-structalias.c (find_func_aliases): Properly handle
10013 2017-04-10 Vladimir Makarov <vmakarov@redhat.com>
10015 PR rtl-optimization/70478
10016 * lra-constraints.c (curr_small_class_check): New.
10017 (update_and_check_small_class_inputs): New.
10018 (process_alt_operands): Update curr_small_class_check. Disfavor
10019 alternative insn memory operands. Check available regs for small
10022 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
10025 * config/mips/mips.opt (-mvirt): Update description.
10026 * doc/invoke.texi (-mvirt): Likewise.
10028 2017-04-10 Richard Biener <rguenther@suse.de>
10030 PR middle-end/80362
10031 * fold-const.c (fold_binary_loc): Look at unstripped ops when
10032 looking for NEGATE_EXPR in -A / -B to A / B folding.
10034 2017-04-10 Martin Liska <mliska@suse.cz>
10036 PR gcov-profile/80224
10037 * gcov.c (print_usage): Fix usage string.
10038 (get_gcov_intermediate_filename): Remove.
10039 (output_gcov_file): Use both for normal and intermediate format.
10040 (generate_results): Do not initialize special file for
10041 intermediate format.
10043 2017-04-10 Richard Biener <rguenther@suse.de>
10045 PR tree-optimization/80304
10046 * tree-ssa-loop-im.c (ref_indep_loop_p_1): Also recurse
10049 2017-04-10 Nathan Sidwell <nathan@acm.org>
10052 * config/rs6000/rs6000.c (rs6000_vector_type): New.
10053 (rs6000_init_builtins): Use it.
10055 2016-04-10 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
10057 * config/arm/arm.md (<mrc>): Add mode to SET source.
10058 (<mrrc>): Likewise.
10060 2017-04-10 Richard Biener <rguenther@suse.de>
10062 PR middle-end/80344
10063 * gimplify.c (is_gimple_mem_rhs_or_call): Allow CLOBBERs.
10065 2017-04-10 Jakub Jelinek <jakub@redhat.com>
10068 * config/i386/avx512fintrin.h (_mm512_reduce_add_epi32,
10069 _mm512_reduce_mul_epi32, _mm512_reduce_and_epi32,
10070 _mm512_reduce_or_epi32, _mm512_mask_reduce_add_epi32,
10071 _mm512_mask_reduce_mul_epi32, _mm512_mask_reduce_and_epi32,
10072 _mm512_mask_reduce_or_epi32, _mm512_reduce_min_epi32,
10073 _mm512_reduce_max_epi32, _mm512_reduce_min_epu32,
10074 _mm512_reduce_max_epu32, _mm512_mask_reduce_min_epi32,
10075 _mm512_mask_reduce_max_epi32, _mm512_mask_reduce_min_epu32,
10076 _mm512_mask_reduce_max_epu32, _mm512_reduce_add_ps,
10077 _mm512_reduce_mul_ps, _mm512_mask_reduce_add_ps,
10078 _mm512_mask_reduce_mul_ps, _mm512_reduce_min_ps, _mm512_reduce_max_ps,
10079 _mm512_mask_reduce_min_ps, _mm512_mask_reduce_max_ps,
10080 _mm512_reduce_add_epi64, _mm512_reduce_mul_epi64,
10081 _mm512_reduce_and_epi64, _mm512_reduce_or_epi64,
10082 _mm512_mask_reduce_add_epi64, _mm512_mask_reduce_mul_epi64,
10083 _mm512_mask_reduce_and_epi64, _mm512_mask_reduce_or_epi64,
10084 _mm512_reduce_min_epi64, _mm512_reduce_max_epi64,
10085 _mm512_mask_reduce_min_epi64, _mm512_mask_reduce_max_epi64,
10086 _mm512_reduce_min_epu64, _mm512_reduce_max_epu64,
10087 _mm512_mask_reduce_min_epu64, _mm512_mask_reduce_max_epu64,
10088 _mm512_reduce_add_pd, _mm512_reduce_mul_pd, _mm512_mask_reduce_add_pd,
10089 _mm512_mask_reduce_mul_pd, _mm512_reduce_min_pd, _mm512_reduce_max_pd,
10090 _mm512_mask_reduce_min_pd, _mm512_mask_reduce_max_pd): New intrinsics.
10092 2017-04-08 Vladimir Makarov <vmakarov@redhat.com>
10094 PR rtl-optimization/70478
10095 * lra-constraints.c: Reverse the last patch.
10097 2017-04-08 Andreas Tobler <andreast@gcc.gnu.org>
10099 * config/aarch64/aarch64-freebsd.h: Define MCOUNT_NAME.
10100 Add comment for WCHAR_T.
10102 2017-04-08 Martin Liska <mliska@suse.cz>
10105 2017-04-07 Martin Liska <mliska@suse.cz>
10108 * ipa-split.c (split_function): Add function part to a same comdat
10111 2017-04-08 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
10114 * config/rs6000/rs6000.c (expand_block_compare): Fix boundary check.
10116 2017-04-07 Pat Haugen <pthaugen@us.ibm.com>
10118 * rs6000/rs6000.c (vec_load_pendulum): Rename...
10119 (vec_pairing): ...to this.
10120 (power9_sched_reorder2): Rewrite code for pairing vector/vecload insns.
10121 (rs6000_sched_init): Adjust for name change.
10122 (struct rs6000_sched_context): Likewise.
10123 (rs6000_init_sched_context): Likewise.
10124 (rs6000_set_sched_context): Likewise.
10126 2017-04-07 Jakub Jelinek <jakub@redhat.com>
10132 * config/i386/avxintrin.h (_mm256_cvtsd_f64, _mm256_cvtss_f32): New
10134 * config/i386/avx512fintrin.h (_mm512_int2mask, _mm512_mask2int,
10135 _mm512_abs_ps, _mm512_mask_abs_ps, _mm512_abs_pd, _mm512_mask_abs_pd,
10136 _mm512_cvtsd_f64, _mm512_cvtss_f32): Likewise.
10138 2017-04-07 Andreas Tobler <andreast@gcc.gnu.org>
10140 * config/aarch64/aarch64-freebsd.h: Define WCHAR_TYPE.
10142 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
10144 PR rtl-optimization/70703
10145 * ira-color.c (update_conflict_hard_regno_costs): Use
10146 int64_t instead of HOST_WIDE_INT.
10148 2017-04-07 Vladimir Makarov <vmakarov@redhat.com>
10150 PR rtl-optimization/70478
10151 * lra-constraints.c (process_alt_operands): Disfavor alternative
10152 insn memory operands.
10154 2017-04-07 Jeff Law <law@redhat.com>
10156 * config/iq2000/iq2000.c (final_prescan_insn): Do not separate a
10157 CALL and NOTE_INSN_CALL_ARG_LOCATION.
10159 2017-04-07 Martin Liska <mliska@suse.cz>
10162 * config/aarch64/aarch64.c (aarch64_process_target_attr):
10163 Show error message instead of an ICE.
10165 2017-04-07 Martin Liska <mliska@suse.cz>
10168 * ipa-split.c (split_function): Add function part to a same comdat
10171 2017-04-07 Richard Biener <rguenther@suse.de>
10173 PR middle-end/80341
10174 * tree.c (get_unwidened): Also handle ! for_type case for
10176 * convert.c (do_narrow): Split out from ...
10177 (convert_to_integer_1): ... here. Do not pass final truncation
10178 type to get_unwidened for TRUNC_DIV_EXPR.
10180 2017-04-07 Richard Biener <rguenther@suse.de>
10182 * tree-affine.c (wide_int_ext_for_comb): Take type rather
10184 (aff_combination_const): Adjust.
10185 (aff_combination_scale): Likewise.
10186 (aff_combination_add_elt): Likewise.
10187 (aff_combination_add_cst): Likewise.
10188 (aff_combination_convert): Likewise.
10189 (add_elt_to_tree): Likewise. Remove unused argument.
10190 (aff_combination_to_tree): Adjust calls to add_elt_to_tree.
10192 2017-04-07 Sebastian Huber <sebastian.huber@embedded-brains.de>
10194 * config/arm/arm.h (ARM_DEFAULT_SHORT_ENUMS): Provide default
10196 * config/arm/arm.c (arm_default_short_enums): Use
10197 ARM_DEFAULT_SHORT_ENUMS.
10198 * config/arm/rtems.h (ARM_DEFAULT_SHORT_ENUMS): Define.
10200 2017-04-06 Jakub Jelinek <jakub@redhat.com>
10203 * dwarf2out.c (gen_member_die): Handle C++17 inline static data
10204 members with redundant out-of-class redeclaration.
10206 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
10209 * config/i386/sse.md (*vec_extractv4si_0_zext_sse4): New pattern.
10210 * config/i386/i386.md (*zero_extendsidi2):
10211 Add (?*x,*x) and (?*v,*v) alternatives.
10213 2017-04-06 Uros Bizjak <ubizjak@gmail.com>
10216 * config/i386/i386.c (ix86_expand_builtin)
10217 <case IX86_BUILTIN_K{,OR}TEST{C,Z}{8,16,32,64}>: Determine insn operand
10218 mode from insn data. Convert operands to insn operand mode.
10219 Copy operands that don't satisfy insn predicate to a register.
10221 2017-04-06 Sam Thursfield <sam.thursfield@codethink.co.uk>
10223 * config/rs6000/x-aix: Increase memory limit for genautomata on AIX.
10226 2017-04-06 Richard Biener <rguenther@suse.de>
10228 PR tree-optimization/80334
10229 * tree-ssa-loop-ivopts.c (rewrite_use_address): Properly
10230 preserve alignment of accesses.
10232 2017-04-06 Richard Biener <rguenther@suse.de>
10234 PR tree-optimization/80262
10235 * tree-sra.c (build_ref_for_offset): Preserve address-space
10237 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address):
10238 Drop useless address-space information on MEM_REF offsets.
10240 2017-04-05 Andreas Schwab <schwab@linux-m68k.org>
10242 * builtins.def (BUILT_IN_UPDATE_SETJMP_BUF): Fix type.
10244 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
10246 PR rtl-optimization/70703
10247 * ira-color.c (update_conflict_hard_regno_costs): Use
10248 HOST_WIDE_INT instead of long.
10250 2017-04-05 Uros Bizjak <ubizjak@gmail.com>
10253 * config/i386/mmintrin.h: Add -msse target option when __SSE__ is
10254 not defined for x86_64 target. Add -mmmx target option when __SSE2__
10256 * config/i386/mm3dnow.h: Add -msse target when __SSE__ is not defined
10257 for x86_64 target. Handle -m3dnowa option.
10259 2017-04-05 Vladimir Makarov <vmakarov@redhat.com>
10261 PR rtl-optimization/70703
10262 * ira-color.c (update_costs_from_allocno): Use the smallest mode.
10263 (update_conflict_hard_regno_costs): Use long instead of unsigned
10264 arithmetic for cost calculation.
10266 2017-04-05 Jakub Jelinek <jakub@redhat.com>
10267 Bernd Edlinger <bernd.edlinger@hotmail.de>
10270 * asan.c (asan_store_shadow_bytes): Fix location of last_chunk_value
10273 2017-04-05 Eric Botcazou <ebotcazou@adacore.com>
10276 * config/aarch64/aarch64.c (aarch64_emit_probe_stack_range): Replace
10277 ptr_mode with Pmode throughout.
10278 * config/aarch64/aarch64.md (probe_stack_range_<PTR:mode): Rename
10279 into probe_stack_range and use DImode.
10281 2017-04-05 Dominik Vogt <vogt@linux.vnet.ibm.com>
10284 * config/s390/s390.c (s390_register_info_gprtofpr): Return if
10285 call_eh_return is true.
10287 2017-04-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10289 * config/s390/s390-c.c (s390_resolve_overloaded_builtin):
10290 Initialize last_match_fntype_index.
10292 2017-04-05 Jakub Jelinek <jakub@redhat.com>
10295 * tree-nvr.c: Include internal-fn.h.
10296 (pass_return_slot::execute): Ignore internal calls without
10299 2017-04-04 Jakub Jelinek <jakub@redhat.com>
10300 Richard Biener <rguenther@suse.de>
10303 * genmatch.c (capture::gen_transform): For GENERIC unshare_expr
10304 captures used multiple times, except for the last use.
10305 * generic-match-head.c: Include gimplify.h.
10307 2017-04-04 Jakub Jelinek <jakub@redhat.com>
10309 PR tree-optimization/79390
10310 * target.h (struct noce_if_info): Declare.
10311 * targhooks.h (default_noce_conversion_profitable_p): Declare.
10312 * target.def (noce_conversion_profitable_p): New target hook.
10313 * ifcvt.h (struct noce_if_info): New type, moved from ...
10314 * ifcvt.c (struct noce_if_info): ... here.
10315 (noce_conversion_profitable_p): Renamed to ...
10316 (default_noce_conversion_profitable_p): ... this. No longer
10318 (noce_try_store_flag_constants, noce_try_addcc,
10319 noce_try_store_flag_mask, noce_try_cmove, noce_try_cmove_arith,
10320 noce_convert_multiple_sets): Use targetm.noce_conversion_profitable_p
10321 instead of noce_conversion_profitable_p.
10322 * config/i386/i386.c: Include ifcvt.h.
10323 (ix86_option_override_internal): Don't override
10324 PARAM_MAX_RTL_IF_CONVERSION_INSNS default.
10325 (ix86_noce_conversion_profitable_p): New function.
10326 (TARGET_NOCE_CONVERSION_PROFITABLE_P): Redefine.
10327 * config/i386/x86-tune.def (X86_TUNE_ONE_IF_CONV_INSN): Adjust comment.
10328 * doc/tm.texi.in (TARGET_NOCE_CONVERSION_PROFITABLE_P): Add.
10329 * doc/tm.texi: Regenerated.
10331 2017-04-04 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10333 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Grammar
10336 2017-04-04 Thomas Preud'homme <thomas.preudhomme@arm.com>
10339 * config/arm/arm.c (thumb1_rtx_costs): Give a cost of 32
10340 instructions for small multiply cores.
10342 2017-04-04 Jeff Law <law@redhat.com>
10344 * config/mips/mips.c (mips_multi_add): Zero initialize the newly
10346 (mips_expand_vec_perm_const): Initialize elements in orig_perm
10347 that are not set by the loop over the elements.
10349 2017-04-04 Jakub Jelinek <jakub@redhat.com>
10352 * config/i386/i386.c (ix86_expand_args_builtin): If op has scalar
10353 int mode, convert_modes it to mode as unsigned, otherwise use
10354 lowpart_subreg to mode rather than SImode.
10355 * config/i386/sse.md (<mask_codefor>ashr<mode>3<mask_name>,
10356 ashr<mode>3, ashr<mode>3<mask_name>, <shift_insn><mode>3<mask_name>):
10357 Use DImode instead of SImode for the shift count operand.
10358 * config/i386/mmx.md (mmx_ashr<mode>3, mmx_<shift_insn><mode>3):
10361 2017-04-04 Richard Biener <rguenther@suse.de>
10363 PR middle-end/80281
10364 * match.pd (A + (-B) -> A - B): Make sure to preserve unsigned
10365 arithmetic done for the negate or the plus. Simplify.
10366 (A - (-B) -> A + B): Likewise.
10367 * fold-const.c (split_tree): Make sure to not negate pointers.
10369 2017-04-04 Segher Boessenkool <segher@kernel.crashing.org>
10371 PR rtl-optimization/60818
10372 * simplify-rtx.c (simplify_binary_operation_1): Do not replace
10373 a compare of comparisons with the thing compared if this results
10374 in a different machine mode.
10376 2017-04-03 Jonathan Wakely <jwakely@redhat.com>
10378 * alias.c (base_alias_check): Fix typo in comment.
10379 * cgraph.h (class ipa_polymorphic_call_context): Likewise.
10380 * cgraphunit.c (symbol_table::compile): Likewise.
10381 * collect2.c (maybe_run_lto_and_relink): Likewise.
10382 * config/arm/arm.c (arm_thumb1_mi_thunk): Likewise.
10383 * config/avr/avr-arch.h (avr_arch_info_t): Likewise.
10384 * config/avr/avr.c (avr_map_op_t): Likewise.
10385 * config/cr16/cr16.h (DATA_ALIGNMENT): Likewise.
10386 * config/epiphany/epiphany.c (TARGET_ARG_PARTIAL_BYTES): Likewise.
10387 * config/epiphany/epiphany.md (movcc): Likewise.
10388 * config/i386/i386.c (legitimize_pe_coff_extern_decl): Likewise.
10389 * config/m68k/m68k.c (struct _sched_ib, m68k_sched_variable_issue):
10391 * config/mips/mips.c (mips_save_restore_reg): Likewise.
10392 * config/rx/rx.c (rx_is_restricted_memory_address): Likewise.
10393 * config/s390/s390.c (Z10_EARLYLOAD_DISTANCE): Likewise.
10394 * config/sh/sh.c (sh_rtx_costs): Likewise.
10395 * fold-const.c (fold_truth_andor): Likewise.
10396 * genautomata.c (collapse_flag): Likewise.
10397 * gengtype.h (struct type::u::s): Likewise.
10398 * gensupport.c (has_subst_attribute, add_mnemonic_string): Likewise.
10399 * input.c (FORMAT_AMOUNT): Likewise.
10400 * ipa-cp.c (class ipcp_lattice, agg_replacements_to_vector)
10401 (known_aggs_to_agg_replacement_list): Likewise.
10402 * ipa-inline-analysis.c: Likewise.
10403 * ipa-inline.h (estimate_edge_time, estimate_edge_hints): Likewise.
10404 * ipa-polymorphic-call.c
10405 (ipa_polymorphic_call_context::restrict_to_inner_class): Likewise.
10406 * loop-unroll.c (analyze_insn_to_expand_var): Likewise.
10407 * lra.c (lra_optional_reload_pseudos, lra_subreg_reload_pseudos):
10409 * modulo-sched.c (apply_reg_moves): Likewise.
10410 * omp-expand.c (build_omp_regions_1): Likewise.
10411 * trans-mem.c (struct tm_wrapper_hasher): Likewise.
10412 * tree-ssa-loop-ivopts.c (may_eliminate_iv): Likewise.
10413 * tree-ssa-loop-niter.c (maybe_lower_iteration_bound): Likewise.
10414 * tree-vect-data-refs.c (vect_enhance_data_refs_alignment): Likewise.
10415 * value-prof.c: Likewise.
10416 * var-tracking.c (val_reset): Likewise.
10418 2017-04-03 Richard Biener <rguenther@suse.de>
10420 PR tree-optimization/80275
10421 * fold-const.c (split_address_to_core_and_offset): Handle
10424 2017-04-03 Eric Botcazou <ebotcazou@adacore.com>
10426 * tree-nested.c (get_descriptor_type): Make sure that the alignment of
10427 descriptors is at least equal to that of functions.
10429 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
10431 * config/i386/sse.md (movdi_to_sse): Add missing DONE.
10433 2017-04-02 Uros Bizjak <ubizjak@gmail.com>
10436 * config/i386/sse.md (mov<IMOD4:mode>): Remove insn pattern.
10437 (mov<IMOD4:mode>): New expander.
10438 (*mov<IMOD4:mode>_internal): New insn and split pattern.
10440 2017-03-31 Segher Boessenkool <segher@kernel.crashing.org>
10442 PR rtl-optimization/79405
10443 * fwprop.c (propagations_left): New variable.
10444 (forward_propagate_into): Decrement it.
10445 (fwprop_init): Initialize it.
10446 (fw_prop): If the variable has reached zero, stop propagating.
10447 (fwprop_addr): Ditto.
10449 2017-03-31 Jakub Jelinek <jakub@redhat.com>
10452 * dwarf2out.c (decls_for_scope): If BLOCK_NONLOCALIZED_VAR is
10453 a FUNCTION_DECL, pass it as decl instead of origin to
10456 2017-03-31 Alexander Monakov <amonakov@ispras.ru>
10458 * config/nvptx/nvptx.c (nvptx_output_softstack_switch): Correct format
10461 2017-03-31 Pat Haugen <pthaugen@us.ibm.com>
10464 * config/rs6000/rs6000.md (extendhi<mode>2): Add test for
10465 TARGET_VSX_SMALL_INTEGER.
10467 2017-03-31 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10469 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Add
10470 reference to the OpenPOWER 64-Bit ELF V2 ABI Specification.
10472 2017-03-31 Matthew Fortune <matthew.fortune@imgtec.com>
10474 * config/mips/mips-msa.md (msa_vec_extract_<msafmt_f>): Update
10475 extraction from odd-numbered MSA register.
10477 2017-03-31 Jakub Jelinek <jakub@redhat.com>
10479 PR middle-end/80173
10480 * expmed.c (store_bit_field_1): Don't attempt to create
10481 a word subreg out of hard registers wider than word if they
10482 have HARD_REGNO_NREGS of 1 for their mode.
10484 PR middle-end/80163
10485 * varasm.c (initializer_constant_valid_p_1): Disallow sign-extending
10486 conversions to integer types wider than word and pointer.
10489 * cselib.h (rtx_equal_for_cselib_1): Add depth argument.
10490 (rtx_equal_for_cselib_p): Pass 0 to it.
10491 * cselib.c (cselib_hasher::equal): Likewise.
10492 (rtx_equal_for_cselib_1): Add depth argument. If depth
10493 is 128, don't look up VALUE locs and punt. Increment
10494 depth in recursive calls when walking VALUE locs.
10496 2017-03-31 Bernd Edlinger <bernd.edlinger@hotmail.de>
10498 * gcov.c (md5sum_to_hex): Fix output of MD5 hex bytes.
10499 (make_gcov_file_name): Use the canonical path name for generating
10501 (read_line): Fix handling of files with ascii null bytes.
10503 2017-03-30 Matthew Fortune <matthew.fortune@imgtec.com>
10505 * config/mips/mips.c (mips_expand_vector_init): Create a const_vector
10506 to initialise a vector register instead
10507 of using a const_int.
10509 2017-03-30 Jakub Jelinek <jakub@redhat.com>
10511 PR translation/80189
10512 * gimplify.c (omp_default_clause): Use %qs instead of %s in
10513 diagnostic messages.
10515 2017-03-30 Peter Bergner <bergner@vnet.ibm.com>
10518 * config/rs6000/dfp.md (dfp_dxex_<mode>): Update mode of operand 0.
10519 (dfp_diex_<mode>): Update mode of operand 1.
10520 * doc/extend.texi (dxex, dxexq): Document change to return type.
10521 (diex, diexq): Document change to argument type.
10523 2017-03-30 Martin Jambor <mjambor@suse.cz>
10526 * cgraph.h (cgraph_build_function_type_skip_args): Declare.
10527 * cgraph.c (redirect_call_stmt_to_callee): Set gimple fntype so that
10528 it reflects the signature changes performed at the callee side.
10529 * cgraphclones.c (build_function_type_skip_args): Make public, renamed
10530 to cgraph_build_function_type_skip_args.
10531 (build_function_decl_skip_args): Adjust call to the above function.
10533 2017-03-30 Jakub Jelinek <jakub@redhat.com>
10536 * config/i386/sse.md
10537 (<extract_type>_vextract<shuffletype><extract_suf>_mask): Use
10538 register as dest whenever it is a MEM not rtx_equal_p to the
10539 corresponding dup operand, and when forcing into reg move the
10540 reg into the memory afterwards.
10541 (<extract_type_2>_vextract<shuffletype><extract_suf_2>_mask):
10542 Likewise. Use <ssehalfvecmode> instead of <ssequartermode>
10543 for the force_reg mode.
10544 (avx512vl_vextractf128<mode>): Use register as dest either
10545 always when a MEM, or when it is a MEM not rtx_equal_p to the
10546 corresponding dup operand, or even not when it is a CONST_VECTOR
10547 depending on the mode and lo vs. hi.
10548 (avx512dq_vextract<shuffletype>64x2_1_maskm): Remove extraneous
10550 (avx512f_vextract<shuffletype>32x4_1_maskm): Likewise.
10551 (<mask_codefor>avx512dq_vextract<shuffletype>64x2_1<mask_name>):
10552 Likewise. Require that operands[2] is even.
10553 (<mask_codefor>avx512f_vextract<shuffletype>32x4_1<mask_name>):
10554 Remove extraneous parens. Require that operands[2] is a multiple
10556 (vec_extract_lo_<mode><mask_name>): Don't bother testing if
10557 operands[0] is a MEM if <mask_applied>, the predicates/constraints
10558 disallow memory then.
10560 2017-03-30 Richard Biener <rguenther@suse.de>
10562 PR tree-optimization/77498
10563 * tree-ssa-pre.c (phi_translate_1): Do not allow simplifications
10564 to non-constants over backedges.
10566 2017-03-29 Segher Boessenkool <segher@kernel.crashing.org>
10568 PR rtl-optimization/80233
10569 * combine.c (combine_instructions): Only take NONDEBUG_INSN_P insns
10570 as last_combined_insn. Do not test for BARRIER_P separately.
10572 2017-03-29 Andreas Schwab <schwab@suse.de>
10575 * calls.c (prepare_call_address): Convert funexp to Pmode before
10576 copying to temp reg.
10578 2017-03-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
10580 PR tree-optimization/80158
10581 * gimple-ssa-strength-reduction.c (replace_mult_candidate):
10582 Handle possible future case of more than one alternate
10584 (replace_rhs_if_not_dup): Likewise.
10585 (replace_one_candidate): Likewise.
10587 2017-03-28 Vladimir Makarov <vmakarov@redhat.com>
10589 PR rtl-optimization/80193
10590 * ira.c (ira): Do not check allocation for LRA.
10592 2017-03-28 Alexander Monakov <amonakov@ispras.ru>
10594 * config/nvptx/nvptx-protos.h (nvptx_output_simt_enter): Declare.
10595 (nvptx_output_simt_exit): Declare.
10596 * config/nvptx/nvptx.c (nvptx_init_unisimt_predicate): Use
10597 cfun->machine->unisimt_location. Handle NULL unisimt_predicate.
10598 (init_softstack_frame): Move initialization of crtl->is_leaf to...
10599 (nvptx_declare_function_name): ...here. Emit declaration of local
10600 memory space buffer for omp_simt_enter insn.
10601 (nvptx_output_unisimt_switch): New.
10602 (nvptx_output_softstack_switch): New.
10603 (nvptx_output_simt_enter): New.
10604 (nvptx_output_simt_exit): New.
10605 * config/nvptx/nvptx.h (struct machine_function): New fields
10606 has_simtreg, unisimt_location, simt_stack_size, simt_stack_align.
10607 * config/nvptx/nvptx.md (UNSPECV_SIMT_ENTER): New unspec.
10608 (UNSPECV_SIMT_EXIT): Ditto.
10609 (omp_simt_enter_insn): New insn.
10610 (omp_simt_enter): New expansion.
10611 (omp_simt_exit): New insn.
10612 * config/nvptx/nvptx.opt (msoft-stack-reserve-local): New option.
10614 * internal-fn.c (expand_GOMP_SIMT_ENTER): New.
10615 (expand_GOMP_SIMT_ENTER_ALLOC): New.
10616 (expand_GOMP_SIMT_EXIT): New.
10617 * internal-fn.def (GOMP_SIMT_ENTER): New internal function.
10618 (GOMP_SIMT_ENTER_ALLOC): Ditto.
10619 (GOMP_SIMT_EXIT): Ditto.
10620 * target-insns.def (omp_simt_enter): New insn.
10621 (omp_simt_exit): Ditto.
10622 * omp-low.c (struct omplow_simd_context): New fields simt_eargs,
10624 (lower_rec_simd_input_clauses): Implement SIMT privatization.
10625 (lower_rec_input_clauses): Likewise.
10626 (lower_lastprivate_clauses): Handle SIMT privatization.
10628 * omp-offload.c: Include langhooks.h, tree-nested.h, stor-layout.h.
10629 (ompdevlow_adjust_simt_enter): New.
10630 (find_simtpriv_var_op): New.
10631 (execute_omp_device_lower): Handle IFN_GOMP_SIMT_ENTER,
10632 IFN_GOMP_SIMT_ENTER_ALLOC, IFN_GOMP_SIMT_EXIT.
10634 * tree-inline.h (struct copy_body_data): New field dst_simt_vars.
10635 * tree-inline.c (expand_call_inline): Handle SIMT privatization.
10636 (copy_decl_for_dup_finish): Ditto.
10638 * tree-ssa.c (execute_update_addresses_taken): Handle GOMP_SIMT_ENTER.
10640 2017-03-28 Uros Bizjak <ubizjak@gmail.com>
10643 * config/i386/i386.c (ix86_option_override_internal): Always
10644 allow -mpreferred-stack-boundary=3 for 64-bit targets.
10646 2017-03-28 Bin Cheng <bin.cheng@arm.com>
10648 * tree-vect-loop.c (optimize_mask_stores): Add bb to the right loop.
10650 2017-03-28 Bin Cheng <bin.cheng@arm.com>
10652 * tree-vect-loop-manip.c (slpeel_add_loop_guard): New param and
10653 mark new edge's irreducible flag accordign to it.
10654 (vect_do_peeling): Check loop preheader edge's irreducible flag
10655 and pass it to function slpeel_add_loop_guard.
10657 2017-03-28 Richard Sandiford <richard.sandiford@arm.com>
10659 PR tree-optimization/80218
10660 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
10661 Update block frequencies and counts.
10663 2017-03-28 Richard Biener <rguenther@suse.de>
10665 PR tree-optimization/78644
10666 * tree-ssa-ccp.c (evaluate_stmt): When we may not use the value
10667 of a simplification result we may not use it at all.
10669 2017-03-28 Richard Biener <rguenther@suse.de>
10672 * tree-inline.c (copy_phis_for_bb): Do not create PHI node
10673 without arguments, generate default definition of a SSA name.
10675 2017-03-28 Richard Biener <rguenther@suse.de>
10677 PR middle-end/80222
10678 * gimple-fold.c (gimple_fold_indirect_ref): Do not touch
10679 TYPE_REF_CAN_ALIAS_ALL references.
10680 * fold-const.c (fold_indirect_ref_1): Likewise.
10682 2017-03-28 Martin Liska <mliska@suse.cz>
10685 * cgraphunit.c (cgraph_node::expand_thunk): Mark argument of a
10686 thunk call as DECL_GIMPLE_REG_P when vector or complex type.
10688 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
10689 Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
10691 * config/arc/arc.h (CPP_SPEC): Add subtarget_cpp_spec.
10692 (EXTRA_SPECS): Define.
10693 (SUBTARGET_EXTRA_SPECS): Likewise.
10694 (SUBTARGET_CPP_SPEC): Likewise.
10695 * config/arc/elf.h (EXTRA_SPECS): Renamed to
10696 SUBTARGET_EXTRA_SPECS.
10697 * config/arc/linux.h (SUBTARGET_CPP_SPEC): Define.
10699 2017-03-28 Claudiu Zissulescu <claziss@synopsys.com>
10701 * config/arc/simdext.md (vst64_insn): Update pattern.
10702 (vld32wh_insn): Likewise.
10703 (vld32wl_insn): Likewise.
10704 (vld64_insn): Likewise.
10705 (vld32_insn): Likewise.
10707 2017-03-28 Marek Polacek <polacek@redhat.com>
10710 * fold-const.c (fold_comparison): Use protected_set_expr_location
10711 instead of SET_EXPR_LOCATION.
10713 2017-03-28 Markus Trippelsdorf <markus@trippelsdorf.de>
10715 * tree.c (add_expr): Avoid name lookup warning.
10717 2017-03-27 Jeff Law <law@redhat.com>
10719 PR tree-optimization/80216
10720 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): Fix typo in
10721 function name. Limit recursion depth.
10722 (record_temporary_equivalences): Corresponding changes.
10724 2017-03-27 Jonathan Wakely <jwakely@redhat.com>
10726 * doc/invoke.texi (-Wno-narrowing): Reorder so default behavior is
10729 2017-03-27 Jakub Jelinek <jakub@redhat.com>
10732 * reg-notes.def (REG_CFA_NOTE): Define. Use it for CFA related
10734 * cfgcleanup.c (reg_note_cfa_p): New array.
10735 (insns_have_identical_cfa_notes): New function.
10736 (old_insns_match_p): Don't cross-jump in between /f
10737 and non-/f instructions. If both i1 and i2 are frame related,
10738 verify all CFA notes, their order and content.
10740 2017-03-27 Michael Meissner <meissner@linux.vnet.ibm.com>
10743 * config/rs6000/rs6000.md (bswaphi2_extenddi): Combine bswap
10744 HImode and SImode with zero extend to DImode to one insn.
10745 (bswap<mode>2_extenddi): Likewise.
10746 (bswapsi2_extenddi): Likewise.
10747 (bswaphi2_extendsi): Likewise.
10748 (bswaphi2): Combine bswap HImode and SImode into one insn.
10749 Separate memory insns from swapping register.
10750 (bswapsi2): Likewise.
10751 (bswap<mode>2): Likewise.
10752 (bswaphi2_internal): Delete, no longer used.
10753 (bswapsi2_internal): Likewise.
10754 (bswap<mode>2_load): Split bswap HImode/SImode into separate load,
10755 store, and gpr<-gpr swap insns.
10756 (bswap<mode>2_store): Likewise.
10757 (bswaphi2_reg): Register only splitter, combine with the splitter.
10758 (bswaphi2 splitter): Likewise.
10759 (bswapsi2_reg): Likewise.
10760 (bswapsi2 splitter): Likewise.
10761 (bswapdi2): If we have the LDBRX and STDBRX instructions, split
10762 the insns into load, store, and register/register insns.
10763 (bswapdi2_ldbrx): Likewise.
10764 (bswapdi2_load): Likewise.
10765 (bswapdi2_store): Likewise.
10766 (bswapdi2_reg): Likewise.
10768 2017-03-27 Gunther Nikl <gnikl@users.sourceforge.net>
10770 * system.h (HAVE_DESIGNATED_INITIALIZERS): Fix non C++ case.
10771 (HAVE_DESIGNATED_UNION_INITIALIZERS): Likewise.
10773 2017-03-27 Kelvin Nilsen <kelvin@gcc.gnu.org>
10776 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Edit and
10778 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
10779 special handling for target option conflicts between dform
10780 options (-mpower9-dform, -mpower9-dform-vector,
10781 -mpower9-dform-scalar) and -mno-direct-move.
10783 2017-03-27 Richard Biener <rguenther@suse.de>
10785 PR tree-optimization/80181
10786 * tree-ssa-ccp.c (likely_value): UNDEFINED ^ X is UNDEFINED.
10788 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
10790 * config/arc/predicates.md (move_double_src_operand): Replace the
10791 call to move_double_src_operand with a call to address_operand.
10793 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
10795 * config/arc/elf.h (ARGET_ARC_TP_REGNO_DEFAULT): Define.
10796 * config/arc/linux.h (ARGET_ARC_TP_REGNO_DEFAULT): Likewise.
10797 * config/arc/arc.opt (mtp-regno): Use ARGET_ARC_TP_REGNO_DEFAULT.
10799 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
10801 * config/arc/predicates.md (long_immediate_loadstore_operand):
10802 Consider scaled addresses cases.
10804 2017-03-27 Claudiu Zissulescu <claziss@synopsys.com>
10806 * config/arc/arc.c (arc_epilogue_uses): BLINK should be also
10807 restored when in interrupt.
10808 * config/arc/arc.md (simple_return): ARCv2 rtie instruction
10809 doesn't have delay slot.
10811 2017-03-27 Richard Biener <rguenther@suse.de>
10814 * tree-ssa-structalias.c (associate_varinfo_to_alias): Skip
10815 inlined thunk clones.
10817 2017-03-27 Jakub Jelinek <jakub@redhat.com>
10820 * asan.c (instrument_derefs): Copy over last operand from
10821 original COMPONENT_REF to the new COMPONENT_REF with
10822 DECL_BIT_FIELD_REPRESENTATIVE.
10823 * ubsan.c (instrument_object_size): Likewise.
10825 2017-03-27 Richard Biener <rguenther@suse.de>
10827 PR tree-optimization/80170
10828 * tree-vect-data-refs.c (vect_compute_data_ref_alignment): Make
10829 sure DR/SCEV didnt fold in constants we do not see when looking
10830 at the reference base alignment.
10832 2017-03-27 Richard Biener <rguenther@suse.de>
10834 PR middle-end/80171
10835 * gimple-fold.c (fold_ctor_reference): Properly guard against
10836 NULL return value from canonicalize_constructor_val.
10838 2017-03-25 Uros Bizjak <ubizjak@gmail.com>
10841 * config/i386/i386.c (ix86_expand_builtin)
10842 <IX86_BUILTIN_RDSEED{16,32,64}_STEP>: Do not expand arg0 between
10843 flags reg setting and flags reg using instructions.
10844 <IX86_BUILTIN_RDRAND{16,32,64}_STEP>: Ditto. Use non-flags reg
10845 clobbering instructions to zero extend op2.
10847 2017-03-25 Gerald Pfeifer <gerald@pfeifer.com>
10849 * doc/install.texi (Configuration) <--with-aix-soname>:
10850 Update link to AIX ld.
10852 2017-03-25 Bernd Schmidt <bschmidt@redhat.com>
10854 PR rtl-optimization/80160
10855 PR rtl-optimization/80159
10856 * lra-assigns.c (must_not_spill_p): Tighten new test to also take
10857 reg_alternate_class into account.
10859 2017-03-24 Vladimir Makarov <vmakarov@redhat.com>
10862 * lra-assigns.c (assign_by_spills): Add spilled non-reload pseudos
10863 to consider in curr_insn_transform.
10865 2017-03-24 Jakub Jelinek <jakub@redhat.com>
10867 * genrecog.c (validate_pattern): Add VEC_SELECT validation.
10868 * genmodes.c (emit_min_insn_modes_c): Call emit_mode_nunits
10869 and emit_mode_inner.
10871 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10873 * config/s390/s390-builtins.def: Add VXE builtins. Add a flags
10874 argument to the overloaded builtin variants. Use the new flag to
10875 deprecate certain builtin variants.
10876 * config/s390/s390-builtin-types.def: Add new builtin types.
10877 * config/s390/s390-builtins.h: Support new flags field for
10878 overloaded builtins.
10879 * config/s390/s390-c.c (OB_DEF_VAR): New flags field.
10880 (s390_macro_to_expand): Enable vector float data type.
10881 (s390_cpu_cpp_builtins_internal): Indicate support of the new
10882 builtins by incrementing the __VEC__ version number.
10883 (s390_expand_overloaded_builtin): Support expansion of vec_xl and
10885 (s390_resolve_overloaded_builtin): Emit error messages depending
10886 on the builtin flags.
10887 * config/s390/s390.c (s390_expand_builtin): Support additional
10888 flags argument. Change error message to match the messages
10889 emitted in s390-c.c.
10890 * config/s390/s390.md: New UNSPEC_* constants.
10891 (op_type): Add new instruction types.
10892 * config/s390/vecintrin.h: Add new builtins and test data class
10894 * config/s390/vx-builtins.md (V_HW_32_64): Add V4SF.
10895 (V_HW_4, VEC_HW, VECF_HW): New mode iterators.
10896 (VEC_INEXACT, VEC_NOINEXACT): New constants.
10897 ("vec_splats<mode>", "vec_insert<mode>", "vec_promote<mode>")
10898 ("vec_insert_and_zero<mode>", "vec_mergeh<mode>")
10899 ("vec_mergel<mode>"): V_HW -> VEC_HW.
10901 ("vlrlrv16qi", "vstrlrv16qi", "vbpermv16qi", "vec_msumv2di")
10902 ("vmslg", "*vftci<mode>_cconly", "vftci<mode>_intcconly")
10903 ("*vftci<mode>", "vftci<mode>_intcc", "vec_double_s64")
10904 ("vec_double_u64", "vfmin<mode>", "vfmax<mode>"): New definition.
10906 ("and_av2df3", "and_cv2df3", "vec_andc_av2df3")
10907 ("vec_andc_cv2df3", "xor_av2df3", "xor_cv2df3", "vec_nor_av2df3")
10908 ("vec_nor_cv2df3", "ior_av2df3", "ior_cv2df3", "vec_nabs")
10909 ("*vftcidb", "*vftcidb_cconly", "vftcidb"): Remove definition.
10911 ("vec_all_<fpcmpcc:code>v2df", "vec_any_<fpcmpcc:code>v2df")
10912 ("vec_scatter_elementv4si_DI", "vec_cmp<fpcmp:code>v2df")
10913 ("vec_di_to_df_s64", "vec_di_to_df_u64", "vec_df_to_di_u64")
10914 ("vfidb", "*vldeb", "*vledb", "*vec_cmp<insn_cmp>v2df_cconly")
10915 ("vec_cmpeqv2df_cc", "vec_cmpeqv2df_cc", "vec_cmphv2df_cc")
10916 ("vec_cmphev2df_cc", "*vec_cmpeqv2df_cc")
10917 ("*vec_cmphv2df_cc", "*vec_cmphev2df_cc"): Enable new modes as ...
10919 ("vec_all_<fpcmpcc:code><mode>", "vec_any_<fpcmpcc:code><mode>")
10920 ("vec_scatter_element<V_HW_4:mode>_DI")
10921 ("vec_cmp<fpcmp:code><mode>", "vcdgb", "vcdlgb", "vclgdb")
10922 ("vec_fpint<mode>", "vflls")
10923 ("vflrd", "*vec_cmp<insn_cmp><mode>_cconly", "vec_cmpeq<mode>_cc")
10924 ("vec_cmpeq<mode>_cc", "vec_cmph<mode>_cc", "vec_cmphe<mode>_cc")
10925 ("*vec_cmpeq<mode>_cc", "*vec_cmph<mode>_cc")
10926 ("*vec_cmphe<mode>_cc"): ... these.
10928 ("vec_ctd_s64", "vec_ctsl", "vec_ctul", "vec_st2f"): Use rounding
10929 mode constant instead of magic value.
10931 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10933 * config/s390/s390.c (s390_expand_vec_compare): Support other
10934 vector floating point modes than just V2DF.
10935 (s390_expand_vcond): Likewise.
10936 (s390_hard_regno_mode_ok): Allow SFmode values in VRs.
10937 (s390_cannot_change_mode_class): Prevent mode changes between TF
10938 and V1TF in vector registers.
10939 * config/s390/s390.md (DF, SF): New mode attributes.
10940 ("*cmp<mode>_ccs", "add<mode>3", "sub<mode>3", "mul<mode>3")
10941 ("fma<mode>4", "fms<mode>4", "div<mode>3", "*neg<mode>2"): Add
10942 SFmode support for VRs.
10943 * config/s390/vector.md (V_HW, V_HW2, VT_HW, ti*, nonvec): Add new
10945 (VFT, VF_HW): New mode iterators.
10946 (vw, sdx): New mode attributes.
10947 ("addv2df3", "subv2df3", "mulv2df3", "divv2df3", "sqrtv2df2")
10948 ("fmav2df4","fmsv2df4", "negv2df2", "absv2df2", "*negabsv2df2")
10949 ("smaxv2df3", "sminv2df3", "*vec_cmp<VFCMP_HW_OP:code>v2df_nocc")
10950 ("vec_cmpuneqv2df", "vec_cmpltgtv2df", "vec_orderedv2df")
10951 ("vec_unorderedv2df"): Adjust the v2df only patterns to support
10952 also the new vector floating point modes. Renaming to ...
10954 ("add<mode>3", "sub<mode>3", "mul<mode>3", "div<mode>3")
10955 ("sqrt<mode>2", "fma<mode>4", "fms<mode>4", "neg<mode>2")
10956 ("abs<mode>2", "negabs<mode>2", "smax<mode>3")
10957 ("smin<mode>3", "*vec_cmp<VFCMP_HW_OP:code><mode>_nocc")
10958 ("vec_cmpuneq<mode>", "vec_cmpltgt<mode>", "vec_ordered<mode>")
10959 ("vec_unordered<mode>"): ... these.
10961 ("neg_fma<mode>4", "neg_fms<mode>4", "*smax<mode>3_vxe")
10962 ("*smin<mode>3_vxe", "*sminv2df3_vx", "*vec_extendv4sf")
10963 ("*vec_extendv2df"): New insn definitions.
10965 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10967 * config/s390/s390.md ("*adddi3_sign", "*subdi3_sign", "mulditi3")
10968 ("mulditi3_2", "*muldi3_sign"): New patterns.
10969 ("muldi3", "*muldi3", "mulsi3", "*mulsi3"): Add an expander and
10970 rename the pattern definition.
10972 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10974 * config/s390/s390.md ("indirect_jump"): Turn insn definition into
10976 ("*indirect_jump", "*indirect2_jump"): New pattern definitions.
10978 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10980 * config/s390/s390.c (s390_expand_vec_init): Use vllezl
10981 instruction if possible.
10982 * config/s390/vector.md (vec_halfnumelts): New mode
10984 ("*vec_vllezlf<mode>"): New pattern.
10986 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10988 * config/s390/vector.md ("popcountv16qi2", "popcountv8hi2")
10989 ("popcountv4si2", "popcountv2di2"): Rename to ...
10990 ("popcount<mode>2", "popcountv8hi2_vx", "popcountv4si2_vx")
10991 ("popcountv2di2_vx"): ... these and add !TARGET_VXE to the
10993 ("popcount<mode>2_vxe"): New pattern.
10995 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
10997 * common/config/s390/s390-common.c (processor_flags_table): Add
10999 * config.gcc: Add arch12.
11000 * config/s390/driver-native.c (s390_host_detect_local_cpu):
11001 Default to arch12 for unknown CPU model numbers.
11002 * config/s390/s390-builtins.def: Add B_VXE builtin flag.
11003 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Adjust
11004 PROCESSOR_max sanity check.
11005 * config/s390/s390-opts.h (enum processor_type): Add
11007 * config/s390/s390.c (processor_table): Add arch12.
11008 (s390_expand_builtin): Add check for B_VXE flag.
11009 (s390_issue_rate): Add PROCESSOR_ARCH12.
11010 (s390_get_sched_attrmask): Likewise.
11011 (s390_get_unit_mask): Likewise.
11012 (s390_sched_score): Enable z13 scheduling for arch12.
11013 (s390_sched_reorder): Likewise.
11014 (s390_sched_variable_issue): Likewise.
11015 * config/s390/s390.h (enum processor_flags): Add PF_ARCH12 and
11017 (s390_tune_attr): Use z13 scheduling also for arch12.
11018 (TARGET_CPU_ARCH12, TARGET_CPU_ARCH12_P, TARGET_CPU_VXE)
11019 (TARGET_CPU_VXE_P, TARGET_ARCH12, TARGET_ARCH12_P, TARGET_VXE)
11020 (TARGET_VXE_P): New macros.
11021 * config/s390/s390.md: Add arch12 to cpu attribute. Add arch12
11022 and vxe to cpu_facility. Add arch12 and vxe to enabled attribute.
11023 * config/s390/s390.opt: Add arch12 as processor_type.
11025 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11027 * config/s390/s390.md
11028 ("fixuns_truncdddi2", "fixuns_trunctddi2")
11029 ("fixuns_trunc<BFP:mode><GPR:mode>2"): Merge into ...
11030 ("fixuns_trunc<FP:mode><GPR:mode>2"): New expander.
11032 ("fixuns_trunc<BFP:mode><GPR:mode>2", "fixuns_trunc<mode>si2"):
11033 Rename expanders to ...
11035 ("fixuns_trunc<BFP:mode><GPR:mode>2_emu")
11036 ("fixuns_truncdddi2_emu"): ... these.
11038 ("fixuns_trunc<mode>si2_emu"): New expander.
11040 ("*fixuns_truncdfdi2_z13"): Rename to ...
11041 ("*fixuns_truncdfdi2_vx"): ... this.
11043 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11045 * config/s390/2964.md: Remove the single element vector compare
11046 instructions which are no longer used.
11047 * config/s390/s390.c (s390_select_ccmode): Remove handling of
11049 (s390_canonicalize_comparison): Remove handling of DFmode
11051 (s390_expand_vec_compare_scalar): Remove function.
11052 (s390_emit_compare): Don't call s390_expand_vec_compare_scalar.
11053 * config/s390/s390.md ("*vec_cmp<insn_cmp>df_cconly"): Remove
11055 ("*cmp<mode>_ccs"): Add wfcdb instruction.
11057 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11059 * config/s390/s390.md ("mov<mode>_64dfp" DD_DF): Use vleig for loading a
11061 ("*mov<mode>_64" DD_DF): Remove the vector instructions. These
11062 will anyway by matched by mov<mode>_64dfp.
11064 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11066 * config/s390/s390.md ("mov<mode>" SD_SF): Change vleg/vsteg to
11067 vlef/vstef. Add missing operand to vleif.
11069 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11071 * config/s390/s390.c (s390_expand_vec_init): Enable vector load
11072 pair for all vector types with 64 bit elements.
11073 * config/s390/vx-builtins.md (V_HW_64): Move mode iterator to ...
11074 * config/s390/vector.md (V_HW_64): ... here.
11075 (V_128_NOSINGLE): New mode iterator.
11076 ("vec_init<V_HW:mode>"): Use V_128 as mode iterator.
11077 ("*vec_splat<mode>"): Use V_128_NOSINGLE mode iterator.
11078 ("*vec_tf_to_v1tf", "*vec_ti_to_v1ti"): New pattern definitions.
11079 ("*vec_load_pairv2di"): Change to ...
11080 ("*vec_load_pair<mode>"): ... this one.
11082 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11084 * config/s390/constraints.md: Add comments.
11085 (jKK): Reject element sizes > 8 bytes.
11086 * config/s390/s390.c (s390_split_ok_p): Enable splitting also for
11088 * config/s390/s390.md: Add the s_operand checks formerly in
11089 s390_split_ok_p to various splitters where they are still
11091 * config/s390/vector.md ("mov<mode>" V_128): Add GPR alternatives
11092 for 128 bit vectors. Plus two splitters.
11094 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11096 * config/s390/s390.md: Rename the cpu facilty vec to vx throughout
11099 2017-03-24 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11102 * config/s390/s390-c.c (s390_adjust_builtin_arglist): Issue an
11103 error if the boundary argument is not constant.
11105 2017-03-24 Jakub Jelinek <jakub@redhat.com>
11107 PR rtl-optimization/80112
11108 * loop-doloop.c (doloop_condition_get): Don't check condition
11109 if cmp isn't SET with IF_THEN_ELSE src.
11111 2017-03-24 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11113 PR tree-optimization/80158
11114 * gimple-ssa-strength-reduction.c (replace_mult_candidate): When
11115 replacing a candidate statement, also replace it for the
11116 candidate's alternate interpretation.
11117 (replace_rhs_if_not_dup): Likewise.
11118 (replace_one_candidate): Likewise.
11120 2017-03-24 Richard Biener <rguenther@suse.de>
11122 PR tree-optimization/80167
11123 * graphite-isl-ast-to-gimple.c
11124 (translate_isl_ast_to_gimple::is_valid_rename): Handle default-defs
11126 (translate_isl_ast_to_gimple::get_rename): Likewise.
11128 2017-03-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
11130 * config/rs6000/rs6000.c (rs6000_option_override_internal): Change
11131 handling of certain combinations of target options, including the
11132 combinations -mpower8-vector vs. -mno-vsx, -mpower9-vector vs.
11133 -mno-power8-vector, and -mpower9_dform vs. -mno-power9-vector.
11135 2017-03-23 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11138 * config/arm/arm.md (*load_multiple): Add reload_completed to
11139 matching condition.
11141 2017-03-23 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11142 Richard Biener <rguenth@suse.de>
11144 PR tree-optimization/79908
11145 PR tree-optimization/80136
11146 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
11147 been cast away, gimplify_and_add suffices.
11149 2017-03-23 Markus Trippelsdorf <markus@trippelsdorf.de>
11151 * tree-vrp.c (identify_jump_threads): Delete avail_exprs.
11153 2017-03-23 Richard Biener <rguenther@suse.de>
11155 PR tree-optimization/80032
11156 * gimplify.c (gimple_push_cleanup): Forced unconditional
11157 cleanups still have to go to the conditional_cleanups
11160 2017-03-22 Jakub Jelinek <jakub@redhat.com>
11162 PR tree-optimization/80072
11163 * tree-ssa-reassoc.c (struct operand_entry): Change id field type
11165 (next_operand_entry_id): Change type to unsigned int.
11166 (sort_by_operand_rank): Make sure to return the right return value
11167 even if unsigned fields are bigger than INT_MAX.
11168 (struct oecount): Change cnt and id type to unsigned int.
11169 (oecount_hasher::equal): Formatting fix.
11170 (oecount_cmp): Make sure to return the right return value
11171 even if unsigned fields are bigger than INT_MAX.
11172 (undistribute_ops_list): Change next_oecount_id type to unsigned int.
11175 * gimplify.c (gimplify_modify_expr_rhs) <case COND_EXPR>: Clear
11176 TREE_READONLY on result if writing it more than once.
11179 * doc/invoke.texi (-fsanitize=thread): Document that with
11180 -fnon-call-exceptions atomics are not able to throw
11184 * tsan.c: Include tree-eh.h.
11185 (instrument_builtin_call): Call maybe_clean_eh_stmt or
11186 maybe_clean_or_replace_eh_stmt where needed.
11187 (instrument_memory_accesses): Add cfg_changed argument.
11188 Call gimple_purge_dead_eh_edges on each block and set *cfg_changed
11189 if it returned true.
11190 (tsan_pass): Adjust caller. Return TODO_cleanup_cfg if cfg_changed.
11192 PR rtl-optimization/63191
11193 * config/i386/i386.c (ix86_delegitimize_address): Turn into small
11194 wrapper function, moved the whole old content into ...
11195 (ix86_delegitimize_address_1): ... this. New inline function.
11196 (ix86_find_base_term): Use ix86_delegitimize_address_1 with
11197 true as last argument instead of ix86_delegitimize_address.
11199 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
11201 * config/aarch64/aarch64.c (generic_branch_cost): Copy
11202 cortexa57_branch_cost.
11204 2017-03-22 Wilco Dijkstra <wdijkstr@arm.com>
11206 * config/aarch64/aarch64.c (generic_tunings): Add AES fusion.
11208 2017-03-21 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11211 * doc/md.texi (Constraints): Document wA constraint.
11212 * config/rs6000/constraints.md (wA): New.
11213 * config/rs6000/rs6000.c (rs6000_debug_reg_global): Add wA reg_class.
11214 (rs6000_init_hard_regno_mode_ok): Init wA constraint.
11215 * config/rs6000/rs6000.h (RS6000_CONSTRAINT_wA): New.
11216 * config/rs6000/vsx.md (vsx_splat_<mode>): Use wA constraint.
11218 2017-03-22 Cesar Philippidis <cesar@codesourcery.com>
11221 * gimplify.c (is_oacc_declared): New function.
11222 (oacc_default_clause): Use it to set default flags for acc declared
11223 variables inside parallel regions.
11224 (gimplify_scan_omp_clauses): Strip firstprivate pointers for acc
11225 declared variables.
11226 (gimplify_oacc_declare): Gimplify the declare clauses. Add the
11227 declare attribute to any decl as necessary.
11229 2017-03-22 Thomas Preud'homme <thomas.preudhomme@arm.com>
11232 * config/arm/arm-isa.h (isa_bit_lpae): New feature bit.
11233 (ISA_ARMv7ve): Add isa_bit_lpae to the definition.
11234 * config/arm/arm-protos.h (arm_arch7ve): Rename into ...
11235 (arm_arch_lpae): This.
11236 * config/arm/arm.c (arm_arch7ve): Rename into ...
11237 (arm_arch_lpae): This. Define it in term of isa_bit_lpae.
11238 * config/arm/arm.h (TARGET_HAVE_LPAE): Redefine in term of
11241 2017-03-22 Martin Liska <mliska@suse.cz>
11244 * config/rs6000/rs6000.c (rs6000_inner_target_options): Show
11245 error message instead of an ICE.
11247 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11249 * doc/extend.texi (6.11 Additional Floating Types): Revise.
11251 2017-03-21 Kelvin Nilsen <kelvin@gcc.gnu.org>
11253 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
11255 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
11258 2017-03-21 Martin Sebor <msebor@redhat.com>
11260 * doc/extend.texi: Use "cannot" instead of "can't."
11261 * doc/hostconfig.texi: Same.
11262 * doc/install.texi: Same.
11263 * doc/invoke.texi: Same.
11264 * doc/loop.texi: Same.
11265 * doc/md.texi: Same.
11266 * doc/objc.texi: Same.
11267 * doc/rtl.texi: Same.
11268 * doc/tm.texi: Same.
11269 * doc/tm.texi.in: Same.
11270 * doc/trouble.texi: Same.
11272 2017-03-21 Alexandre Oliva <aoliva@redhat.com>
11275 * dwarf2out.c (struct checksum_attributes): Add at_alignment.
11276 (collect_checksum_attributes): Set it.
11277 (die_checksum_ordered): Use it.
11279 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11281 PR tree-optimization/79908
11282 * tree-stdarg.c (expand_ifn_va_arg_1): Revert the following
11283 change: For a VA_ARG whose LHS has been cast away, use
11284 force_gimple_operand to construct the side effects.
11286 2017-03-21 David Malcolm <dmalcolm@redhat.com>
11288 PR translation/80001
11289 * omp-offload.c (oacc_loop_fixed_partitions): Make diagnostics
11290 more amenable to translation.
11291 (oacc_loop_auto_partitions): Likewise.
11293 2017-03-21 Marek Polacek <polacek@redhat.com>
11294 Martin Sebor <msebor@redhat.com>
11296 PR tree-optimization/80109
11297 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call get_range_info
11298 on INTEGRAL_TYPE_P.
11300 2017-03-21 Jakub Jelinek <jakub@redhat.com>
11301 Segher Boessenkool <segher@kernel.crashing.org>
11304 * combine.c (can_combine_p): Revert the 2017-03-20 change, only
11305 check reg_used_between_p between insn and one of succ or succ2
11306 depending on if succ is artificial insn not inserted into insn
11309 2017-03-21 Martin Liska <mliska@suse.cz>
11311 PR gcov-profile/80081
11312 * Makefile.in: Add gcov-dump and fix installation of gcov-tool.
11313 * doc/gcc.texi: Include gcov-dump stuff.
11314 * doc/gcov-dump.texi: New file.
11316 2017-03-21 Toma Tabacu <toma.tabacu@imgtec.com>
11318 PR rtl-optimization/79150
11319 * config/mips/mips.c (mips_block_move_loop): Emit a NOP after the
11320 conditional jump, if the jump is the last insn of the loop.
11322 2017-03-21 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11323 Richard Biener <rguenth@suse.de>
11325 PR tree-optimization/79908
11326 * tree-stdarg.c (expand_ifn_va_arg_1): For a VA_ARG whose LHS has
11327 been cast away, use force_gimple_operand to construct the side
11330 2017-03-21 Martin Liska <mliska@suse.cz>
11332 PR libfortran/79956
11333 * simplify-rtx.c (simplify_immed_subreg): Initialize a variable
11336 2017-03-21 Brad Spengler <spender@grsecurity.net>
11339 * plugin.c (htab_hash_plugin): New function.
11340 (add_new_plugin): Use it and adjust.
11341 (parse_plugin_arg_opt): Adjust.
11342 (init_one_plugin): Likewise.
11344 2017-03-21 Richard Biener <rguenther@suse.de>
11346 PR tree-optimization/80032
11347 * gimplify.c (gimple_push_cleanup): Add force_uncond parameter,
11348 if set force the cleanup to happen unconditionally.
11349 (gimplify_target_expr): Push inserted clobbers with force_uncond
11350 to avoid them being removed by control-dependent DCE.
11352 2017-03-21 Richard Biener <rguenther@suse.de>
11354 PR tree-optimization/80122
11355 * tree-inline.c (copy_bb): Do not expans va-arg packs or
11356 va_arg_pack_len when the inlined call stmt requires pack
11358 * tree-inline.h (struct copy_body_data): Make call_stmt a gcall *.
11360 2017-03-21 Jakub Jelinek <jakub@redhat.com>
11363 * tsan.c (instrument_builtin_call): If the memory model argument
11364 is not a constant, assume it is valid.
11367 * fold-const.c (round_up_loc): Negate divisor in unsigned type to
11370 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
11372 PR rtl-optimization/79910
11373 * combine.c (can_combine_p): Do not allow combining an I0 or I1
11374 if its dest is used by an insn before I2 (other than the combined
11375 insns themselves, which are properly handled already).
11377 2017-03-20 Segher Boessenkool <segher@kernel.crashing.org>
11380 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
11382 * combine.c (record_used_regs): New static function.
11383 (try_combine): Handle situations where there is an additional
11384 instruction between I2 and I3 which needs to have a LOG_LINK
11388 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
11390 * combine.c (try_combine): Delete redundant i1 test. Call
11391 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
11393 2017-03-20 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
11396 * config/rs6000/rs6000.md (*movsi_internal1): Fix constraints for
11397 alternatives 13/14.
11399 2017-03-20 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11401 PR tree-optimization/80054
11402 * gimple-ssa-strength-reduction.c (all_phi_incrs_profitable): Fail
11403 the optimization if a PHI or any of its arguments is not dominated
11404 by the candidate's basis. Use gphi* rather than gimple* as
11406 (replace_profitable_candidates): Clean up a gimple* variable that
11407 should be a gphi* variable.
11409 2017-03-20 Martin Sebor <msebor@redhat.com>
11412 * doc/extend.texi (attribute constructor): Document present limitation.
11414 2017-03-20 Kelvin Nilsen <kelvin@gcc.gnu.org>
11417 * config/rs6000/altivec.h (vec_all_ne): Under __cplusplus__ and
11418 __POWER9_VECTOR__ #ifdef control, change template definition to
11419 use Power9-specific built-in function.
11420 (vec_any_eq): Likewise.
11421 * config/rs6000/vector.md (vector_ae_v2di_p): Change the flag used
11422 to control outcomes from this test.
11423 (vector_ae_<mode>p): For VEC_F modes, likewise.
11425 2017-03-20 Ian Lance Taylor <iant@google.com>
11427 * config/i386/i386.c (ix86_function_regparm): Save an extra
11428 register for -fsplit-stack with DECL_STATIC_CHAIN.
11430 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
11433 * config/riscv/riscv.c (riscv_preferred_reload_class): Remove.
11434 (TARGET_PREFERRED_RELOAD_CLASS): Likewise.
11436 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com>
11438 * config/riscv/riscv.c (riscv_print_operand): Use "fence
11440 * config/riscv/sync.mc (mem_thread_fence_1): Use "fence
11443 2017-03-20 Marek Polacek <polacek@redhat.com>
11446 * asan.c (DEF_SANITIZER_BUILTIN): Use do { } while (0).
11448 2017-03-20 Richard Biener <rguenther@suse.de>
11450 PR tree-optimization/80113
11451 * graphite-isl-ast-to-gimple.c (copy_loop_phi_nodes): Do not
11452 allocate extra SSA name for PHI def.
11453 (add_close_phis_to_outer_loops): Likewise.
11454 (add_close_phis_to_merge_points): Likewise.
11455 (copy_loop_close_phi_args): Likewise.
11456 (copy_cond_phi_nodes): Likewise.
11458 2017-03-20 Martin Liska <mliska@suse.cz>
11460 PR middle-end/79753
11461 * tree-chkp.c (chkp_build_returned_bound): Do not build
11462 returned bounds for a LHS that's not a BOUNDED_P type.
11464 2017-03-20 Martin Liska <mliska@suse.cz>
11468 * tree-chkp.c (chkp_find_bounds_1): Handle REAL_CST,
11469 COMPLEX_CST and VECTOR_CST.
11471 2017-03-20 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
11474 * config/s390/s390.md ("cmp<mode>_ccs_0"): Add a clobber of the
11475 target operand. A new splitter adds the clobber statement in case
11476 the target operand is dead anyway.
11478 2017-03-19 Gerald Pfeifer <gerald@pfeifer.com>
11480 * doc/install.texi (Specific) <sparc-*-linux*>: No longer refer
11481 to age-old versions of binutils and glibc.
11483 2017-03-18 Segher Boessenkool <segher@kernel.crashing.org>
11485 * doc/contrib.texi (Contributors): Remove duplicate entry for myself.
11487 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
11489 * doc/contrib.texi (Contributors): Add Segher Boessenkool.
11491 2017-03-18 Gerald Pfeifer <gerald@pfeifer.com>
11493 * doc/install.texi (Specific) <arm-*-eabi>: Remove old
11494 requirement for binutils 2.13.
11496 2017-03-17 Jim Wilson <jim.wilson@linaro.org>
11498 * combine.c (try_combine): Delete redundant i1 test. Call
11499 prev_nonnote_nondebug_insn instead of prev_nonnote_insn.
11501 2017-03-17 Palmer Dabbelt <palmer@dabbelt.com
11503 * doc/install.texi (Specific) <riscv32-*-elf>: Add riscv32-*-elf,
11504 riscv32-*-linux, riscv64-*-elf, riscv64-*-linux to the table of
11506 <riscv64-*-elf>: Re-arrange section
11507 <riscv32-*-elf>: Add a note about requiring binutils 2.28.
11508 <riscv32-*-linux>: Likewise.
11509 <riscv64-*-elf>: Likewise
11510 <riscv64-*-linux>: Likewise.
11512 2017-03-17 Richard Earnshaw <rearnsha@arm.com>
11515 * aarch64.opt(verbose-cost-dump): Fix typo.
11517 2017-03-17 Pat Haugen <pthaugen@us.ibm.com>
11520 * config/rs6000/rs6000.md (copysign<mode>3_fcpsgn): Test
11521 for VECTOR_UNIT_VSX_P (<MODE>mode) too.
11523 2017-03-17 Bernd Schmidt <bschmidt@redhat.com>
11525 * reload.c (find_reloads): When reloading a nonoffsettable address,
11526 use RELOAD_OTHER for it and its address reloads.
11528 PR rtl-optimization/79910
11529 * combine.c (record_used_regs): New static function.
11530 (try_combine): Handle situations where there is an additional
11531 instruction between I2 and I3 which needs to have a LOG_LINK
11534 2017-03-17 Jeff Law <law@redhat.com>
11536 PR tree-optimization/71437
11537 * tree-vrp.c (simplify_stmt_for_jump_threading): Lookup the
11538 conditional in the hash table first.
11539 (vrp_dom_walker::before_dom_children): Extract condition from
11540 ASSERT_EXPR. Record condition, its inverion and any implied
11541 conditions as well.
11543 2017-03-17 Marek Polacek <polacek@redhat.com>
11544 Markus Trippelsdorf <markus@trippelsdorf.de>
11546 PR tree-optimization/80079
11547 * gimple-ssa-store-merging.c (class pass_store_merging): Initialize
11550 2017-03-17 Richard Biener <rguenther@suse.de>
11552 PR middle-end/80075
11553 * tree-eh.c (stmt_could_throw_1_p): Only handle gimple assigns.
11554 Properly verify the LHS before the RHS possibly claims to be
11556 (stmt_could_throw_p): Hande gimple conds fully here. Clobbers
11559 2017-03-17 Martin Jambor <mjambor@suse.cz>
11561 * doc/invoke.texi (Option Options): Include -fipa-vrp in the list.
11562 (List of -O2 options): Likewise.
11563 (-fipa-bit-cp): Replace "ipa" with "interprocedural."
11566 2017-03-17 Tom de Vries <tom@codesourcery.com>
11568 * gcov-dump.c (print_usage): Print bug_report_url.
11570 2017-03-17 Richard Biener <rguenther@suse.de>
11572 PR middle-end/80050
11573 * genmatch.c (parser::next): Remove pointless check for CPP_EOF.
11574 (parser::peek): Likewise.
11576 2017-03-17 Richard Biener <rguenther@suse.de>
11578 PR tree-optimization/80048
11579 * sese.c (free_sese_info): Properly release rename_map and
11580 copied_bb_map elements.
11582 2017-03-16 Alexandre Oliva <aoliva@redhat.com>
11584 * gimple-ssa-store-merging.c (struct imm_store_chain_info):
11585 Add linked-list forward and backlinks. Insert on
11586 construction, remove on destruction.
11587 (class pass_store_merging): Add m_stores_head field.
11588 (pass_store_merging::terminate_and_process_all_chains):
11589 Iterate over m_stores_head list.
11590 (pass_store_merging::terminate_all_aliasing_chains):
11592 (pass_store_merging::execute): Check for debug stmts first.
11593 Push new chains onto the m_stores_head stack.
11595 2017-03-16 Michael Meissner <meissner@linux.vnet.ibm.com>
11598 * config/rs6000/vsx.md (vsx_splat_<mode>, VSX_D iterator): Allow a
11599 SPLAT operation on ISA 2.07 64-bit systems that have direct move,
11600 but no MTVSRDD support, by doing MTVSRD and XXPERMDI.
11602 2017-03-16 Jeff Law <law@redhat.com>
11604 PR tree-optimization/71437
11605 * tree-ssa-dom.c (dom_opt_dom_walker): Remove thread_across_edge
11606 member function. Implementation moved into after_dom_children
11607 member function and into the threader's thread_outgoing_edges
11609 (dom_opt_dom_walker::after_dom_children): Simplify by moving
11610 some code into new thread_outgoing_edges.
11611 * tree-ssa-threadedge.c (thread_across_edge): Make static and simplify
11612 definition. Simplify marker handling (do it here). Assume we always
11613 have the available expression and the const/copies tables.
11614 (thread_outgoing_edges): New function extracted from tree-ssa-dom.c
11616 * tree-ssa-threadedge.h (thread_outgoing_edges): Declare.
11617 * tree-vrp.c (equiv_stack): No longer file scoped.
11618 (vrp_dom_walker): New class.
11619 (vrp_dom_walker::before_dom_children): New member function.
11620 (vrp_dom_walker::after_dom_children): Likewise.
11621 (identify_jump_threads): Setup domwalker. Use it rather than
11622 walking edges in a random order by hand. Simplify setup/finalization.
11623 (finalize_jump_threads): Remove.
11624 (vrp_finalize): Do not call identify_jump_threads here.
11625 (execute_vrp): Do it here instead and call thread_through_all_blocks
11628 PR tree-optimization/71437
11629 * tree-ssa-dom.c (pfn_simplify): Add basic_block argument. All
11631 (simplify_stmt_for_jump_threading): Add basic_block argument. All
11633 (lhs_of_dominating_assert): Moved from here into tree-vrp.c.
11634 (dom_opt_dom_walker::thread_across_edge): Remove
11635 handle_dominating_asserts argument. All callers changed.
11636 (record_temporary_equivalences_from_stmts_at_dest): Corresponding
11637 changes. Remove calls to lhs_of_dominating_assert. Other
11638 uses of handle_dominating_asserts turn into unconditional code
11639 (simplify_control_stmt_condition_1): Likewise.
11640 (simplify_control_stmt_condition): Likewise.
11641 (thread_through_normal_block, thread_across_edge): Likewise.
11642 * tree-ssa-threadedge.h (thread_across_edge): Corresponding changes.
11643 * tree-vrp.c (lhs_of_dominating_assert): Move here. Return original
11644 object if it is not an SSA_NAME.
11645 (simplify_stmt_for_jump_threading): Call lhs_of_dominating_assert
11646 before calling into the VRP specific simplifiers.
11647 (identify_jump_threads): Remove handle_dominating_asserts
11650 2017-03-16 Jakub Jelinek <jakub@redhat.com>
11653 * tree-diagnostic.c (default_tree_printer): No longer static.
11654 * tree-diagnostic.h (default_tree_printer): New prototype.
11656 2017-03-16 Tamar Christina <tamar.christina@arm.com>
11658 * config/aarch64/aarch64-simd.md (*aarch64_simd_mov<mode>)
11659 Change ins into fmov.
11661 2017-03-16 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11663 * config/aarch64/iterators.md (h_con): Return "x" for V4HF and V8HF.
11664 * config/aarch64/aarch64-simd.md (*aarch64_fma4_elt_from_dup<mode>):
11665 Use h_con constraint for operand 1.
11666 (*aarch64_fnma4_elt_from_dup<mode>): Likewise.
11667 (*aarch64_mulx_elt_from_dup<mode>): Likewise for operand 2.
11669 2017-03-15 Jeff Law <law@redhat.com>
11671 PR tree-optimization/71437
11672 * tree-ssa-dom.c (derive_equivalences_from_bit_ior): New function.
11673 (record_temporary_equivalences): Use it.
11675 PR tree-optimization/71437
11676 * tree-ssa-dom.c (struct cond_equivalence): Moved from here into
11677 tree-ssa-scopedtables.
11678 (lookup_avail_expr, build_and_record_new_cond): Likewise.
11679 (record_conditions, record_cond, vuse_eq): Likewise.
11680 (record_edge_info): Adjust to API tweak of record_conditions.
11681 (simplify_stmt_for_jump_threading): Similarly for lookup_avail_expr.
11682 (record_temporary_equivalences, optimize_stmt): Likewise.
11683 (eliminate_redundant_computations): Likewise.
11684 (record_equivalences_from_stmt): Likewise.
11685 * tree-ssa-scopedtables.c: Include options.h and params.h.
11686 (vuse_eq): New function, moved from tree-ssa-dom.c
11687 (build_and_record_new_cond): Likewise.
11688 (record_conditions): Likewise. Accept vector of conditions rather
11689 than edge_equivalence structure for first argument.
11690 for the first argument.
11691 (avail_exprs_stack::lookup_avail_expr): New member function, moved
11692 from tree-ssa-dom.c.
11693 (avail_exprs_stack::record_cond): Likewise.
11694 * tree-ssa-scopedtables.h (struct cond_equivalence): Moved here
11695 from tree-ssa-dom.c.
11696 (avail_exprs_stack): Add new member functions lookup_avail_expr
11698 (record_conditions): Declare.
11700 2017-03-15 Vladimir Makarov <vmakarov@redhat.com>
11703 * lra-constraints.c (process_alt_operands): Increase reject for
11704 reloading an input/output operand.
11706 2017-03-15 Michael Meissner <meissner@linux.vnet.ibm.com>
11709 * config/rs6000/rs6000.md (float<QHI:mode><IEEE128:mode>2): Define
11710 insns to convert from signed/unsigned char/short to IEEE 128-bit
11712 (floatuns<QHI:mode><IEEE128:mode>2): Likewise.
11714 2017-03-15 Uros Bizjak <ubizjak@gmail.com>
11717 * config/i386/i386.c (ix86_vector_duplicate_value): Create
11718 subreg of inner mode for values already in registers.
11720 2017-03-15 Bernd Schmidt <bschmidt@redhat.com>
11722 * config/c6x/c6x.c (hwloop_optimize): Handle case where the old
11723 iteration reg is used after the loop.
11725 2017-03-14 Martin Sebor <msebor@redhat.com>
11727 PR tree-optimization/79800
11728 * gimple-ssa-sprintf.c (format_floating: Add argument. Handle
11729 precision in negative-positive range.
11730 (format_floating): Call non-const overload with adjusted precision.
11732 2017-03-14 Michael Meissner <meissner@linux.vnet.ibm.com>
11735 * config/rs6000/rs6000.h (TARGET_FRSQRTES): Add check for
11738 2017-03-14 Martin Sebor <msebor@redhat.com>
11740 PR middle-end/80020
11741 * builtin-attrs.def (ATTR_ALLOC_SIZE_2_NOTHROW_LIST): New macro.
11742 * builtins.def (aligned_alloc): Use it.
11745 * Makefile.in (GTFILES): Add calls.c.
11746 * calls.c: Include "gt-calls.h".
11748 2017-03-14 Bernd Schmidt <bschmidt@redhat.com>
11750 PR rtl-optimization/79728
11751 * regs.h (struct target_regs): New field
11752 x_contains_allocatable_regs_of_mode.
11753 (contains_allocatable_regs_of_mode): New macro.
11754 * reginfo.c (init_reg_sets_1): Initialize it, and change
11755 contains_reg_of_mode so it includes global regs as well.
11756 * reload.c (push_reload): Use contains_allocatable_regs_of_mode
11757 rather than contains_regs_of_mode.
11759 2017-03-14 Martin Liska <mliska@suse.cz>
11761 * doc/invoke.texi: Document options that can't be combined with
11762 -fcheck-pointer-bounds.
11764 2017-03-14 Martin Liska <mliska@suse.cz>
11766 PR middle-end/79831
11767 * doc/invoke.texi (-Wchkp): Document the option.
11769 2017-03-14 Martin Liska <mliska@suse.cz>
11771 * Makefile.in: Install gcov-dump.
11773 2017-03-14 Martin Liska <mliska@suse.cz>
11775 * multiple_target.c (expand_target_clones): Bail out for
11776 an invalid attribute.
11778 2017-03-14 Richard Biener <rguenther@suse.de>
11780 * alias.c (struct alias_set_entry): Pack properly.
11781 * cfgloop.h (struct loop): Likewise.
11782 * cse.c (struct set): Likewise.
11783 * ipa-utils.c (struct searchc_env): Likewise.
11784 * loop-invariant.c (struct invariant): Likewise.
11785 * lra-remat.c (struct cand): Likewise.
11786 * recog.c (struct change_t): Likewise.
11787 * rtl.h (struct address_info): Likewise.
11788 * symbol-summary.h (function_summary): Likewise.
11789 * tree-loop-distribution.c (struct partition): Likewise.
11790 * tree-object-size.c (struct object_size_info): Likewise.
11791 * tree-ssa-loop-ivopts.c (struct cost_pair): Likewise.
11792 * tree-ssa-threadupdate.c (struct ssa_local_info_t): Likewise.
11793 * tree-vect-data-refs.c (struct _vect_peel_info): Likewise.
11794 * tree-vect-slp.c (struct _slp_oprnd_info): Likewise.
11795 * tree-vect-stmts.c (struct simd_call_arg_info): Likewise.
11796 * tree-vectorizer.h (struct _loop_vec_info): Likewise.
11797 (struct _stmt_vec_info): Likewise.
11799 2017-03-14 Martin Liska <mliska@suse.cz>
11802 * multiple_target.c (create_dispatcher_calls): Check that
11803 a target can create a function dispatcher.
11805 2017-03-14 Martin Liska <mliska@suse.cz>
11808 * multiple_target.c (expand_target_clones): Drop local.local
11809 flag for default implementation.
11811 2017-03-14 Richard Biener <rguenther@suse.de>
11813 PR tree-optimization/80030
11814 * tree-vect-stmts.c (vectorizable_store): Plug memleak.
11816 2017-03-13 Kito Cheng <kito.cheng@gmail.com>
11818 * config/riscv/riscv.c (riscv_emit_float_compare>: Use
11819 gcc_fallthrough() instead of __attribute__((fallthrough));
11821 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
11823 * doc/gcc.texi: Remove "up" link to (DIR).
11824 * doc/gccint.texi: Ditto.
11826 2017-03-13 Gerald Pfeifer <gerald@pfeifer.com>
11828 * doc/install.texi (Specific) <avr>: Remove reference to
11831 2017-03-13 Jeff Law <law@redhat.com>
11833 * config/riscv/riscv.c (riscv_emit_float_compare): Use fallthru
11834 attribute rather than comments.
11836 * config/pdp11/pdp11.md (movmemhi): Adjust operand numbers to
11837 match_scratch operand is highest.
11839 2017-03-13 Martin Liska <mliska@suse.cz>
11841 PR middle-end/78339
11842 * ipa-pure-const.c (warn_function_noreturn): If the declarations
11843 is a CHKP clone, use original declaration.
11845 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
11847 * config/arc/arc.c (arc_init): Use multiplier whenever we have it.
11848 (arc_conditional_register_usage): Use a different allocation order
11849 when optimizing for size.
11850 * common/config/arc/arc-common.c (arc_option_optimization_table):
11851 Section anchors default on when optimizing for size.
11853 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
11855 * config/arc/arc.md (*tst_bitfield_tst): Fix pattern.
11857 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
11859 * config/arc/arc.c (arc_output_addsi): Emit code density adds.
11860 * config/arc/arc.md (cpu_facility): Add cd variant.
11861 (*movqi_insn): Add code density variant.
11862 (*movhi_insn): Likewise.
11863 (*movqi_insn): Likewise.
11864 (*addsi3_mixed): Likewise.
11865 (subsi3_insn): Likewise.
11867 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
11869 * config/arc/arc.md (movsi_cond_exec): Update constraint.
11871 2017-03-13 Claudiu Zissulescu <claziss@synopsys.com>
11873 * config/arc/arc.c (arc_legitimize_pic_address): Handle PIC
11874 expressions with MINUS and UNARY ops.
11876 2017-03-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
11879 * config/arm/neon.md (vec_sel_widen_ssum_lo<VQI:mode><VW:mode>3):
11881 (vec_sel_widen_ssum_lo<mode><V_half>3): ... This. Avoid mismatch
11882 between vec_select and vector argument.
11883 (vec_sel_widen_ssum_hi<VQI:mode><VW:mode>3): Rename to...
11884 (vec_sel_widen_ssum_hi<mode><V_half>3): ... This. Likewise.
11885 (vec_sel_widen_usum_lo<VQI:mode><VW:mode>3): Rename to...
11886 (vec_sel_widen_usum_lo<mode><V_half>3): ... This.
11887 (vec_sel_widen_usum_hi<VQI:mode><VW:mode>3): Rename to...
11888 (vec_sel_widen_usum_hi<mode><V_half>3): ... This.
11890 2017-03-13 Richard Biener <rguenther@suse.de>
11893 * params.def (vect-max-peeling-for-alignment): Fix typo.
11895 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
11897 * doc/install.texi (Specific) <mips-*-*>: Remove description of
11898 issue that only occurred with binutils below 2.18.
11900 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
11902 * doc/install.texi (Specific) <cris-axis-elf>: No longer
11903 refer to binutils 2.11/2.12 minimum.
11905 2017-03-12 Gerald Pfeifer <gerald@pfeifer.com>
11907 * doc/install.texi (Specific) <powerpc-*-*>: Remove link to
11908 ftp.kernel.org and simplify binutils requirement.
11910 2017-03-11 Gerald Pfeifer <gerald@pfeifer.com>
11912 * doc/invoke.texi (Warning Options): Fix spelling of link-time
11914 (Optimize Options): Ditto. Also remove redundancy.
11916 2017-03-10 David Malcolm <dmalcolm@redhat.com>
11918 PR translation/79848
11919 * ipa-devirt.c (warn_types_mismatch): Simplify uses of "%<%s%>" to
11921 * ipa-pure-const.c (suggest_attribute): Likewise. Convert _
11922 to G_ to avoid double translation.
11924 2017-03-10 David Malcolm <dmalcolm@redhat.com>
11926 PR translation/79923
11927 * auto-profile.c (get_combined_location): Convert leading
11928 character of diagnostics to lower case and remove trailing period.
11929 (read_profile): Likewise for various diagnostics.
11930 * config/arm/arm.c (arm_option_override): Remove trailing period
11931 from various diagnostics.
11932 * config/msp430/msp430.c (msp430_expand_delay_cycles): Likewise.
11933 (msp430_expand_delay_cycles): Likewise.
11935 2017-03-10 David Malcolm <dmalcolm@redhat.com>
11938 * config/aarch64/aarch64.c (aarch64_validate_mcpu): Quote the
11939 full command-line argument, rather than just "str".
11940 (aarch64_validate_march): Likewise.
11941 (aarch64_validate_mtune): Likewise.
11943 2017-03-10 Bernd Schmidt <bschmidt@redhat.com>
11945 PR rtl-optimization/78911
11946 * lra-assigns.c (must_not_spill_p): New function.
11947 (spill_for): Use it.
11949 2017-03-10 Jakub Jelinek <jakub@redhat.com>
11951 PR tree-optimization/79981
11952 * tree-vrp.c (extract_range_basic): Handle IMAGPART_EXPR of
11953 ATOMIC_COMPARE_EXCHANGE ifn result.
11954 (stmt_interesting_for_vrp, vrp_visit_stmt): Handle
11955 IFN_ATOMIC_COMPARE_EXCHANGE.
11957 2017-03-10 David Malcolm <dmalcolm@redhat.com>
11960 * opts.c (parse_sanitizer_options): Add missing question mark to
11961 "did you mean" message.
11963 2017-03-10 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
11965 * config/rs6000/rs6000-builtin.def (VMULEUB_UNS): Remove orphaned
11967 (VMULEUH_UNS): Likewise.
11968 (VMULOUB_UNS): Likewise.
11969 (VMULOUH_UNS): Likewise.
11970 * config/rs6000/rs6000.c (builtin_function_type): Remove
11971 references to ALTIVEC_BUILTIN_VMUL[EO]U[BH]_UNS.
11973 2017-03-10 David Malcolm <dmalcolm@redhat.com>
11976 * read-rtl-function.c (function_reader::read_rtx_operand): Update
11977 x with result of extra_parsing_for_operand_code_0.
11978 (function_reader::extra_parsing_for_operand_code_0): Convert
11979 return type from void to rtx, returning x. When reading
11980 SYMBOL_REF with SYMBOL_FLAG_HAS_BLOCK_INFO, reallocate x to the
11981 larger size containing struct block_symbol.
11983 2017-03-10 Segher Boessenkool <segher@kernel.crashing.org>
11985 * config/rs6000/rs6000.c (rs6000_option_override_internal): Disallow
11986 -mfloat128-hardware without -m64.
11988 2017-03-10 Will Schmidt <will_schmidt@vnet.ibm.com>
11991 * config/rs6000/rs6000.c (builtin_function_type): Add VMUL*U[HB]
11992 entries to the case statement that marks unsigned arguments to
11993 overloaded functions.
11995 2017-03-10 Kelvin Nilsen <kelvin@gcc.gnu.org>
11997 * config/rs6000/rs6000.c (rs6000_option_override_internal): Fix
11998 two typographic errors in the handling of TARGET_UPPER_REGS_DI.
12000 2017-03-10 Pat Haugen <pthaugen@us.ibm.com>
12003 * config/rs6000/rs6000.c (rs6000_init_hard_regno_mode_ok): Test
12004 TARGET_UPPER_REGS_DI when setting 'wi' constraint regclass.
12006 2017-03-10 Martin Liska <mliska@suse.cz>
12010 * toplev.c (process_options): Enable MPX with LSAN and UBSAN.
12011 * tree-chkp.c (chkp_walk_pointer_assignments): Verify that
12014 2017-03-10 Olivier Hainque <hainque@adacore.com>
12016 * tree-switch-conversion (array_value_type): Start by resetting
12017 candidate type to it's main variant.
12019 2017-03-10 Jakub Jelinek <jakub@redhat.com>
12021 PR rtl-optimization/79909
12022 * combine.c (try_combine): Use simplify_replace_rtx on individual
12023 CALL_INSN_FUNCTION_USAGE elements instead of replace_rtx on copy_rtx
12024 of the whole CALL_INSN_FUNCTION_USAGE.
12026 PR tree-optimization/79972
12027 * gimple-ssa-warn-alloca.c (alloca_call_type): Only call
12028 get_range_info on SSA_NAMEs. Formatting fixes.
12030 2017-03-10 Richard Biener <rguenther@suse.de>
12031 Jakub Jelinek <jakub@redhat.com>
12033 PR tree-optimization/77975
12034 * tree-ssa-loop-niter.c (get_base_for): Allow phi argument from latch
12035 edge to be constant.
12036 (get_val_for): For constant x return it. Formatting fix.
12037 (loop_niter_by_eval): Avoid pointless looping if the next iteration
12038 would use the same bases as the current one.
12040 2017-03-09 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12042 * config/rs6000/rs6000.c (rs6000_gen_le_vsx_permute): Use rotate
12043 instead of vec_select for V1TImode.
12044 * conifg/rs6000/vsx.md (VSX_LE): Remove mode iterator that is no
12046 (VSX_LE_128): Add V1TI to this mode iterator.
12047 (*vsx_le_perm_load_<mode>): Change to use VSX_D mode iterator.
12048 (*vsx_le_perm_store_<mode>): Likewise.
12049 (pre-reload splitter for VSX stores): Likewise.
12050 (post-reload splitter for VSX stores): Likewise.
12051 (*vsx_xxpermdi2_le_<mode>): Likewise.
12052 (*vsx_lxvd2x2_le_<mode>): Likewise.
12053 (*vsx_stxvd2x2_le_<mode>): Likewise.
12055 2017-03-09 Michael Eager <eager@eagercon.com>
12057 Correct failures with --enable-checking=yes,rtl.
12059 * config/microblaze/microblaze.c (microblaze_expand_shift):
12060 Replace GET_CODE test with CONST_INT_P and INTVAL test with
12061 test for const0_rtx.
12062 * config/microblaze/microblaze.md (ashlsi3_byone, ashrsi3_byone,
12063 lshrsi3_byone): Replace INTVAL with test for const1_rtx.
12065 2017-03-09 Richard Biener <rguenther@suse.de>
12067 PR tree-optimization/79977
12068 * graphite-scop-detection.c (scop_detection::merge_sese):
12069 Handle the case of extra exits to blocks dominating the entry.
12071 2017-03-09 Toma Tabacu <toma.tabacu@imgtec.com>
12073 * doc/sourcebuild.texi (Effective-Target Keywords, Other attributes):
12076 2017-03-09 Vladimir Makarov <vmakarov@redhat.com>
12078 PR rtl-optimization/79949
12079 * lra-constraints.c (process_alt_operands): Check memory when
12080 trying to predict a cycle. Print about the overall increase.
12082 2017-03-09 Richard Biener <rguenther@suse.de>
12084 PR middle-end/79971
12085 * gimple-expr.c (useless_type_conversion_p): Preserve
12086 TYPE_SATURATING for fixed-point types.
12088 2017-03-09 Richard Biener <rguenther@suse.de>
12091 * ipa-prop.c (ipa_modify_formal_parameters): Avoid changing
12092 alignment of BLKmode params.
12094 2017-03-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12097 * config/aarch64/iterators.md (VALL_F16_NO_V2Q): New mode iterator.
12098 (VALL_NO_V2Q): Likewise.
12100 * config/aarch64/aarch64-simd.md
12101 (aarch64_dup_lane_<vswap_width_name><mode>): Use VALL_F16_NO_V2Q
12103 (*aarch64_simd_vec_copy_lane_<vswap_width_name><mode>): Use
12104 VALL_NO_V2Q mode iterator.
12105 (*aarch64_vgetfmulx<mode>): Use VDQF iterator.
12107 2017-03-09 Martin Liska <mliska@suse.cz>
12109 PR tree-optimization/79631
12110 * tree-chkp-opt.c (chkp_is_constant_addr): Call
12111 tree_int_cst_sign_bit just for INTEGER constants.
12113 2017-03-09 Martin Liska <mliska@suse.cz>
12117 * toplev.c (process_options): Disable -fcheck-pointer-bounds with
12120 2017-03-09 Marek Polacek <polacek@redhat.com>
12123 * tree.c (inchash::add_expr): Handle TREE_VEC.
12125 2017-03-09 Martin Liska <mliska@suse.cz>
12128 (chkp_narrow_size_and_offset): New function.
12129 (chkp_parse_array_and_component_ref): Support BIT_FIELD_REF.
12130 (void chkp_parse_bit_field_ref): New function.
12131 (chkp_make_addressed_object_bounds): Add case for BIT_FIELD_REF.
12132 (chkp_process_stmt): Use chkp_parse_bit_field_ref.
12134 2017-03-09 Martin Liska <mliska@suse.cz>
12137 * tree-chkp.c (chkp_get_bound_for_parm): Get bounds for a param.
12138 (chkp_find_bounds_1): Remove gcc_unreachable.
12140 2017-03-09 Jakub Jelinek <jakub@redhat.com>
12143 * asan.c (get_mem_refs_of_builtin_call): For BUILT_IN_ATOMIC* and
12144 BUILT_IN_SYNC*, determine the access type from the size suffix and
12145 always build a MEM_REF with that type. Handle forgotten
12146 BUILT_IN_SYNC_FETCH_AND_NAND_16 and BUILT_IN_SYNC_NAND_AND_FETCH_16.
12149 * config/i386/avx512vlintrin.h (_mm256_cmpge_epi32_mask,
12150 _mm256_cmpge_epi64_mask, _mm256_cmpge_epu32_mask,
12151 _mm256_cmpge_epu64_mask, _mm256_cmple_epi32_mask,
12152 _mm256_cmple_epi64_mask, _mm256_cmple_epu32_mask,
12153 _mm256_cmple_epu64_mask, _mm256_cmplt_epi32_mask,
12154 _mm256_cmplt_epi64_mask, _mm256_cmplt_epu32_mask,
12155 _mm256_cmplt_epu64_mask, _mm256_cmpneq_epi32_mask,
12156 _mm256_cmpneq_epi64_mask, _mm256_cmpneq_epu32_mask,
12157 _mm256_cmpneq_epu64_mask, _mm256_mask_cmpge_epi32_mask,
12158 _mm256_mask_cmpge_epi64_mask, _mm256_mask_cmpge_epu32_mask,
12159 _mm256_mask_cmpge_epu64_mask, _mm256_mask_cmple_epi32_mask,
12160 _mm256_mask_cmple_epi64_mask, _mm256_mask_cmple_epu32_mask,
12161 _mm256_mask_cmple_epu64_mask, _mm256_mask_cmplt_epi32_mask,
12162 _mm256_mask_cmplt_epi64_mask, _mm256_mask_cmplt_epu32_mask,
12163 _mm256_mask_cmplt_epu64_mask, _mm256_mask_cmpneq_epi32_mask,
12164 _mm256_mask_cmpneq_epi64_mask, _mm256_mask_cmpneq_epu32_mask,
12165 _mm256_mask_cmpneq_epu64_mask, _mm_cmpge_epi32_mask,
12166 _mm_cmpge_epi64_mask, _mm_cmpge_epu32_mask, _mm_cmpge_epu64_mask,
12167 _mm_cmple_epi32_mask, _mm_cmple_epi64_mask, _mm_cmple_epu32_mask,
12168 _mm_cmple_epu64_mask, _mm_cmplt_epi32_mask, _mm_cmplt_epi64_mask,
12169 _mm_cmplt_epu32_mask, _mm_cmplt_epu64_mask, _mm_cmpneq_epi32_mask,
12170 _mm_cmpneq_epi64_mask, _mm_cmpneq_epu32_mask, _mm_cmpneq_epu64_mask,
12171 _mm_mask_cmpge_epi32_mask, _mm_mask_cmpge_epi64_mask,
12172 _mm_mask_cmpge_epu32_mask, _mm_mask_cmpge_epu64_mask,
12173 _mm_mask_cmple_epi32_mask, _mm_mask_cmple_epi64_mask,
12174 _mm_mask_cmple_epu32_mask, _mm_mask_cmple_epu64_mask,
12175 _mm_mask_cmplt_epi32_mask, _mm_mask_cmplt_epi64_mask,
12176 _mm_mask_cmplt_epu32_mask, _mm_mask_cmplt_epu64_mask,
12177 _mm_mask_cmpneq_epi32_mask, _mm_mask_cmpneq_epi64_mask,
12178 _mm_mask_cmpneq_epu32_mask, _mm_mask_cmpneq_epu64_mask): Move
12179 definitions outside of __OPTIMIZE__ guarded section.
12182 * config/i386/avx512bwintrin.h (_mm512_packs_epi32,
12183 _mm512_maskz_packs_epi32, _mm512_mask_packs_epi32,
12184 _mm512_packus_epi32, _mm512_maskz_packus_epi32,
12185 _mm512_mask_packus_epi32): Move definitions outside of __OPTIMIZE__
12188 2017-03-09 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
12190 * config/s390/vx-builtins.md ("vfee<mode>", "vfeez<mode>")
12191 ("vfenez<mode>"): Add missing constraints.
12193 2017-03-08 Martin Sebor <msebor@redhat.com>
12196 * config/nds32/nds32.c (nds32_option_override):
12197 Fix misspelled diagnostic.
12199 2017-03-08 Jakub Jelinek <jakub@redhat.com>
12202 * gimplify.c (gimplify_omp_for): Replace index var in outer
12203 taskloop statement with an artificial variable and add
12204 OMP_CLAUSE_PRIVATE clause for it.
12206 2017-03-08 Richard Biener <rguenther@suse.de>
12208 PR tree-optimization/79955
12209 * tree-ssa-uninit.c (warn_uninitialized_vars): Do not warn
12210 for accesses that are completely outside of the variable.
12212 2017-03-08 Andrew Haley <aph@redhat.com>
12214 PR tree-optimization/79943
12215 * tree-ssa-loop-split.c (compute_new_first_bound): When
12216 calculating the new upper bound, (END-BEG) should be added, not
12219 2017-03-08 Jakub Jelinek <jakub@redhat.com>
12221 * config/avr/avr.md (setmemhi): Make sure match_dup
12222 operand number comes before match_scratch.
12224 2017-03-08 Richard Biener <rguenther@suse.de>
12226 PR tree-optimization/79920
12227 * tree-vect-slp.c (vect_create_mask_and_perm): Remove and inline
12228 with ncopies == 1 to ...
12229 (vect_transform_slp_perm_load): ... here. Properly compute
12230 all element loads by iterating VF times over the group. Do
12231 not handle ncopies (computed in a broken way) in
12232 vect_create_mask_and_perm.
12234 2017-03-08 Jakub Jelinek <jakub@redhat.com>
12237 * internal-fn.c (expand_vector_ubsan_overflow): If arg0 or arg1
12238 is a uniform vector, use uniform_vector_p return value instead of
12239 building ARRAY_REF on folded VIEW_CONVERT_EXPR to array type.
12241 2017-03-07 Marek Polacek <polacek@redhat.com>
12243 PR middle-end/79809
12244 * gimple-ssa-warn-alloca.c (pass_walloca::gate): Use HOST_WIDE_INT.
12245 (alloca_call_type): Likewise.
12247 2017-03-07 Martin Liska <mliska@suse.cz>
12249 * gcov.c (process_args): Put comment to correct location.
12251 2017-03-07 Martin Liska <mliska@suse.cz>
12253 PR middle-end/68270
12254 * tree-chkp.c (chkp_may_narrow_to_field): Add new argument ref.
12255 Use array_at_struct_end_p instead of DECL_CHAIN (field).
12256 (chkp_narrow_bounds_for_field): Likewise.
12257 (chkp_parse_array_and_component_ref): Pass one more argument to
12260 2017-03-07 Richard Biener <rguenther@suse.de>
12262 * tree-vect-loop-manip.c (slpeel_add_loop_guard): Preserve
12265 2017-03-07 Segher Boessenkool <segher@kernel.crashing.org>
12267 * config/i386/i386.c (ix86_local_alignment): Align most aggregates
12268 of 16 bytes and more to 16 bytes, not those of 16 bits and more.
12270 2017-03-07 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12273 * params.def (PARAM_STORE_MERGING_ALLOW_UNALIGNED): Add full stop
12274 to end of description.
12275 (PARAM_MAX_STORES_TO_MERGE): Likewise.
12277 2017-03-07 Jakub Jelinek <jakub@redhat.com>
12279 PR rtl-optimization/79901
12280 * config/i386/sse.md (*avx512bw_<code><mode>3<mask_name>): Renamed to
12282 (*avx512f_<code><mode>3<mask_name>): ... this.
12283 (<code><mode>3 with maxmin code iterator): Use VI8_AVX2_AVX512F
12284 iterator instead of VI8_AVX2_AVX512BW.
12286 PR rtl-optimization/79901
12287 * expr.c (expand_expr_real_2): For vector MIN/MAX, if there is no
12288 min/max expander, expand it using expand_vec_cond_expr.
12291 * ubsan.c (ubsan_encode_value): Call mark_addressable on the
12294 2017-03-06 Jakub Jelinek <jakub@redhat.com>
12297 * dwarf2out.h (dw_vec_const): Change array type from unsigned char *
12298 to void * for PCH reasons.
12299 * dwarf2out.c (output_loc_operands, output_die): Cast
12300 v.val_vec.array to unsigned char *.
12302 2017-03-06 John David Anglin <danglin@gcc.gnu.org>
12305 * config/pa/pa-64.h (PAD_VARARGS_DOWN): Don't pad down complex and
12308 2017-03-06 Vladimir Makarov <vmakarov@redhat.com>
12310 PR rtl-optimization/79571
12311 * lra-constraints.c (process_alt_operands): Calculate static
12312 reject and subtract it from overall when only addresses will be
12315 2017-03-06 Julia Koval <julia.koval@intel.com>
12318 * config/i386/i386.c (ix86_minimum_incoming_stack_boundary): Set
12319 incoming stack boundary to 128 for 64-bit targets.
12321 2017-03-06 Richard Biener <rguenther@suse.de>
12323 PR tree-optimization/79894
12324 * tree-vectorizer.c (vectorize_loops): Set loop_vectorized_call
12325 to NULL after folding it.
12327 2017-03-06 Richard Biener <rguenther@suse.de>
12329 PR tree-optimization/79824
12330 * tree-vect-stmts.c (get_group_load_store_type): Fix alignment
12331 check disabling peeling for gaps.
12333 2017-03-06 Toma Tabacu <toma.tabacu@imgtec.com>
12335 * doc/sourcebuild.texi (Effective-Target Keywords, Environment
12336 attributes): Document gettimeofday.
12338 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
12340 * config/s390/s390.c (s390_option_override_internal): Set
12341 PARAM_MIN_VECT_LOOP_BOUND
12343 2017-03-06 Robin Dapp <rdapp@linux.vnet.ibm.com>
12345 * config/s390/s390.c (s390_asm_output_function_label): Use nopr %r0.
12346 * config/s390/s390.md: Likewise.
12348 2017-03-06 Jakub Jelinek <jakub@redhat.com>
12351 * config/i386/sse.md (VI8F_256_512): Remove mode iterator.
12352 (<avx2_avx512>_perm<mode>): Rename to ...
12353 (avx2_perm<mode>): ... this. Use VI8F_256 iterator instead
12355 (<avx512>_perm<mode>_mask): Rename to ...
12356 (avx512vl_perm<mode>_mask): ... this. Use VI8F_256 iterator instead
12358 (<avx2_avx512>_perm<mode>_1<mask_name>): Rename to ...
12359 (avx2_perm<mode>_1<mask_name): ... this. Use VI8F_256 iterator
12360 instead of VI8F_256_512.
12361 (avx512f_perm<mode>): New define_expand.
12362 (avx512f_perm<mode>_mask): Likewise.
12363 (avx512f_perm<mode>_1<mask_name>): New define_insn.
12364 (<avx512>_vec_dup<mode>_1): Fix up vec_select mode.
12366 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
12368 * config/mips/mips-msa.md (msa_fmax_a_<msafmt>, msa_fmin_a_<msafmt>,
12369 msa_max_a_<msafmt>, msa_min_a_<msafmt>): Introduce mode interator for
12371 (smin<mode>3, smax<mode>3): Change operand print code from 'B' to 'E'.
12373 2017-03-06 Martin Liska <mliska@suse.cz>
12376 * asan.c (asan_expand_poison_ifn): Do not expand ASAN_POISON
12377 when having a SSA NAME w/o VAR_DECL assigned to it.
12379 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
12381 * config/mips/mips-msa.md (msa_dotp_<su>_d, msa_dpadd_<su>_d,
12382 msa_dpsub_<su>_d): Fix MODE for vec_select.
12384 2017-03-06 Prachi Godbole <prachi.godbole@imgtec.com>
12386 * config/mips/mips.c (mips_gen_const_int_vector): Change type of last
12388 * config/mips/mips-protos.h (mips_gen_const_int_vector): Likewise.
12390 2017-03-06 Richard Biener <rguenther@suse.de>
12392 * lto-streamer.c (lto_check_version): Use %qs in diagnostics.
12393 * plugin.c (register_plugin_info): Likewise.
12394 * tree-chkp.c (chkp_make_static_const_bounds): Likewise.
12396 2017-03-05 Jakub Jelinek <jakub@redhat.com>
12398 * config/i386/sse.md (sse_storehps, sse_storelps,
12399 avx_<castmode><avxsizesuffix>_<castmode>,
12400 avx512f_<castmode><avxsizesuffix>_<castmode>,
12401 avx512f_<castmode><avxsizesuffix>_256<castmode>): Require
12402 in condition that at least one operand is not a MEM.
12404 2017-03-03 Jakub Jelinek <jakub@redhat.com>
12406 PR middle-end/79805
12407 * internal-fn.def (ATOMIC_BIT_TEST_AND_SET, ATOMIC_BIT_TEST_AND_RESET,
12408 ATOMIC_BIT_TEST_AND_COMPLEMENT, ATOMIC_COMPARE_EXCHANGE): Remove
12410 * gimple-fold.c (fold_builtin_atomic_compare_exchange): Set
12411 gimple_call_nothrow_p flag based on whether original builtin can throw.
12412 If it can, emit following stmts on the fallthrough edge.
12413 * tree-ssa-ccp.c (optimize_atomic_bit_test_and): Similarly, except
12414 don't create new bb if inserting just debug stmts on the edge, try to
12415 insert them on the fallthru bb or just reset debug stmts.
12417 2017-03-03 Segher Boesssenkool <segher@kernel.crashing.org>
12420 * config/rs6000/rs6000.c (rs6000_final_prescan_insn): Save and
12421 restore recog_data (including the operand rtxes inside it) around
12422 the call to get_insn_template.
12424 2017-03-03 Martin Sebor <msebor@redhat.com>
12426 PR tree-optimization/79699
12427 * context.c (context::~context): Free MPFR caches to avoid
12428 a memory leak on program exit.
12430 2017-03-03 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
12432 * config/aarch64/aarch64.c (aarch64_float_const_representable_p):
12433 Use wide_int::ulow () instead of .elt (0).
12435 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
12437 * config/i386/i386.md (*pushtf): Change *roF constraint to *roC.
12438 (*pushxf): Limit oF constraint to 32bit targets and add oC
12439 constraint for 64bit targets.
12440 (pushxf splitter): Use PUSH_ROUNDING to calculate stack adjustment.
12441 (*pushdf): Change rmF constraint to rmC.
12443 2017-03-03 Martin Liska <mliska@suse.cz>
12445 * tree-ssa-loop-prefetch.c (pass_loop_prefetch::execute):
12446 Remove unused variable.
12448 2017-03-03 Jakub Jelinek <jakub@redhat.com>
12451 * config/i386/i386.c (ix86_expand_multi_arg_builtin): If target
12452 is a memory operand, increase num_memory.
12453 (ix86_expand_args_builtin): Likewise.
12455 2017-03-03 Jan Hubicka <jh@suse.cz>
12458 * ipa-devirt.c (maybe_record_node): Properly handle
12459 __cxa_pure_virtual visibility.
12461 2017-03-03 Martin Liska <mliska@suse.cz>
12463 PR tree-optimization/79803
12464 * tree-ssa-loop-prefetch.c (tree_ssa_prefetch_arrays): Remove
12466 (pass_loop_prefetch::execute): Disabled optimization if an
12467 assumption about L1 cache size is not met.
12469 2017-03-03 Martin Liska <mliska@suse.cz>
12471 PR rtl-optimization/79574
12472 * gcse.c (struct gcse_expr): Use HOST_WIDE_INT instead of int.
12473 (hash_scan_set): Likewise.
12474 (dump_hash_table): Likewise.
12475 (hoist_code): Likewise.
12477 2017-03-03 Richard Biener <rguenther@suse.de>
12479 * fixed-value.c (fixed_from_string): Restore use of elt (1)
12480 in place of uhigh ().
12481 (fixed_convert_from_real): Likewise.
12483 2017-03-03 Uros Bizjak <ubizjak@gmail.com>
12486 * config/i386/i386.md (*pushxf_rounded): Use Pmode instead of DImode.
12488 2017-03-03 Richard Biener <rguenther@suse.de>
12490 PR middle-end/79818
12491 * match.pd ( X +- C1 CMP C2 -> X CMP C2 -+ C1): Add missing
12492 TYPE_OVERFLOW_UNDEFINED check.
12494 2017-03-02 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12496 * config/rs6000/vector.md (vector_ne_<mode>_p): Correct operand
12498 (vector_ae_<mode>_p): Likewise.
12499 (vector_nez_<mode>_p): Likewise.
12500 (vector_ne_v2di_p): Likewise.
12501 (vector_ae_v2di_p): Likewise.
12502 (vector_ne_<mode>_p): Likewise.
12503 * config/rs6000/vsx.md (vsx_tsqrt<mode>2_fg): Correct operand
12505 (vsx_tsqrt<mode>2_fe): Likewise.
12507 2017-03-02 Uros Bizjak <ubizjak@gmail.com>
12510 * config/i386/i386.md (*pushxf_rounded): New insn_and_split pattern.
12512 2017-03-02 Jakub Jelinek <jakub@redhat.com>
12514 PR rtl-optimization/79780
12515 * cprop.c (one_cprop_pass): When second and further conditional trap
12516 in a single basic block is turned into an unconditional trap, turn it
12517 into a deleted note to avoid RTL verification failures.
12519 2017-03-02 Richard Biener <rguenther@suse.de>
12521 * fold-const.c (const_binop): Use ulow () instead of elt (0).
12523 2017-03-02 Richard Biener <rguenther@suse.de>
12525 PR tree-optimization/79345
12527 * tree-ssa-alias.c (walk_aliased_vdefs_1): Take a limit
12528 param and abort the walk, returning -1 if it is hit.
12529 (walk_aliased_vdefs): Take a limit param and pass it on.
12530 * tree-ssa-alias.h (walk_aliased_vdefs): Add a limit param,
12531 defaulting to 0 and return a signed int.
12532 * tree-ssa-uninit.c (struct check_defs_data): New struct.
12533 (check_defs): New helper.
12534 (warn_uninitialized_vars): Use walk_aliased_vdefs to warn
12535 about uninitialized memory.
12536 * fixed-value.c (fixed_from_string): Use ulow/uhigh to avoid
12537 bogus uninitialized warning.
12538 (fixed_convert_from_real): Likewise.
12540 2017-03-02 Bin Cheng <bin.cheng@arm.com>
12542 PR tree-optimization/66768
12543 * tree-ssa-loop-ivopts.c (find_interesting_uses_address): Skip addr
12544 iv_use if base object can't be determined.
12546 2017-03-02 Jakub Jelinek <jakub@redhat.com>
12548 PR tree-optimization/79345
12549 * gensupport.h (struct pattern_stats): Add min_scratch_opno field.
12550 * gensupport.c (get_pattern_stats_1) <case MATCH_SCRATCH>: Update it.
12551 (get_pattern_stats): Initialize it.
12552 * genemit.c (gen_expand): Verify match_scratch numbers come after
12553 match_operand/match_dup numbers.
12554 * config/i386/i386.md (<s>mul<mode>3_highpart): Swap match_dup and
12555 match_scratch numbers.
12556 * config/i386/sse.md (avx2_gathersi<mode>, avx2_gatherdi<mode>):
12558 * config/s390/s390.md (trunctdsd2): Likewise.
12560 2017-03-02 Richard Biener <rguenther@suse.de>
12562 * wide-int.h (wide_int_storage::operator=): Implement in terms
12565 2017-03-02 Richard Biener <rguenther@suse.de>
12567 PR tree-optimization/79777
12568 * tree-ssa-pre.c (eliminate_insert): Give up if we simplify
12569 the to insert expression to sth existing.
12571 2017-03-01 Martin Sebor <msebor@redhat.com>
12573 PR middle-end/79692
12574 * gimple-ssa-sprintf.c
12575 (directive::known_width_and_precision): New function.
12576 (format_integer): Use it.
12577 (get_mpfr_format_length): Consider the full range of precision
12578 when computing %g output with the # flag. Set the likely byte
12579 count to 3 rather than 1 when precision is indeterminate.
12580 (format_floating): Correct the lower bound of precision.
12582 2017-03-01 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
12584 * doc/invoke.texi: Document default code model for 64-bit Linux.
12586 2017-03-01 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
12589 * config/rs6000/rs6000.md (peephole2 for udiv/umod): Should emit
12590 udiv rather than div since input pattern is unsigned.
12592 2017-03-01 Uros Bizjak <ubizjak@gmail.com>
12594 * config/i386/i386.c (print_reg): Warn for values of
12595 unsupported size in integer register.
12597 2017-03-01 Michael Meissner <meissner@linux.vnet.ibm.com>
12600 * config/rs6000/predicates.md (current_file_function_operand): Do
12601 not allow self calls to be local if the function is replaceable.
12603 2017-03-01 Kelvin Nilsen <kelvin@gcc.gnu.org>
12606 * config/rs6000/altivec.h (vec_ctz and others): Change the
12607 preprocessor macro that controls conditional compilation from
12608 _ARCH_PWR9 to __POWER9_VECTOR__.
12609 (vec_all_ne): Change parameterization of __altivec_scalar_pred
12610 macro expansion under preprocessor #ifdef __POWER9_VECTOR__
12611 control (instead of _ARCH_PWR9 control) so that template
12612 definition uses power9-specific function.
12613 (vec_any_eq): Likewise.
12614 (vec_all_ne): Change macro definition to use a power9-specific
12615 expansion under #ifdef __POWER9_VECTOR__ control (instead of
12616 _ARCH_PWR9 control).
12617 (vec_any_eq) Likewise.
12618 * config/rs6000/rs6000-builtin.def (CMPNEF): Remove BU_P9V_AV_2
12619 expansion for CMPNEF to remove support for xvcmpnesp instruction.
12620 (CMPNED): Remove BU_P9V_AV2 expansion for CMPNED to remove
12621 support for xvcmpnedp instruction.
12622 (VCMPNEB_P): Replace BU_P9V_AV_P macro expansion with BU_P9V_AV_2
12623 macro expansion so that Power9 implementation of vec_all_ne does
12624 not use the AltiVec predicate framework.
12625 (VCMPNEH_P): Likewise.
12626 (VCMPNEW_P): Likewise.
12627 (VCMPNED_P): Likewise.
12628 (VCMPNEFP_P): Likewise.
12629 (VCMPNEDP_P): Likewise.
12630 (VCMPAEB_P): Add BU_P9V_AV_2 macro expansion to change
12631 implementation of vec_any_eq to not use AltiVec predicate
12633 (VCMPAEH_P): Likewise.
12634 (VCMPAEW_P): Likewise.
12635 (VCMPAED_P): Likewise.
12636 (VCMPAEFP_P): Likewise.
12637 (VCMPAEDP_P): Likewise.
12638 (VCMPNE_P): Replace BU_P9V_OVERLOAD_P macro expansion with
12639 BU_P9V_OVERLOAD_2 so that Power9 implementation of vec_all_ne does
12640 not use the AltiVec predicate framework.
12641 (VCMPAE_P): Add BU_P9V_OVERLOAD_2 macro to change implementation
12642 of vec_any_eq to not use AltiVec predicate framework.
12643 * config/rs6000/rs6000-c.c (rs6000_target_modify_macros): Add
12644 support for predefined __POWER9_VECTOR__ macro to indicate that
12645 Power9 instruction selection is enabled.
12646 (altivec_overloaded_builtins): Remove extraneous
12647 ALTIVEC_BUILTIN_VEC_CMPNE entry for overloaded
12648 function argument types RS6000_BTI_bool_V16QI and
12649 RS6000_BTI_bool_V16QI. Remove erroneous ALTIVEC_BUILTIN_VEC_CMPNE
12650 entry for overloaded function argument types RS6000_BTI_bool_V4SI
12651 andRS6000_BTI_bool_V4SI, mapping to P9V_BUILTIN_CMPNEB. Remove
12652 two entries mapping to P9V_BUITIN_CMPNED and one entry mapping to
12653 P9V_BUILTIN_CMPNEF to force use of instructions not specific to
12654 Power9 for implementations of vec_cmpne. Change the signature for
12655 all definitions of the overloaded P9V_BUILTIN_VEC_CMPNE_P function
12656 (representing vec_all_ne) to remove the previously described first
12657 argument of type RS6000_BTI_INTSI, as this was an artifact of
12658 reliance on the AltiVec predicate framework, which is no longer
12659 used in the implementation of these functions. Add
12660 P9V_BUILTIN_VEC_VCMPAE_P entries (representing the vec_anyeq
12661 function) to match all of the P9V_BUILTIN_VEC_VCMNE_P entries
12662 since, unlike the AltiVec predicate framework implementation, we
12663 do not share function descriptors between vec_alle and vec_anyeq.
12664 (altivec_resolve_overloaded_builtin): Add SFmode and DFmode to the
12665 set of modes that receive special treatment even when
12666 TARGET_P9_VECTOR is true. The special treatment emits code that
12667 does not depend on Power9 instructions.
12668 * config/rs6000/vector.md (vector_ne_<mode>_p): Change this
12669 define_expand to not rely on AltiVec predicate framework.
12670 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
12672 (vector_ne_v2di_p): Change this define_expand to not rely on
12673 AltiVec predicate framework.
12674 (vector_ae_v2di_p): New define_expand to represent vec_any_eq
12676 (vector_ne_<mode>_p): Change this define_expand to not rely on
12677 AltiVec predicate framework.
12678 (vector_ae_<mode>p): New define_expand to represent vec_any_eq
12680 * config/rs6000/vsx.md (*vsx_ne_<mode>_p): For modes VSX_EXTRACT_I
12681 (V16QI, V8HI, V4SI), correct a typo in the code emitted for this
12682 define_insn pattern.
12683 (*vsx_ne_<mode>_p): For modes VSX_F (V4SF and V2DF), remove this
12684 define_insn pattern because the xvcmpne<VSs>. instruction is not
12686 (vcmpne<VSs>): Remove this define_insn because xvcmpne<VSs>
12687 instruction is not supported.
12689 2017-03-01 Jakub Jelinek <jakub@redhat.com>
12691 * config/nvptx/nvptx.c: Include intl.h.
12693 2017-03-01 Martin Jambor <mjambor@suse.cz>
12696 * ipa-prop.h (ipa_bits): Removed field known.
12697 (ipa_jump_func): Removed field vr_known. Changed fields bits and m_vr
12698 to pointers. Adjusted their comments to warn about their sharing.
12699 (ipcp_transformation_summary): Change bits to a vector of pointers.
12700 (ipa_check_create_edge_args): Moved to ipa-prop.c, declare.
12701 (ipa_get_ipa_bits_for_value): Declare.
12702 * tree-vrp.h (value_range): Mark as GTY((for_user)).
12703 * ipa-prop.c (ipa_bit_ggc_hash_traits): New.
12704 (ipa_bits_hash_table): Likewise.
12705 (ipa_vr_ggc_hash_traits): Likewise.
12706 (ipa_vr_hash_table): Likewise.
12707 (ipa_print_node_jump_functions_for_edge): Adjust for bits and m_vr
12708 being pointers and vr_known being removed.
12709 (ipa_set_jf_unknown): Likewise.
12710 (ipa_get_ipa_bits_for_value): New function.
12711 (ipa_set_jfunc_bits): Likewise.
12712 (ipa_get_value_range): New overloaded functions.
12713 (ipa_set_jfunc_vr): Likewise.
12714 (ipa_compute_jump_functions_for_edge): Use the above functions to
12715 construct bits and vr parts of jump functions.
12716 (ipa_check_create_edge_args): Move here from ipa-prop.h, also allocate
12717 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
12719 (ipcp_grow_transformations_if_necessary): Also allocate
12720 ipa_bits_hash_table and ipa_vr_hash_table if they do not already
12722 (ipa_node_params_t::duplicate): Do not copy bits, just pointers to
12723 them. Fix too long lines.
12724 (ipa_write_jump_function): Adjust for bits and m_vr being pointers and
12725 vr_known being removed.
12726 (ipa_read_jump_function): Use new setter functions to construct bits
12727 and vr parts of jump functions or set them to NULL.
12728 (write_ipcp_transformation_info): Adjust for bits being pointers.
12729 (read_ipcp_transformation_info): Likewise.
12730 (ipcp_update_bits): Likewise. Fix excessively long lines a trailing
12732 Include gt-ipa-prop.h.
12733 * ipa-cp.c (propagate_bits_across_jump_function): Adjust for bits
12735 (ipcp_store_bits_results): Likewise.
12736 (propagate_vr_across_jump_function): Adjust for m_vr being a pointer.
12737 Do not write to existing jump functions but use a temporary instead.
12739 2017-03-01 Jakub Jelinek <jakub@redhat.com>
12742 * fold-const.c (make_bit_field_ref): If orig_inner is COMPONENT_REF,
12743 attempt to use its first operand as BIT_FIELD_REF base.
12745 2017-03-01 Richard Biener <rguenther@suse.de>
12747 PR middle-end/79721
12748 * tree-chrec.c (chrec_evaluate): Perform computation of Newtons
12749 interpolating formula in wrapping arithmetic.
12750 (chrec_apply): Convert chrec_evaluate return value to wanted type.
12752 2017-03-01 Jakub Jelinek <jakub@redhat.com>
12754 PR tree-optimization/79734
12755 * tree-vect-generic.c (expand_vector_condition): Optimize
12756 AVX512 vector boolean VEC_COND_EXPRs into bitwise operations.
12757 Handle VEC_COND_EXPR where comparison has different inner width from
12758 type's inner width.
12760 2017-02-28 Sandra Loosemore <sandra@codesourcery.com>
12762 * doc/invoke.texi (ARC Options): Copy-edit to fix punctuation,
12763 markup, and similar issues. Remove @opindex entries for things
12764 that aren't options. Add missing -mmpy-option entries.
12766 2017-02-28 Jakub Jelinek <jakub@redhat.com>
12768 PR tree-optimization/79737
12769 * gimple-ssa-store-merging.c (encode_tree_to_bitpos): If bitlen is
12770 a multiple of BITS_PER_UNIT and !BYTES_BIG_ENDIAN, clear
12771 tmpbuf[byte_size - 1]. Call natice_encode_expr with byte_size - 1
12772 instead of byte_size. Formatting fix.
12773 (shift_bytes_in_array_right): Formatting fix.
12775 2017-02-28 Eric Botcazou <ebotcazou@adacore.com>
12778 * config/sparc/sparc.c (sparc_frame_pointer_required): Add missing
12779 condition on optimize for the leaf function test.
12781 2017-02-28 Martin Liska <mliska@suse.cz>
12784 * read-rtl-function.c (function_reader::handle_unknown_directive):
12785 Bail out when one uses -flto.
12787 2017-02-28 Martin Liska <mliska@suse.cz>
12789 * common.opt: Replace space with tabular for options of <number>
12791 * config/i386/i386.opt: Show <number> value for
12792 -mlarge-data-threshold.
12793 * opts.c (print_filtered_help): Do not display number in hexadecimal
12796 2017-02-28 Martin Liska <mliska@suse.cz>
12798 * common.opt: Fix --help=option -Q for options which are of
12801 2017-02-28 Uros Bizjak <ubizjak@gmail.com>
12803 * config/i386/i386.c (print_reg): Error out for values
12804 of 8-bit size in invalid integer register.
12806 2017-02-28 Martin Sebor <msebor@redhat.com>
12808 PR tree-optimization/79691
12809 * passes.def (pass_all_optimizations_g): Enable pass_sprintf_length.
12811 2017-02-28 Jakub Jelinek <jakub@redhat.com>
12814 * config/i386/i386.c (ix86_print_operand) <case 'R'>: Replace
12815 gcc_unreachable with output_operand_lossage.
12817 2017-02-28 Richard Biener <rguenther@suse.de>
12819 PR tree-optimization/79740
12820 * tree-ssa-sccvn.c (vn_nary_op_insert_into): Allow redundant
12822 (visit_nary_op): Insert the nary into the hashtable if we
12823 pattern-matched sth.
12824 * tree-ssa-pre.c (eliminate_insert): Robustify.
12826 2017-02-28 Richard Biener <rguenther@suse.de>
12828 PR middle-end/79731
12829 * fold-const.c (decode_field_reference): Reject out-of-bound
12832 2017-02-28 Jakub Jelinek <jakub@redhat.com>
12834 * config/i386/i386.c: Include intl.h.
12835 (ix86_option_override_internal): Use cond ? G_("...") : G_("...")
12836 instead of just cond ? "..." : "...".
12837 * config/nvptx/nvptx.c (nvptx_goacc_validate_dims): Likewise.
12838 * coverage.c (read_counts_file): Likewise.
12839 * omp-offload.c: Include intl.h.
12840 (oacc_loop_fixed_partitions): Use cond ? G_("...") : G_("...") instead
12841 of just cond ? "..." : "...".
12842 * gcov.c (read_count_file): Use cond ? N_("...") : N_("...") instead
12843 of just cond ? "..." : "...".
12845 2017-02-28 Richard Earnshaw <rearnsha@arm.com>
12848 * config/arm/parsecpu.awk (gen_data): Set tuning target to 'tune for'
12850 * config/arm/arm-cpus.in (cortex-m0plus.small-multiply): Correct
12851 'tune for' CPU name.
12852 * config/arm/arm-cpu-data.h: Regenerated.
12854 2017-02-28 Richard Biener <rguenther@suse.de>
12856 PR tree-optimization/79732
12857 * tree-inline.c (expand_call_inline): Do not shadow var.
12859 2017-02-28 Richard Biener <rguenther@suse.de>
12861 PR tree-optimization/79723
12862 * tree-vect-stmts.c (get_vectype_for_scalar_type_and_size): Preserve
12863 address-space properly.
12865 2017-02-28 Thomas Schwinge <thomas@codesourcery.com>
12867 * doc/optinfo.texi (Optimization groups): Fix option used for
12869 * doc/invoke.texi (-fopt-info): Document "omp".
12870 * dumpfile.h: Sort OPTGROUP_OMP before OPTGROUP_VEC.
12871 (OPTGROUP_ALL): Add OPTGROUP_OMP.
12872 * hsa-gen.c (pass_data_gen_hsail): Use OPTGROUP_OMP.
12873 * ipa-hsa.c (pass_data_ipa_hsa): Likewise.
12874 * omp-simd-clone.c (pass_data_omp_simd_clone): Likewise.
12876 * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
12878 * dumpfile.c (optgroup_options): Instead of "openmp", associate
12879 OPTGROUP_OMP with "omp".
12881 2017-02-27 Pat Haugen <pthaugen@us.ibm.com>
12884 * config/rs6000/rs6000-c.c (struct altivec_builtin_types): Use VSRAD
12885 for arithmetic shift of unsigned V2DI.
12887 2017-02-27 Claudiu Zissulescu <claziss@synopsys.com>
12889 * config.gcc (arc*-): Clean up, use arc/big.h, arc/elf.h, and
12890 arc/linux.h headers.
12891 * config/arc/arc.h (TARGET_OS_CPP_BUILTINS): Remove.
12892 (LINK_SPEC): Likewise.
12893 (ARC_TLS_EXTRA_START_SPEC): Likewise.
12894 (EXTRA_SPECS): Likewise.
12895 (STARTFILE_SPEC): Likewise.
12896 (ENDFILE_SPEC): Likewise.
12897 (LIB_SPEC): Likewise.
12898 (TARGET_SDATA_DEFAULT): Likewise.
12899 (TARGET_MMEDIUM_CALLS_DEFAULT): Likewise.
12900 (MULTILIB_DEFAULTS): Likewise.
12901 (DWARF2_UNWIND_INFO): Likewise.
12902 * config/arc/big.h: New file.
12903 * config/arc/elf.h: Likewise.
12904 * config/arc/linux.h: Likewise.
12905 * config/arc/t-uClibc: Remove.
12907 2017-02-27 Bin Cheng <bin.cheng@arm.com>
12909 PR tree-optimization/77536
12910 * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
12911 (tree_transform_and_unroll_loop): Use above function to compute the
12912 estimated niter of unrolled loop and use it when scaling profile.
12913 Also use count info rather than frequency if it's non-zero.
12914 * tree-ssa-loop-manip.h niter_for_unrolled_loop(): New declaration.
12915 * tree-vect-loop.c (scale_profile_for_vect_loop): New function.
12916 (vect_transform_loop): Call above function.
12918 2017-02-27 Richard Biener <rguenther@suse.de>
12920 PR tree-optimization/45397
12921 * tree-ssa-pre.c (eliminate_insert): Handle BIT_AND_EXPR.
12922 * tree-ssa-sccvn.c (valueized_wider_op): New helper.
12923 (visit_nary_op): Add pattern matching for CSEing sign-changed
12924 or truncated operations with wider ones.
12926 2017-02-27 Richard Biener <rguenther@suse.de>
12928 PR tree-optimization/79690
12929 * tree-vect-stmts.c (vectorizable_store): Use vector type
12930 built from the DR with address-space.
12932 2017-02-26 Gerald Pfeifer <gerald@pfeifer.com>
12934 * doc/invoke.texi (Optimize Options): Refine the description
12935 of asan-use-after-return.
12937 2017-02-25 Alan Modra <amodra@gmail.com>
12939 PR rtl-optimization/79584
12940 * lra-constraints.c (base_to_reg): Reload ad->base, the entire
12941 base, not ad->base_term, the reg within base. Remove assertion
12942 that ad->base == ad->base_term. Replace gen_int_mode using
12943 bogus mode with const0_rtx.
12945 2017-02-25 Jakub Jelinek <jakub@redhat.com>
12947 PR middle-end/79396
12948 * tree-eh.c (operation_could_trap_p, stmt_could_throw_1_p): Handle
12949 FMA_EXPR like tcc_binary or tcc_unary.
12951 * tree-ssa-loop-niter.c (number_of_iterations_exit): Simplify warning.
12954 * dwarf2out.c (struct dw_loc_list_struct): Add noted_variable_value
12956 (size_of_loc_descr): Handle DW_OP_GNU_variable_value.
12957 (output_loc_operands): Handle DW_OP_call_ref and
12958 DW_OP_GNU_variable_value.
12959 (struct variable_value_struct): New type.
12960 (struct variable_value_hasher): Likewise.
12961 (variable_value_hash): New variable.
12962 (string_types): Remove.
12963 (copy_loc_descr): New function.
12964 (add_loc_descr_to_each): Clarify comment. Use copy_loc_descr.
12965 (prepend_loc_descr_to_each): New function.
12966 (add_loc_list): Fix comment typo. Use prepend_loc_descr_to_each
12967 instead of add_loc_descr_to_each if the first argument is single
12968 location list and the second has multiple.
12969 (resolve_args_picking_1): Handle DW_OP_GNU_variable_value.
12970 (loc_list_from_tree_1): For early_dwarf, emit DW_OP_GNU_variable_value
12971 when looking for variable value which doesn't have other location info.
12972 (loc_list_from_tree): Formatting fix.
12973 (gen_array_type_die): Simplify DW_AT_string_length handling.
12974 (adjust_string_types): Remove.
12975 (gen_subprogram_die): Don't call adjust_string_types nor test/set
12976 string_types. Call resolve_variable_values.
12977 (prune_unused_types_walk_loc_descr): Handle DW_OP_GNU_variable_value.
12978 (resolve_addr_in_expr): Likewise. Add A argument.
12979 (copy_deref_exprloc): Remove deref argument. Adjust for the
12980 original expression being DW_OP_GNU_variable_value with optionally
12981 DW_OP_stack_value after it instead of DW_OP_call4 with DW_OP_deref
12982 optionally after it.
12983 (optimize_string_length): Rework for DW_OP_GNU_variable_value.
12984 (resolve_addr): Adjust optimize_string_length and resolve_addr_in_expr
12985 callers. Set remove_AT_byte_size if removing DW_AT_string_length.
12986 (variable_value_hasher::hash, variable_value_hasher::equal): New
12988 (resolve_variable_value_in_expr, resolve_variable_value,
12989 resolve_variable_values, note_variable_value_in_expr,
12990 note_variable_value): New functions.
12991 (dwarf2out_early_finish): Call note_variable_value on all toplevel
12994 2017-02-24 Jakub Jelinek <jakub@redhat.com>
12997 * opts.h (handle_generated_option): Add GENERATED_P argument.
12998 * opts-common.c (handle_option): Adjust function comment.
12999 (handle_generated_option): Add GENERATED_P argument, pass it to
13001 (control_warning_option): Pass false to handle_generated_option
13003 * opts.c (maybe_default_option): Pass true to handle_generated_option
13005 * optc-gen.awk: Likewise.
13007 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
13009 * config/sh/sh.md (tstsi_t): If operands[0] is a SUBREG instead of
13010 a REG, look at the REG it is a SUBREG of.
13011 (splitter for cmpeqsi_t): Ditto.
13013 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
13015 * config/pa/pa.c (pa_combine_instructions): Do not share RTL. Make
13016 the special USEs with the pattern of the insn, not the insn itself.
13018 2017-02-24 Matthew Fortune <matthew.fortune@imgtec.com>
13021 * doc/invoke.texi: Document -mload-store-pairs.
13023 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
13024 Sandra Loosemore <sandra@codesourcery.com>
13026 * config/nios2/nios2.c (nios2_simple_const_p): Returns false if the
13027 argument isn't a CONST_INT.
13028 (nios2_alternate_compare_const): Assert op is a CONST_INT.
13029 (nios2_valid_compare_const_p): Assert op is a CONST_INT.
13030 (nios2_validate_compare): Bypass alternate compare logic if *op2
13031 is not a CONST_INT.
13032 (ldstwm_operation_p): Return false if first_base is not a REG or
13033 if first_offset is not a CONST_INT.
13035 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
13037 * config/cris/cris.md: Use correct operand in a define_peephole2.
13039 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
13041 * config/c6x/c6x.c (predicate_insn): Do not incorrectly share RTL.
13043 2017-02-24 Segher Boessenkool <segher@kernel.crashing.org>
13045 * config/arc/arc.c (arc_ccfsm_advance): Only take the PATTERN of
13046 this_insn if it is an INSN or JUMP_INSN.
13047 (force_offsettable): Look at base, not at addr.
13048 * config/arc/predicates.md (brcc_nolimm_operator): Don't call INTVAL
13049 on things that aren't necessarily CONST_INTs.
13051 2017-02-24 Uros Bizjak <ubizjak@gmail.com>
13053 * doc/invoke.texi (x86 Options, -mfpmath=sse): Mention that
13054 -mfpmath=sse is the default also for x86-32 targets with SSE2
13055 instruction set when @option{-ffast-math} is enabled
13057 2017-02-24 Jeff Law <law@redhat.com>
13059 PR rtl-optimizatoin/79286
13060 * ira.c (update_equiv_regs): Drop may_trap_p exception to
13063 2017-02-24 Richard Biener <rguenther@suse.de>
13065 PR tree-optimization/79389
13066 * gimple-ssa-split-paths.c (is_feasible_trace): Properly skip
13069 2017-02-24 Aldy Hernandez <aldyh@redhat.com>
13071 * tree-ssa-loop-niter.c (number_of_iterations_exit): Update
13072 function comment to reflect reality.
13073 (loop_exits_before_overflow): Fix typo in function description.
13075 2017-02-24 Richard Biener <rguenther@suse.de>
13077 PR tree-optimization/79389
13078 * gimple-ssa-split-paths.c (is_feasible_trace): Verify more
13079 properly that a threading opportunity exists. Detect conditional
13080 copy/constant propagation opportunities.
13082 2017-02-23 Eric Botcazou <ebotcazou@adacore.com>
13084 * config/visium/visium.md (type): Add trap.
13085 (b): New mode attribute.
13086 (*btst): Rename into...
13087 (*btst<mode>): ...this and adjust.
13088 (*cbranchsi4_btst_insn): Rename into...
13089 (*cbranch<mode>4_btst_insn): ...this and adjust.
13090 (trap): New define_insn.
13092 2017-02-23 Jakub Jelinek <jakub@redhat.com>
13094 PR tree-optimization/79389
13095 * ifcvt.c (struct noce_if_info): Add rev_cond field.
13096 (noce_reversed_cond_code): New function.
13097 (noce_emit_store_flag): Use rev_cond if non-NULL instead of
13098 reversed_comparison_code. Formatting fix.
13099 (noce_try_store_flag): Test rev_cond != NULL in addition to
13100 reversed_comparison_code.
13101 (noce_try_store_flag_constants): Likewise.
13102 (noce_try_store_flag_mask): Likewise.
13103 (noce_try_addcc): Use rev_cond if non-NULL instead of
13104 reversed_comparison_code.
13105 (noce_try_cmove_arith): Likewise. Formatting fixes.
13106 (noce_try_minmax, noce_try_abs): Clear rev_cond.
13107 (noce_find_if_block): Initialize rev_cond.
13108 (find_cond_trap): Call noce_get_condition with then_bb == trap_bb
13109 instead of false as last argument never attempt to reverse it
13112 2017-02-23 Bin Cheng <bin.cheng@arm.com>
13114 PR tree-optimization/79663
13115 * tree-predcom.c (combine_chains): Process refs in reverse order
13116 only for ZERO length chains, and add explaining comment.
13118 2017-02-23 Jeff Law <law@redhat.com>
13120 PR tree-optimization/79578
13121 * tree-ssa-dse.c (clear_bytes_written_by): Use OEP_ADDRESS_OF
13122 in call to operand_equal_p.
13124 2017-01-23 Dominique d'Humieres <dominiq@lps.ens.fr>
13127 * config/i386/cpuid.h: Fix another undefined behavior.
13129 2017-02-23 Richard Biener <rguenther@suse.de>
13131 PR tree-optimization/79683
13132 * tree-vect-stmts.c (vect_analyze_stmt): Do not overwrite
13133 vector types for data-refs.
13135 2017-02-23 Martin Liska <mliska@suse.cz>
13137 * params.def (PARAM_MIN_NONDEBUG_INSN_UID): Change default to 0.
13139 2017-02-23 Jakub Jelinek <jakub@redhat.com>
13141 PR middle-end/79665
13142 * internal-fn.c (get_range_pos_neg): Moved to ...
13143 * tree.c (get_range_pos_neg): ... here. No longer static.
13144 * tree.h (get_range_pos_neg): New prototype.
13145 * expr.c (expand_expr_real_2) <case TRUNC_DIV_EXPR>: If both arguments
13146 are known to be in between 0 and signed maximum inclusive, try to
13147 expand both unsigned and signed divmod and use the cheaper one from
13150 2017-02-22 Jeff Law <law@redhat.com>
13152 PR tree-optimization/79578
13153 * tree-ssa-dse.c (clear_bytes_written_by): Use operand_equal_p
13154 to compare base operands.
13156 2017-02-22 Segher Boessenkool <segher@kernel.crashing.org>
13159 * config/rs6000/rs6000.md (*fsel<SFDF:mode><SFDF2:mode>4): Use
13160 gpc_reg_operand instead of fpr_reg_operand.
13162 2017-02-22 Sameera Deshpande <sameera.deshpande@imgtec.com>
13164 * config/mips/mips.c (mips_return_in_memory): Force FP
13165 vector types to be returned in memory for o32 ABI.
13167 2017-02-22 Jakub Jelinek <jakub@redhat.com>
13169 * dwarf2out.c (gen_variable_die): For -gdwarf-5, use DW_TAG_variable
13170 instead of DW_TAG_member for static data member declarations and don't
13171 set no_linkage_name for static inline data members.
13172 (gen_member_die): For -gdwarf-5 don't change DW_TAG_variable
13175 2017-02-22 Martin Liska <mliska@suse.cz>
13177 * doc/invoke.texi: Replace inequality signs with square brackets
13180 2017-02-22 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13182 PR tree-optimization/68644
13183 * gcc.dg/tree-ssa/ivopts-lt-2.c: Skip for powerpc*-*-*.
13185 2017-02-22 Matthew Fortune <matthew.fortune@imgtec.com>
13188 * lra-constraints.c (simplify_operand_subreg): Handle
13189 WORD_REGISTER_OPERATIONS targets.
13191 2017-02-22 Jakub Jelinek <jakub@redhat.com>
13194 * reg-stack.c (emit_swap_insn): Treat (float_extend:?F (mem:?F))
13195 and (const_double:?F) like (mem:?F) for the purpose of fxch %st(1)
13196 elimination by swapping fld*.
13198 2017-02-22 Richard Biener <rguenther@suse.de>
13200 PR tree-optimization/79673
13201 * tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
13202 convert the [TARGET_]MEM_REF offset INTEGER_CST, scrapping off
13203 irrelevant address-space qualifiers and avoiding a
13204 ADDR_SPACE_CONVERT_EXPR from fold_convert.
13206 2017-02-22 Richard Biener <rguenther@suse.de>
13208 PR tree-optimization/79666
13209 * tree-vrp.c (extract_range_from_binary_expr_1): Make sure
13210 to not symbolically negate if that may introduce undefined
13213 2017-02-22 Martin Liska <mliska@suse.cz>
13216 * data-streamer-in.c (streamer_read_gcov_count): Remove assert.
13217 * data-streamer-out.c (streamer_write_gcov_count_stream):
13219 * value-prof.c (stream_out_histogram_value): Make assert more
13220 precise based on type of counter.
13222 2017-02-21 Uros Bizjak <ubizjak@gmail.com>
13225 * config/i386/i386.md (standard_x87sse_constant_load splitter):
13226 Use nonimmediate_operand instead of memory_operand for operand 1.
13227 (float-extend standard_x87sse_constant_load splitter): Ditto.
13229 2017-02-21 Jeff Law <law@redhat.com>
13231 PR tree-optimization/79621
13232 * gimple-ssa-isolate-paths.c (find_implicit_erroneous_behavior): Ignore
13233 blocks with edges to themselves.
13235 2017-02-21 Jakub Jelinek <jakub@redhat.com>
13238 * tree-chkp-opt.c (chkp_optimize_string_function_calls): Use
13239 is_gimple_call instead of comparing gimple_code with GIMPLE_CALL.
13240 Use gimple_call_builtin_p.
13243 * sel-sched.c (moveup_expr_cached): Don't call sel_bb_head
13244 on temporarily removed DEBUG_INSNs.
13246 PR tree-optimization/79649
13247 * tree-loop-distribution.c (classify_partition): Give up on
13248 non-generic address space loads/stores.
13250 2017-02-21 Aldy Hernandez <aldyh@redhat.com>
13252 * doc/loop.texi (Loop manipulation): Remove nonexistent
13253 tree_ssa_loop_version from the documentation.
13254 * cfgloopmanip.c (loop_version): Document CONDITION_BB argument.
13256 2017-02-21 Jakub Jelinek <jakub@redhat.com>
13259 * config/i386/i386.c (ix86_expand_split_stack_prologue): Call
13260 make_reg_eh_region_note_nothrow_nononlocal on call_insn.
13261 * config/rs6000/rs6000.c: Include except.h.
13262 (rs6000_expand_split_stack_prologue): Call
13263 make_reg_eh_region_note_nothrow_nononlocal on the call insn.
13265 2017-02-21 Martin Jambor <mjambor@suse.cz>
13268 * ipa-prop.c (ipa_prop_write_jump_functions): Bail out if no edges
13269 have been analyzed.
13271 2017-02-21 Martin Jambor <mjambor@suse.cz>
13273 * common.opt (-fipa-cp-alignment): Mark as ignored and preserved
13274 for backward compatibility only.
13275 * doc/invoke.texi (Option Summary): Remove all references to
13276 -fipa-cp-alignment.
13278 2017-02-21 Matthew Fortune <matthew.fortune@imgtec.com>
13282 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
13284 * lra-constraints.c (curr_insn_transform): Handle
13285 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
13287 2017-02-21 Martin Liska <mliska@suse.cz>
13289 * config/i386/i386.opt: Replace -masm-dialect with -masm.
13291 2017-02-21 Thomas Schwinge <thomas@codesourcery.com>
13293 PR translation/79638
13294 * config/nvptx/nvptx.c (ENTRY_TEMPLATE): Single out "%ntid.y".
13296 2017-02-21 Eric Botcazou <ebotcazou@adacore.com>
13299 * config/arm/arm.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define.
13300 (arm_function_ok_for_sibcall): Return false for an indirect call by
13301 descriptor if all the argument registers are used.
13302 (arm_relayout_function): Use FUNCTION_ALIGNMENT macro to adjust the
13303 alignment of the function.
13305 2017-02-21 Jakub Jelinek <jakub@redhat.com>
13307 PR tree-optimization/61441
13308 * simplify-rtx.c (simplify_const_unary_operation): For
13309 -fsignaling-nans and sNaN operand, return NULL_RTX rather than
13310 the sNaN unmodified.
13312 2017-02-20 Bernd Edlinger <bernd.edlinger@hotmail.de>
13314 * Makefile.in (BUILD_SYSTEM_HEADER_DIR): New make variabe.
13315 (LIMITS_H_TEST, if_multiarch, stmp-fixinc): Use BUILD_SYSTEM_HEADER_DIR
13316 instead of SYSTEM_HEADER_DIR.
13318 2017-02-20 Gerald Pfeifer <gerald@pfeifer.com>
13319 Martin Liška <mliska@suse.cz>
13321 * doc/invoke.texi (use-after-scope-direct-emission-threshold):
13322 Fix typos and grammar, use active voice, and clarify.
13324 2017-02-20 Marek Polacek <polacek@redhat.com>
13326 PR middle-end/79537
13327 * gimplify.c (gimplify_expr): Handle unused *&&L;.
13330 * ubsan.c (ubsan_type_descriptor): Check if TYPE_MAX_VALUE is null.
13332 2017-02-20 Jakub Jelinek <jakub@redhat.com>
13335 * config/i386/i386.c (ix86_expand_builtin): Handle
13336 OPTION_MASK_ISA_AVX512VL and OPTION_MASK_ISA_64BIT in
13337 ix86_builtins_isa[fcode].isa as a requirement of those
13338 flags and any other flag in the bitmask.
13339 (ix86_init_mmx_sse_builtins): Use 0 instead of
13340 ~OPTION_MASK_ISA_64BIT as mask.
13341 * config/i386/i386-builtin.def (__builtin_ia32_rdtsc,
13342 __builtin_ia32_rdtscp, __builtin_ia32_pause, __builtin_ia32_bsrsi,
13343 __builtin_ia32_rdpmc, __builtin_ia32_rolqi, __builtin_ia32_rolhi,
13344 __builtin_ia32_rorqi, __builtin_ia32_rorhi): Likewise.
13346 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
13349 * lra-constraints.c (split_reg): Check requested split mode
13350 is supported by the register.
13352 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
13354 * lra-constraints.c (simplify_operand_subreg): Remove early
13357 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
13360 * lra-constraints.c (curr_insn_transform): Tighten condition
13361 for converting SUBREG reloads from OP_OUT to OP_INOUT.
13363 2017-02-20 Matthew Fortune <matthew.fortune@imgtec.com>
13366 * lra-constraints.c (curr_insn_transform): Handle
13367 WORD_REGISTER_OPERATIONS requirements when reloading SUBREGs.
13369 2017-02-19 Uros Bizjak <ubizjak@gmail.com>
13372 2016-05-30 Uros Bizjak <ubizjak@gmail.com>
13374 * config/i386/sync.md (mfence_nosse): Use "lock orl $0, -4(%esp)".
13376 2017-02-19 Jonathan Wakely <jwakely@redhat.com>
13379 * doc/invoke.texi (C++ Dialect Options) [-Wliteral-suffix]: Update
13382 2017-02-19 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13384 * gimple-pretty-print.c (dump_ternary_rhs): Adjust gimple dump format
13387 2017-02-18 Jakub Jelinek <jakub@redhat.com>
13389 * final.c (last_columnnum, override_columnnum): New variables.
13390 (final_start_function): Set last_columnnum, pass it to begin_prologue
13391 hook and pass 0 to dwarf2out_begin_prologue.
13392 (final_scan_insn): Update override_columnnum. Pass last_columnnum
13393 to source_line debug hook.
13394 (notice_source_line): Compute last_columnnum and for debug_column_info
13395 return true on column changes.
13396 * debug.h (struct gcc_debug_hooks): Add column argument to
13397 source_line and begin_prologue hooks.
13398 (debug_nothing_int_charstar_int_bool): Remove prototype.
13399 (debug_nothing_int_int_charstar,
13400 debug_nothing_int_int_charstar_int_bool): New prototypes.
13401 (dwarf2out_begin_prologue): Add column argument.
13402 * debug.c (do_nothing_debug_hooks): Adjust source_line and
13403 begin_prologue hooks.
13404 (debug_nothing_int_charstar_int_bool): Remove.
13405 (debug_nothing_int_int_charstar,
13406 debug_nothing_int_int_charstar_int_bool): New functions.
13407 * dwarf2out.c (dwarf2out_begin_prologue): Add column argument, pass it
13408 through to dwarf2out_source_line.
13409 (dwarf2_lineno_debug_hooks): Adjust begin_prologue hook.
13410 (dwarf2out_source_line): Add column argument, emit it if requested.
13411 * sdbout.c (sdbout_source_line, sdbout_begin_prologue): Add column
13413 * xcoffout.h (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
13414 * xcoffout.c (xcoffout_begin_prologue, xcoffout_source_line): Likewise.
13415 * vmsdbgout.c (vmsdbgout_begin_prologue): Add column argument, pass it
13416 through to dwarf2out_begin_prologue.
13417 (vmsdbgout_source_line): Add column argument, pass it through to
13418 dwarf2out_source_line.
13419 * dbxout.c (dbxout_begin_prologue): Add column argument, adjust
13420 dbxout_source_line caller.
13421 (dbxout_source_line): Add column argument.
13423 * common.opt (gno-column-info, gcolumn-info): New options.
13424 * dwarf2out.c (dwarf2_lineno_debug_hooks): Formatting fix.
13425 (check_die): Also test for multiple DW_AT_decl_column attributes.
13426 (add_src_coords_attributes, dwarf2out_imported_module_or_decl_1): Add
13427 DW_AT_decl_column if requested.
13428 (gen_subprogram_die): Compare and/or add also DW_AT_decl_column
13430 (gen_variable_die): Likewise.
13431 (add_call_src_coords_attributes): Add DW_AT_call_column if requested.
13432 * doc/invoke.texi (-gcolumn-info, -gno-column-info): Document.
13435 * config/i386/i386.opt (m3dnowa): Replace Undocumented with Report.
13436 * common/config/i386/i386-common.c (OPTION_MASK_ISA_3DNOW_A_SET): Define.
13437 (ix86_handle_option): Handle OPT_m3dnowa.
13438 * doc/invoke.texi (-m3dnowa): Document.
13439 * doc/extend.texi (__builtin_ia32_pmulhuw, __builtin_ia32_pf2iw): Use
13440 -m3dnowa instead of -m3dnow -march=athlon.
13443 * config/i386/i386.c (ix86_print_operand): Use output_operand_lossage
13444 instead of gcc_assert for K, r and R code checks. Formatting fixes.
13446 2017-02-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
13449 * config/rs6000/rs6000.c (rs6000_expand_ternop_builtin): Add
13450 support for CODE_FOR_vsx_xxpermdi_v2d[fi]_be.
13451 * config/rs6000/rs6000.md (reload_gpr_from_vsx<mode>): Call
13452 generator for vsx_xxpermdi_<mode>_be.
13453 * config/rs6000/vsx.md (vsx_xxpermdi_<mode>): Remove logic to
13454 force big-endian semantics.
13455 (vsx_xxpermdi_<mode>_be): New define_expand with same
13456 implementation as previous version of vsx_xxpermdi_<mode>.
13458 2017-02-17 Jakub Jelinek <jakub@redhat.com>
13460 PR tree-optimization/79327
13461 * gimple-ssa-sprintf.c (format_integer): Remove likely_adjust
13462 variable, its initialization and use.
13464 2017-02-17 Julia Koval <julia.koval@intel.com>
13466 * common/config/i386/i386-common.c (OPTION_MASK_ISA_RDPID_SET): New.
13467 (OPTION_MASK_ISA_PKU_UNSET): New.
13468 (ix86_handle_option): Handle -mrdpid.
13469 * config/i386/cpuid.h (bit_RDPID): New.
13470 * config/i386/driver-i386.c (host_detect_local_cpu):
13471 Detect RDPID feature.
13472 * config/i386/i386-builtin.def (__builtin_ia32_rdpid): New.
13473 * config/i386/i386-c.c (ix86_target_macros_internal):
13475 * config/i386/i386.c (ix86_target_string): Add -mrdpid to isa2_opts.
13476 (ix86_valid_target_attribute_inner_p): Add "rdpid".
13477 (ix86_expand_builtin): Handle IX86_BUILTIN_RDPID.
13478 * config/i386/i386.h (TARGET_RDPID, TARGET_RDPID_P): New.
13479 * config/i386/i386.md (define_insn "rdpid"): New.
13480 * config/i386/i386.opt Add -mrdpid.
13481 * config/i386/immintrin.h (_rdpid_u32): New.
13483 2017-02-17 Vladimir Makarov <vmakarov@redhat.com>
13485 PR rtl-optimization/79541
13486 * lra-constraints.c (curr_insn_transform): Remove wrong asm insn
13487 instead of transforming it into USE.
13489 2017-02-17 Segher Boessenkool <segher@kernel.crashing.org>
13491 * config/rs6000/rs6000.md (extendsfdf2): Remove default arguments.
13492 If HONOR_SNANS (SFmode) force the input to a register.
13493 (*extendsfdf2_fpr): Add !HONOR_SNANS (SFmode) condition.
13494 (*extendsfdf2_snan): New pattern, used when using SNaNs; it generates
13495 an frsp or similar insn.
13497 2017-02-17 Martin Liska <mliska@suse.cz>
13499 PR rtl-optimization/79577
13500 * params.def (selsched-max-sched-times): Increase minimum to 1.
13502 2017-02-17 Martin Liska <mliska@suse.cz>
13504 PR rtl-optimization/79574
13505 * gcse.c (want_to_gcse_p): Prevent integer overflow.
13507 2017-02-17 Martin Liska <mliska@suse.cz>
13509 PR tree-optimization/79529
13510 * tree-ssa-loop-unswitch.c (is_maybe_undefined): Use
13511 ssa_defined_default_def_p to handle cases which are implicitly
13513 * tree-ssa.c (ssa_defined_default_def_p): New function.
13514 (ssa_undefined_value_p): Use ssa_defined_default_def_p to handle cases
13515 which are implicitly defined.
13516 * tree-ssa.h (ssa_defined_default_def_p): Declare.
13518 2017-02-17 Richard Biener <rguenther@suse.de>
13520 PR middle-end/79576
13521 * params.def (max-ssa-name-query-depth): Limit to 10.
13523 2017-02-17 Richard Biener <rguenther@suse.de>
13525 PR tree-optimization/79552
13526 * tree-ssa-structalias.c (visit_loadstore): Properly verify
13529 2017-02-17 Richard Biener <rguenther@suse.de>
13532 * genmatch.c (output_line_directive): Handle DIR_SEPARATOR_2.
13534 2017-02-17 Marek Polacek <polacek@redhat.com>
13536 PR middle-end/79536
13537 * fold-const.c (fold_negate_expr_1): Renamed from fold_negate_expr.
13538 (fold_negate_expr): New wrapper.
13540 2017-02-16 Sandra Loosemore <sandra@codesourcery.com>
13542 * doc/invoke.texi (C++ Dialect Options) [-Wno-non-template-friend]:
13543 Correct terminology and de-emphasize pre-standard behavior.
13545 2017-02-16 Alan Modra <amodra@gmail.com>
13547 PR rtl-optimization/79286
13548 * ira.c (def_dominates_uses): New function.
13549 (update_equiv_regs): Don't create an equivalence for insns that
13550 may trap where the register def does not dominate the use.
13552 2017-02-16 Vladimir Makarov <vmakarov@redhat.com>
13554 PR rtl-optimization/78127
13555 * lra.c (lra): Call lra_eliminate before finish the loop after
13558 2017-02-16 Richard Biener <rguenther@suse.de>
13560 * graphite.h: Do not include isl/isl_val_gmp.h, instead include
13562 * graphite-isl-ast-to-gimple.c (gmp_cst_to_tree): Remove.
13563 (gcc_expression_from_isl_expr_int): Use generic isl_val interface.
13564 * graphite-sese-to-poly.c: Do not include isl/isl_val_gmp.h.
13565 (isl_val_int_from_wi): New function.
13566 (extract_affine_gmp): Rename to ...
13567 (extract_affine_wi): ... this, take a widest_int.
13568 (extract_affine_int): Just wrap extract_affine_wi.
13569 (add_param_constraints): Use isl_val_int_from_wi.
13570 (add_loop_constraints): Likewise, and extract_affine_wi.
13572 2017-02-15 Jeff Law <law@redhat.com>
13574 PR middle-end/79521
13575 * ira-costs.c (scan_one_insn): Check have_regs_of_mode before calling
13576 ira_init_register_move_cost_if_necessary.
13578 2017-02-15 Martin Sebor <msebor@redhat.com>
13580 PR middle-end/32003
13581 * doc/invoke.texi (-fdump-final-insns): Replace option accidentally
13582 removed in a prior commit.
13584 2017-02-15 Bin Cheng <bin.cheng@arm.com>
13586 PR tree-optimization/79347
13587 * tree-vect-loop-manip.c (vect_do_peeling): Maintain profile
13588 counters during peeling.
13590 2017-02-15 Thomas Schwinge <thomas@codesourcery.com>
13592 * Makefile.in (site.exp): Remove "set ISLVER".
13594 2017-02-15 Jakub Jelinek <jakub@redhat.com>
13597 * real.c (real_from_integer): Call real_convert even for decimal.
13599 2017-02-15 Dominik Vogt <vogt@linux.vnet.ibm.com>
13602 * config/s390/s390.c: define TARGET_CUSTOM_FUNCTION_DESCRIPTORS.
13604 2017-02-14 Andrew Pinski <apinski@cavium.com>
13606 * config/aarch64/aarch64-cores.def (thunderx2t99): Move to under 'C"
13607 cores and change the partno/implementer to be correct.
13608 (thunderx2t99p1): New core which replaces thunderx2t99 and still has
13609 the 'B" as the implementer.
13610 * config/aarch64/aarch64-tune.md: Regenerate.
13612 2017-02-14 Carl Love <cel@us.ibm.com>
13614 * config/rs6000/rs6000.c: Add case statement entry to make the
13615 xvcvuxdsp built-in argument unsigned.
13616 * config/rs6000/vsx.md: Fix the source and return operand types so they
13617 match the instruction definitions from the ISA document. Fix typo
13618 in the instruction generation for the (define_insn "vsx_xvcvuxdsp"
13621 2017-02-14 Vladimir Makarov <vmakarov@redhat.com>
13624 * lra-int.h (struct lra_operand_data, struct lra_insn_reg): Add
13625 member early_clobber_alts.
13626 * lra-lives.c (reg_early_clobber_p): New.
13627 (process_bb_lives): Use it.
13628 * lra.c (new_insn_reg): New arg early_clobber_alts. Use it.
13629 (debug_operand_data): Initialize early_clobber_alts.
13630 (setup_operand_alternative): Set up early_clobber_alts.
13631 (collect_non_operand_hard_regs): Ditto. Pass early clobber
13632 alternatives to new_insn_reg.
13633 (add_regs_to_insn_regno_info): Add arg early_clobber_alts. Use
13635 (lra_update_insn_regno_info): Pass the new arg.
13637 2017-02-14 Jakub Jelinek <jakub@redhat.com>
13639 PR middle-end/79505
13640 * omp-offload.c (free_oacc_loop): Release loop->ifns vector.
13641 (new_oacc_loop_raw): Don't clear already cleared fields.
13644 * config/i386/avx512pfintrin.h (_mm512_prefetch_i32gather_pd,
13645 _mm512_prefetch_i32gather_ps, _mm512_prefetch_i64gather_pd,
13646 _mm512_prefetch_i64gather_ps): New inline functions and macros.
13648 2017-02-14 Uros Bizjak <ubizjak@gmail.com>
13651 * config/i386/i386.md (*movxf_internal): Add (o,rC) alternative.
13653 2017-02-14 H.J. Lu <hongjiu.lu@intel.com>
13656 * config/i386/i386.c (timode_scalar_chain::convert_insn): Insert
13657 the extra instruction to the right place to store 128-bit constant
13660 2017-02-14 Martin Sebor <msebor@redhat.com>
13662 PR middle-end/79448
13663 * gimple-ssa-sprintf.c (format_directive): Avoid issuing INT_MAX
13664 warning for strings of unknown length.
13666 2017-02-13 Segher Boessenkool <segher@kernel.crashing.org>
13668 * config.gcc (supported_defaults) [powerpc*-*-*]: Update.
13670 2017-02-14 Jeff Law <law@redhat.com>
13673 * ira-costs.c (scan_one_insn): Initialize register move costs
13674 for pseudos seen in USE/CLOBBER insns.
13676 PR tree-optimization/79095
13677 * tree-vrp.c (extract_range_from_binary_expr_1): For EXACT_DIV_EXPR,
13678 if the numerator has the range ~[0,0] make the resultant range ~[0,0].
13679 (extract_range_from_binary_expr): For MINUS_EXPR with no derived range,
13680 if the operands are known to be not equal, then the resulting range
13682 (intersect_ranges): If the new range is ~[0,0] and the old range is
13683 wide, then prefer ~[0,0].
13684 * tree-vrp.c (overflow_comparison_p_1): New function.
13685 (overflow_comparison_p): New function.
13686 * tree-vrp.c (register_edge_assert_for_2): Register additional asserts
13687 if NAME is used in an overflow test.
13688 (vrp_evaluate_conditional_warnv_with_ops): If the ops represent an
13689 overflow check that can be expressed as an equality test, then adjust
13690 ops to be that equality test.
13692 2017-02-14 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
13694 * config/s390/s390-builtin-types.def: Remove flags argument.
13695 * config/s390/s390.c (s390_init_builtins): Likewise.
13697 2017-02-14 Martin Liska <mliska@suse.cz>
13699 * tree-ssa-loop-unswitch.c (hoist_guard): Release get_loop_body
13700 vector. Fix trailing white spaces.
13702 2017-02-14 James Greenhalgh <james.greenhalgh@arm.com>
13704 * config/aarch64/aarch64.c (aarch64_simd_container_mode): Handle
13707 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13709 PR rtl-optimization/68664
13710 * config/arm/arm.c (arm_sched_can_speculate_insn):
13711 New function. Declare prototype.
13712 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
13714 2017-02-14 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13716 PR rtl-optimization/68664
13717 * config/aarch64/aarch64.c (aarch64_sched_can_speculate_insn):
13719 (TARGET_SCHED_CAN_SPECULATE_INSN): Define.
13721 2017-02-14 Amit Pawar <amit.pawar@amd.com>
13723 * config/i386/i386.c (znver1_cost): Fix the alignment for function and
13724 max skip bytes for function, loop and jump.
13726 2017-02-14 Prathamesh Kulkarni <prathamesh.kulkarni@linaro.org>
13728 * gimple-pretty-print.c (dump_unary_rhs): Adjust dump format for
13729 ABS_EXPR for gimple dump.
13731 2017-02-14 Jakub Jelinek <jakub@redhat.com>
13734 * config/sh/sh.c (expand_cbranchdi4): Don't clear operands[4].
13736 PR tree-optimization/79408
13737 * tree-vrp.c (simplify_div_or_mod_using_ranges): Handle also the
13738 case when on TRUNC_MOD_EXPR op0 is INTEGER_CST.
13739 (simplify_stmt_using_ranges): Call simplify_div_or_mod_using_ranges
13740 also if rhs1 is INTEGER_CST.
13742 2017-02-14 Richard Biener <rguenther@suse.de>
13744 PR middle-end/79432
13745 * tree-into-ssa.c (insert_phi_nodes): When the function can
13746 have abnormal edges rewrite SSA names with broken use-def
13747 dominance out of SSA and register them for PHI insertion.
13749 2017-02-13 Martin Sebor <msebor@redhat.com>
13751 PR middle-end/79496
13752 * gimple-ssa-sprintf.c (pass_sprintf_length::handle_gimple_call): Avoid
13753 clearing info.nowrite flag when snprintf size argument is a range.
13755 2017-02-13 Jakub Jelinek <jakub@redhat.com>
13757 * cprop.c (cprop_jump): Add missing space in string literal.
13758 * tree-ssa-structalias.c (rewrite_constraints): Likewise.
13759 (get_constraint_for_component_ref): Likewise.
13760 * df-core.c (df_worklist_dataflow_doublequeue): Likewise.
13761 * tree-outof-ssa.c (insert_partition_copy_on_edge): Likewise.
13762 * lra-constraints.c (process_alt_operands): Likewise.
13763 * ipa-inline.c (inline_small_functions): Likewise.
13764 * tree-ssa-sccvn.c (visit_reference_op_store): Likewise.
13765 * cgraph.c (cgraph_edge::redirect_call_stmt_to_callee): Likewise.
13766 * trans-mem.c (diagnose_tm_1_op): Likewise.
13767 * omp-grid.c (grid_find_single_omp_among_assignments): Likewise.
13768 (grid_parallel_clauses_gridifiable): Likewise.
13770 * config/nvptx/mkoffload.c (process): Add space in between
13773 * config/i386/i386.h (REG_CLASS_NAMES): Add , in between
13774 "MOD4_SSE_REGS" and "ALL_REGS".
13776 * spellcheck.c (test_data): Add , in between "foo" and "food".
13778 2017-02-13 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13781 * config/rs6000/rs6000.c (expand_block_compare): Make sure runtime
13782 boundary crossing check and subsequent code generation agree.
13784 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13786 * config/aarch64/aarch64.c (has_memory_op): Delete.
13787 (aarch64_madd_needs_nop): Use contains_mem_rtx_p instead of
13790 2017-02-13 Jakub Jelinek <jakub@redhat.com>
13792 PR rtl-optimization/79388
13793 PR rtl-optimization/79450
13794 * combine.c (distribute_notes): When removing TEM_INSN for which
13795 corresponding dest has last value recorded, invalidate that last
13798 2017-02-13 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
13800 * config/arm/arm.c (arm_print_tune_info): Use ASM_COMMENT_START instead
13801 of explicit '@'. Add missing assembly comment marker on branch costs
13804 2017-02-13 Nathan Sidwell <nathan@acm.org>
13806 * gengtype-lex.l (<in_struct>): Add '/'.
13808 2017-02-13 Martin Liska <mliska@suse.cz>
13811 * calls.c (expand_call): Replace XALLOCAVEC with XCNEWVEC.
13813 2017-02-13 Richard Biener <rguenther@suse.de>
13815 * configure.ac (HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS):
13817 * configure: Re-generate.
13818 * config.in: Likewise.
13819 * graphite-dependences.c: Simplify as if
13820 HAVE_ISL_OPTIONS_SET_SCHEDULE_SERIALIZE_SCCS was defined.
13821 * graphite-isl-ast-to-gimple.c: Likewise.
13822 * graphite-optimize-isl.c: Likewise.
13823 * graphite-poly.c: Likewise.
13824 * graphite-sese-to-poly.c: Likewise.
13825 * graphite.h: Likewise.
13826 * toplev.c: Include isl/version.h and use isl_version () for
13827 printing the ISL version.
13828 * doc/install.texi: Update ISL requirement.
13830 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
13832 * doc/standards.texi (Standards): Update reference to
13835 2017-02-12 Gerald Pfeifer <gerald@pfeifer.com>
13837 * doc/extend.texi (Named Address Spaces): sourceware.org now
13839 * doc/install.texi (Binaries): Ditto.
13842 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
13844 * doc/cpp.texi: Replace "stringify"/"stringification" with C
13845 standard terminology "stringize"/"stringizing" throughout.
13846 * doc/cppinternals.texi: Likewise.
13848 2017-02-11 Sandra Loosemore <sandra@codesourcery.com>
13850 * doc/extend.texi: Fix some spelling mistakes and typos.
13851 * doc/invoke.texi: Likewise.
13853 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
13856 * params.def (inline-min-speedup) Change from 10 to 8.
13858 2017-02-11 Jakub Jelinek <jakub@redhat.com>
13860 * doc/invoke.texi (fopenmp): Bump OpenMP version from 4.0 to
13863 2017-02-11 Jan Hubicka <hubicka@ucw.cz>
13866 * ipa-inline-analysis.c (get_minimal_bb): New function.
13867 (record_modified): Use it.
13868 (remap_edge_change_prob): Handle also ancestor functions.
13870 2017-02-11 Gerald Pfeifer <gerald@pfeifer.com>
13872 * doc/contrib.texi (Contributors): Remove broken link into
13873 the Mauve CVS repository.
13875 2017-02-11 Jakub Jelinek <jakub@redhat.com>
13877 PR middle-end/79454
13878 * internal-fn.c (expand_vector_ubsan_overflow): Use piece-wise
13879 result computation whenever lhs doesn't have vector mode, not
13880 just when it has BLKmode.
13882 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
13884 * doc/makefile.texi (profiledbootstrap): Refer to the
13885 installation instructions only in textual form.
13887 2017-02-10 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
13890 * config/rs6000/altivec.md (bcd<bcd_add_sub>): Fix constraints.
13892 2017-02-10 Gerald Pfeifer <gerald@pfeifer.com>
13894 * doc/install.texi (Specific): Use https for blackfin.uclinux.org.
13895 (Specific): Update mingw-w64 reference.
13897 (Specific): Remove broken link to Renesas RX processor.
13899 2017-02-10 Richard Biener <rguenther@suse.de>
13901 * toplev.c (process_options): Do not mention obsolete graphite
13902 options when printing sorry message about missing graphite support.
13903 Mention -floop-nest-optimize.
13905 2017-02-10 Christophe Lyon <christophe.lyon@linaro.org>
13907 * config/aarch64/arm_neon.h (vtst_p8): Rewrite without asm.
13908 (vtst_p16): Likewise.
13909 (vtstq_p8): Likewise.
13910 (vtstq_p16): Likewise.
13912 (vtstq_p64): Likewise.
13913 * config/arm/arm_neon.h (vgetq_lane_p64): New.
13914 (vset_lane_p64): New.
13915 (vsetq_lane_p64): New.
13917 2017-02-10 Jakub Jelinek <jakub@redhat.com>
13919 PR tree-optimization/79411
13920 * tree-ssa-reassoc.c (is_reassociable_op): Return false if
13921 stmt operands are SSA_NAMEs used in abnormal phis.
13922 (can_reassociate_p): Return false if op is SSA_NAME used in abnormal
13925 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
13928 * cgraphunit.c (cgraph_node::add_new_function): Set externally_visible
13931 2017-02-09 Jan Hubicka <hubicka@ucw.cz>
13933 * tree-ssa-loop-unswitch.c (hoist_guard): Update profile.
13935 2017-02-09 Jakub Jelinek <jakub@redhat.com>
13937 * omp-offload.c (oacc_loop_auto_partitions): Use || instead of |
13941 * gimplify.h (is_gimple_sizepos): Only test for INTEGER_CST constants,
13942 not arbitrary TREE_CONSTANT.
13945 * gimplify.c (gimplify_adjust_omp_clauses): Ignore
13946 "omp declare target link" attribute unless is_global_var.
13947 * omp-offload.c (find_link_var_op): Likewise.
13949 2017-02-09 Nathan Sidwell <nathan@codesourcery.com>
13950 Chung-Lin Tang <cltang@codesourcery.com>
13952 * gimplify.c (gimplify_scan_omp_clauses): No special handling for
13954 (gimplify_adjust_omp_clauses): Don't delete TILE.
13955 (gimplify_omp_for): Deal with TILE.
13956 * internal-fn.c (expand_GOACC_TILE): New function.
13957 * internal-fn.def (GOACC_DIM_POS): Comment may be overly conservative.
13959 * omp-expand.c (struct oacc_collapse): Add tile and outer fields.
13960 (expand_oacc_collapse_init): Add LOC paramter. Initialize tile
13962 (expand_oacc_collapse_vars): Add INNER parm, adjust for tiling,
13963 avoid DIV for outermost collapse var.
13964 (expand_oacc_for): Insert tile element loop as needed. Adjust.
13965 Remove out of date comments, fix whitespace.
13966 * omp-general.c (omp_extract_for_data): Deal with tiling.
13967 * omp-general.h (enum oacc_loop_flags): Add OLF_TILE flag,
13968 adjust OLF_DIM_BASE value.
13969 (struct omp_for_data): Add tiling field.
13970 * omp-low.c (scan_sharing_clauses): Allow OMP_CLAUSE_TILE.
13971 (lower_oacc_head_mark): Add OLF_TILE as appropriate. Ensure 2 levels
13972 for auto loops. Remove default auto determining, moved to
13973 oacc_loop_fixed_partitions.
13974 * omp-offload.c (struct oacc_loop): Change 'ifns' to vector of call
13975 stmts, add e_mask field.
13976 (oacc_dim_call): New function, abstracted out from oacc_thread_numbers.
13977 (oacc_thread_numbers): Use oacc_dim_call.
13978 (oacc_xform_tile): New.
13979 (new_oacc_loop_raw): Initialize e_mask, adjust for ifns vector.
13980 (finish_oacc_loop): Adjust for ifns vector.
13981 (oacc_loop_discover_walk): Append loop abstraction sites to list,
13982 add case for GOACC_TILE fns.
13983 (oacc_loop_xform_loop): Delete.
13984 (oacc_loop_process): Iterate over call list directly, and add
13985 handling for GOACC_TILE fns.
13986 (oacc_loop_fixed_partitions): Determine default auto, deal with TILE,
13988 (oacc_loop_auto_partitions): Add outer_assign parm. Assign all but
13989 vector partitioning to outer loops. Assign 2 partitions to loops
13990 when available. Add TILE handling.
13991 (oacc_loop_partition): Adjust oacc_loop_auto_partitions call.
13992 (execite_oacc_device_lower): Process GOACC_TILE fns,
13993 ignore unknown specs.
13994 * tree-nested.c (convert_nonlocal_omp_clauses): Allow OMP_CLAUSE_TILE.
13995 * tree.c (omp_clause_num_ops): Adjust TILE ops.
13996 * tree.h (OMP_CLAUSE_TILE_ITERVAR, OMP_CLAUSE_TILE_COUNT): New.
13998 2017-02-09 Gerald Pfeifer <gerald@pfeifer.com>
14000 * configure.ac (ACX_BUGURL): Update.
14001 * configure: Regenerate.
14003 2017-02-09 Richard Biener <rguenther@suse.de>
14005 PR tree-optimization/69823
14006 * graphite-scop-detection.c (scop_detection::harmful_loop_in_region):
14007 Properly enumerate all BBs in the region. Use auto_vec/auto_bitmap.
14009 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
14011 * config/arc/arc-c.def: Add __NPS400__ definition.
14012 * config/arc/arc.h (CPP_SPEC): Don't define __NPS400__ here.
14013 (TARGET_NPS400): Define.
14015 2017-02-09 Andrew Burgess <andrew.burgess@embecosm.com>
14017 * config/arc/arc-arch.h (arc_arch_t): Move unchanged to earlier in
14019 (arc_cpu_t): Change base_architecture field, arch, to a arc_arc_t
14020 pointer, arch_info.
14021 (arc_cpu_types): Fill the arch_info field with a pointer into the
14022 arc_arch_types table.
14023 (arc_selected_cpu): Declare.
14024 * config/arc/arc.c (arc_selected_cpu): Make global.
14025 (arc_selected_arch): Delete.
14026 (arc_base_cpu): Delete.
14027 (arc_override_options): Remove references to deleted variables,
14028 update access to arch information.
14029 (ARC_OPT): Update access to arch information.
14030 (ARC_OPTX): Likewise.
14031 * config/arc/arc.h (arc_base_cpu): Remove declaration.
14032 (TARGET_ARC600): Update access to arch information.
14033 (TARGET_ARC601): Likewise.
14034 (TARGET_ARC700): Likewise.
14035 (TARGET_EM): Likewise.
14036 (TARGET_HS): Likewise.
14037 * config/arc/driver-arc.c (arc_cpu_to_as): Update access to arch
14040 2017-02-08 Pat Haugen <pthaugen@us.ibm.com>
14043 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Invert
14044 condition/operands for integer GE/LE/GEU/LEU operations.
14046 2017-02-08 Segher Boessenkool <segher@kernel.crashing.org>
14048 PR translation/79397
14049 * config/rs6000/rs6000.opt (maltivec=le, maltivec=be): Fix spelling
14052 2017-02-08 Martin Jambor <mjambor@suse.cz>
14055 * ipa-prop.c (ipa_alloc_node_params): Make static, return bool
14056 whether allocation happened.
14057 (ipa_initialize_node_params): Do not call ipa_alloc_node_params if
14058 nothing was allocated.
14060 2017-02-08 Jakub Jelinek <jakub@redhat.com>
14062 PR tree-optimization/79408
14063 * tree-vrp.c (simplify_div_or_mod_using_ranges): If op1 is not
14064 constant, but SSA_NAME with a known integer range, use the minimum
14065 of that range instead of op1 to determine if modulo can be replaced
14066 with its first operand.
14068 2016-02-08 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
14070 * config/riscv/riscv.c (riscv_build_integer_1): Avoid use of INT16_MAX.
14072 2017-02-08 Richard Biener <rguenther@suse.de>
14074 PR tree-optimization/71824
14075 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
14076 Check all loops contained in the merged region.
14078 2017-02-07 Andrew Pinski <apinski@cavium.com>
14080 * config/aarch64/aarch64.md (popcount<mode>2): New pattern.
14082 2017-02-07 Andrew Pinski <apinski@cavium.com>
14084 * config/aarch64/aarch64-cores.def (thunderx): Disable LSE.
14085 (thunderxt88): Likewise.
14086 (thunderxt81): Disable LSE and change v8.1 to v8.
14087 (thunderxt83): Likewise.
14089 2017-02-07 Jakub Jelinek <jakub@redhat.com>
14090 Richard Biener <rguenther@suse.de>
14092 PR middle-end/79399
14093 * ira-int.h (struct target_ira_int): Change x_max_struct_costs_size
14094 type from int to size_t.
14095 * ira-costs.c (struct_costs_size): Change type from int to size_t.
14097 2017-02-07 Jakub Jelinek <jakub@redhat.com>
14099 PR rtl-optimization/79386
14100 * cprop.c (bypass_conditional_jumps): Initialize
14101 bypass_last_basic_block already before splitting bbs after
14102 unconditional traps...
14103 (bypass_conditional_jumps): ... rather than here.
14106 * config/i386/sse.md (xtg_mode, gatherq_mode): New mode attrs.
14107 (*avx512f_gathersi<mode>, *avx512f_gathersi<mode>_2,
14108 *avx512f_gatherdi<mode>, *avx512f_gatherdi<mode>_2): Use them,
14109 fix -masm=intel patterns.
14111 2017-02-07 Richard Biener <rguenther@suse.de>
14113 PR tree-optimization/79256
14114 PR middle-end/79278
14115 * builtins.c (get_object_alignment_2): Use min_align_of_type
14116 to extract alignment for MEM_REFs to honor BIGGEST_FIELD_ALIGNMENT
14117 and ADJUST_FIELD_ALIGN.
14119 * doc/tm.texi.in (ADJUST_FIELD_ALIGN): Adjust to take additional
14121 * doc/tm.texi: Regenerate.
14122 * stor-layout.c (layout_decl): Adjust.
14123 (update_alignment_for_field): Likewise.
14124 (place_field): Likewise.
14125 (min_align_of_type): Likewise.
14126 * config/arc/arc.h (ADJUST_FIELD_ALIGN): Adjust.
14127 * config/epiphany/epiphany.h (ADJUST_FIELD_ALIGN): Likewise.
14128 * config/epiphany/epiphany.c (epiphany_adjust_field_align): Likewise.
14129 * config/frv/frv.h (ADJUST_FIELD_ALIGN): Likewise.
14130 * config/frv/frv.c (frv_adjust_field_align): Likewise.
14131 * config/i386/i386.h (ADJUST_FIELD_ALIGN): Likewise.
14132 * config/i386/i386.c (x86_field_alignment): Likewise.
14133 * config/rs6000/aix.h (ADJUST_FIELD_ALIGN): Likewise.
14134 * config/rs6000/darwin.h (ADJUST_FIELD_ALIGN): Likewise.
14135 * config/rs6000/freebsd64.h (ADJUST_FIELD_ALIGN): Likewise.
14136 * config/rs6000/linux64.h (ADJUST_FIELD_ALIGN): Likewise.
14137 * config/rs6000/sysv4.h (ADJUST_FIELD_ALIGN): Likewise.
14138 * config/rs6000/rs6000.c (rs6000_special_adjust_field_align_p):
14142 2017-01-30 Richard Biener <rguenther@suse.de>
14144 PR tree-optimization/79256
14145 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
14146 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
14149 2017-02-07 Toma Tabacu <toma.tabacu@imgtec.com>
14151 * config/mips/mips.c (mips_expand_builtin_insn): Convert the QImode
14152 argument of the pshufh, psllh, psllw, psrah, psraw, psrlh, psrlw
14153 builtins to SImode and emit a zero-extend, if necessary.
14155 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
14157 * docs/invoke.texi (RISC-V Options): Alphabetize.
14159 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
14161 * doc/invoke.texi (RISC-V Options): Use two spaces to separate
14164 2017-02-06 Palmer Dabbelt <palmer@dabbelt.com>
14166 * config/riscv/riscv.c: New file.
14167 * gcc/common/config/riscv/riscv-common.c: Likewise.
14168 * config.gcc: Likewise.
14169 * config/riscv/constraints.md: Likewise.
14170 * config/riscv/elf.h: Likewise.
14171 * config/riscv/generic.md: Likewise.
14172 * config/riscv/linux.h: Likewise.
14173 * config/riscv/multilib-generator: Likewise.
14174 * config/riscv/peephole.md: Likewise.
14175 * config/riscv/pic.md: Likewise.
14176 * config/riscv/predicates.md: Likewise.
14177 * config/riscv/riscv-builtins.c: Likewise.
14178 * config/riscv/riscv-c.c: Likewise.
14179 * config/riscv/riscv-ftypes.def: Likewise.
14180 * config/riscv/riscv-modes.def: Likewise.
14181 * config/riscv/riscv-opts.h: Likewise.
14182 * config/riscv/riscv-protos.h: Likewise.
14183 * config/riscv/riscv.h: Likewise.
14184 * config/riscv/riscv.md: Likewise.
14185 * config/riscv/riscv.opt: Likewise.
14186 * config/riscv/sync.md: Likewise.
14187 * config/riscv/t-elf-multilib: Likewise.
14188 * config/riscv/t-linux: Likewise.
14189 * config/riscv/t-linux-multilib: Likewise.
14190 * config/riscv/t-riscv: Likewise.
14191 * configure.ac: Likewise.
14192 * doc/contrib.texi: Add Kito Cheng, Palmer Dabbelt, and Andrew
14193 Waterman as RISC-V maintainers.
14194 * doc/install.texi: Add RISC-V entries.
14195 * doc/invoke.texi: Add RISC-V options section.
14196 * doc/md.texi: Add RISC-V constraints section.
14197 * configure: Regenerated.
14199 2017-02-06 Michael Meissner <meissner@linux.vnet.ibm.com>
14202 * config/rs6000/vector.md (vcond<mode><mode>): Allow the true and
14203 false values to be constant vectors with all 0 or all 1 bits set.
14204 (vcondu<mode><mode>): Likewise.
14205 * config/rs6000/predicates.md (vector_int_reg_or_same_bit): New
14207 (fpmask_comparison_operator): Update comment.
14208 (vecint_comparison_operator): New predicate.
14209 * config/rs6000/rs6000.c (rs6000_emit_vector_cond_expr): Optimize
14210 vector conditionals when the true and false values are constant
14211 vectors with all 0 bits or all 1 bits set.
14213 2017-02-06 Martin Sebor <msebor@redhat.com>
14215 PR tree-optimization/79376
14216 * gimple-fold.c (get_range_strlen): Set the minimum length to zero.
14218 2017-02-06 Uros Bizjak <ubizjak@gmail.com>
14220 * config/i386/sse.md (vector modes -> vec_extract* splitter): Use
14221 explicit subreg RTX with operand 1. Use VECTOR_MODE_P predicate
14222 to simplify split condition.
14224 2017-02-06 Jakub Jelinek <jakub@redhat.com>
14226 * omp-expand.c (oxpand_omp_atomic_fetch_op,
14227 expand_omp_atomic_pipeline): Return false if can_atomic_load_p is
14230 2017-02-06 Segher Boessenkool <segher@kernel.crashing.org>
14232 PR rtl-optimization/68664
14233 * target.def (can_speculate_insn): New hook.
14234 * doc/tm.texi.in (TARGET_SCHED_CAN_SPECULATE_INSN): New hook.
14235 * doc/tm.texi: Regenerate.
14236 * sched-rgn.c (can_schedule_ready_p): Use the new hook.
14237 * config/rs6000/rs6000.c (TARGET_SCHED_CAN_SPECULATE_INSN): New macro.
14238 (rs6000_sched_can_speculate_insn): New function.
14240 2017-02-06 Jakub Jelinek <jakub@redhat.com>
14242 PR tree-optimization/79284
14243 * tree-vectorizer.h (VECT_SCALAR_BOOLEAN_TYPE_P): Define.
14244 * tree-vect-stmts.c (vect_get_vec_def_for_operand,
14245 vectorizable_mask_load_store, vectorizable_operation,
14246 vect_is_simple_cond, get_same_sized_vectype): Use it instead
14247 of comparing TREE_CODE of a type against BOOLEAN_TYPE.
14248 * tree-vect-patterns.c (check_bool_pattern, search_type_for_mask_1,
14249 vect_recog_bool_pattern, vect_recog_mask_conversion_pattern): Likewise.
14250 * tree-vect-slp.c (vect_get_constant_vectors): Likewise.
14251 * tree-vect-loop.c (vect_determine_vectorization_factor): Likewise.
14252 Remove redundant gimple_code (stmt) == GIMPLE_ASSIGN test after
14253 is_gimple_assign (stmt). Replace another such test with
14254 is_gimple_assign (stmt).
14256 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
14259 * config/avr/avr.c (rtl-iter.h): Include it.
14260 (TARGET_LEGITIMATE_COMBINED_INSN): New hook define...
14261 (avr_legitimate_combined_insn): ...and implementation.
14263 2017-02-06 Dominik Vogt <vogt@linux.vnet.ibm.com>
14265 * config/s390/predicates.md ("larl_operand"): Use macros from hwint.h.
14266 * config/s390/s390.c (s390_const_operand_ok)
14267 (s390_canonicalize_comparison, s390_extract_part)
14268 (s390_single_part, s390_contiguous_bitmask_nowrap_p)
14269 (s390_contiguous_bitmask_p, s390_rtx_costs)
14270 (legitimize_pic_address): Likewise.
14271 * config/s390/s390.md ("clzdi2", "clztidi2"): Likewise.
14272 * config/s390/vx-builtins.md ("vec_genbytemaskv16qi")
14273 ("vec_permi<mode>", "vfae<mode>", "*vfaes<mode>", "vstrc<mode>")
14274 ("*vstrcs<mode>"): Use UINTVAL() to set unsigned HOST_WIDE_INT.
14275 * config/s390/vector.md ("vec_vfenes<mode>"): Likewise.
14277 2017-02-06 Georg-Johann Lay <avr@gjlay.de>
14279 * config/avr/avr.md (*addhi3_zero_extend): Add alternative where
14280 REGNO($0) == REGNO($1).
14282 2017-02-06 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14284 * config/s390/linux.h(SIZE_TYPE): Add comment.
14286 2017-02-06 Julian Brown <julian@codesourcery.com>
14287 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14288 Virendra Pathak <virendra.pathak@broadcom.com>
14290 * config/aarch64/aarch64-cores.def: Change the scheduler
14292 * config/aarch64/aarch64.md: Include thunderx2t99.md.
14293 * config/aarch64/thunderx2t99.md: New file.
14295 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
14297 * doc/standards.texi (Go Language): Update link to language
14300 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
14302 * tree-eh.c (lower_resx): Sanitize profile.
14303 (cleanup_empty_eh_move_lp): Likewise.
14305 2017-02-05 Jan Hubicka <hubicka@ucw.cz>
14308 * cfgloopmanip.c (lv_adjust_loop_entry_edge, loop_version): Add
14310 * cfgloopmanip.h (loop_version): Update prototype.
14311 * modulo-sched.c (sms_schedule): Update call of loop_version.
14312 * tree-if-conv.c(version_loop_for_if_conversion): Likewise.
14313 * tree-parloops.c (gen_parallel_loop): Likewise.
14314 * tree-ssa-loop-manip.c (tree_transform_and_unroll_loop): Likewise.
14315 * tree-ssa-loop-split.c (split_loop): Likewise.
14316 * tree-ssa-loop-unswitch.c (tree_unswitch_loop): Likewise.
14317 * tree-vect-loop-manip.c (vect_loop_versioning): Likewise.
14319 2017-02-05 Martin Liska <mliska@suse.cz>
14322 * config/s390/s390.c (s390_gimplify_va_arg): Initialize local
14324 (print_operand_address): Initialize a struct to zero.
14326 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
14328 * doc/contrib.texi (Contributors): Refer to Hans Boehm's
14329 garbage collector only in textual form.
14331 2017-02-05 Gerald Pfeifer <gerald@pfeifer.com>
14333 * doc/extend.texi (x86 specific memory model extensions for
14334 transactional memory): Simplify a phrase.
14336 2017-02-05 Eric Botcazou <ebotcazou@adacore.com>
14339 * config/sparc/sync.md (atomic_loaddi_1): Replace 'U' constraint with
14340 'r', 'm' constraint with 'T' and !TARGET_ARCH64 with TARGET_ARCH32.
14341 (atomic_storedi_1): Likewise.
14343 2017-02-04 Jakub Jelinek <jakub@redhat.com>
14345 PR tree-optimization/79338
14346 * tree-parloops.c (gather_scalar_reductions): Don't call
14347 vect_analyze_loop_form for loop->inner before destroying loop's
14350 2017-02-03 Martin Sebor <msebor@redhat.com>
14352 PR tree-optimization/79327
14353 * gimple-ssa-sprintf.c (tree_digits): Avoid adding the base prefix
14354 when precision has resulted in leading zeros.
14355 (format_integer): Adjust the likely counter to assume an unknown
14356 argument that may be zero is non-zero.
14358 2017-02-03 Jason Merrill <jason@redhat.com>
14361 * tree-inline.c (copy_tree_body_r) [COND_EXPR]: Revert change to
14362 avoid copying non-taken branch.
14364 2017-02-03 Jakub Jelinek <jakub@redhat.com>
14366 PR tree-optimization/79340
14367 * tree-vect-loop.c (vectorizable_reduction): Release
14368 vec_defs elements after safe_splicing them into other vectors.
14371 PR tree-optimization/79327
14372 * gimple-ssa-sprintf.c (adjust_range_for_overflow): If returning
14373 true, always set *argmin and *argmax to TYPE_{MIN,MAX}_VALUE of
14375 (format_integer): Use wide_int_to_tree instead of build_int_cst
14376 + to_?hwi. If argmin is NULL, just set argmin and argmax to
14377 TYPE_{MIN,MAX}_VALUE of argtype. Simplify and fix computation
14378 of shortest and longest sequence.
14380 2017-02-03 Uros Bizjak <ubizjak@gmail.com>
14382 * config/i386/i386.c (dimode_scalar_chain::convert_reg):
14383 Use pextrd for TARGET_SSE4_1 when creating scalar copy.
14385 2017-02-03 Walter Lee <walt@tilera.com>
14388 * config/tilegx/tilegx.md (tilegx_expand_prologue): Add blockage
14389 after initial stackframe link reg save.
14390 * config/tilepro/tilepro.md (tilepro_expand_prologue): Likewise.
14392 2017-02-03 Jakub Jelinek <jakub@redhat.com>
14395 * config/rs6000/rs6000.md (movsi_from_sf): Use wb constraint instead of
14396 wu for stxssp alternative.
14398 2017-02-03 Martin Sebor <msebor@redhat.com>
14400 PR tree-optimization/79352
14401 * gimple-fold.c (get_range_strlen): Add argument.
14402 (get_range_strlen): Change return type to bool.
14403 (get_maxval_strlen): Pass in a dummy argument.
14404 * gimple-fold.h (get_range_strlen): Change return type to bool.
14405 * gimple-ssa-sprintf.c (get_string_length): Set unlikely counter.
14406 * tree.h (array_at_struct_end_p): Add argument.
14407 * tree.c (array_at_struct_end_p): Handle it.
14409 2017-02-03 Martin Liska <mliska@suse.cz>
14412 * multiple_target.c (create_dispatcher_calls): Redirect edge
14413 from a caller of a dispatcher.
14414 (expand_target_clones): Make the clones local.
14415 (ipa_target_clone): Do both target clones and resolvers.
14416 (ipa_dispatcher_calls): Remove the pass.
14417 (pass_dispatcher_calls::gate): Likewise.
14418 (make_pass_dispatcher_calls): Likewise.
14419 * passes.def (pass_target_clone): Put as very first IPA early
14422 2017-02-03 Martin Liska <mliska@suse.cz>
14424 * symtab.c (symtab_node::binds_to_current_def_p): Bail out
14425 in case of a function with ifunc attribute.
14427 2017-02-03 Martin Liska <mliska@suse.cz>
14429 * cgraph.c (cgraph_node::dump): Dump function version info.
14430 * symtab.c (symtab_node::dump_base): Add missing new line.
14432 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
14434 * tree-ssa-ifcombine.c (update_profile_after_ifcombine): New function.
14435 (ifcombine_ifandif): Use it.
14437 2017-02-03 Martin Liska <mliska@suse.cz>
14439 * doc/invoke.texi: Document default value for
14440 use-after-scope-direct-emission-threshold.
14442 2017-02-03 Martin Liska <mliska@suse.cz>
14444 PR tree-optimization/79339
14445 * gimple-ssa-sprintf.c (format_floating_max): Call mpfr_clear.
14446 (format_floating): Likewise.
14448 2017-02-03 Martin Liska <mliska@suse.cz>
14451 * ipa-prop.c (ipa_node_params_t::insert): Remove current
14453 (ipa_node_params_t::remove): Likewise.
14454 * ipa-prop.h (ipa_node_params::ipa_node_params): Make default
14455 initialization from removed ipa_node_params_t::insert.
14456 (ipa_node_params::~ipa_node_params): Move from removed
14457 ipa_node_params_t::release.
14458 * symbol-summary.h (symbol_summary::m_released): New member.
14459 Do not release a summary twice. Do not allow to call finalizer
14460 for types of a summary that live in GGC memory.
14462 2017-02-02 Naveen H.S <Naveen.Hurugalawadi@cavium.com>
14464 * config/aarch64/aarch64.c (thunderx2t99_tunings): Enable AES and
14467 2017-02-02 Martin Sebor <msebor@redhat.com>
14469 PR middle-end/79275
14470 * gimple-ssa-sprintf.c (get_string_length): Set lower bound to zero.
14471 (format_string): Tighten up the range of output for non-constant
14472 strings and correct the expected range for wide non-constant strings.
14474 2017-02-02 Martin Sebor <msebor@redhat.com>
14476 * doc/invoke.texi (-maccumulate-args): Fix bad grammar.
14478 PR middle-end/32003
14479 * doc/invoke.texi (-fdump-tree-): Remove pass-specific options from
14481 (-fdump-tree-@var): Add to index and document how to come up
14482 with pass-specific option and dump file names.
14483 (-fdump-passes): Clarify where to look for output.
14485 2017-02-02 Jan Hubicka <hubicka@ucw.cz>
14487 PR middle-end/77445
14488 * tree-ssa-threadbackward.c (profitable_jump_thread_path): Dump
14489 statistics of the analyzed path; allow threading for speed when
14490 any of BBs along the path are optimized for speed.
14492 2017-02-02 Eric Botcazou <ebotcazou@adacore.com>
14494 PR middle-end/78468
14495 * emit-rtl.c (init_emit): Add ??? comment for problematic alignment
14496 settings of the virtual registers.
14499 2016-08-23 Dominik Vogt <vogt@linux.vnet.ibm.com>
14501 * explow.c (get_dynamic_stack_size): Take known alignment of stack
14502 pointer + STACK_DYNAMIC_OFFSET into account when calculating the size
14505 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14507 * config/s390/vx-builtins.md ("vec_ceil", "vec_floor")
14508 ("vec_trunc", "vec_roundc", "vec_round"): Remove expanders.
14510 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14512 * config/s390/s390.md: Add missing comments with the expanded
14514 * config/s390/vector.md: Likewise.
14515 * config/s390/vx-builtins.md: Likewise.
14517 2017-02-02 Jakub Jelinek <jakub@redhat.com>
14520 * config/rs6000/rs6000.md (*fixuns_trunc<mode>di2_fctiduz): Rename to ...
14521 (fixuns_trunc<mode>di2): ... this, remove previous expander. Put all
14522 conditions on a single line.
14524 2017-02-02 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14526 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
14527 __S390_VX__ to __VX__.
14529 2017-02-01 Andrew Pinski <apinski@cavium.com>
14531 * tree-vect-loop.c (vect_compute_single_scalar_iteration_cost): Pass
14532 stmt_info to record_stmt_cost.
14533 (vect_get_known_peeling_cost): Pass stmt_info if known to
14535 * config/aarch64/aarch64-protos.h (cpu_vector_cost): Split
14536 cpu_vector_cost field into
14537 scalar_int_stmt_cost and scalar_fp_stmt_cost. Split vec_stmt_cost
14538 field into vec_int_stmt_cost and vec_fp_stmt_cost.
14539 * config/aarch64/aarch64.c (generic_vector_cost): Update for the
14540 splitting of scalar_stmt_cost and vec_stmt_cost.
14541 (thunderx_vector_cost): Likewise.
14542 (cortexa57_vector_cost): LIkewise.
14543 (exynosm1_vector_cost): Likewise.
14544 (xgene1_vector_cost): Likewise.
14545 (thunderx2t99_vector_cost): Improve after the splitting of the two
14547 (aarch64_builtin_vectorization_cost): Update for the splitting of
14548 scalar_stmt_cost and vec_stmt_cost.
14550 2017-02-01 Torvald Riegel <triegel@redhat.com>
14551 Richard Henderson <rth@redhat.com>
14553 * builtins.c (fold_builtin_atomic_always_lock_free): Make "lock-free"
14554 conditional on existance of a fast atomic load.
14555 * optabs-query.c (can_atomic_load_p): New function.
14556 * optabs-query.h (can_atomic_load_p): Declare it.
14557 * optabs.c (expand_atomic_exchange): Always delegate to libatomic if
14558 no fast atomic load is available for the particular size of access.
14559 (expand_atomic_compare_and_swap): Likewise.
14560 (expand_atomic_load): Likewise.
14561 (expand_atomic_store): Likewise.
14562 (expand_atomic_fetch_op): Likewise.
14563 * testsuite/lib/target-supports.exp
14564 (check_effective_target_sync_int_128): Remove x86 because it provides
14565 no fast atomic load.
14566 (check_effective_target_sync_int_128_runtime): Likewise.
14568 2017-02-01 Richard Biener <rguenther@suse.de>
14570 * graphite.c: Include tree-vectorizer.h for find_loop_location.
14571 (graphite_transform_loops): Provide opt-info for optimized nests.
14572 * tree-parloop.c (parallelize_loops): Provide opt-info for
14573 parallelized loops.
14575 2017-02-01 Richard Biener <rguenther@suse.de>
14577 PR middle-end/79315
14578 * tree-cfg.c (move_stmt_op): Never set TREE_BLOCK when it
14579 was not set before.
14581 2017-02-01 Richard Biener <rguenther@suse.de>
14583 PR tree-optimization/71824
14584 * graphite-scop-detection.c (scop_detection::build_scop_breadth):
14585 Verify the loops are valid in the merged SESE region.
14586 (scop_detection::can_represent_loop_1): Check analyzing the
14587 evolution of the number of iterations in the region succeeds.
14589 2017-01-31 Ian Lance Taylor <iant@golang.org>
14591 * config/i386/i386.c (ix86_expand_split_stack_prologue): Add
14592 REG_ARGS_SIZE note to 32-bit push insns and call insn.
14594 2017-01-31 David Malcolm <dmalcolm@redhat.com>
14596 PR preprocessor/79210
14597 * input.c (get_substring_ranges_for_loc): Replace line_width
14598 assertion with error-handling.
14600 2017-01-31 Richard Biener <rguenther@suse.de>
14602 PR tree-optimization/77318
14603 * graphite-sese-to-poly.c (extract_affine): Fix assert.
14604 (create_pw_aff_from_tree): Take loop parameter.
14605 (add_condition_to_pbb): Pass loop of the condition to
14606 create_pw_aff_from_tree.
14608 2017-01-31 Jakub Jelinek <jakub@redhat.com>
14610 * config/s390/s390.c (s390_asan_shadow_offset): New function.
14611 (TARGET_ASAN_SHADOW_OFFSET): Redefine.
14613 2017-01-31 Michael Meissner <meissner@linux.vnet.ibm.com>
14617 * config/rs6000/rs6000-protos.h (convert_float128_to_int): Delete,
14619 (convert_int_to_float128): Likewise.
14620 * config/rs6000/rs6000.c (convert_float128_to_int): Likewise.
14621 (convert_int_to_float128): Likewise.
14622 * config/rs6000/rs6000.md (UNSPEC_IEEE128_MOVE): Likewise.
14623 (UNSPEC_IEEE128_CONVERT): Likewise.
14624 (floatsi<mode>2, FLOAT128 iterator): Bypass calling
14625 rs6000_expand_float128_convert if we have IEEE 128-bit hardware.
14626 Use local variables for IBM extended format.
14627 (fix_trunc<mode>si2, FLOAT128 iterator): Likewise.
14628 (fix_trunc<mode>si2_fprs): Likewise.
14629 (fixuns_trunc<IEEE128:mode><SDI:mode>2): Likewise.
14630 (floatuns<IEEE128:mode>2, IEEE128 iterator): Likewise.
14631 (fix<uns>_<mode>si2_hw): Rework the IEEE 128-bt hardware support
14632 to know that we can now have integers of all sizes in vector
14634 (fix<uns>_<mode>di2_hw): Likewise.
14635 (float<uns>_<mode>si2_hw): Likewise.
14636 (fix_<mode>si2_hw): Likewise.
14637 (fixuns_<mode>si2_hw): Likewise.
14638 (float<uns>_<mode>di2_hw): Likewise.
14639 (float_<mode>di2_hw): Likewise.
14640 (float_<mode>si2_hw): Likewise.
14641 (floatuns_<mode>di2_hw): Likewise.
14642 (floatuns_<mode>si2_hw): Likewise.
14643 (xscvqp<su>wz_<mode>): Delete, no longer used.
14644 (xscvqp<su>dz_<mode>): Likewise.
14645 (xscv<su>dqp_<mode>): Likewise.
14646 (ieee128_mfvsrd_64bit): Likewise.
14647 (ieee128_mfvsrd_32bit): Likewise.
14648 (ieee128_mfvsrwz): Likewise.
14649 (ieee128_mtvsrw): Likewise.
14650 (ieee128_mtvsrd_64bit): Likewise.
14651 (ieee128_mtvsrd_32bit): Likewise.
14653 2017-01-31 Martin Liska <mliska@suse.cz>
14656 * ipa-prop.c (ipa_free_all_node_params): Call release method
14657 instead of ~sumbol_summary to not to trigger double times
14660 2017-01-31 Aldy Hernandez <aldyh@redhat.com>
14662 PR tree-optimization/71691
14663 * bitmap.h (class auto_bitmap): New.
14664 * tree-ssa-loop-unswitch.c (tree_may_unswitch_on): Call
14665 is_maybe_undefined instead of ssa_undefined_value_p.
14667 2017-01-31 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
14669 * config/s390/s390-c.c (s390_cpu_cpp_builtins_internal): Rename
14670 __S390_ARCH_LEVEL__ to __ARCH__.
14672 2017-01-31 Jakub Jelinek <jakub@redhat.com>
14674 PR tree-optimization/79267
14675 * value-prof.c (gimple_ic): Only drop lhs for noreturn calls
14676 if should_remove_lhs_p is true.
14678 2017-01-30 Alexandre Oliva <aoliva@redhat.com>
14681 * dwarf2out.c (clone_as_declaration): Drop DW_AT_alignment.
14682 (add_alignment_attribute): New.
14683 (base_type_die): Add alignment attribute.
14684 (subrange_type_die): Likewise.
14685 (modified_type_die): Likewise.
14686 (gen_array_type_die): Likewise.
14687 (gen_descr_array_type_die: Likewise.
14688 (gen_enumeration_type_die): Likewise.
14689 (gen_subprogram_die): Likewise.
14690 (gen_variable_die): Likewise.
14691 (gen_field_die): Likewise.
14692 (gen_ptr_to_mbr_type_die): Likewise.
14693 (gen_struct_or_union_type_die): Likewise.
14694 (gen_subroutine_type_die): Likewise.
14695 (gen_typedef_die): Likewise.
14696 (base_type_cmp): Compare alignment attribute.
14698 2017-01-30 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
14701 * config/rs6000/altivec.md (*setb_internal): Rename to setb_signed.
14702 (setb_unsigned) New pattern for setb with CCUNS.
14703 * config/rs6000/rs6000.c (expand_block_compare): Use a different
14704 subfc./subfe sequence to avoid overflow problems. Generate a
14705 shorter sequence with cmpld/setb for power9.
14706 * config/rs6000/rs6000.md (subf<mode>3_carry_dot2): Add a new pattern
14707 for generating subfc. instruction.
14708 (cmpstrsi): Add TARGET_POPCNTD predicate as the generate sequence
14709 now uses this instruction.
14711 2017-01-30 Ian Lance Taylor <iant@google.com>
14714 * dwarf2out.c (gen_type_die_with_usage): When picking a variant
14715 for FUNCTION_TYPE/METHOD_TYPE, use the first matching one.
14717 2017-01-30 Martin Sebor <msebor@redhat.com>
14719 * gimple-ssa-sprintf.c (fmtresult::adjust_for_width_or_precision):
14720 Move constant to the right of a relational operator.
14721 (get_mpfr_format_length, format_character, format_string): Ditto.
14722 (should_warn_p, maybe_warn): Same.
14724 * doc/invoke.texi (-Wformat-truncation=1): Fix typo.
14726 2017-01-30 Maxim Ostapenko <m.ostapenko@samsung.com>
14729 * asan.c (get_translation_unit_decl): Remove function.
14730 (asan_add_global): Force has_dynamic_init to zero in LTO mode.
14732 2017-01-30 Martin Liska <mliska@suse.cz>
14734 PR gcov-profile/79259
14735 * opts.c (common_handle_option): Enable flag_ipa_bit_cp w/
14736 -fprofile-generate.
14738 2017-01-30 Martin Liska <mliska@suse.cz>
14741 * config/aarch64/cortex-a57-fma-steering.c (func_fma_steering::analyze):
14742 Initialize variables with NULL value.
14744 2017-01-30 Richard Earnshaw <rearnsh@arm.com>
14747 * config.gcc (arm*-*-*): Add arm/arm-flags.h and arm/arm-isa.h to
14749 * arm/arm-protos.h: Don't directly include arm-flags.h and arm-isa.h.
14751 2017-01-30 Richard Biener <rguenther@suse.de>
14753 PR tree-optimization/79276
14754 * tree-vrp.c (process_assert_insertions): Properly adjust common
14755 when removing a duplicate.
14757 * gcc.dg/torture/pr79276.c: New testcase.
14759 2017-01-30 Richard Biener <rguenther@suse.de>
14761 PR tree-optimization/79256
14762 * targhooks.c (default_builtin_vector_alignment_reachable): Honor
14763 BIGGEST_FIELD_ALIGNMENT and ADJUST_FIELD_ALIGN to fix up bogus
14765 * tree.c (build_aligned_type): Set TYPE_USER_ALIGN.
14767 2017-01-30 Dominik Vogt <vogt@linux.vnet.ibm.com>
14770 * config/s390/s390.md ("*r<noxa>sbg_<mode>_srl_bitmask")
14771 ("*r<noxa>sbg_<mode>_sll_bitmask")
14772 ("*extzv_<mode>_srl<clobbercc_or_nocc>")
14773 ("*extzv_<mode>_sll<clobbercc_or_nocc>"):
14774 Use contiguous_bitmask_nowrap_operand.
14776 2017-01-29 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
14779 * config/rs6000/altivec.h (vec_xl): Revise #define.
14780 (vec_xst): Likewise.
14782 2017-01-27 Uros Bizjak <ubizjak@gmail.com>
14784 * config/i386/i386.c (print_reg): Use REGNO instead of true_regnum.
14786 2017-01-27 Bernd Schmidt <bschmidt@redhat.com>
14788 PR rtl-optimization/79194
14789 * cprop.c (one_cprop_pass): Move deletion of code after unconditional
14790 traps before call to bypass_conditional_jumps.
14792 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
14794 PR tree-optimization/71374
14795 * lra-constraints.c (check_conflict_input_operands): New.
14796 (match_reload): Use it.
14798 2017-01-27 Vladimir Makarov <vmakarov@redhat.com>
14801 * lra-assigns.c (find_hard_regno_for_1): Take endianess for into
14802 account to calculate conflict_set.
14804 2017-01-27 Bin Cheng <bin.cheng@arm.com>
14806 PR rtl-optimization/78559
14807 * combine.c (try_combine): Discard REG_EQUAL and REG_EQUIV for
14808 other_insn in combine.
14810 2017-01-27 Pekka Jääskeläinen <pekka.jaaskelainen@parmance.com>
14812 * builtin-types.def: Use unsigned_char_type_node for BT_UINT8. Use
14813 uint16_type_node for BT_UINT16.
14815 2017-01-27 David Malcolm <dmalcolm@redhat.com>
14817 * doc/sourcebuild.texi (Testsuites): Add "GIMPLE Tests" and
14818 "RTL Tests" to menu.
14819 (GIMPLE Tests): New node.
14820 (RTL Tests): New node.
14822 2017-01-27 Richard Biener <rguenther@suse.de>
14824 PR tree-optimization/79245
14825 * tree-loop-distribution.c (distribute_loop): Apply cost
14826 modeling also to detected patterns.
14828 2017-01-27 Richard Biener <rguenther@suse.de>
14830 PR tree-optimization/71433
14831 * tree-vrp.c (register_new_assert_for): Revert earlier changes.
14832 (compare_assert_loc): New function.
14833 (process_assert_insertions): Sort and optimize assert locations
14834 to remove duplicates and push down identical assertions on
14835 edges to their destination block.
14837 2017-01-27 Richard Biener <rguenther@suse.de>
14839 PR tree-optimization/79244
14840 * tree-vrp.c (remove_range_assertions): Forcefully propagate
14841 out SSA names even if abnormal.
14843 2017-01-27 Jakub Jelinek <jakub@redhat.com>
14845 * realmpfr.h: Poison MPFR_RND{N,Z,U,D}.
14846 * gimple-ssa-sprintf.c (format_floating_max): Use GMP_RNDN
14847 instead of MPFR_RNDN.
14849 2017-01-27 Richard Earnshaw <rearnsha@arm.com>
14852 * arm.c (arm_option_override): Don't call build_target_option_node
14853 until after doing all option overrides.
14854 (arm_valid_target_attribute_tree): Likewise.
14856 2017-01-27 Martin Liska <mliska@suse.cz>
14858 * doc/invoke.texi (-fprofile-arcs): Document profiling support
14859 for {cd}tors and C++ {cd}tors.
14861 2017-01-27 Dominik Vogt <vogt@linux.vnet.ibm.com>
14863 * config/s390/s390.md ("*setmem_long_and")
14864 ("*setmem_long_and_31z"): Use zero_extend instead of and.
14866 2017-01-26 Martin Sebor <msebor@redhat.com>
14868 * gimple-ssa-sprintf.c (format_floating): Simplify the computation
14871 2017-01-26 Martin Sebor <msebor@redhat.com>
14873 * gimple-ssa-sprintf.c (format_floating): Test HAVE_XFmode and
14874 HAVE_DFmode before using XFmode or DFmode.
14875 (parse_directive): Avoid using the z length modifier to avoid
14876 the ISO C++98 does not support the ‘z’ gnu_printf length modifier.
14878 PR middle-end/78703
14879 * gimple-ssa-sprintf.c (adjust_for_width_or_precision): Change
14880 to accept adjustment as an array.
14881 (get_int_range): New function.
14882 (struct directive): Make width and prec arrays.
14883 (directive::set_width, directive::set_precision): Call get_int_range.
14884 (format_integer, format_floating): Handle width and precision ranges.
14885 (format_string, parse_directive): Same.
14887 2017-01-26 Jakub Jelinek <jakub@redhat.com>
14890 * dwarf2out.c (generate_skeleton_bottom_up): For children with
14891 comdat_type_p set, just clone them, but keep the children in the
14895 * dwarf2out.c (prune_unused_types): Mark all functions with DIEs
14896 which have direct callers with -fvar-tracking-assignments enabled
14898 (resolve_addr): Avoid adding skeleton DIEs for DW_AT_call_origin
14899 inside of type units.
14901 2017-01-26 Martin Sebor <msebor@redhat.com>
14903 PR middle-end/78703
14904 * gimple-ssa-sprintf.c (struct result_range): Add likely and
14906 (struct format_result): Replace number_chars, number_chars_min,
14907 and number_chars_max with a single member of struct result_range.
14909 (format_result::operator+=): Adjust.
14910 (struct fmtresult): Remove bounded. Handle likely and unlikely
14912 (fmtresult::adjust_for_width_or_precision): New function.
14913 (fmtresult:type_max_digits): New function.
14914 (bytes_remaining): Handle likely and unlikely counters.
14915 (min_bytes_remaining): Remove.
14916 (format_percent): Simplify.
14917 (format_integer, format_floating): Set likely and unlikely counters.
14918 (get_string_length, format_character, format_string): Same.
14919 (format_plain, should_warn_p): New function.
14920 (maybe_warn): Call should_warn_p. Update diagnostic messages
14921 and handle those for all directives, including plain strings.
14922 (format_directive): Handle likely and unlikely counters.
14923 Remove unnecessary quoting from diagnostics. Add an informational
14925 (add_bytes): Remove.
14926 (pass_sprintf_length::compute_format_length): Simplify.
14927 (try_substitute_return_value): Handle likely and unlikely counters.
14929 2017-01-26 Carl Love <cel@us.ibm.com>
14931 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Remove
14932 bogus entries for the P8V_BUILTIN_VEC_VGBBD built-ins
14934 2017-01-26 Vladimir Makarov <vmakarov@redhat.com>
14937 * lra-assigns.c (setup_live_pseudos_and_spill_after_risky): Take
14938 endianess for subregs into account.
14939 * lra-constraints.c (lra_constraints): Do risky transformations
14940 always on the first iteration.
14941 * lra-lives.c (check_pseudos_live_through_calls): Add arg
14942 last_call_used_reg_set.
14943 (process_bb_lives): Define and use last_call_used_reg_set.
14944 * lra.c (lra): Always continue after lra_constraints on the first
14947 2017-01-26 Kirill Yukhin <kirill.yukhin@gmail.com>
14949 * gcc.target/i386/avx512bw-kshiftlq-2.c: Use unsigned long long
14951 * gcc.target/i386/avx512bw-kshiftrq-2.c: Ditto.
14953 2017-01-26 Jakub Jelinek <jakub@redhat.com>
14955 * config/i386/avx512fintrin.h (_ktest_mask16_u8,
14956 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): Move to ...
14957 * config/i386/avx512dqintrin.h (_ktest_mask16_u8,
14958 _ktestz_mask16_u8, _ktestc_mask16_u8, _kadd_mask16): ... here.
14959 * config/i386/i386-builtin.def (__builtin_ia32_ktestchi,
14960 __builtin_ia32_ktestzhi, __builtin_ia32_kaddhi): Use
14961 OPTION_MASK_ISA_AVX512DQ instead of OPTION_MASK_ISA_AVX512F.
14962 * config/i386/sse.md (SWI1248_AVX512BWDQ2): New mode iterator.
14963 (kadd<mode>, ktest<mode>): Use it instead of SWI1248_AVX512BWDQ.
14965 2017-01-26 Marek Polacek <polacek@redhat.com>
14968 * fold-const.c (operand_equal_p) [COND_EXPR]: Use OP_SAME_WITH_NULL
14969 for the third operand.
14971 2017-01-26 Jakub Jelinek <jakub@redhat.com>
14973 PR middle-end/79236
14974 * omp-low.c (struct omp_context): Add simt_stmt field.
14975 (scan_omp_for): Return omp_context *.
14976 (scan_omp_simd): Set simt_stmt on the non-_simt_ SIMD
14977 context to the _simt_ SIMD stmt.
14978 (lower_omp_for): For combined SIMD with sibling _simt_
14979 SIMD, make sure to use the same decls in _looptemp_
14980 clauses as in the sibling.
14982 2017-01-26 David Sherwood <david.sherwood@arm.com>
14984 PR middle-end/79212
14985 * gimplify.c (omp_notice_variable): Add GOVD_SEEN flag to variables in
14988 2017-01-26 Jakub Jelinek <jakub@redhat.com>
14991 * reg-stack.c (emit_swap_insn): Instead of fld a; fld b; fxchg %st(1);
14992 emit fld b; fld a; if possible.
14994 * brig-builtins.def: Update copyright years.
14995 * config/arm/arm_acle_builtins.def: Update copyright years.
14997 2017-01-25 Michael Meissner <meissner@linux.vnet.ibm.com>
15000 * config/rs6000/vsx.md (vsx_extract_<mode>_store): Use wY
15001 constraint instead of o for the stxsd instruction.
15003 2017-01-25 Carl Love <cel@us.ibm.com>
15005 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Fix order
15006 of entries for ALTIVEC_BUILTIN_VEC_PACKS and P8V_BUILTIN_VEC_VGBBD.
15008 2017-01-25 Jonathan Wakely <jwakely@redhat.com>
15010 * doc/invoke.texi (C++ Dialect Options): Fix typo.
15012 2017-01-25 Richard Biener <rguenther@suse.de>
15014 PR tree-optimization/69264
15015 * target.def (vector_alignment_reachable): Improve documentation.
15016 * doc/tm.texi: Regenerate.
15017 * targhooks.c (default_builtin_vector_alignment_reachable): Simplify
15019 * tree-vect-data-refs.c (vect_supportable_dr_alignment): Revert
15020 earlier changes with respect to TYPE_USER_ALIGN.
15021 (vector_alignment_reachable_p): Likewise. Improve dumping.
15023 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15026 * config/arm/arm.md (xordi3): Force constant operand into a register
15029 2016-01-25 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15031 * doc/invoke.texi (-fstore-merging): Correct default optimization
15032 levels at which it is enabled.
15033 (-O): Move -fstore-merging from list to...
15036 2017-01-25 Richard Biener <rguenther@suse.de>
15039 * omp-expand.c: Include debug.h.
15040 (expand_omp_taskreg): Make sure to generate early debug before
15041 outlining anything from a function.
15042 (expand_omp_target): Likewise.
15043 (grid_expand_target_grid_body): Likewise.
15045 2017-01-25 Maxim Ostapenko <m.ostapenko@samsung.com>
15048 * asan.c (get_translation_unit_decl): New function.
15049 (asan_add_global): Extract modules file name from globals
15050 TRANSLATION_UNIT_DECL name.
15052 2017-01-24 Eric Botcazou <ebotcazou@adacore.com>
15055 * config/arm/arm.c (arm_function_ok_for_sibcall): Add back restriction
15056 for long calls with APCS frame and VFP.
15058 2017-01-24 David Malcolm <dmalcolm@redhat.com>
15060 * cfg.c (original_copy_tables_initialized_p): New function.
15061 * cfg.h (original_copy_tables_initialized_p): New decl.
15062 * cfgrtl.c (relink_block_chain): Guard the call to
15063 free_original_copy_tables with a call to
15064 original_copy_tables_initialized_p.
15065 * cgraph.h (symtab_node::native_rtl_p): New decl.
15066 * cgraphunit.c (symtab_node::native_rtl_p): New function.
15067 (symtab_node::needed_p): Don't assert for early assembly output
15068 for __RTL functions.
15069 (cgraph_node::finalize_function): Set "force_output" for __RTL
15071 (cgraph_node::analyze): Bail out early for __RTL functions.
15072 (analyze_functions): Update assertion to support __RTL functions.
15073 (cgraph_node::expand): Bail out early for __RTL functions.
15074 * final.c (rest_of_clean_state): Don't call delete_tree_ssa for
15076 * function.h (struct function): Update comment for field
15078 * gimple-expr.c: Include "tree-pass.h".
15079 (gimple_has_body_p): Return false for __RTL functions.
15080 * Makefile.in (OBJS): Add run-rtl-passes.o.
15081 * pass_manager.h (gcc::pass_manager::get_rest_of_compilation): New
15083 (gcc::pass_manager::get_clean_slate): New accessor.
15084 * passes.c: Include "insn-addr.h".
15085 (should_skip_pass_p): Add logging. Update logic for running
15086 "expand" to be compatible with both __GIMPLE and __RTL. Guard
15087 property-provider override so it is only done for gimple passes.
15089 (skip_pass): New function.
15090 (execute_one_pass): Call skip_pass when skipping passes.
15091 * read-md.c (md_reader::read_char): Support filtering
15092 the input to a subset of line numbers.
15093 (md_reader::md_reader): Initialize fields
15094 m_first_line and m_last_line.
15095 (md_reader::read_file_fragment): New function.
15096 * read-md.h (md_reader::read_file_fragment): New decl.
15097 (md_reader::m_first_line): New field.
15098 (md_reader::m_last_line): New field.
15099 * read-rtl-function.c (function_reader::create_function): Only
15100 create cfun if it doesn't already exist. Set PROP_rtl on cfun's
15101 curr_properties. Set DECL_INITIAL to a dummy block.
15102 (read_rtl_function_body_from_file_range): New function.
15103 * read-rtl-function.h (read_rtl_function_body_from_file_range):
15105 * run-rtl-passes.c: New file.
15106 * run-rtl-passes.h: New file.
15108 2017-01-24 Jeff Law <law@redhat.com>
15110 * config/microblaze/microblaze.h (ASM_FORMAT_PRIVATE_NAME): Increase
15113 2017-01-24 Bin Cheng <bin.cheng@arm.com>
15115 PR tree-optimization/79159
15116 * tree-ssa-loop-niter.c (get_cst_init_from_scev): New function.
15117 (record_nonwrapping_iv): Improve boundary using above function if no
15118 value range information.
15120 2017-01-24 Pekka Jääskeläinen <pekka@parmance.com>
15121 Martin Jambor <mjambor@suse.cz>
15123 * brig-builtins.def: New file.
15124 * builtins.def (DEF_HSAIL_BUILTIN): New macro.
15125 (DEF_HSAIL_ATOMIC_BUILTIN): Likewise.
15126 (DEF_HSAIL_SAT_BUILTIN): Likewise.
15127 (DEF_HSAIL_INTR_BUILTIN): Likewise.
15128 (DEF_HSAIL_CVT_ZEROI_SAT_BUILTIN): Likewise.
15129 * builtin-types.def (BT_INT8): New.
15130 (BT_INT16): Likewise.
15131 (BT_UINT8): Likewise.
15132 (BT_UINT16): Likewise.
15133 (BT_FN_ULONG): Likewise.
15134 (BT_FN_UINT_INT): Likewise.
15135 (BT_FN_UINT_ULONG): Likewise.
15136 (BT_FN_UINT_LONG): Likewise.
15137 (BT_FN_UINT_PTR): Likewise.
15138 (BT_FN_ULONG_PTR): Likewise.
15139 (BT_FN_INT8_FLOAT): Likewise.
15140 (BT_FN_INT16_FLOAT): Likewise.
15141 (BT_FN_UINT32_FLOAT): Likewise.
15142 (BT_FN_UINT16_FLOAT): Likewise.
15143 (BT_FN_UINT8_FLOAT): Likewise.
15144 (BT_FN_UINT64_FLOAT): Likewise.
15145 (BT_FN_UINT16_UINT32): Likewise.
15146 (BT_FN_UINT32_UINT16): Likewise.
15147 (BT_FN_UINT16_UINT16_UINT16): Likewise.
15148 (BT_FN_INT_PTR_INT): Likewise.
15149 (BT_FN_UINT_PTR_UINT): Likewise.
15150 (BT_FN_LONG_PTR_LONG): Likewise.
15151 (BT_FN_ULONG_PTR_ULONG): Likewise.
15152 (BT_FN_VOID_UINT64_UINT64): Likewise.
15153 (BT_FN_UINT8_UINT8_UINT8): Likewise.
15154 (BT_FN_INT8_INT8_INT8): Likewise.
15155 (BT_FN_INT16_INT16_INT16): Likewise.
15156 (BT_FN_INT_INT_INT): Likewise.
15157 (BT_FN_UINT_FLOAT_UINT): Likewise.
15158 (BT_FN_FLOAT_UINT_UINT): Likewise.
15159 (BT_FN_ULONG_UINT_UINT): Likewise.
15160 (BT_FN_ULONG_UINT_PTR): Likewise.
15161 (BT_FN_ULONG_ULONG_ULONG): Likewise.
15162 (BT_FN_UINT_UINT_UINT): Likewise.
15163 (BT_FN_VOID_UINT_PTR): Likewise.
15164 (BT_FN_UINT_UINT_PTR: Likewise.
15165 (BT_FN_UINT32_UINT64_PTR): Likewise.
15166 (BT_FN_INT_INT_UINT_UINT): Likewise.
15167 (BT_FN_UINT_UINT_UINT_UINT): Likewise.
15168 (BT_FN_UINT_UINT_UINT_PTR): Likewise.
15169 (BT_FN_UINT_ULONG_ULONG_UINT): Likewise.
15170 (BT_FN_ULONG_ULONG_ULONG_ULONG): Likewise.
15171 (BT_FN_LONG_LONG_UINT_UINT): Likewise.
15172 (BT_FN_ULONG_ULONG_UINT_UINT): Likewise.
15173 (BT_FN_VOID_UINT32_UINT64_PTR): Likewise.
15174 (BT_FN_VOID_UINT32_UINT32_PTR): Likewise.
15175 (BT_FN_UINT_UINT_UINT_UINT_UINT): Likewise.
15176 (BT_FN_UINT_FLOAT_FLOAT_FLOAT_FLOAT): Likewise.
15177 (BT_FN_ULONG_ULONG_ULONG_UINT_UINT): Likewise.
15178 * doc/frontends.texi: List BRIG FE.
15179 * doc/install.texi (Testing): Add BRIG tesring requirements.
15180 * doc/invoke.texi (Overall Options): Mention BRIG.
15181 * doc/standards.texi (Standards): Doucment BRIG HSA version.
15183 2017-01-24 Richard Biener <rguenther@suse.de>
15185 PR translation/79208
15186 * ipa-devirt.c (odr_types_equivalent_p): Fix typo in diagnostic.
15188 2017-01-24 Martin Jambor <mjambor@suse.cz>
15191 * ipa-prop.c (ipa_free_all_node_params): Call summary destructor.
15192 * ipa-prop.c (ipa_node_params_t::insert): Initialize fields known_csts
15193 and known_contexts.
15195 2017-01-24 Aldy Hernandez <aldyh@redhat.com>
15197 PR middle-end/79123
15198 * gimple-ssa-warn-alloca.c (alloca_call_type): Make sure
15199 casts from signed to unsigned really don't have a range.
15201 2017-01-24 Markus Trippelsdorf <markus@trippelsdorf.de>
15203 * gimple-ssa-sprintf.c (format_floating): Change MPFR_RNDx to
15204 GMP_RNDx for compatiblity.
15206 2017-01-24 Martin Liska <mliska@suse.cz>
15209 * tree-ssa-reassoc.c (rewrite_expr_tree_parallel): Insert assert
15210 that would prevent us to call alloca with -1 as argument.
15212 2017-01-24 Jakub Jelinek <jakub@redhat.com>
15214 * dwarf2out.c (output_compilation_unit_header, output_file_names):
15215 Avoid -Wformat-security warning.
15217 2017-01-23 Andrew Pinski <apinski@cavium.com>
15219 * config/aarch64/aarch64.c (thunderx2t99_addrcost_table): Improve
15222 2017-01-23 Martin Sebor <msebor@redhat.com>
15224 PR middle-end/78703
15225 * gimple-ssa-sprintf.c (warn_level): New global.
15226 (format_integer): Use it here and throughout the rest of the file.
15227 Use the same switch to compute sign as base.
15228 (maybe_warn): New function.
15229 (format_directive): Factor out warnings into maybe_warn.
15230 Add debugging output. Use warn_level.
15231 (add_bytes): Use warn_level.
15232 (pass_sprintf_length::compute_format_length): Add debugging output.
15233 (try_substitute_return_value): Same.
15234 (pass_sprintf_length::handle_gimple_call): Set and use warn_level.
15236 PR middle-end/78703
15237 * gimple-ssa-sprintf.c (struct format_result): Remove constant member.
15238 (struct fmtresult, format_integer, format_floating): Adjust.
15239 (fmtresult::fmtresult): Set max correctly in two argument ctor.
15240 (get_string_length, format_string,format_directive): Same.
15241 (pass_sprintf_length::compute_format_length): Same.
15242 (try_substitute_return_value): Simplify slightly.
15244 PR middle-end/78703
15245 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust formatting.
15246 (fmtresult::operator+=): Outlined.
15247 (struct fmtresult): Add ctors.
15248 (struct conversion_spec): Rename...
15249 (struct directive): ...to this. Add and remove data members.
15250 (directive::set_width, directive::set_precision): New functions.
15251 (format_percent): Use fmtresult ctor.
15252 (get_width_and_precision): Remove.
15253 (format_integer): Make naming changes. Avoid computing width and
15255 (format_floating): Same. Adjust indentation.
15256 (format_character, format_none): New functions.
15257 (format_string): Moved character handling to format_character.
15258 (format_directive): Remove arguments, change return type.
15259 (parse_directive): New function.
15260 (pass_sprintf_length::compute_format_length): Move directive
15261 parsing to parse_directive.
15263 2017-01-23 Jakub Jelinek <jakub@redhat.com>
15265 * tree.h (assign_assembler_name_if_neeeded): Rename to ...
15266 (assign_assembler_name_if_needed): ... this.
15267 * tree.c (assign_assembler_name_if_neeeded): Rename to ...
15268 (assign_assembler_name_if_needed): ... this.
15269 (free_lang_data_in_cgraph): Adjust callers.
15270 * cgraphunit.c (cgraph_node::analyze): Likewise.
15271 * omp-expand.c (expand_omp_taskreg, expand_omp_target):
15274 2017-01-23 Richard Biener <rguenther@suse.de>
15276 PR tree-optimization/79088
15277 PR tree-optimization/79188
15278 * tree-ssa-threadupdate.c (mark_threaded_blocks): Move code
15279 resetting loop bounds after last path deletion. Reset loop
15280 bounds of the target loop, make code match the comments.
15281 * tree-ssa-threadbackwards.c (pass_early_thread_jumps::execute):
15282 Make sure loops need no fixups.
15284 2017-01-23 Kelvin Nilsen <kelvin@gcc.gnu.org>
15286 * config/rs6000/rs6000-builtin.def (VSIEDPF): Add scalar insert
15287 exponent support with double type for first argument.
15288 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Changed
15289 type returned by __builtin_vec_extract_sig,
15290 __builtin_vec_extract_sig_sp, and __builtin_vec_extract_sig_dp
15291 functions from "vector int" to "vector unsigned int" or from
15292 "vector long long int" to "vector unsigned long long int".
15293 Changed type returned by __builtin_vec_extract_exp,
15294 __builtin_vec_extract_exp_sp, and __builtin_vec_extract_exp_dp
15295 functions from "vector int" to "vector unsigned int" or from
15296 "vector long long int" to "vector unsigned long long int".
15297 Changed return type of __builtin_vec_test_data_class,
15298 __builtin_vec_test_data_class_sp, and
15299 __builtin_vec_test_data_class_dp from "vector int" to
15300 "vector bool int" or from "vector long long int" to "vector bool
15301 long long int" and changed second argument type from "unsigned
15302 int" to "int". Added new overloaded function forms "vector float
15303 __builtin_vec_insert_exp (vector float, vector unsigned int)" and
15304 "vector float __builtin_vec_insert_exp_sp (vector float, vector
15305 unsigned int)" and "vector double __builtin_vec_insert_exp (vector
15306 double, vector unsigned long long int)" and "vector double
15307 __builtin_vec_insert_exp_dp (vector double, vector unsigned long
15308 long int)". Changed return type of
15309 __builtin_scalar_test_data_class and
15310 __builtin_scalar_test_data_class_sp and
15311 __builtin_scalar_test_data_class_dp from "unsigned int" to "bool
15312 int" and changed second argument from "unsigned int" to "int".
15313 Changed type returned by __builtin_scalar_test_neg,
15314 __builtin_scalar_test_neg_sp, and __builtin_scalar_test_neg_dp
15315 from "int" to "bool int". Added new overloaded function form
15316 "double __builtin_scalar_insert_exp (double, unsigned long long int)".
15317 * config/rs6000/vsx.md (xsiexpdpf): New insn for scalar insert
15318 exponent double-precision with floating point first argument.
15319 * doc/extend.texi (PowerPC AltiVec Built-in Functions): Adjust
15320 documentation of scalar_test_data_class, scalar_test_neg,
15321 scalar_extract_sig, scalar_extract_exp, scalar_insert_exp,
15322 vector_extract_exp, vec_extract_sig, vec_insert_exp, and
15323 vec_test_data_class built-in functions to reflect refinements in
15324 their type signatures.
15326 2017-01-23 Andreas Tobler <andreast@gcc.gnu.org>
15328 * config/aarch64/aarch64.c (aarch64_elf_asm_constructor): Increase
15330 (aarch64_elf_asm_destructor): Likewise.
15332 2017-01-23 Bernd Schmidt <bschmidt@redhat.com>
15334 PR rtl-optimization/78634
15335 * config/i386/i386.c (ix86_max_noce_ifcvt_seq_cost): New function.
15336 (TARGET_MAX_NOCE_IFCVT_SEQ_COST): Define.
15337 * ifcvt.c (noce_try_cmove): Add missing cost check.
15339 PR rtl-optimization/71724
15340 * combine.c (if_then_else_cond): Look for situations where it is
15341 beneficial to undo the work of one of the recursive calls.
15343 2017-01-23 Bin Cheng <bin.cheng@arm.com>
15345 PR tree-optimization/70754
15346 * tree-predcom.c (stmt_combining_refs): New parameter INSERT_BEFORE.
15347 (reassociate_to_the_same_stmt): New parameter INSERT_BEFORE. Insert
15348 combined stmt before it if not NULL.
15349 (combine_chains): Process refs reversely and compute dominance point
15352 2017-01-23 Martin Liska <mliska@suse.cz>
15354 PR tree-optimization/79196
15355 * tree-ssa-strlen.c (fold_strstr_to_memcmp): Rename to ...
15356 (fold_strstr_to_strncmp): ... this. Fold the pattern to strncmp
15358 (strlen_optimize_stmt): Call the renamed function.
15360 2017-01-23 Michael Matz <matz@suse.de>
15362 PR tree-optimization/78384
15363 * tree-ssa-loop-split.c (patch_loop_exit): Use correct edge.
15365 2017-01-23 Richard Biener <rguenther@suse.de>
15367 PR tree-optimization/79186
15368 * tree-vrp.c (register_new_assert_for): Make sure we've seen
15369 both incoming edges before moving an assert.
15371 2017-01-23 Martin Jambor <mjambor@suse.cz>
15373 * ipa-prop.c (load_from_param_1): Removed.
15374 (load_from_unmodified_param): Bits from load_from_param_1 put back
15376 (load_from_param): Removed.
15377 (compute_complex_assign_jump_func): Removed stmt2 and just replaced it
15378 with stmt. Reverted back to use of load_from_unmodified_param.
15380 2017-01-23 Martin Jambor <mjambor@suse.cz>
15383 * ipa-prop.h (ipa_param_descriptor): Anotate with with GTY(()).
15384 (ipa_node_params): Annotate with GTY((for_user)). Make descriptors
15385 field a pointer to garbage collected vector, mark lattices and
15386 ipcp_orig_node with GTY((skip)).
15387 (ipa_get_param_count): Adjust to descriptors being a pointer.
15388 (ipa_get_param): Likewise.
15389 (ipa_get_type): Likewise.
15390 (ipa_get_param_move_cost): Likewise.
15391 (ipa_set_param_used): Likewise.
15392 (ipa_get_controlled_uses): Likewise.
15393 (ipa_set_controlled_uses): Likewise.
15394 (ipa_is_param_used): Likewise.
15395 (ipa_node_params_t): Move into garbage collector. New methods insert
15397 (ipa_node_params_sum): Annotate wth GTY(()).
15398 (ipa_check_create_node_params): Adjust to ipa_node_params_sum being
15400 (ipa_load_from_parm_agg): Adjust declaration.
15401 * ipa-icf.c (param_used_p): Adjust to descriptors being a pointer.
15402 * ipa-profile.c (ipa_profile): Likewise.
15403 * ipa-prop.c (ipa_get_param_decl_index_1): Likewise.
15404 (ipa_populate_param_decls): Make descriptors parameter garbage
15406 (ipa_dump_param): Adjust to descriptors being a pointer.
15407 (ipa_alloc_node_params): Likewise.
15408 (ipa_initialize_node_params): Likewise.
15409 (load_from_param_1): Make descriptors parameter garbage collected.
15410 (load_from_unmodified_param): Likewise.
15411 (load_from_param): Likewise.
15412 (ipa_load_from_parm_agg): Likewise.
15413 (ipa_node_params::~ipa_node_params): Removed.
15414 (ipa_free_all_node_params): Remove call to delete operator.
15415 (ipa_node_params_t::insert): New.
15416 (ipa_node_params_t::remove): Likewise.
15417 (ipa_node_params_t::duplicate): Adjust to descriptors being a pointer,
15418 copy known_csts and known_contexts vectors.
15419 (ipa_read_node_info): Adjust to descriptors being a pointer.
15420 (ipcp_modif_dom_walker): Make m_descriptors field garbage
15422 (ipcp_transform_function): Make descriptors variable garbage
15425 2017-01-23 Andrew Senkevich <andrew.senkevich@intel.com>
15427 * config/i386/avx512bwintrin.h: Add k-mask test, kortest intrinsics.
15428 * config/i386/avx512dqintrin.h: Ditto.
15429 * config/i386/avx512fintrin.h: Ditto.
15430 * gcc/config/i386/i386.c: Handle new builtins.
15431 * config/i386/i386-builtin.def: Add new builtins.
15432 * config/i386/sse.md (ktest<mode>, kortest<mode>): New.
15433 (UNSPEC_KORTEST, UNSPEC_KTEST): New.
15435 2017-01-23 Jakub Jelinek <jakub@redhat.com>
15436 Martin Liska <mliska@suse.cz>
15438 * asan.h: Define ASAN_USE_AFTER_SCOPE_ATTRIBUTE.
15439 * asan.c (asan_expand_poison_ifn): Support stores and use
15440 appropriate ASAN report function.
15441 * internal-fn.c (expand_ASAN_POISON_USE): New function.
15442 * internal-fn.def (ASAN_POISON_USE): Declare.
15443 * tree-into-ssa.c (maybe_add_asan_poison_write): New function.
15444 (maybe_register_def): Create ASAN_POISON_USE when sanitizing.
15445 * tree-ssa-dce.c (eliminate_unnecessary_stmts): Remove
15446 ASAN_POISON calls w/o LHS.
15447 * tree-ssa.c (execute_update_addresses_taken): Create clobber
15448 for ASAN_MARK (UNPOISON, &x, ...) in order to prevent usage of a LHS
15449 from ASAN_MARK (POISON, &x, ...) coming to a PHI node.
15450 * gimplify.c (asan_poison_variables): Add attribute
15451 use_after_scope_memory to variables that really needs to live
15453 * tree-ssa.c (is_asan_mark_p): Do not rewrite into SSA when
15454 having the attribute.
15456 2017-01-23 Martin Liska <mliska@suse.cz>
15458 * asan.c (create_asan_shadow_var): New function.
15459 (asan_expand_poison_ifn): Likewise.
15460 * asan.h (asan_expand_poison_ifn): New declaration.
15461 * internal-fn.c (expand_ASAN_POISON): Likewise.
15462 * internal-fn.def (ASAN_POISON): New builtin.
15463 * sanopt.c (pass_sanopt::execute): Expand
15464 asan_expand_poison_ifn.
15465 * tree-inline.c (copy_decl_for_dup_finish): Make function
15467 * tree-inline.h (copy_decl_for_dup_finish): Likewise.
15468 * tree-ssa.c (is_asan_mark_p): New function.
15469 (execute_update_addresses_taken): Rewrite local variables
15470 (identified just by use-after-scope as addressable) into SSA.
15472 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
15474 * doc/install.texi (Specific): opensource.apple.com uses https
15475 now. Remove trailing slash.
15477 2017-01-22 Gerald Pfeifer <gerald@pfeifer.com>
15479 * README.Portability: Remove note on an Irix compatibility issue.
15481 2017-01-22 Dimitry Andric <dim@FreeBSD.org>
15483 * gcov.c (INCLUDE_ALGORITHM): Define.
15484 (INCLUDE_VECTOR): Define.
15485 No longer include <vector> and <algorithm> directly.
15487 2017-01-21 Gerald Pfeifer <gerald@pfeifer.com>
15489 * doc/extend.texi (Thread-Local): Change www.akkadia.org reference
15491 * doc/invoke.texi (Code Gen Options): Ditto.
15493 2017-01-21 Jan Hubicka <hubicka@ucw.cz>
15496 * cfg.c (update_bb_profile_for_threading): Fix updating of probablity.
15498 2017-01-21 Bernd Schmidt <bschmidt@redhat.com>
15500 rtl-optimization/79125
15501 * cprop.c (local_cprop_pass): Handle cases where we make an
15502 unconditional trap.
15504 2017-01-20 Segher Boessenkool <segher@kernel.crashing.org>
15508 * config/rs6000/rs6000.c (rs6000_gimplify_va_arg): Adjust address to
15509 read from, for big endian.
15511 2017-01-20 Jiong Wang <jiong.wang@arm.com>
15513 * config/aarch64/aarch64-builtins.c (aarch64_init_builtins): Register
15514 register pauth builtins for LP64 only.
15516 2017-01-20 Marek Polacek <polacek@redhat.com>
15519 * gimplify.c (should_warn_for_implicit_fallthrough): Handle consecutive
15522 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
15524 * omp-expand.c (expand_omp_simd): Clear PROP_gimple_lomp_dev regardless
15526 * omp-offload.c (pass_omp_device_lower::gate): Use PROP_gimple_lomp_dev.
15527 * passes.c (dump_properties): Handle PROP_gimple_lomp_dev.
15528 * tree-inline.c (expand_call_inline): Propagate PROP_gimple_lomp_dev.
15530 2017-01-20 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15533 * config/arm/arm.c (neon_valid_immediate): Reject vector constants
15534 in big-endian mode when they are not a single duplicated value.
15536 2017-01-20 Richard Biener <rguenther@suse.de>
15538 * BASE-VER: Bump to 7.0.1.
15540 2017-01-20 Alexander Monakov <amonakov@ispras.ru>
15542 * omp-low.c (omplow_simd_context): New struct. Use it...
15543 (lower_rec_simd_input_clauses): ...here and...
15544 (lower_rec_input_clauses): ...here to hold common data. Adjust all
15545 references to idx, lane, max_vf, is_simt.
15547 2017-01-20 Graham Markall <graham.markall@embecosm.com>
15549 * config/arc/arc.h (LINK_SPEC): Use arclinux_nps emulation when
15552 2017-01-20 Martin Jambor <mjambor@suse.cz>
15554 * hsa.h: Renaed to hsa-common.h. Adjusted a comment.
15555 * hsa.c: Renaed to hsa-common.c. Change include of gt-hsa.h to
15557 * Makefile.in (OBJS): Rename hsa.o to hsa-common.o.
15558 (GTFILES): Rename hsa.c to hsa-common.c.
15559 * hsa-brig.c: Change include of hsa.h to hsa-common.h.
15560 * hsa-dump.c: Likewise.
15561 * hsa-gen.c: Likewise.
15562 * hsa-regalloc.c: Likewise.
15563 * ipa-hsa.c: Likewise.
15564 * omp-expand.c: Likewise.
15565 * omp-low.c: Likewise.
15566 * toplev.c: Likewise.
15568 2017-01-20 Marek Polacek <polacek@redhat.com>
15571 * doc/invoke.texi: Document -Wduplicated-branches.
15572 * fold-const.c (operand_equal_p): Handle MODIFY_EXPR, INIT_EXPR,
15573 COMPOUND_EXPR, PREDECREMENT_EXPR, PREINCREMENT_EXPR,
15574 POSTDECREMENT_EXPR, POSTINCREMENT_EXPR, CLEANUP_POINT_EXPR, EXPR_STMT,
15575 STATEMENT_LIST, and RETURN_EXPR. For non-pure non-const functions
15576 return 0 only when not OEP_LEXICOGRAPHIC.
15577 (fold_build_cleanup_point_expr): Use the expression
15578 location when building CLEANUP_POINT_EXPR.
15579 * tree-core.h (enum operand_equal_flag): Add OEP_LEXICOGRAPHIC.
15580 * tree.c (add_expr): Handle error_mark_node.
15582 2017-01-20 Martin Liska <mliska@suse.cz>
15585 * tree-profile.c (init_ic_make_global_vars): Do not call
15587 (gimple_init_gcov_profiler): Likewise.
15589 2017-01-20 Martin Liska <mliska@suse.cz>
15592 * cgraph.h (maybe_create_reference): Remove argument and
15594 * cgraphclones.c (cgraph_node::create_virtual_clone): Remove one
15596 * ipa-cp.c (create_specialized_node): Likewise.
15597 * symtab.c (symtab_node::maybe_create_reference): Handle
15598 VAR_DECLs and ADDR_EXPRs and select ipa_ref_use type.
15600 2017-01-20 Martin Liska <mliska@suse.cz>
15602 * read-rtl-function.c (function_reader::create_function): Use
15603 build_decl instread of build_decl_stat.
15605 2017-01-20 Andrew Senkevich <andrew.senkevich@intel.com>
15607 * config/i386/avx512bwintrin.h: Add k-mask registers shift intrinsics.
15608 * config/i386/avx512dqintrin.h: Ditto.
15609 * config/i386/avx512fintrin.h: Ditto.
15610 * config/i386/i386-builtin-types.def: Add new types.
15611 * gcc/config/i386/i386.c: Handle new types.
15612 * config/i386/i386-builtin.def (__builtin_ia32_kshiftliqi)
15613 (__builtin_ia32_kshiftlihi, __builtin_ia32_kshiftlisi)
15614 (__builtin_ia32_kshiftlidi, __builtin_ia32_kshiftriqi)
15615 (__builtin_ia32_kshiftrihi, __builtin_ia32_kshiftrisi)
15616 (__builtin_ia32_kshiftridi): New.
15617 * config/i386/sse.md (k<code><mode>): Rename *k<code><mode>.
15619 2017-01-19 Segher Boessenkool <segher@kernel.crashing.org>
15623 * config/rs6000/rs6000.c (TARGET_STACK_PROTECT_GUARD): Unconditionally
15624 define to rs6000_init_stack_protect_guard.
15625 (rs6000_init_stack_protect_guard): New function.
15627 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
15628 Yunqiang Su <yunqiang.su@imgtec.com>
15630 * config.gcc (supported_defaults): Add madd4.
15631 (with_madd4): Add validation.
15632 (all_defaults): Add madd4.
15633 * config/mips/mips.opt (mmadd4): New option.
15634 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
15636 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
15638 (ISA_HAS_UNFUSED_MADD4): Gate with mips_madd4.
15639 (ISA_HAS_FUSED_MADD4): Likewise.
15640 * gcc/doc/invoke.texi (-mmadd4): Document the new option.
15641 * gcc/doc/install.texi (--with-madd4): Document the new option.
15643 2017-01-19 Jiong Wang <jiong.wang@arm.com>
15645 * config/aarch64/aarch64-builtins.c (enum aarch64_builtins): New
15646 entries for AARCH64_PAUTH_BUILTIN_XPACLRI,
15647 AARCH64_PAUTH_BUILTIN_PACIA1716, AARCH64_PAUTH_BUILTIN_AUTIA1716.
15648 (aarch64_init_pauth_hint_builtins): New.
15649 (aarch64_init_builtins): Call aarch64_init_pauth_hint_builtins.
15650 (aarch64_expand_builtin): Expand new builtins.
15652 2017-01-19 Jiong Wang <jiong.wang@arm.com>
15654 * reg-notes.def (CFA_TOGGLE_RA_MANGLE): New reg-note.
15655 * combine-stack-adj.c (no_unhandled_cfa): Handle
15656 REG_CFA_TOGGLE_RA_MANGLE.
15657 * dwarf2cfi.c (dwarf2out_frame_debug): Handle REG_CFA_TOGGLE_RA_MANGLE.
15658 * config/aarch64/aarch64.c (aarch64_expand_prologue): Generates DWARF
15659 info for return address signing.
15660 (aarch64_expand_epilogue): Likewise.
15662 2017-01-19 Jiong Wang <jiong.wang@arm.com>
15664 * config/aarch64/aarch64-opts.h (aarch64_function_type): New enum.
15665 * config/aarch64/aarch64-protos.h
15666 (aarch64_return_address_signing_enabled): New declaration.
15667 * config/aarch64/aarch64.c (aarch64_return_address_signing_enabled):
15669 (aarch64_expand_prologue): Sign return address before it's pushed onto
15671 (aarch64_expand_epilogue): Authenticate return address fetched from
15673 (aarch64_override_options): Sanity check for ILP32 and ISA level.
15674 (aarch64_attributes): New function attributes for "sign-return-address".
15675 * config/aarch64/aarch64.md (UNSPEC_AUTI1716, UNSPEC_AUTISP,
15676 UNSPEC_PACI1716, UNSPEC_PACISP, UNSPEC_XPACLRI): New unspecs.
15677 ("*do_return"): Generate combined instructions according to key index.
15678 ("<pauth_mnem_prefix>sp", "<pauth_mnem_prefix1716", "xpaclri"): New.
15679 * config/aarch64/iterators.md (PAUTH_LR_SP, PAUTH_17_16): New integer
15681 (pauth_mnem_prefix, pauth_hint_num_a): New integer attributes.
15682 * config/aarch64/aarch64.opt (msign-return-address=): New.
15683 * doc/extend.texi (AArch64 Function Attributes): Documents
15684 "sign-return-address=".
15685 * doc/invoke.texi (AArch64 Options): Documents "-msign-return-address=".
15687 2017-01-19 Matthew Fortune <matthew.fortune@imgtec.com>
15689 * doc/invoke.texi: Add missing -mlxc1-sxc1 options to
15690 overall option summary.
15692 2017-01-19 Jiong Wang <jiong.wang@arm.com>
15694 * config/aarch64/aarch64-arches.def: New entry for "armv8.3-a".
15695 * config/aarch64/aarch64.h (AARCH64_FL_V8_3, AARCH64_FL_FOR_ARCH8_3,
15696 AARCH64_ISA_V8_3, TARGET_ARMV8_3): New.
15697 * doc/invoke.texi (AArch64 Options): Document "armv8.3-a".
15699 2017-01-19 Michael Meissner <meissner@linux.vnet.ibm.com>
15701 * config/rs6000/rs6000-cpus.def (ISA_3_0_MASKS_SERVER): Enable
15702 -mpower9-minmax by default for -mcpu=power9.
15703 (ISA_3_MASKS_IEEE): Require -mvsx-small-integer to enable IEEE
15704 128-bit floating point.
15706 2017-01-20 Alan Modra <amodra@gmail.com>
15708 * config/rs6000/rs6000.md (cmpstrnsi, cmpstrsi): Fail if
15709 optimizing for size.
15711 2017-01-20 Alan Modra <amodra@gmail.com>
15714 * config/rs6000/rs6000.c (expand_strn_compare): Get the asm name
15715 for strcmp and strncmp from corresponding builtin decl.
15717 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
15719 * config.gcc (x86_64-*-rtems*): Use i386/rtemself.h
15720 instead of i386/rtems-64.h.
15721 * config/i386/rtems-64.h: Remove.
15723 2017-01-19 Uros Bizjak <ubizjak@gmail.com>
15727 2013-11-05 Uros Bizjak <ubizjak@gmail.com>
15729 * config/i386/rtemself.h (LONG_DOUBLE_TYPE_SIZE): New define.
15731 2017-01-19 Tamar Christina <tamar.christina@arm.com>
15733 * config/aarch64/aarch64.c (aarch64_simd_gen_const_vector_dup):
15734 Change int to HOST_WIDE_INT.
15735 * config/aarch64/aarch64-protos.h
15736 (aarch64_simd_gen_const_vector_dup): Likewise.
15737 * config/aarch64/aarch64-simd.md: Add copysign<mode>3.
15739 2017-01-19 David Malcolm <dmalcolm@redhat.com>
15741 * langhooks-def.h (lhd_type_for_size): New decl.
15742 (LANG_HOOKS_TYPE_FOR_SIZE): Define as lhd_type_for_size.
15743 * langhooks.c (lhd_type_for_size): New function, taken from
15746 2017-01-19 Pat Haugen <pthaugen@us.ibm.com>
15748 * config/rs6000/power9.md (power9-alu): Remove 'cmp' type and add
15749 define_bypass for CR latency.
15750 (power9-cracked-alu): Update bypass latency and remove power9-branch.
15751 (power9-alu2): Add define_bypass for CR latency.
15753 (power9-mul): Update insn latency.
15754 (power9-mul-compare): Update insn latency, bypass latency and remove
15757 2016-01-19 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
15759 * config/aarch64/aarch64-protos.h (aarch64_nopcrelative_literal_loads):
15761 * config/aarch64/aarch64.md
15762 (aarch64_reload_movcp<GPF_TF:mode><P:mode>): Delete reference to
15763 aarch64_nopcrelative_literal_loads.
15764 (aarch64_reload_movcp<VALL:mode><P:mode>): Likewise.
15766 2017-01-19 Chenghua Xu <paul.hua.gm@gmail.com>
15768 * config/mips/mips.h (ISA_HAS_FUSED_MADD4): Enable for
15769 TARGET_LOONGSON_3A.
15770 (ISA_HAS_UNFUSED_MADD4): Exclude TARGET_LOONGSON_3A.
15772 2017-01-19 Doug Gilmore <doug.gilmore@imgtec.com>
15775 * config.gcc (supported_defaults): Add lxc1-sxc1.
15776 (with_lxc1_sxc1): Add validation.
15777 (all_defaults): Add lxc1-sxc1.
15778 * config/mips/mips.opt (mlxc1-sxc1): New option.
15779 * gcc/config/mips/mips.h (OPTION_DEFAULT_SPECS): Add a default for
15781 (TARGET_CPU_CPP_BUILTINS): Add builtin_define for
15782 __mips_no_lxc1_sxc1.
15783 (ISA_HAS_LXC1_SXC1): Gate with mips_lxc1_sxc1.
15784 * gcc/doc/invoke.texi (-mlxc1-sxc1): Document the new option.
15785 * doc/install.texi (--with-lxc1-sxc1): Document the new option.
15787 2017-01-19 Richard Biener <rguenther@suse.de>
15789 PR tree-optimization/72488
15790 * tree-ssa-sccvn.c (run_scc_vn): When we abort the VN make
15791 sure to restore SSA info.
15792 * tree-ssa.c (verify_ssa): Verify SSA info is not shared.
15794 2017-01-19 Richard Earnshaw <rearnsha@arm.com>
15796 PR rtl-optimization/79121
15797 * expr.c (expand_expr_real_2, case LSHIFT_EXPR): Look at the signedness
15798 of the inner type when shifting an extended value.
15800 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
15803 * symtab.c (symtab_node::equal_address_to): Fix comparing of
15804 interposable aliases.
15806 2017-01-18 Peter Bergner <bergner@vnet.ibm.com>
15809 * config/rs6000/spe.md (mov_si<mode>_e500_subreg0): Fix constraints.
15810 Use the evmergelohi instruction.
15811 (mov_si<mode>_e500_subreg4_2_le): Likewise.
15812 (mov_sitf_e500_subreg8_2_be): Likewise.
15813 (mov_sitf_e500_subreg12_2_le): Likewise.
15814 (mov_si<mode>_e500_subreg0_2_le): Fix constraints.
15815 (mov_si<mode>_e500_subreg4_2_be): Likewise.
15816 (mov_sitf_e500_subreg8_2_le): Likewise.
15817 (mov_sitf_e500_subreg12_2_be): Likewise.
15819 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15821 * config/rs6000/altivec.md (altivec_vbpermq): Change "type"
15822 attribute from vecsimple to vecperm.
15823 (altivec_vbpermq2): Likewise.
15825 2017-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15828 * config/rs6000/altivec.h: Fix typo of vec_cntlz to vec_cnttz.
15830 2017-01-18 Aaron Sawdey <acsawdey@linux.vnet.ibm.com>
15831 * config/rs6000/rs6000-protos.h (expand_strn_compare): Add arg.
15832 * config/rs6000/rs6000.c (expand_strn_compare): Add ability to expand
15833 strcmp. Fix bug where comparison didn't stop with zero byte. Fix
15834 case where N arg is SIZE_MAX.
15835 * config/rs6000/rs6000.md (cmpstrnsi): Args to expand_strn_compare.
15836 (cmpstrsi): Add pattern.
15838 2017-01-18 Michael Meissner <meissner@linux.vnet.ibm.com>
15840 * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Add
15841 __builtin_vec_revb builtins.
15842 * config/rs6000/rs6000-builtins.def (P9V_BUILTIN_XXBRQ_V16QI): Add
15843 built-in functions to support generation of the ISA 3.0 XXBR<x>
15844 vector byte reverse instructions.
15845 (P9V_BUILTIN_XXBRQ_V1TI): Likewise.
15846 (P9V_BUILTIN_XXBRD_V2DI): Likewise.
15847 (P9V_BUILTIN_XXBRD_V2DF): Likewise.
15848 (P9V_BUILTIN_XXBGW_V4SI): Likewise.
15849 (P9V_BUILTIN_XXBGW_V4SF): Likewise.
15850 (P9V_BUILTIN_XXBGH_V8HI): Likewise.
15851 (P9V_BUILTIN_VEC_REVB): Likewise.
15852 * config/rs6000/vsx.md (p9_xxbrq_v1ti): New insns/expanders to
15853 generate the ISA 3.0 XXBR<x> vector byte reverse instructions.
15854 (p9_xxbrq_v16qi): Likewise.
15855 (p9_xxbrd_<mode>, VSX_D iterator): Likewise.
15856 (p9_xxbrw_<mode>, VSX_W iterator): Likewise.
15857 (p9_xxbrh_v8hi): Likewise.
15858 * config/rs6000/altivec.h (vec_revb): Define if ISA 3.0.
15859 * doc/extend.texi (RS/6000 Altivec Built-ins): Document the
15860 vec_revb built-in functions.
15862 2017-01-18 Uros Bizjak <ubizjak@gmail.com>
15864 PR rtl-optimization/78952
15865 * config/i386/i386.md (any_extract): New code iterator.
15866 (*insvqi_2): Use any_extract for source operand.
15867 (*insvqi_3): Use any_shiftrt for source operand.
15869 2017-01-18 Wilco Dijkstra <wdijkstr@arm.com>
15871 * config/aarch64/aarch64.c (aarch64_sched_adjust_priority)
15873 (TARGET_SCHED_ADJUST_PRIORITY): Define target hook.
15875 2017-01-18 Matthias Klose <doko@ubuntu.com>
15877 * doc/install.texi: Allow default for --with-target-bdw-gc-include.
15879 2016-01-18 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15881 * config/rs6000/altivec.h (vec_bperm): Change #define.
15882 * config/rs6000/altivec.md (UNSPEC_VBPERMD): New enum constant.
15883 (altivec_vbpermq2): New define_insn.
15884 (altivec_vbpermd): Likewise.
15885 * config/rs6000/rs6000-builtin.def (VBPERMQ2): New monomorphic
15886 function interface.
15887 (VBPERMD): Likewise.
15888 (VBPERM): New polymorphic function interface.
15889 * config/rs6000/r6000-c.c (altivec_overloaded_builtins_table):
15890 Add entries for P9V_BUILTIN_VEC_VBPERM.
15891 * doc/extend.texi: Add interfaces for vec_bperm.
15893 2017-01-18 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
15895 * config/s390/s390-c.c (s390_expand_overloaded_builtin): Downcase
15896 first letter of error messages.
15897 (s390_resolve_overloaded_builtin): Likewise.
15898 * config/s390/s390.c (s390_expand_builtin): Likewise.
15899 (s390_invalid_arg_for_unprototyped_fn): Likewise.
15900 (s390_valid_target_attribute_inner_p): Likewise.
15901 * config/s390/s390.md ("tabort"): Likewise.
15903 2017-01-18 Toma Tabacu <toma.tabacu@imgtec.com>
15905 * config/mips/mips.h (ISA_HAS_DIV3): Remove unused macro.
15906 (ISA_AVOID_DIV_HILO): New macro.
15907 (ISA_HAS_DIV): Use new ISA_AVOID_DIV_HILO macro.
15908 (ISA_HAS_DDIV): Likewise.
15910 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
15912 * doc/invoke.texi (fabi-version): Correct number of occurrences.
15914 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
15916 * doc/invoke.texi (fabi-version): Spelling fix.
15918 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
15921 * doc/invoke.texi (fabi-version): Mention mangling fix for
15924 2017-01-18 Markus Trippelsdorf <markus@trippelsdorf.de>
15927 * doc/invoke.texi (fabi-version): Document discriminator mangling.
15929 2017-01-17 Segher Boessenkool <segher@kernel.crashing.org>
15932 * config/rs6000/rs6000-opts.h (stack_protector_guard): New enum.
15933 * config/rs6000/rs6000.c (rs6000_option_override_internal): Handle
15935 * config/rs6000/rs6000.md (stack_protect_set): Handle the new more
15937 (stack_protect_test): Ditto.
15938 * config/rs6000/rs6000.opt (mstack-protector-guard=,
15939 mstack-protector-guard-reg=, mstack-protector-guard-offset=): New
15941 * doc/invoke.texi (Option Summary) [RS/6000 and PowerPC Options]:
15942 Add -mstack-protector-guard=, -mstack-protector-guard-reg=, and
15943 -mstack-protector-guard-offset=.
15944 (RS/6000 and PowerPC Options): Ditto.
15946 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
15948 * config/i386/i386.h (MASK_CLASS_P): New define.
15949 * config/i386/i386.c (inline_secondary_memory_needed): Ensure that
15950 there are no registers from different register sets also when
15951 mask registers are used. Update function comment.
15952 * config/i386/i386.md (*movsi_internal): Split (*k/*krm) alternative
15953 to (*k/*r) and (*k/*km) alternatives.
15955 2017-01-17 Wilco Dijkstra <wdijkstr@arm.com>
15957 * config/aarch64/aarch64.md (eh_return): Remove pattern and splitter.
15958 * config/aarch64/aarch64.h (AARCH64_EH_STACKADJ_REGNUM): Remove.
15959 (EH_RETURN_HANDLER_RTX): New define.
15960 * config/aarch64/aarch64.c (aarch64_frame_pointer_required):
15961 Force frame pointer in EH return functions.
15962 (aarch64_expand_epilogue): Add barrier for eh_return.
15963 (aarch64_final_eh_return_addr): Remove.
15964 (aarch64_eh_return_handler_rtx): New function.
15965 * config/aarch64/aarch64-protos.h (aarch64_final_eh_return_addr):
15967 (aarch64_eh_return_handler_rtx): New prototype.
15969 2017-01-17 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
15971 * config/rs6000/altivec.h (vec_rlmi): New #define.
15972 (vec_vrlnm): Likewise.
15973 (vec_rlnm): Likewise.
15974 * config/rs6000/altivec.md (UNSPEC_VRLMI): New UNSPEC enum value.
15975 (UNSPEC_VRLNM): Likewise.
15976 (VIlong): New mode iterator.
15977 (altivec_vrl<VI_char>mi): New define_insn.
15978 (altivec_vrl<VI_char>nm): Likewise.
15979 * config/rs6000/rs6000-builtin.def (VRLWNM): New monomorphic
15981 (VRLDNM): Likewise.
15982 (RLNM): New polymorphic function entry.
15983 (VRLWMI): New monomorphic function entry.
15984 (VRLDMI): Likewise.
15985 (RLMI): New polymorphic function entry.
15986 * config/rs6000/r6000-c.c (altivec_overloaded_builtin_table): Add
15987 new entries for P9V_BUILTIN_VEC_RLMI and P9V_BUILTIN_VEC_RLNM.
15988 * doc/extend.texi: Add description of vec_rlmi, vec_rlnm, and
15991 2017-01-17 Jakub Jelinek <jakub@redhat.com>
15994 * dwarf2out.c (field_byte_offset): Restore the
15995 PCC_BITFIELD_TYPE_MATTERS behavior for INTEGER_CST DECL_FIELD_OFFSET
15996 and DECL_FIELD_BIT_OFFSET. Use fold_build2 instead of build2 + fold.
15997 (analyze_variants_discr, gen_variant_part): Use fold_build2 instead
16000 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
16003 * config/aarch64/aarch64.c (TARGET_CUSTOM_FUNCTION_DESCRIPTORS): Define
16005 2017-01-17 Jakub Jelinek <jakub@redhat.com>
16008 * dwarf2out.c (add_data_member_location_attribute): For constant
16009 offset bitfield emit for -gdwarf-5 DW_AT_data_bit_offset attribute
16010 instead of DW_AT_data_member_location, DW_AT_bit_offset and
16011 DW_AT_byte_size attributes.
16013 2017-01-17 Eric Botcazou <ebotcazou@adacore.com>
16015 * config/rs6000/rs6000.c (rs6000_emit_move): Also use a TOC reference
16016 after forcing to constant memory when the code model is medium.
16018 2017-01-17 Julia Koval <julia.koval@intel.com>
16021 * config/i386/avx512fintrin.h
16022 (_mm512_i32gather_ps): Change __addr type to void const*.
16023 (_mm512_mask_i32gather_ps): Ditto.
16024 (_mm512_i32gather_pd): Ditto.
16025 (_mm512_mask_i32gather_pd): Ditto.
16026 (_mm512_i64gather_ps): Ditto.
16027 (_mm512_mask_i64gather_ps): Ditto.
16028 (_mm512_i64gather_pd): Ditto.
16029 (_mm512_mask_i64gather_pd): Ditto.
16030 (_mm512_i32gather_epi32): Ditto.
16031 (_mm512_mask_i32gather_epi32): Ditto.
16032 (_mm512_i32gather_epi64): Ditto.
16033 (_mm512_mask_i32gather_epi64): Ditto.
16034 (_mm512_i64gather_epi32): Ditto.
16035 (_mm512_mask_i64gather_epi32): Ditto.
16036 (_mm512_i64gather_epi64): Ditto.
16037 (_mm512_mask_i64gather_epi64): Ditto.
16038 (_mm512_i32scatter_ps): Change __addr type to void*.
16039 (_mm512_mask_i32scatter_ps): Ditto.
16040 (_mm512_i32scatter_pd): Ditto.
16041 (_mm512_mask_i32scatter_pd): Ditto.
16042 (_mm512_i64scatter_ps): Ditto.
16043 (_mm512_mask_i64scatter_ps): Ditto.
16044 (_mm512_i64scatter_pd): Ditto.
16045 (_mm512_mask_i64scatter_pd): Ditto.
16046 (_mm512_i32scatter_epi32): Ditto.
16047 (_mm512_mask_i32scatter_epi32): Ditto.
16048 (_mm512_i32scatter_epi64): Ditto.
16049 (_mm512_mask_i32scatter_epi64): Ditto.
16050 (_mm512_i64scatter_epi32): Ditto.
16051 (_mm512_mask_i64scatter_epi32): Ditto.
16052 (_mm512_i64scatter_epi64): Ditto.
16053 (_mm512_mask_i64scatter_epi64): Ditto.
16054 * config/i386/avx512pfintrin.h
16055 (_mm512_mask_prefetch_i32gather_pd): Change __addr type to void const*.
16056 (_mm512_mask_prefetch_i32gather_ps): Ditto.
16057 (_mm512_mask_prefetch_i64gather_pd): Ditto.
16058 (_mm512_mask_prefetch_i64gather_ps): Ditto.
16059 (_mm512_prefetch_i32scatter_pd): Change __addr type to void*.
16060 (_mm512_prefetch_i32scatter_ps): Ditto.
16061 (_mm512_mask_prefetch_i32scatter_pd): Ditto.
16062 (_mm512_mask_prefetch_i32scatter_ps): Ditto.
16063 (_mm512_prefetch_i64scatter_pd): Ditto.
16064 (_mm512_prefetch_i64scatter_ps): Ditto.
16065 (_mm512_mask_prefetch_i64scatter_pd): Ditto.
16066 (_mm512_mask_prefetch_i64scatter_ps): Ditto.
16067 * config/i386/avx512vlintrin.h
16068 (_mm256_mmask_i32gather_ps): Change __addr type to void const*.
16069 (_mm_mmask_i32gather_ps): Ditto.
16070 (_mm256_mmask_i32gather_pd): Ditto.
16071 (_mm_mmask_i32gather_pd): Ditto.
16072 (_mm256_mmask_i64gather_ps): Ditto.
16073 (_mm_mmask_i64gather_ps): Ditto.
16074 (_mm256_mmask_i64gather_pd): Ditto.
16075 (_mm_mmask_i64gather_pd): Ditto.
16076 (_mm256_mmask_i32gather_epi32): Ditto.
16077 (_mm_mmask_i32gather_epi32): Ditto.
16078 (_mm256_mmask_i32gather_epi64): Ditto.
16079 (_mm_mmask_i32gather_epi64): Ditto.
16080 (_mm256_mmask_i64gather_epi32): Ditto.
16081 (_mm_mmask_i64gather_epi32): Ditto.
16082 (_mm256_mmask_i64gather_epi64): Ditto.
16083 (_mm_mmask_i64gather_epi64): Ditto.
16084 (_mm256_i32scatter_ps): Change __addr type to void*.
16085 (_mm256_mask_i32scatter_ps): Ditto.
16086 (_mm_i32scatter_ps): Ditto.
16087 (_mm_mask_i32scatter_ps): Ditto.
16088 (_mm256_i32scatter_pd): Ditto.
16089 (_mm256_mask_i32scatter_pd): Ditto.
16090 (_mm_i32scatter_pd): Ditto.
16091 (_mm_mask_i32scatter_pd): Ditto.
16092 (_mm256_i64scatter_ps): Ditto.
16093 (_mm256_mask_i64scatter_ps): Ditto.
16094 (_mm_i64scatter_ps): Ditto.
16095 (_mm_mask_i64scatter_ps): Ditto.
16096 (_mm256_i64scatter_pd): Ditto.
16097 (_mm256_mask_i64scatter_pd): Ditto.
16098 (_mm_i64scatter_pd): Ditto.
16099 (_mm_mask_i64scatter_pd): Ditto.
16100 (_mm256_i32scatter_epi32): Ditto.
16101 (_mm256_mask_i32scatter_epi32): Ditto.
16102 (_mm_i32scatter_epi32): Ditto.
16103 (_mm_mask_i32scatter_epi32): Ditto.
16104 (_mm256_i32scatter_epi64): Ditto.
16105 (_mm256_mask_i32scatter_epi64): Ditto.
16106 (_mm_i32scatter_epi64): Ditto.
16107 (_mm_mask_i32scatter_epi64): Ditto.
16108 (_mm256_i64scatter_epi32): Ditto.
16109 (_mm256_mask_i64scatter_epi32): Ditto.
16110 (_mm_i64scatter_epi32): Ditto.
16111 (_mm_mask_i64scatter_epi32): Ditto.
16112 (_mm256_i64scatter_epi64): Ditto.
16113 (_mm256_mask_i64scatter_epi64): Ditto.
16114 (_mm_i64scatter_epi64): Ditto.
16115 (_mm_mask_i64scatter_epi64): Ditto.
16116 * config/i386/i386-builtin-types.def (V16SF_V16SF_PCFLOAT_V16SI_HI_INT)
16117 (V8DF_V8DF_PCDOUBLE_V8SI_QI_INT, V8SF_V8SF_PCFLOAT_V8DI_QI_INT)
16118 (V8DF_V8DF_PCDOUBLE_V8DI_QI_INT, V16SI_V16SI_PCINT_V16SI_HI_INT)
16119 (V8DI_V8DI_PCINT64_V8SI_QI_INT, V8SI_V8SI_PCINT_V8DI_QI_INT)
16120 (V8DI_V8DI_PCINT64_V8DI_QI_INT, V2DF_V2DF_PCDOUBLE_V4SI_QI_INT)
16121 (V4DF_V4DF_PCDOUBLE_V4SI_QI_INT, V2DF_V2DF_PCDOUBLE_V2DI_QI_INT)
16122 (V4DF_V4DF_PCDOUBLE_V4DI_QI_INT, V4SF_V4SF_PCFLOAT_V4SI_QI_INT)
16123 (V8SF_V8SF_PCFLOAT_V8SI_QI_INT, V4SF_V4SF_PCFLOAT_V2DI_QI_INT)
16124 (V4SF_V4SF_PCFLOAT_V4DI_QI_INT, V2DI_V2DI_PCINT64_V4SI_QI_INT)
16125 (V4DI_V4DI_PCINT64_V4SI_QI_INT, V2DI_V2DI_PCINT64_V2DI_QI_INT)
16126 (V4DI_V4DI_PCINT64_V4DI_QI_INT, V4SI_V4SI_PCINT_V4SI_QI_INT)
16127 (V8SI_V8SI_PCINT_V8SI_QI_INT, V4SI_V4SI_PCINT_V2DI_QI_INT)
16128 (V4SI_V4SI_PCINT_V4DI_QI_INT, VOID_PFLOAT_HI_V16SI_V16SF_INT)
16129 (VOID_PFLOAT_QI_V8SI_V8SF_INT, VOID_PFLOAT_QI_V4SI_V4SF_INT)
16130 (VOID_PDOUBLE_QI_V8SI_V8DF_INT, VOID_PDOUBLE_QI_V4SI_V4DF_INT)
16131 (VOID_PDOUBLE_QI_V4SI_V2DF_INT, VOID_PFLOAT_QI_V8DI_V8SF_INT)
16132 (VOID_PFLOAT_QI_V4DI_V4SF_INT, VOID_PFLOAT_QI_V2DI_V4SF_INT)
16133 (VOID_PDOUBLE_QI_V8DI_V8DF_INT, VOID_PDOUBLE_QI_V4DI_V4DF_INT)
16134 (VOID_PDOUBLE_QI_V2DI_V2DF_INT, VOID_PINT_HI_V16SI_V16SI_INT)
16135 (VOID_PINT_QI_V8SI_V8SI_INT, VOID_PINT_QI_V4SI_V4SI_INT)
16136 (VOID_PLONGLONG_QI_V8SI_V8DI_INT, VOID_PLONGLONG_QI_V4SI_V4DI_INT)
16137 (VOID_PLONGLONG_QI_V4SI_V2DI_INT, VOID_PINT_QI_V8DI_V8SI_INT)
16138 (VOID_PINT_QI_V4DI_V4SI_INT, VOID_PINT_QI_V2DI_V4SI_INT)
16139 (VOID_PLONGLONG_QI_V8DI_V8DI_INT, VOID_QI_V8SI_PCINT64_INT_INT)
16140 (VOID_PLONGLONG_QI_V4DI_V4DI_INT, VOID_PLONGLONG_QI_V2DI_V2DI_INT)
16141 (VOID_HI_V16SI_PCINT_INT_INT, VOID_QI_V8DI_PCINT64_INT_INT)
16142 (VOID_QI_V8DI_PCINT_INT_INT): Remove.
16143 (V16SF_V16SF_PCVOID_V16SI_HI_INT, V8DF_V8DF_PCVOID_V8SI_QI_INT)
16144 (V8SF_V8SF_PCVOID_V8DI_QI_INT, V8DF_V8DF_PCVOID_V8DI_QI_INT)
16145 (V16SI_V16SI_PCVOID_V16SI_HI_INT, V8DI_V8DI_PCVOID_V8SI_QI_INT)
16146 (V8SI_V8SI_PCVOID_V8DI_QI_INT, V8DI_V8DI_PCVOID_V8DI_QI_INT)
16147 (VOID_PVOID_HI_V16SI_V16SF_INT, VOID_PVOID_QI_V8SI_V8DF_INT)
16148 (VOID_PVOID_QI_V8DI_V8SF_INT, VOID_PVOID_QI_V8DI_V8DF_INT)
16149 (VOID_PVOID_HI_V16SI_V16SI_INT, VOID_PVOID_QI_V8SI_V8DI_INT)
16150 (VOID_PVOID_QI_V8DI_V8SI_INT, VOID_PVOID_QI_V8DI_V8DI_INT)
16151 (V2DF_V2DF_PCVOID_V4SI_QI_INT, V4DF_V4DF_PCVOID_V4SI_QI_INT)
16152 (V2DF_V2DF_PCVOID_V2DI_QI_INT, V4DF_V4DF_PCVOID_V4DI_QI_INT
16153 (V4SF_V4SF_PCVOID_V4SI_QI_INT, V8SF_V8SF_PCVOID_V8SI_QI_INT)
16154 (V4SF_V4SF_PCVOID_V2DI_QI_INT, V4SF_V4SF_PCVOID_V4DI_QI_INT)
16155 (V2DI_V2DI_PCVOID_V4SI_QI_INT, V4DI_V4DI_PCVOID_V4SI_QI_INT)
16156 (V2DI_V2DI_PCVOID_V2DI_QI_INT, V4DI_V4DI_PCVOID_V4DI_QI_INT)
16157 (V4SI_V4SI_PCVOID_V4SI_QI_INT, V8SI_V8SI_PCVOID_V8SI_QI_INT)
16158 (V4SI_V4SI_PCVOID_V2DI_QI_INT, V4SI_V4SI_PCVOID_V4DI_QI_INT)
16159 (VOID_PVOID_QI_V8SI_V8SF_INT, VOID_PVOID_QI_V4SI_V4SF_INT)
16160 (VOID_PVOID_QI_V4SI_V4DF_INT, VOID_PVOID_QI_V4SI_V2DF_INT)
16161 (VOID_PVOID_QI_V4DI_V4SF_INT, VOID_PVOID_QI_V2DI_V4SF_INT)
16162 (VOID_PVOID_QI_V4DI_V4DF_INT, VOID_PVOID_QI_V2DI_V2DF_INT)
16163 (VOID_PVOID_QI_V8SI_V8SI_INT, VOID_PVOID_QI_V4SI_V4SI_INT)
16164 (VOID_PVOID_QI_V4SI_V4DI_INT, VOID_PVOID_QI_V4SI_V2DI_INT)
16165 (VOID_PVOID_QI_V4DI_V4SI_INT, VOID_PVOID_QI_V2DI_V4SI_INT)
16166 (VOID_PVOID_QI_V4DI_V4DI_INT, VOID_PVOID_QI_V2DI_V2DI_INT)
16167 (VOID_QI_V8SI_PCVOID_INT_INT, VOID_HI_V16SI_PCVOID_INT_INT)
16168 (VOID_QI_V8DI_PCVOID_INT_INT): Add.
16169 * config/i386/i386.c (ix86_init_mmx_sse_builtins): Adjust builtin
16170 definitions accordingly.
16172 2017-01-17 Kito Cheng <kito.cheng@gmail.com>
16173 Kuan-Lin Chen <kuanlinchentw@gmail.com>
16176 * internal-fn.c (expand_mul_overflow): Use convert_modes instead of
16179 2017-01-17 Vladimir Makarov <vmakarov@redhat.com>
16182 * ira-conflicts.c (ira_build_conflicts): Update total conflict
16183 hard regs for inner regno.
16185 2017-01-17 Martin Liska <mliska@suse.cz>
16188 * ipa-polymorphic-call.c (contains_type_p): Fix wrong
16189 assumption and add comment.
16191 2017-01-17 Nathan Sidwell <nathan@acm.org>
16193 * ipa-visibility.c (localize_node): New function, broken out of ...
16194 (function_and_variable_visibility): ... here. Call it.
16196 2017-01-17 Jan Hubicka <hubicka@ucw.cz>
16198 PR middle-end/77445
16199 * tree-ssa-threadupdate.c (remove_ctrl_stmt_and_useless_edges):
16200 correctly set frequency of oudgoing edge.
16201 (duplicate_thread_path): Fix profile updating.
16203 2017-01-17 Jakub Jelinek <jakub@redhat.com>
16206 * configure.ac: Add GCC_BASE_VER.
16207 * Makefile.in (version): Use @get_gcc_base_ver@ instead of cat to get
16208 version from BASE-VER file.
16209 (CFLAGS-gcc.o): Add -DBASEVER=$(BASEVER_s).
16210 (gcc.o): Depend on $(BASEVER).
16211 * common.opt (dumpfullversion): New option.
16212 * gcc.c (driver_handle_option): Handle OPT_dumpfullversion.
16213 * doc/invoke.texi: Document -dumpfullversion.
16214 * doc/install.texi: Document --with-gcc-major-version-only.
16215 * configure: Regenerated.
16217 2017-01-17 Richard Biener <rguenther@suse.de>
16219 PR tree-optimization/71433
16220 * tree-vrp.c (register_new_assert_for): Merge same asserts
16221 on all incoming edges.
16222 (process_assert_insertions_for): Handle insertions at the
16225 2017-01-17 Gerald Pfeifer <gerald@pfeifer.com>
16227 * config/i386/cygwin.h (LIBGCJ_SONAME): Remove.
16228 * config/i386/mingw32.h (LIBGCJ_SONAME): Remove.
16230 2017-01-17 Kaz Kojima <kkojima@gcc.gnu.org>
16233 * config/sh/sh.md (cmpeqsi_t+1): Call copy_rtx to avoid invalid
16236 2017-01-17 Alan Modra <amodra@gmail.com>
16239 * config/rs6000/rs6000.md (elf_high, elf_low): Disable when pic.
16240 * config/rs6000/rs6000.c (rs6000_emit_allocate_stack): Don't allow
16241 symbolic stack limit when pic.
16243 2017-01-16 Martin Sebor <msebor@redhat.com>
16245 PR tree-optimization/78608
16246 * gimple-ssa-sprintf.c (tree_digits): Avoid negating TYPE_MIN.
16248 2017-01-16 Jeff Law <law@redhat.com>
16251 2016-12-02 Tadek Kijkowski <tkijkowski@gmail.com>
16252 * Makefile.in (PREPROCESSOR_DEFINES): Add a level of indirection
16253 for several include directories that may be relative to sysroot.
16254 * config/i386/x-mingw32 (gplus_includedir): Define.
16255 (gplus_tool_includedir, gplus_backward_include_dir): Likewise.
16256 (native_system_includedir): Likewise.
16257 * config/i386/mingw32.h (STANDARD_STARTFILE_PREFIX_1): Do not
16258 override if TARGET_SYSTEM_ROOT is defined.
16259 (NATIVE_SYSTEM_HEADER_DIR): Likewise.
16261 PR tree-optimization/79090
16262 PR tree-optimization/33562
16263 PR tree-optimization/61912
16264 PR tree-optimization/77485
16265 * tree-ssa-dse.c (compute_trims): Accept STMT argument. Dump STMT
16266 and computed trims into the dump file.
16268 2017-01-17 Uros Bizjak <ubizjak@gmail.com>
16270 * config/i386/i386.h (LIMIT_RELOAD_CLASS): Remove.
16272 2017-01-16 Jakub Jelinek <jakub@redhat.com>
16275 * gimplify.c (gimplify_init_constructor): If want_value and
16276 object == lhs, unshare lhs to avoid invalid tree sharing. Formatting
16280 * loop-doloop.c (doloop_modify): Call unshare_all_rtl_in_chain on
16281 sequence. Formatting fixes.
16282 (doloop_optimize): Formatting fixes.
16285 * gcc.c (debug_level_greater_than_spec_func): New function.
16286 (static_spec_functions): Add debug-level-gt spec function.
16287 (ASM_DEBUG_SPEC, cpp_options): Use %:debug-level-gt(0) instead of
16289 * config/darwin.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
16290 * config/darwin9.h (DSYMUTIL_SPEC, ASM_DEBUG_SPEC): Likewise.
16291 * common.opt (g, gcoff, gdwarf, gdwarf-, ggdb, gno-pubnames,
16292 gpubnames, ggnu-pubnames, gno-record-gcc-switches,
16293 grecord-gcc-switches, gno-strict-dwarf, gstrict-dwarf, gstabs,
16294 gstabs+, gtoggle, gvms, gxcoff, gxcoff+): Add Driver flag.
16296 2017-01-16 Uros Bizjak <ubizjak@gmail.com>
16298 * config/i386/i386.h (HARD_REGNO_CALLER_SAVE_MODE): Apply HImode and
16299 QImode fixups to general and mask registers only.
16301 2017-01-16 Carl Love <cel@us.ibm.com>
16303 * config/rs6000/rs6000-c (altivec_overloaded_builtins): Add support
16304 for built-in functions
16305 vector signed char vec_nabs (vector signed char)
16306 vector signed short vec_nabs (vector signed short)
16307 vector signed int vec_nabs (vector signed int)
16308 vector signed long long vec_nabs (vector signed long long)
16309 vector float vec_nabs (vector float)
16310 vector double vec_nabs (vector double)
16311 * config/rs6000/rs6000-builtin.def: Add definitions for NABS functions
16313 * config/rs6000/altivec.md: New define_expand nabs<mode>2 types
16314 * config/rs6000/altivec.h: New define for vec_nabs built-in function.
16315 * doc/extend.texi: Update the documentation file for the new built-in
16318 2017-01-16 Martin Sebor <msebor@redhat.com>
16320 * gimple-ssa-sprintf.c (format_directive): Correct a typo in a warning
16323 2017-01-16 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16325 * config/rs6000/rs6000.c (rtx_is_swappable_p): Change
16326 UNSPEC_VSX__XXSPLTD to require special splat handling.
16328 2017-01-16 David Malcolm <dmalcolm@redhat.com>
16331 * system.h: Poison strndup.
16333 2017-01-16 Alan Modra <amodra@gmail.com>
16336 * config/rs6000/rs6000.c (rs6000_legitimate_combined_insn): Don't
16339 2017-01-16 Georg-Johann Lay <avr@gjlay.de>
16341 * config/avr/avr.h (BRANCH_COST) [reload_completed]: Increase by 4.
16343 2017-01-15 Uros Bizjak <ubizjak@gmail.com>
16345 * config/i386/i386.c (ix86_legitimate_combined_insn): Do not
16346 call recog here. Assert that INSN_CODE (insn) is non-negative.
16348 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
16351 * cfgrtl.c (rtl_split_edge): Also patch jump insns that jump to the
16353 * haifa-sched.c (dump_insn_stream): Don't crash if there is a label
16354 in the currently scheduled RTL fragment.
16356 2017-01-15 Segher Boessenkool <segher@kernel.crashing.org>
16358 PR rtl-optimization/78751
16359 * ifcvt.c (find_cond_trap): If we generated a non-existing insn,
16362 2017-01-14 Jeff Law <law@redhat.com>
16364 PR tree-optimization/79090
16365 * tree-ssa-dse.c (valid_ao_ref_for_dse): Reject zero length and
16366 variable length stores.
16367 (compute_trims): Delete dead assignment to *trim_tail.
16368 (dse_dom_walker::dse_optimize_stmt): Optimize mem* calls with
16371 2017-01-14 Bernd Schmidt <bschmidt@redhat.com>
16373 PR rtl-optimization/78626
16374 PR rtl-optimization/78727
16375 * cprop.c (one_cprop_pass): Collect unconditional traps in the middle
16376 of a block, and split such blocks after everything else is finished.
16378 2017-01-14 Alan Modra <amodra@gmail.com>
16381 * combine.c (recog_for_combine_1): Set INSN_CODE before calling
16382 target legitimate_combined_insn.
16383 * config/rs6000/rs6000.c (TARGET_LEGITIMATE_COMBINED_INSN): Define.
16384 (rs6000_legitimate_combined_insn): New function.
16385 * config/rs6000/rs6000.md (UNSPEC_DOLOOP): Delete, and remove
16387 (ctr<mode>_internal3): Rename from *ctr<mode>_internal5.
16388 (ctr<mode>_internal4): Rename from *ctr<mode>_internal6.
16389 (ctr<mode>_internal1, ctr<mode>_internal2): Remove '*' from name.
16391 2017-01-14 Gerald Pfeifer <gerald@pfeifer.com>
16393 * doc/frontends.texi (G++ and GCC): Remove references to Java.
16395 2017-01-13 Jeff Law <law@redhat.com>
16397 PR tree-optimization/33562
16398 PR tree-optimization/61912
16399 PR tree-optimization/77485
16400 * tree-ssa-dse.c (delete_dead_call): Accept gsi rather than
16402 (delete_dead_assignment): Likewise.
16403 (dse_dom_walker::dse_optimize_stmt): Pass in the gsi rather than
16404 statement to delete_dead_call and delete_dead_assignment.
16406 2017-01-13 David Malcolm <dmalcolm@redhat.com>
16409 * substring-locations.c (format_warning_va): Strengthen case 1 so
16410 that both endpoints of the substring must be within the format
16411 range for just the substring to be printed.
16413 2017-01-13 Uros Bizjak <ubizjak@gmail.com>
16415 * config/i386/i386.opt (msgx): Use ix86_isa_flags2 variable.
16416 * config/i386/i386.c (ix86_target_string): Add missing options
16417 to isa_opts and reorder options by implied ISAs. Rename isa_opts2 to
16418 isa2_opts, ix86_flag_opts to flag2_opts, ix86_target_other to
16419 flags_other and ix86_target_other to flags2_other. Display unknown
16421 (ix86_valid_target_attribute_inner_p): Add missing options and
16422 reorder options by implied ISAs, as in ix86_target_string.
16424 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
16426 * hash-table.h (hash_table::too_empty_p): New function.
16427 (hash_table::expand): Use it.
16428 (hash_table::traverse): Likewise.
16429 (hash_table::empty_slot): Use sizeof (value_type) instead of
16430 sizeof (PTR) to convert bytes to elements. Shrink the table
16431 if the current size is excessive for the current number of
16434 2017-01-13 Richard Sandiford <richard.sandiford@arm.com>
16436 * ira-costs.c (record_reg_classes): Break from the inner loop
16437 early once alt_fail is known to be true. Update outer loop
16438 handling accordingly.
16440 2017-01-13 Jeff Law <law@redhat.com>
16442 * tree-ssa-dse.c (decrement_count): New function.
16443 (increment_start_addr, maybe_trim_memstar_call): Likewise.
16444 (dse_dom_walker::optimize_stmt): Call maybe_trim_memstar_call directly
16445 when we know the partially dead statement is a mem* function.
16447 PR tree-optimization/61912
16448 PR tree-optimization/77485
16449 * tree-ssa-dse.c: Include expr.h.
16450 (maybe_trim_constructor_store): New function.
16451 (maybe_trim_partially_dead_store): Call maybe_trim_constructor_store.
16453 PR tree-optimization/33562
16454 PR tree-optimization/61912
16455 PR tree-optimization/77485
16456 * doc/invoke.texi: Document new dse-max-object-size param.
16457 * params.def (PARM_DSE_MAX_OBJECT_SIZE): New PARAM.
16458 * tree-ssa-dse.c: Include params.h.
16459 (dse_store_status): New enum.
16460 (initialize_ao_ref_for_dse): New, partially extracted from
16462 (valid_ao_ref_for_dse, normalize_ref): New.
16463 (setup_live_bytes_from_ref, compute_trims): Likewise.
16464 (clear_bytes_written_by, maybe_trim_complex_store): Likewise.
16465 (maybe_trim_partially_dead_store): Likewise.
16466 (maybe_trim_complex_store): Likewise.
16467 (dse_classify_store): Renamed from dse_possibly_dead_store_p.
16468 Track what bytes live from the original store. Return tri-state
16469 for dead, partially dead or live.
16470 (dse_dom_walker): Add constructor, destructor and new private members.
16471 (delete_dead_call, delete_dead_assignment): New extracted from
16473 (dse_optimize_stmt): Make a member of dse_dom_walker.
16474 Use initialize_ao_ref_for_dse.
16476 PR tree-optimization/33562
16477 PR tree-optimization/61912
16478 PR tree-optimization/77485
16479 * sbitmap.h (bitmap_count_bits): Prototype.
16480 (bitmap_clear_range, bitmap_set_range): Likewise.
16481 * sbitmap.c (bitmap_clear_range): New function.
16482 (bitmap_set_range, sbitmap_popcount, bitmap_count_bits): Likewise.
16484 2017-01-13 Martin Liska <mliska@suse.cz>
16487 * function.c (set_cfun): Add new argument force.
16488 * function.h (set_cfun): Likewise.
16489 * ipa-inline-transform.c (inline_call): Use the function when
16490 strict alising from is dropped for function we inline to.
16492 2017-01-13 Richard Biener <rguenther@suse.de>
16494 * tree-pretty-print.c (dump_generic_node): Fix inverted condition
16495 for dumping GIMPLE INTEGER_CSTs.
16497 2017-01-13 Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
16499 * config/sol2.h (TARGET_OS_CPP_BUILTINS): Define __STDC_VERSION__
16500 to 201112L since C++17.
16502 2017-01-13 Maxim Ostapenko <m.ostapenko@samsung.com>
16505 * asan.c (asan_needs_odr_indicator_p): Don't emit ODR indicators
16506 if -fsanitize=kernel-address is present.
16508 2017-01-13 Richard Biener <rguenther@suse.de>
16510 * tree-pretty-print.c (dump_generic_node): Dump INTEGER_CSTs
16511 as _Literal ( type ) number in case usual suffixes do not
16512 preserve all information.
16514 2017-01-13 Richard Biener <rguenther@suse.de>
16516 PR tree-optimization/77283
16517 * gimple-ssa-split-paths.c: Include gimple-ssa.h, tree-phinodes.h
16518 and ssa-iterators.h.
16519 (is_feasible_trace): Implement a cost model based on joiner
16522 2017-01-12 Michael Meissner <meissner@linux.vnet.ibm.com>
16525 * config/rs6000/rs6000.md (FP_ISA3): Do not optimize converting
16526 char or short to __float128/_Float128 directly.
16528 2017-01-12 Martin Sebor <msebor@redhat.com>
16530 to -Wformat-overflow.
16531 * gimple-ssa-sprintf.c (pass_sprintf_length::gate): Adjust.
16532 (min_bytes_remaining): Same.
16533 (get_string_length): Same.
16534 (format_string): Same.
16535 (format_directive): Same.
16537 (pass_sprintf_length::handle_gimple_call): Same.
16539 2017-01-12 Jakub Jelinek <jakub@redhat.com>
16541 * gimple-ssa-sprintf.c (try_substitute_return_value): Remove
16542 info.nowrite calls with no lhs that can't throw. Return bool
16543 whether gsi_remove has been called or not.
16544 (pass_sprintf_length::handle_gimple_call): Return bool whether
16545 try_substitute_return_value called gsi_remove. Formatting fix.
16546 (pass_sprintf_length::execute): Don't use gsi_remove if
16547 handle_gimple_call returned true.
16550 * cfgrtl.c (rtl_tidy_fallthru_edge): For any_uncondjump_p that can't
16551 be removed due to side-effects, don't remove following barrier nor
16552 turn the successor edge into fallthru edge.
16554 2017-01-12 Bill Schmidt <wschmidt@linux.vnet.ibm.com>
16557 * config/rs6000/rs6000.c (insn_is_swappable_p): Mark
16558 element-reversing loads and stores as not swappable.
16560 2017-01-12 Nathan Sidwell <nathan@acm.org>
16561 Nicolai Stange <nicstange@gmail.com>
16563 * combine.c (try_combine): Don't ignore result of overlap checking
16564 loop. Combine overlap & asm check into single loop.
16566 2017-01-12 Richard Biener <rguenther@suse.de>
16568 * tree-pretty-print.c (dump_generic_node): Provide -gimple
16569 variant for MEM_REF. Sanitize INTEGER_CST for -gimple.
16571 2017-01-12 Richard Biener <rguenther@suse.de>
16573 * tree.c (initialize_tree_contains_struct): Make TS_OPTIMIZATION
16574 and TS_TARGET_OPTION directly derive from TS_BASE.
16575 * tree-core.h (tree_optimization_option): Derive from tree_base.
16576 (tree_target_option): Likewise.
16578 2017-01-11 Uros Bizjak <ubizjak@gmail.com>
16580 * config/i386/i386.c (memory_address_length): Increase len
16581 only when rip_relative_addr_p returns false.
16583 2017-01-11 Julia Koval <julia.koval@intel.com>
16585 * common/config/i386/i386-common.c (OPTION_MASK_ISA_SGX_UNSET): New.
16586 (OPTION_MASK_ISA_SGX_SET): New.
16587 (ix86_handle_option): Handle OPT_msgx.
16588 * config.gcc: Added sgxintrin.h.
16589 * config/i386/driver-i386.c (host_detect_local_cpu): Detect sgx.
16590 * config/i386/i386-c.c (ix86_target_macros_internal): Define __SGX__.
16591 * config/i386/i386.c (ix86_target_string): Add -msgx.
16593 (ix86_option_override_internal): Handle new options.
16594 (ix86_valid_target_attribute_inner_p): Add sgx.
16595 * config/i386/i386.h (TARGET_SGX, TARGET_SGX_P): New.
16596 * config/i386/i386.opt: Add msgx.
16597 * config/i386/sgxintrin.h: New file.
16598 * config/i386/x86intrin.h: Add sgxintrin.h.
16600 2017-01-11 Jakub Jelinek <jakub@redhat.com>
16603 * fold-const.c (maybe_nonzero_address): Return 1 for function
16605 (tree_single_nonzero_warnv_p): Don't handle function local objects
16609 * gcc.c (default_compilers): Don't add -o %g.s for -S -save-temps
16612 2017-01-11 David Malcolm <dmalcolm@redhat.com>
16615 * opts.c: Include "spellcheck.h"
16616 (struct string_fragment): New struct.
16617 (struct edit_distance_traits<const string_fragment &>): New
16619 (get_closest_sanitizer_option): New function.
16620 (parse_sanitizer_options): Offer suggestions for unrecognized arguments.
16622 2017-01-11 Jakub Jelinek <jakub@redhat.com>
16624 * dwarf2out.c (DWARF_COMPILE_UNIT_HEADER_SIZE): For DWARF5 decrease
16626 (DWARF_COMDAT_TYPE_UNIT_HEADER_SIZE): Always
16627 DWARF_COMPILE_UNIT_HEADER_SIZE plus 12.
16628 (DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE): Define.
16629 (calc_base_type_die_sizes): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE
16630 for initial die_offset if dwarf_split_debug_info.
16631 (output_comp_unit): Use DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE for
16632 initial next_die_offset if dwo_id is non-NULL. Don't emit padding
16634 (output_skeleton_debug_sections): Formatting fix. Use
16635 DWARF_COMPILE_UNIT_SKELETON_HEADER_SIZE instead of
16636 DWARF_COMPILE_UNIT_HEADER_SIZE. Don't emit padding.
16638 2017-01-11 Wilco Dijkstra <wdijkstr@arm.com>
16640 * config/arm/cortex-a53.md: Add bypasses for
16641 cortex_a53_r2f_cvt.
16642 (cortex_a53_r2f): Only use for transfers.
16643 (cortex_a53_f2r): Likewise.
16644 (cortex_a53_r2f_cvt): Add reservation for conversions.
16645 (cortex_a53_f2r_cvt): Likewise.
16647 2017-01-11 Tamar Christina <tamar.christina@arm.com>
16649 * config/arm/arm_neon.h: Add __artificial__ and gnu_inline
16650 to all inlined functions, change static to extern.
16652 2017-01-11 Christophe Lyon <christophe.lyon@linaro.org>
16655 * config/arm/arm.c (legitimize_pic_address): Handle reference to
16657 (arm_assemble_integer): Likewise.
16659 2017-01-11 Richard Earnshaw <rearnsha@arm.com>
16661 * config.gcc: Use new awk script to check CPU, FPU and architecture
16662 parameters for --with-... options.
16663 * config/arm/parsecpu.awk: New file
16664 * config/arm/arm-cpus.in: New file.
16665 * config/arm/arm-opts.h: Include arm-cpu.h instead of processing .def
16667 * config/arm/arm.c: Include arm-cpu-data.h instead of processing .def
16669 * config/arm/t-arm: Update dependency rules.
16670 * common/config/arm/arm-common.c: Include arm-cpu-cdata.h instead
16671 of processing .def files.
16672 * config/arm/genopt.sh: Deleted.
16673 * config/arm/gentune.sh: Deleted.
16674 * config/arm/arm-cores.def: Deleted.
16675 * config/arm/arm-arches.def: Deleted.
16676 * config/arm/arm-fpus.def: Deleted.
16677 * config/arm/arm-tune.md: Regenerated.
16678 * config/arm/arm-tables.opt: Regenerated.
16679 * config/arm/arm-cpu.h: New generated file.
16680 * config/arm/arm-cpu-data.h: New generated file.
16681 * config/arm/arm-cpu-cdata.h: New generated file.
16683 2017-01-11 Maxim Ostapenko <m.ostapenko@samsung.com>
16686 * lto-cgraph.c (lto_output_varpool_node): Pack dynamically_initialized
16688 (input_varpool_node): Unpack dynamically_initialized bit.
16690 2017-01-11 Eric Botcazou <ebotcazou@adacore.com>
16692 PR rtl-optimization/79032
16693 * lra-constraints.c (simplify_operand_subreg): In the MEM case, test
16694 the alignment of the adjusted memory reference against that of MODE,
16695 instead of the alignment of the original memory reference.
16697 2017-01-11 Martin Jambor <mjambor@suse.cz>
16699 * hsa.c (hsa_callable_function_p): Revert addition of DECL_ARTIFICIAL
16701 * ipa-hsa.c (process_hsa_functions): Only duplicate non-artificial
16702 decorated functions.
16704 2017-01-11 Richard Biener <rguenther@suse.de>
16706 * tree-vrp.c (evrp_dom_walker::before_dom_children): Also
16707 set range/nonnull info for PHI results. Do not set it on
16708 stmts marked for removal.
16710 2017-01-10 Eric Botcazou <ebotcazou@adacore.com>
16712 * expr.c (store_field): In the bitfield case, fetch the return value
16713 from the registers before applying a single big-endian adjustment.
16714 Always do a final load for a BLKmode value not larger than a word.
16716 2017-01-10 David Malcolm <dmalcolm@redhat.com>
16719 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
16720 that we correctly handle column numbers greater than
16721 LINE_MAP_MAX_COLUMN_NUMBER.
16723 2017-01-10 Martin Sebor <msebor@redhat.com>
16725 PR middle-end/78245
16726 * gimple-ssa-sprintf.c (get_destination_size): Call
16727 {init,fini}object_sizes.
16728 * tree-object-size.c (addr_object_size): Adjust.
16729 (pass_through_call): Adjust.
16730 (pass_object_sizes::execute): Adjust.
16731 * tree-object-size.h (fini_object_sizes): Declare.
16733 2017-01-10 Martin Sebor <msebor@redhat.com>
16735 PR tree-optimization/78775
16736 * builtins.c (get_size_range): Move...
16737 * calls.c: ...to here.
16738 (alloc_max_size): Accept zero argument.
16739 (operand_signed_p): Remove.
16740 (maybe_warn_alloc_args_overflow): Call get_size_range.
16741 * calls.h (get_size_range): Declare.
16743 2017-01-10 Joe Seymour <joe.s@somniumtech.com>
16745 * config/msp430/driver-msp430.c (msp430_mcu_data): Sync with data
16746 from TI's devices.csv file as of September 2016.
16747 * config/msp430/msp430.c (msp430_mcu_data): Likewise.
16749 2017-01-10 Sandra Loosemore <sandra@codesourcery.com>
16751 * doc/extend.texi: Tweak formatting to fix overfull hbox warnings.
16752 * doc/invoke.texi: Likewise.
16753 * doc/md.texi: Likewise.
16754 * doc/objc.texi: Likewise.
16756 2017-01-10 Joshua Conner <joshconner@google.com>
16758 * config/arm/fuchsia-elf.h: New file.
16759 * config/fuchsia.h: New file.
16760 * config.gcc (*-*-fuchsia*): Set native_system_header_dir.
16761 (aarch64*-*-fuchsia*, arm*-*-fuchsia*, x86_64-*-fuchsia*): Add to
16763 * config.host: (aarch64*-*-fuchsia*, arm*-*-fuchsia*): Add to hosts.
16765 2016-01-10 Richard Biener <rguenther@suse.de>
16767 PR tree-optimization/79034
16768 * tree-call-cdce.c (shrink_wrap_one_built_in_call_with_conds):
16769 Propagate out degenerate PHIs in the joiner.
16771 2017-01-10 Martin Liska <mliska@suse.cz>
16773 * ipa-icf.c (sort_sem_items_by_decl_uid): New function.
16774 (sort_congruence_classes_by_decl_uid): Likewise.
16775 (sort_congruence_class_groups_by_decl_uid): Likewise.
16776 (sem_item_optimizer::merge_classes): Sort class, groups in these
16777 classes and members in the groups by DECL_UID of declarations.
16778 This would make merge operations stable.
16780 2017-01-10 Martin Liska <mliska@suse.cz>
16782 * ipa-icf.c (sem_item_optimizer::sem_item_optimizer): Remove
16783 usage of m_classes_vec.
16784 (sem_item_optimizer::~sem_item_optimizer): Likewise.
16785 (sem_item_optimizer::get_group_by_hash): Likewise.
16786 (sem_item_optimizer::subdivide_classes_by_equality): Likewise.
16787 (sem_item_optimizer::subdivide_classes_by_sensitive_refs): Likewise.
16788 (sem_item_optimizer::verify_classes): Likewise.
16789 (sem_item_optimizer::process_cong_reduction): Likewise.
16790 (sem_item_optimizer::dump_cong_classes): Likewise.
16791 (sem_item_optimizer::merge_classes): Likewise.
16792 * ipa-icf.h (congruence_class_hash): Rename from
16793 congruence_class_group_hash. Remove declaration of m_classes_vec.
16795 2017-01-10 Andrew Senkevich <andrew.senkevich@intel.com>
16797 * common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX512VPOPCNTDQ_SET,
16798 OPTION_MASK_ISA_AVX512VPOPCNTDQ_UNSET): New.
16799 * config.gcc: Add avx512vpopcntdqintrin.h.
16800 * config/i386/avx512vpopcntdqintrin.h: New.
16801 * config/i386/cpuid.h (bit_AVX512VPOPCNTDQ): New.
16802 * config/i386/i386-builtin-types.def: Add new types.
16803 * config/i386/i386-builtin.def (__builtin_ia32_vpopcountd_v16si,
16804 __builtin_ia32_vpopcountd_v16si_mask, __builtin_ia32_vpopcountq_v8di,
16805 __builtin_ia32_vpopcountq_v8di_mask): New.
16806 * config/i386/i386-c.c (ix86_target_macros_internal): Define
16807 __AVX512VPOPCNTDQ__.
16808 * config/i386/i386.c (ix86_target_string): Add -mavx512vpopcntdq.
16809 (PTA_AVX512VPOPCNTDQ): Define.
16810 * config/i386/i386.h (TARGET_AVX512VPOPCNTDQ,
16811 TARGET_AVX512VPOPCNTDQ_P): Define.
16812 * config/i386/i386.opt: Add mavx512vpopcntdq.
16813 * config/i386/immintrin.h: Include avx512vpopcntdqintrin.h.
16814 * config/i386/sse.md (define_insn "vpopcount<mode><mask_name>"): New.
16816 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
16818 PR middle-end/77484
16819 * predict.def (PRED_CALL): Set to 67.
16821 2017-01-09 Eric Botcazou <ebotcazou@adacore.com>
16823 * expr.c (store_field): In the bitfield case, if the value comes from
16824 a function call and is of an aggregate type returned in registers, do
16825 not modify the field mode; extract the value in all cases if the mode
16826 is BLKmode and the size is not larger than a word.
16828 2017-01-09 Dominique d'Humieres <dominiq@lps.ens.fr>
16831 * config/i386/cpuid.h: Fix undefined behavior.
16833 2017-01-04 Jeff Law <law@redhat.com>
16835 PR tree-optimization/79007
16836 PR tree-optimization/67955
16837 * tree-ssa-alias.c (same_addr_size_stores_p): Only need to be
16838 conservative for pt.null when flag_non_call_exceptions is on.
16840 2017-01-09 Jakub Jelinek <jakub@redhat.com>
16842 PR translation/79019
16843 PR translation/79020
16844 * params.def (PARAM_INLINE_MIN_SPEEDUP,
16845 PARAM_IPA_CP_SINGLE_CALL_PENALTY,
16846 PARAM_USE_AFTER_SCOPE_DIRECT_EMISSION_THRESHOLD): Fix typos
16848 * config/avr/avr.opt (maccumulate-args): Likewise.
16849 * config/msp430/msp430.opt (mwarn-mcu): Likewise.
16850 * common.opt (freport-bug): Likewise.
16851 * cif-code.def (CIF_FINAL_ERROR): Likewise.
16852 * doc/invoke.texi (ipa-cp-single-call-penalty): Likewise.
16853 * config/s390/s390.c (s390_invalid_binary_op): Fix spelling in
16854 translatable string.
16855 * config/i386/i386.c (function_value_32): Likewise.
16856 * config/nios2/nios2.c (nios2_valid_target_attribute_rec): Likewise.
16857 * config/msp430/msp430.c (msp430_option_override, msp430_attr):
16859 * config/msp430/driver-msp430.c (msp430_select_hwmult_lib): Likewise.
16860 * common/config/msp430/msp430-common.c (msp430_handle_option):
16862 * symtab.c (symtab_node::verify_base): Likewise.
16863 * opts.c (set_debug_level): Likewise.
16864 * tree.c (verify_type_variant): Likewise. Fix typo in comment.
16865 * config/rs6000/rs6000-c.c (altivec_resolve_overloaded_builtin): Add
16866 missing whitespace to translatable strings.
16867 * config/avr/avr.md (bswapsi2): Fix typo in comment.
16868 * config/sh/superh.h: Likewise.
16869 * config/i386/xopintrin.h: Likewise.
16870 * config/i386/znver1.md: Likewise.
16871 * config/rs6000/rs6000.c (struct rs6000_opt_mask): Likewise.
16872 * ipa-inline-analysis.c (compute_inline_parameters): Likewise.
16873 * double-int.h (struct double_int): Likewise.
16874 * double-int.c (div_and_round_double): Likewise.
16875 * wide-int.cc: Likewise.
16876 * tree-ssa.c (non_rewritable_mem_ref_base): Likewise.
16877 * tree-ssa-sccvn.c (vn_reference_lookup_3): Likewise.
16878 * cfgcleanup.c (crossjumps_occured): Renamed to ...
16879 (crossjumps_occurred): ... this.
16880 (try_crossjump_bb, try_head_merge_bb, try_optimize_cfg, cleanup_cfg):
16883 PR tree-optimization/78899
16884 * tree-if-conv.c (version_loop_for_if_conversion): Instead of
16885 returning bool return struct loop *, NULL for failure and the new
16887 (versionable_outer_loop_p): Don't version outer loop if it has
16888 dont_vectorized bit set.
16889 (tree_if_conversion): When versioning outer loop, ensure
16890 tree_if_conversion is performed also on the inner loop of the
16891 non-vectorizable outer loop copy.
16892 * tree-vectorizer.c (set_uid_loop_bbs): Formatting fix. Fold
16893 LOOP_VECTORIZED in inner loop of the scalar outer loop and
16894 prevent vectorization of it.
16895 (vectorize_loops): For outer + inner LOOP_VECTORIZED, ensure
16896 the outer loop vectorization of the non-scalar version is attempted
16897 before vectorization of the inner loop in scalar version. If
16898 outer LOOP_VECTORIZED guarded loop is not vectorized, prevent
16899 vectorization of its inner loop.
16900 * tree-vect-loop-manip.c (rename_variables_in_bb): If outer_loop
16901 has 2 inner loops, rename also on edges from bb whose single pred
16902 is outer_loop->header. Fix typo in function comment.
16904 2017-01-09 Martin Sebor <msebor@redhat.com>
16907 * asan.c (asan_emit_stack_protection): Increase local buffer size
16908 to avoid snprintf truncation warning.
16910 2017-01-09 Andrew Pinski <apinski@cavium.com>
16912 * config/aarch64/aarch64-cores.def: Add thunderx2t99. Change vulcan
16913 to reference thunderx2t99 for the tuning structure
16914 * config/aarch64/aarch64-cost-tables.h (vulcan_extra_costs):
16916 (thunderx2t99_extra_costs): This.
16917 * config/aarch64/aarch64-tune.md: Regenerate.
16918 * config/aarch64/aarch64.c (vulcan_addrcost_table): Rename to ...
16919 (vulcan_addrcost_table): This.
16920 (vulcan_regmove_cost): Rename to ...
16921 (thunderx2t99_regmove_cost): This.
16922 (vulcan_vector_cost): Rename to ...
16923 (thunderx2t99_vector_cost): this.
16924 (vulcan_branch_cost): Rename to ...
16925 (thunderx2t99_branch_cost): This.
16926 (vulcan_tunings): Rename to ...
16927 (thunderx2t99_tunings): This and s/vulcan/thunderx2t99 .
16928 * doc/invoke.texi (AARCH64/mtune): Add thunderx2t99.
16930 2017-01-09 Martin Jambor <mjambor@suse.cz>
16934 * ipa-prop.h (ipa_jump_func): Swap positions of vr_known and m_vr.
16935 * ipa-cp.c (ipa_vr_operation_and_type_effects): New function.
16936 (propagate_vr_accross_jump_function): Use the above function for all
16937 value range computations for pass-through jump functions and type
16938 converasion from explicit value range values.
16939 (ipcp_propagate_stage): Do not attempt to deduce types of formal
16940 parameters from TYPE_ARG_TYPES.
16941 * ipa-prop.c (ipa_write_jump_function): Remove trailing whitespace.
16942 (ipa_write_node_info): Stream type of the actual argument.
16943 (ipa_read_node_info): Likewise. Also remove trailing whitespace.
16945 2017-01-09 Martin Liska <mliska@suse.cz>
16948 * gcc.c (driver_handle_option): Handle OPT_E and set have_E.
16949 (lookup_compiler): Do not show error message with have_E.
16951 2017-01-09 Jakub Jelinek <jakub@redhat.com>
16953 PR tree-optimization/78938
16954 * tree-vect-stmts.c (vectorizable_condition): For non-masked COND_EXPR
16955 where comp_vectype is VECTOR_BOOLEAN_TYPE_P, use
16956 BIT_{NOT,XOR,AND,IOR}_EXPR on the comparison operands instead of
16957 {EQ,NE,GE,GT,LE,LT}_EXPR directly inside of VEC_COND_EXPR. Formatting
16960 2017-01-09 Kyrylo Tkachov <kyrylo.tkachov@arm.com>
16962 * tree-ssa-address.c (gen_addr_rtx): Don't handle index if it
16965 2017-01-09 Richard Biener <rguenther@suse.de>
16967 PR tree-optimization/78997
16968 * tree-vect-slp.c (vect_mask_constant_operand_p): Handle SSA
16969 name condition properly.
16971 2017-01-09 Richard Biener <rguenther@suse.de>
16974 * dwarf2out.c (is_cxx): New overload with context.
16975 (is_naming_typedef_decl): Use it.
16977 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
16979 * invoke.texi (Option Summary): Correct spacing in option lists
16980 and add line breaks to fix over-long lines.
16982 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
16984 PR middle-end/17660
16986 * extend.texi (Common Variable Attributes): Add xref to GCC
16987 Internals manual to explain mode attribute keywords.
16989 2017-01-08 Sandra Loosemore <sandra@codesourcery.com>
16992 * doc/invoke.texi (Option Summary): Move -pthread to Linker Options
16993 and Preprocessor Options.
16994 (Options for Linking): Document -pthread here....
16995 (RS/6000 and PowerPC Options): ...not here.
16996 (Solaris 2 Options): ...or here.
16997 * doc/cppopts.texi: Document -pthread.
16999 2017-01-08 Martin Sebor <msebor@redhat.com>
17001 PR middle-end/77708
17002 * doc/invoke.texi (Warning Options): Document -Wformat-truncation.
17003 * gimple-ssa-sprintf.c (call_info::reval_used, call_info::warnopt):
17004 New member functions.
17005 (format_directive): Used them.
17007 (pass_sprintf_length::handle_gimple_call): Same.
17008 * graphite-sese-to-poly.c (tree_int_to_gmp): Increase buffer size
17009 to avoid truncation for any argument.
17010 (extract_affine_mul): Same.
17011 * tree.c (get_file_function_name): Same.
17013 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
17015 PR middle-end/77484
17016 * predict.def (PRED_INDIR_CALL): Set to 86.
17018 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
17020 PR preprocessor/54124
17021 * doc/cppopts.texi: Reformat -d subtable to list the full name
17022 of the options. Add cross-reference to the docs for the general
17023 compiler -d options.
17024 * doc/invoke.texi (Developer Options): Add cross-reference to the
17025 preprocessor-specific -d option documentation.
17027 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
17029 PR preprocessor/13498
17030 * doc/cpp.texi (Search Path): Rewrite to remove obsolete and
17031 redudant material, and reflect new command-line options.
17032 (System Headers): Likewise.
17034 2017-01-07 Sandra Loosemore <sandra@codesourcery.com>
17036 * doc/cppdiropts.texi: Merge documentation of -I, -iquote,
17037 -isystem, and -idirafter. Copy-edit.
17038 * doc/cppopts.texi: Copy-edit. Remove contradiction about
17039 default for -ftrack-macro-expansion. Delete obsolete and
17040 badly-formatted implementation details about -fdebug-cpp output.
17041 * doc/cppwarnopts.texi: Copy-edit.
17043 2017-01-07 David Malcolm <dmalcolm@redhat.com>
17046 * input.c (selftest::test_accessing_ordinary_linemaps): Verify
17047 that the transition from a max line width >= 1<<10 to narrower
17048 lines works correctly.
17050 2017-01-07 Alexandre Oliva <aoliva@redhat.com>
17052 * doc/options.texi (PerFunction): New.
17053 * opt-functions.awk (switch_flags): Map both Optimization and
17054 PerFunction to CL_OPTIMIZATION.
17055 * opth-gen.awk: Test for PerFunction flag along with
17057 * optc-save-gen.awk: Likewise. Introduce var_opt_hash and set
17058 it only when the latter is present. Skip those that don't in
17059 the hash function generator.
17060 * common.opt (fvar-tracking): Mark as PerFunction instead of
17062 (fvar-tracking-assignments): Likewise.
17063 (fvar-tracking-assignments-toggle): Likewise.
17064 (fvar-tracking-uninit): Likewise.
17066 2017-01-07 Jakub Jelinek <jakub@redhat.com>
17068 PR translation/79018
17069 * params.def (PARAM_MAX_STORES_TO_MERGE): Add missing space between
17072 2017-01-06 Mikael Pettersson <mikpelinux@gmail.com>
17075 * config/m68k/m68k.opt (LONG_JUMP_TABLE_OFFSETS): New option.
17076 * config/m68k/linux.h (ASM_RETURN_CASE_JUMP): Handle
17077 TARGET_LONG_JUMP_TABLE_OFFSETS.
17078 * config/m68k/m68kelf.h (ASM_RETURN_CASE_JUMP): Likewise.
17079 * config/m68k/netbsd-elf.h (ASM_RETURN_CASE_JUMP): Likewise.
17080 * config/m68k/m68k.h (CASE_VECTOR_MODE): Likewise.
17081 (ASM_OUTPUT_ADDR_DIFF_ELF): Likewise.
17082 * config/m68k/m68k.md (tablejump expander): Likewise.
17083 (*tablejump_pcrel_hi): Renamed from unnamed insn, reject
17084 TARGET_LONG_JUMP_TABLE_OFFSETS.
17085 (*tablejump_pcrel_si): New insn, handle TARGET_LONG_JUMP_TABLE_OFFSETS.
17086 * doc/invoke.texi (M68K options): Add -mlong-jump-table-offsets.
17088 2017-01-06 Edgar E. Iglesias <edgar.iglesias@xilinx.com>
17089 David Holsgrove <david.holsgrove@xilinx.com>
17091 * common/config/microblaze/microblaze-common.c
17092 (TARGET_EXCEPT_UNWIND_INFO): Remove.
17093 * config/microblaze/microblaze-protos.h (microblaze_eh_return):
17095 * config/microblaze/microblaze.c (microblaze_must_save_register)
17096 (microblaze_expand_epilogue, microblaze_return_addr): Handle
17098 (microblaze_eh_return): New function.
17099 * config/microblaze/microblaze.h (RETURN_ADDR_OFFSET)
17100 (EH_RETURN_DATA_REGNO, MB_EH_STACKADJ_REGNUM)
17101 (EH_RETURN_STACKADJ_RTX, ASM_PREFERRED_EH_DATA_FORMAT): New macros.
17102 * config/microblaze/microblaze.md (eh_return): New pattern.
17104 2017-01-06 Jakub Jelinek <jakub@redhat.com>
17106 * system.h (GCC_DIAGNOSTIC_PUSH_IGNORED, GCC_DIAGNOSTIC_POP,
17107 GCC_DIAGNOSTIC_STRINGIFY): Define.
17109 * read-rtl.c (rtx_reader::read_rtx_code): Avoid -Wsign-compare warning.
17111 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
17113 * config/arm/arm.md (<mcrr>): New.
17115 * config/arm/arm.c (arm_arch5te): New.
17116 (arm_option_override): Set arm_arch5te.
17117 (arm_coproc_builtin_available): Add support for mcrr, mcrr2, mrrc
17119 * config/arm/arm-builtins.c (MCRR_QUALIFIERS): Define to...
17120 (arm_mcrr_qualifiers): ... this. New.
17121 (MRRC_QUALIFIERS): Define to...
17122 (arm_mrrc_qualifiers): ... this. New.
17123 * config/arm/arm_acle.h (__arm_mcrr, __arm_mcrr2, __arm_mrrc,
17125 * config/arm/arm_acle_builtins.def (mcrr, mcrr2, mrrc, mrrc2): New.
17126 * config/arm/iterators.md (MCRRI, mcrr, MCRR): New.
17127 (MRRCI, mrrc, MRRC): New.
17128 * config/arm/unspecs.md (VUNSPEC_MCRR, VUNSPEC_MCRR2, VUNSPEC_MRRC,
17129 VUNSPEC_MRRC2): New.
17131 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
17133 * config/arm/arm.md (<mcr>): New.
17135 * config/arm/arm.c (arm_coproc_builtin_available): Add
17136 support for mcr, mrc, mcr2 and mrc2.
17137 * config/arm/arm-builtins.c (MCR_QUALIFIERS): Define to...
17138 (arm_mcr_qualifiers): ... this. New.
17139 (MRC_QUALIFIERS): Define to ...
17140 (arm_mrc_qualifiers): ... this. New.
17141 (MCR_QUALIFIERS): Define to ...
17142 (arm_mcr_qualifiers): ... this. New.
17143 * config/arm/arm_acle.h (__arm_mcr, __arm_mrc, __arm_mcr2,
17145 * config/arm/arm_acle_builtins.def (mcr, mcr2, mrc, mrc2): New.
17146 * config/arm/iterators.md (MCRI, mcr, MCR, MRCI, mrc, MRC): New.
17147 * config/arm/unspecs.md (VUNSPEC_MCR, VUNSPEC_MCR2, VUNSPEC_MRC,
17148 VUNSPEC_MRC2): New.
17150 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
17152 * config/arm/arm.md (*ldc): New.
17156 * config/arm/arm.c (arm_coproc_builtin_available): Add
17157 support for ldc,ldcl,stc,stcl,ldc2,ldc2l,stc2 and stc2l.
17158 (arm_coproc_ldc_stc_legitimate_address): New.
17159 * config/arm/arm-builtins.c (arm_type_qualifiers): Add
17160 'qualifier_const_pointer'.
17161 (LDC_QUALIFIERS): Define to...
17162 (arm_ldc_qualifiers): ... this. New.
17163 (STC_QUALIFIERS): Define to...
17164 (arm_stc_qualifiers): ... this. New.
17165 * config/arm/arm-protos.h
17166 (arm_coproc_ldc_stc_legitimate_address): New.
17167 * config/arm/arm_acle.h (__arm_ldc, __arm_ldcl, __arm_stc,
17168 __arm_stcl, __arm_ldc2, __arm_ldc2l, __arm_stc2, __arm_stc2l): New.
17169 * config/arm/arm_acle_builtins.def (ldc, ldc2, ldcl, ldc2l, stc,
17170 stc2, stcl, stc2l): New.
17171 * config/arm/constraints.md (Uz): New.
17172 * config/arm/iterators.md (LDCI, STCI, ldc, stc, LDC STC): New.
17173 * config/arm/unspecs.md (VUNSPEC_LDC, VUNSPEC_LDC2, VUNSPEC_LDCL,
17174 VUNSPEC_LDC2L, VUNSPEC_STC, VUNSPEC_STC2, VUNSPEC_STCL,
17175 VUNSPEC_STC2L): New.
17177 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
17179 * config/arm/arm.md (<cdp>): New.
17180 * config/arm/arm.c (neon_const_bounds): Rename this ...
17181 (arm_const_bounds): ... this.
17182 (arm_coproc_builtin_available): New.
17183 * config/arm/arm-builtins.c (SIMD_MAX_BUILTIN_ARGS): Increase.
17184 (arm_type_qualifiers): Add 'qualifier_unsigned_immediate'.
17185 (CDP_QUALIFIERS): Define to...
17186 (arm_cdp_qualifiers): ... this. New.
17188 (arm_expand_builtin_args): Add case for 6 arguments.
17189 * config/arm/arm-protos.h (neon_const_bounds): Rename this ...
17190 (arm_const_bounds): ... this.
17191 (arm_coproc_builtin_available): New.
17192 * config/arm/arm_acle.h (__arm_cdp): New.
17194 * config/arm/arm_acle_builtins.def (cdp): New.
17196 * config/arm/iterators.md (CDPI,CDP,cdp): New.
17197 * config/arm/neon.md: Rename all 'neon_const_bounds' to
17198 'arm_const_bounds'.
17199 * config/arm/types.md (coproc): New.
17200 * config/arm/unspecs.md (VUNSPEC_CDP, VUNSPEC_CDP2): New.
17201 * gcc/doc/extend.texi (ACLE): Add a mention of Coprocessor intrinsics.
17202 * gcc/doc/sourcebuild.texi (arm_coproc1_ok, arm_coproc2_ok,
17203 arm_coproc3_ok, arm_coproc4_ok): Document new effective targets.
17205 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
17207 * config/arm/arm-builtins.c (arm_unsigned_binop_qualifiers): New.
17208 (UBINOP_QUALIFIERS): New.
17210 (acle_builtin_data): New. Change comment.
17211 (arm_builtins): Remove ARM_BUILTIN_CRC32B, ARM_BUILTIN_CRC32H,
17212 ARM_BUILTIN_CRC32W, ARM_BUILTIN_CRC32CB, ARM_BUILTIN_CRC32CH,
17213 ARM_BUILTIN_CRC32CW. Add ARM_BUILTIN_ACLE_BASE and include
17214 arm_acle_builtins.def.
17215 (ARM_BUILTIN_ACLE_PATTERN_START): Define.
17216 (arm_init_acle_builtins): New.
17217 (CRC32_BUILTIN): Remove.
17218 (bdesc_2arg): Remove entries for crc32b, crc32h, crc32w,
17219 crc32cb, crc32ch and crc32cw.
17220 (arm_init_crc32_builtins): Remove.
17221 (arm_init_builtins): Use arm_init_acle_builtins rather
17222 than arm_init_crc32_builtins.
17223 (arm_expand_acle_builtin): New.
17224 (arm_expand_builtin): Use 'arm_expand_acle_builtin'.
17225 * config/arm/arm_acle_builtins.def: New.
17227 2017-01-06 Andre Vieira <andre.simoesdiasvieira@arm.com>
17229 * config/arm/arm-builtins.c (neon_builtin_datum): Rename to ..
17230 (arm_builtin_datum): ... this.
17231 (arm_init_neon_builtin): Rename to ...
17232 (arm_init_builtin): ... this. Add a new parameters PREFIX
17233 and USE_SIG_IN_NAME.
17234 (arm_init_neon_builtins): Replace 'arm_init_neon_builtin' with
17235 'arm_init_builtin'. Replace type 'neon_builtin_datum' with
17236 'arm_builtin_datum'.
17237 (arm_init_vfp_builtins): Likewise.
17238 (builtin_arg): Rename enum's replacing 'NEON_ARG' with
17239 'ARG_BUILTIN' and add a 'ARG_BUILTIN_NEON_MEMORY.
17240 (arm_expand_neon_args): Rename to ...
17241 (arm_expand_builtin_args): ... this. Rename builtin_arg
17242 enum values and differentiate between ARG_BUILTIN_MEMORY
17243 and ARG_BUILTIN_NEON_MEMORY.
17244 (arm_expand_neon_builtin_1): Rename to ...
17245 (arm_expand_builtin_1): ... this. Rename builtin_arg enum
17246 values, arm_expand_builtin_args and add bool parameter NEON.
17247 (arm_expand_neon_builtin): Use arm_expand_builtin_1.
17248 (arm_expand_vfp_builtin): Likewise.
17249 (NEON_MAX_BUILTIN_ARGS): Remove, it was unused.
17251 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
17253 PR middle-end/77484
17254 * predict.def (PRED_POLYMORPHIC_CALL): Set to 59.
17255 * predict.c (tree_estimate_probability_bb): Reverse direction of
17256 polymorphic call predictor.
17258 2017-01-06 David Malcolm <dmalcolm@redhat.com>
17260 * passes.c (execute_one_pass): Split out pass-skipping logic into...
17261 (determine_pass_name_match): ...this new function and...
17262 (should_skip_pass_p): ...this new function.
17264 2017-01-06 Nathan Sidwell <nathan@acm.org>
17266 * ipa-visibility.c (function_and_variable_visibility): Reformat
17267 comments and long lines. Remove extrneous if.
17268 * symtab.c (symtab_node::make_decl_local): Fix code format.
17269 (symtab_node::set_section_for_node): Fix comment typo.
17271 2017-01-06 Martin Liska <mliska@suse.cz>
17274 * lra-constraints.c: Rename invariant to lra_invariant.
17275 * predict.c (set_even_probabilities): Initialize e to NULL.
17277 2017-01-05 Martin Sebor <msebor@redhat.com>
17279 PR tree-optimization/78910
17280 * gimple-ssa-sprintf.c (tree_digits): Add an argument.
17281 (format_integer): Correct off-by-one error in the handling
17282 of precision with negative numbers in signed conversions..
17284 2017-01-05 Eric Botcazou <ebotcazou@adacore.com>
17286 * doc/invoke.texi (C Dialect Options): Adjust -fsso-struct entry.
17288 2017-01-05 Jakub Jelinek <jakub@redhat.com>
17290 PR tree-optimization/71016
17291 * tree-ssa-phiopt.c (tree_ssa_phiopt_worker): Pass cond_stmt to
17292 factor_out_conditional_conversion. Formatting fix.
17293 (factor_out_conditional_conversion): Add cond_stmt argument.
17294 If arg1 is INTEGER_CST, punt if new_arg0 is not any operand of
17295 cond_stmt and if arg0_def_stmt is not the only stmt in its bb.
17298 2017-01-05 David Malcolm <dmalcolm@redhat.com>
17300 * Makefile.in (OBJS): Add read-md.o, read-rtl.o,
17301 read-rtl-function.o, and selftest-rtl.o.
17302 * config/aarch64/aarch64.c: Include selftest.h and selftest-rtl.h.
17303 (selftest::aarch64_test_loading_full_dump): New function.
17304 (selftest::aarch64_run_selftests): New function.
17305 (TARGET_RUN_TARGET_SELFTESTS): Wire it up to
17306 selftest::aarch64_run_selftests.
17307 * config/i386/i386.c
17308 (selftest::ix86_test_loading_dump_fragment_1): New function.
17309 (selftest::ix86_test_loading_call_insn): New function.
17310 (selftest::ix86_test_loading_full_dump): New function.
17311 (selftest::ix86_test_loading_unspec): New function.
17312 (selftest::ix86_run_selftests): Call the new functions.
17313 * emit-rtl.c (maybe_set_max_label_num): New function.
17314 * emit-rtl.h (maybe_set_max_label_num): New decl.
17315 * function.c (instantiate_decls): Guard call to
17316 instantiate_decls_1 with if (DECL_INITIAL (fndecl)).
17317 * function-tests.c (selftest::verify_three_block_rtl_cfg): Remove
17319 * gensupport.c (gen_reader::gen_reader): Pass "false"
17320 for new "compact" param of rtx_reader.
17321 * print-rtl.c (rtx_writer::print_rtx_operand): Print "(nil)"
17322 rather than an empty string for NULL strings.
17323 * read-md.c: Potentially include config.h rather than bconfig.h.
17324 Wrap include of errors.h with #ifdef GENERATOR_FILE.
17325 (have_error): New global, copied from errors.c.
17326 (md_reader::read_name): Rename to...
17327 (md_reader::read_name_1): ...this, adding "out_loc" param,
17328 and converting "missing name or number" to returning false, rather
17330 (md_reader::read_name): Reimplement in terms of read_name_1.
17331 (md_reader::read_name_or_nil): New function.
17332 (md_reader::read_string): Handle "(nil)" by returning NULL.
17333 (md_reader::md_reader): Add new param "compact".
17334 (md_reader::read_md_files): Wrap with #ifdef GENERATOR_FILE.
17335 (md_reader::read_file): New method.
17336 * read-md.h (md_reader::md_reader): Add new param "compact".
17337 (md_reader::read_file): New method.
17338 (md_reader::is_compact): New accessor.
17339 (md_reader::read_name): Convert return type from void to file_location.
17340 (md_reader::read_name_or_nil): New decl.
17341 (md_reader::read_name_1): New decl.
17342 (md_reader::m_compact): New field.
17343 (noop_reader::noop_reader): Pass "false" for new "compact" param
17345 (rtx_reader::rtx_reader): Add new "compact" param.
17346 (rtx_reader::read_rtx_operand): Make virtual and convert return
17347 type from void to rtx.
17348 (rtx_reader::read_until): New decl.
17349 (rtx_reader::handle_any_trailing_information): New virtual function.
17350 (rtx_reader::postprocess): New virtual function.
17351 (rtx_reader::finalize_string): New virtual function.
17352 (rtx_reader::m_in_call_function_usage): New field.
17353 (rtx_reader::m_reuse_rtx_by_id): New field.
17354 * read-rtl-function.c: New file.
17355 * selftest-rtl.c (selftest::assert_rtx_ptr_eq_at): New function.
17356 * selftest-rtl.h (ASSERT_RTX_PTR_EQ): New macro.
17357 (selftest::verify_three_block_rtl_cfg): New decl.
17358 * read-rtl-function.h: New file.
17359 * read-rtl.c: Potentially include config.h rather than bconfig.h.
17360 For host, include function.h, memmodel.h, and emit-rtl.h.
17361 (one_time_initialization): New function.
17362 (struct compact_insn_name): New struct.
17363 (compact_insn_names): New array.
17364 (find_code): Handle insn codes in compact dumps.
17365 (apply_subst_iterator): Wrap with #ifdef GENERATOR_FILE.
17366 (bind_subst_iter_and_attr): Likewise.
17367 (add_condition_to_string): Likewise.
17368 (add_condition_to_rtx): Likewise.
17369 (apply_attribute_uses): Likewise.
17370 (add_current_iterators): Likewise.
17371 (apply_iterators): Likewise.
17372 (initialize_iterators): Guard usage of apply_subst_iterator with
17373 #ifdef GENERATOR_FILE.
17374 (read_conditions): Wrap with #ifdef GENERATOR_FILE.
17375 (md_reader::read_mapping): Likewise.
17376 (add_define_attr_for_define_subst): Likewise.
17377 (add_define_subst_attr): Likewise.
17378 (read_subst_mapping): Likewise.
17379 (check_code_iterator): Likewise.
17380 (rtx_reader::read_rtx): Likewise. Move one-time initialization
17382 (one_time_initialization): New function.
17383 (rtx_reader::read_until): New method.
17384 (read_flags): New function.
17385 (parse_reg_note_name): New function.
17386 (rtx_reader::read_rtx_code): Initialize "iterator" to NULL.
17387 Handle reuse_rtx ids.
17388 Wrap iterator lookup within #ifdef GENERATOR_FILE.
17389 Add parsing support for RTL dumps, mirroring the special-cases in
17390 print_rtx, by calling read_flags, reading REG_NOTE names, INSN_UID
17391 values, and calling handle_any_trailing_information.
17392 (rtx_reader::read_rtx_operand): Convert return type from void
17393 to rtx, returning return_rtx. Handle case 'e'. Call
17394 finalize_string on XSTR and XTMPL fields.
17395 (rtx_reader::read_nested_rtx): Handle dumps in which trailing
17396 "(nil)" values were omitted. Call the postprocess vfunc on the
17398 (rtx_reader::rtx_reader): Add new "compact" param and pass to base
17399 class ctor. Initialize m_in_call_function_usage. Call
17400 one_time_initialization.
17401 * rtl-tests.c (selftest::test_uncond_jump): Call
17402 set_new_first_and_last_insn.
17403 * rtl.h (read_rtx): Wrap decl with #ifdef GENERATOR_FILE.
17404 * selftest-rtl.c: New file.
17405 * selftest-rtl.h (class selftest::rtl_dump_test): New class.
17406 (selftest::get_insn_by_uid): New decl.
17407 * selftest-run-tests.c (selftest::run_tests): Call
17408 read_rtl_function_c_tests.
17409 * selftest.h (selftest::read_rtl_function_c_tests): New decl.
17410 * tree-dfa.c (ssa_default_def): Return NULL_TREE for rtl function
17413 2017-01-05 Uros Bizjak <ubizjak@gmail.com>
17415 * config/i386/i386.md (*testqi_ext_3): No need to handle memory
17416 operands in a special way. Assert that pos+len <= mode precision.
17418 2017-01-05 Jakub Jelinek <jakub@redhat.com>
17420 * common.opt (fvect-cost-model): Remove RejectNegative flag, use
17421 3 argument Alias with unlimited for the negative form.
17422 (fno-vect-cost-model): Removed.
17424 2017-01-05 Martin Liska <mliska@suse.cz>
17426 * hsa-gen.c (gen_hsa_divmod): New function.
17427 (gen_hsa_insn_for_internal_fn_call): Use the function for IFN_DIVMOD.
17429 2017-01-05 Martin Liska <mliska@suse.cz>
17432 * gcc.c (lookup_compiler): Reject '-' filename for a precompiled
17435 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17437 * config/s390/s390.c (s390_expand_setmem): Unroll the loop for
17438 small constant length operands.
17440 2017-01-05 Andreas Krebbel <krebbel@linux.vnet.ibm.com>
17442 * config/s390/s390.c (s390_expand_setmem): Avoid overlapping bytes
17443 between loop iterations.
17445 2017-01-05 Martin Liska <mliska@suse.cz>
17448 * gimplify.c (gimplify_decl_expr): Compare to
17449 asan_poisoned_variables instread of checking flags.
17450 (gimplify_target_expr): Likewise.
17451 (gimplify_expr): Likewise.
17452 (gimplify_function_tree): Conditionally initialize
17453 asan_poisoned_variables.
17455 2017-01-04 Jeff Law <law@redhat.com>
17457 PR tree-optimizatin/78812
17458 * rtl.h (contains_mem_rtx_p): Prototype.
17459 * ifcvt.c (containts_mem_rtx_p): Move from here to...
17460 * rtlanal.c (contains_mem_rtx_p): Here and remove static linkage.
17461 * gcse.c (prune_expressions): Use contains_mem_rtx_p to discover
17462 and prune MEMs that are not at the toplevel of a SET_SRC rtx. Look
17463 through ZERO_EXTEND and SIGN_EXTEND when trying to avoid pruning MEMs.
17465 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
17467 * input.c (assert_char_at_range): Default-initialize actual_range.
17469 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
17471 * df-scan.c (df_ref_create_structure): Make regno unsigned,
17472 to match the caller.
17474 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
17476 * cfgexpand.c (expand_gimple_basic_block): Disregard debug
17477 insns after final jump in test to emit dummy move.
17479 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
17481 * gimple-iterator.h (gsi_one_nondebug_before_end_p): New.
17482 * tree-eh.c (cleanup_empty_eh): Skip more debug stmts.
17484 2017-01-04 Alexandre Oliva <aoliva@redhat.com>
17486 * multiple_target.c (create_dispatcher_calls): Init e_next.
17487 * tree-ssa-loop-split.c (split_loop): Init border.
17488 * tree-vect-loop.c (vect_determine_vectorization_factor): Init
17491 2017-01-04 Michael Meissner <meissner@linux.vnet.ibm.com>
17496 * config/rs6000/predicates.md (sf_subreg_operand): New predicate.
17497 (altivec_register_operand): Do not return true if the operand
17498 contains a SUBREG mixing SImode and SFmode.
17499 (vsx_register_operand): Likewise.
17500 (vsx_reg_sfsubreg_ok): New predicate.
17501 (vfloat_operand): Do not return true if the operand contains a
17502 SUBREG mixing SImode and SFmode.
17503 (vint_operand): Likewise.
17504 (vlogical_operand): Likewise.
17505 (gpc_reg_operand): Likewise.
17506 (int_reg_operand): Likewise.
17507 * config/rs6000/rs6000-protos.h (valid_sf_si_move): Add declaration.
17508 * config/rs6000/rs6000.c (valid_sf_si_move): New function to
17509 determine if a MOVSI or MOVSF operation contains SUBREGs that mix
17511 (rs6000_emit_move_si_sf_subreg): New helper function.
17512 (rs6000_emit_move): Call rs6000_emit_move_si_sf_subreg to possbily
17513 fixup SUBREGs involving SImode and SFmode.
17514 * config/rs6000/vsx.md (SFBOOL_*): New constants that are operand
17515 numbers for the new peephole2 optimization.
17516 (peephole2 for SFmode unions): New peephole2 to optimize cases in
17517 the GLIBC math library that do AND/IOR/XOR operations on single
17518 precision floating point.
17519 * config/rs6000/rs6000.h (TARGET_NO_SF_SUBREG): New internal
17520 target macros to say whether we need to avoid SUBREGs mixing
17522 (TARGET_ALLOW_SF_SUBREG): Likewise.
17523 * config/rs6000/rs6000.md (UNSPEC_SF_FROM_SI): New unspecs.
17524 (UNSPEC_SI_FROM_SF): Likewise.
17525 (iorxor): Change spacing.
17526 (and_ior_xor): New iterator for AND, IOR, and XOR.
17527 (movsi_from_sf): New insns for SImode/SFmode SUBREG support.
17528 (movdi_from_sf_zero_ext): Likewise.
17529 (mov<mode>_hardfloat, FMOVE32 iterator): Use register_operand
17530 instead of gpc_reg_operand. Add SImode/SFmode SUBREG support.
17531 (movsf_from_si): New insn for SImode/SFmode SUBREG support.
17532 (fma<mode>4): Use gpc_reg_operand instead of register_operand.
17533 (fms<mode>4): Likewise.
17534 (fnma<mode>4): Likewise.
17535 (fnms<mode>4): Likewise.
17536 (nfma<mode>4): Likewise.
17537 (nfms<mode>4): Likewise.
17539 2017-01-04 Marek Polacek <polacek@redhat.com>
17542 * doc/invoke.texi: Document -Wpointer-compare.
17544 2017-01-04 Jakub Jelinek <jakub@redhat.com>
17546 * optc-gen.awk: Emit #error for -W*/-f*/-m* Enum without
17549 * dwarf2out.c (output_loc_list): Don't throw away 64K+ location
17550 descriptions for -gdwarf-5 and emit them as uleb128 instead of
17553 2017-01-04 Kelvin Nilsen <kelvin@gcc.gnu.org>
17556 * doc/sourcebuild.texi (PowerPC-specific attributes): Add
17557 documentation of the powerpc_popcntb_ok attribute.
17558 * config/rs6000/rs6000.c (rs6000_option_override_internal): Add
17559 code to issue warning messages if a requested CPU configuration is
17560 not supported by the binary (assembler and loader) toolchain.
17561 (spe_init_builtins): Add two assertions to prevent ICE if attempt is
17562 made to define a built-in function that has been disabled.
17563 (paired_init_builtins): Add assertion to prevent ICE if attempt is
17564 made to define a built-in function that has been disabled.
17565 (altivec_init_builtins): Add comment explaining why definition
17566 of the DST built-in functions is not preceded by an assertion
17567 check. Add assertions to prevent ICE if attempts are made to
17568 define an altivec predicate or an abs* built-in function that has
17570 (htm_init_builtins): Add comment explaining why definition of the
17571 htm built-in functions is not preceded by an assertion check.
17573 2017-01-04 Jeff Law <law@redhat.com>
17575 PR tree-optimizatin/67955
17576 * tree-ssa-alias.c (same_addr_size_stores_p): Check offsets first.
17577 Allow any SSA_VAR_P as the base objects. Use integer_zerop. Verify
17578 the points-to solution does not include pt_null. Use DECL_PT_UID
17581 2017-01-04 Uros Bizjak <ubizjak@gmail.com>
17583 * config/i386/i386.md (HI/SImode test with imm to QImode splitters):
17584 Use gen_int_mode instead of gen_lopwart for const_int operands.
17586 2017-01-04 Jakub Jelinek <jakub@redhat.com>
17588 PR tree-optimization/71563
17589 * match.pd: Simplify X << Y into X if Y is known to be 0 or
17590 out of range value - has low bits known to be zero.
17592 2017-01-04 Alan Modra <amodra@gmail.com>
17594 * Makefile.in (aclocal_deps): Update and order as per aclocal.m4.
17595 * configure: Regenerate.
17596 * config.in: Regenerate.
17598 2017-01-04 Jakub Jelinek <jakub@redhat.com>
17601 * input.c (ebcdic_execution_charset::on_error): Don't use strstr for
17602 a substring of the message, but strcmp with the whole message. Ifdef
17603 ENABLE_NLS, translate the message first using dgettext.
17605 2017-01-03 Jeff Law <law@redhat.com>
17607 PR tree-optimizatin/78856
17608 * tree-ssa-threadupdate.c: Include tree-vectorizer.h.
17609 (mark_threaded_blocks): Remove code to truncate thread paths that
17610 cross multiple loop headers. Instead invalidate the cached loop
17611 iteration information and handle case of a thread path walking
17612 into an irreducible region.
17614 2017-01-03 Michael Meissner <meissner@linux.vnet.ibm.com>
17617 * config/rs6000/rs6000.c (rs6000_split_signbit): Change some
17618 assertions. Add support for doing the signbit if the IEEE 128-bit
17619 floating point value is in a GPR.
17620 * config/rs6000/rs6000.md (Fsignbit): Delete.
17621 (signbit<mode>2_dm): Delete using <Fsignbit> and just use "wa".
17622 Update the length attribute if the value is in a GPR.
17623 (signbit<mode>2_dm_<su>ext): Add combiner pattern to eliminate
17624 the sign or zero extension instruction, since the value is always 0/1.
17625 (signbit<mode>2_dm2): Delete using <Fsignbit>.
17628 * config/rs6000/vsx.md (vsx_extract_<mode>_store_p9): If we are
17629 extracting SImode to a GPR register so that we can generate a
17630 store, limit the vector to be in a traditional Altivec register
17631 for the vextuwrx instruction.
17633 2017-01-03 Ian Lance Taylor <iant@google.com>
17635 * godump.c (go_format_type): Treat ENUMERAL_TYPE like INTEGER_TYPE.
17637 2017-01-03 Martin Sebor <msebor@redhat.com>
17639 PR tree-optimization/78696
17640 * gimple-ssa-sprintf.c (format_floating): Correct handling of
17641 precision. Use MPFR for %f for greater fidelity. Correct handling
17643 (pass_sprintf_length::compute_format_length): Set width and precision
17644 specified by asrerisk to void_node for vararg functions.
17645 (try_substitute_return_value): Adjust dump output.
17647 2017-01-03 David Edelsohn <dje.gcc@gmail.com>
17649 * doc/invoke.texi (RS6000 options): LRA is enabled by default.
17651 2017-01-03 Eric Botcazou <ebotcazou@adacore.com>
17653 * doc/invoke.texi (SPARC options): Document -mlra as the default.
17654 * config/sparc/sparc.c (sparc_option_override): Force LRA unless
17655 -mlra/-mno-lra was passed to the compiler.
17657 2017-01-03 James Cowgill <James.Cowgill@imgtec.com>
17659 PR rtl-optimization/65618
17660 * emit-rtl.c (try_split): Move initialization of "before" and
17661 "after" to just before the call to emit_insn_after_setloc.
17663 2017-01-03 Gerald Pfeifer <gerald@pfeifer.com>
17665 * doc/md.texi (Standard Names): Remove reference to Java frontend.
17667 2017-01-03 Pierre-Marie de Rodat <derodat@adacore.com>
17669 * dwarf2out.c (gen_enumeration_type_die): When
17670 -gno-strict-dwarf, add a DW_AT_encoding attribute.
17672 2017-01-03 Jakub Jelinek <jakub@redhat.com>
17674 PR tree-optimization/78965
17675 * gimple-ssa-sprintf.c (pass_sprintf_length::compute_format_length):
17676 Change first argument from const call_info & to call_info &. For %n
17677 set info.nowrite to false.
17679 PR middle-end/78901
17680 * gimple-ssa-sprintf.c (try_substitute_return_value): Don't change
17681 possibly throwing calls.
17683 * genmatch.c (dt_node::gen_kids_1): If generic_exprs include SSA_NAME
17684 and exprs_len || fns_len, emit the code for SSA_NAME next to the exprs
17685 and fns handling, rather than in a separate case SSA_NAME.
17687 2017-01-02 Jeff Law <law@redhat.com>
17689 * config/darwin-driver.c (darwin_driver_init): Const-correctness
17690 fixes for first_period and second_period variables.
17692 2017-01-02 Uros Bizjak <ubizjak@gmail.com>
17695 * config/i386/i386.md (UNSPEC_NOREX_MEM): New unspec.
17696 (*insvqi_1): New insn pattern.
17697 (*insvqi_1_mem_rex64): Ditto.
17698 (*insvqi_2): Ditto.
17699 (*insvqi_3): Rename from *insvqi.
17701 (*extzvqi_mem_rex64): Add UNSPEC_NOREX_MEM tag.
17703 2017-01-02 Gerald Pfeifer <gerald@pfeifer.com>
17705 * doc/cfg.texi (Edges): Remove reference to Java.
17706 (Maintaining the CFG): Ditto.
17708 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
17710 PR middle-end/77674
17711 * symtab.c (symtab_node::binds_to_current_def_p): Fix handling of
17712 transparent aliases.
17714 2017-01-01 Jan Hubicka <hubicka@ucw.cz>
17716 PR middle-end/77484
17717 * predict.def (PRED_CALL): Update hitrate.
17718 (PRED_INDIR_CALL, PRED_POLYMORPHIC_CALL): New predictors.
17719 * predict.c (tree_estimate_probability_bb): Split CALL predictor
17720 into direct/indirect/polymorphic variants.
17722 2017-01-01 Jakub Jelinek <jakub@redhat.com>
17724 Update copyright years.
17726 * gcc.c (process_command): Update copyright notice dates.
17727 * gcov-dump.c (print_version): Ditto.
17728 * gcov.c (print_version): Ditto.
17729 * gcov-tool.c (print_version): Ditto.
17730 * gengtype.c (create_file): Ditto.
17731 * doc/cpp.texi: Bump @copying's copyright year.
17732 * doc/cppinternals.texi: Ditto.
17733 * doc/gcc.texi: Ditto.
17734 * doc/gccint.texi: Ditto.
17735 * doc/gcov.texi: Ditto.
17736 * doc/install.texi: Ditto.
17737 * doc/invoke.texi: Ditto.
17739 Copyright (C) 2017 Free Software Foundation, Inc.
17741 Copying and distribution of this file, with or without modification,
17742 are permitted in any medium without royalty provided the copyright
17743 notice and this notice are preserved.